Optimization of glycopeptide enrichment techniques for the identification of Clinical Biomarkers

Abstract

Introduction:

The identification and characterization of glycopeptides through LC-MS/MS and advanced enrichment techniques are crucial for advancing clinical glycoproteomics, significantly impacting the discovery of disease biomarkers and therapeutic targets. Despite progress in enrichment methods like Lectin Affinity Chromatography (LAC), Hydrophilic Interaction Liquid Chromatography (HILIC), and Electrostatic Repulsion Hydrophilic Interaction Chromatography (ERLIC), issues with specificity, efficiency, and scalability remain, impeding thorough analysis of complex glycosylation patterns crucial for disease understanding.

Areas Covered:

This review explores the current challenges and innovative solutions in glycopeptide enrichment and mass spectrometry analysis, highlighting the importance of novel materials and computational advances for improving sensitivity and specificity. It outlines the potential future directions of these technologies in clinical glycoproteomics, emphasizing their transformative impact on medical diagnostics and therapeutic strategies.

Expert Opinion:

The application of innovative materials such as Metal-Organic Frameworks (MOFs), Covalent Organic Frameworks (COFs), functional nanomaterials, and online enrichment shows promise in addressing challenges associated with glycoproteomics analysis by providing more selective and robust enrichment platforms. Moreover, the integration of artificial intelligence and machine learning is revolutionizing glycoproteomics by enhancing the processing and interpretation of extensive data from LC-MS/MS, boosting biomarker discovery, and improving predictive accuracy, thus supporting personalized medicine.

1. Introduction

The quest for precision medicine, which involves a tailored approach to disease prevention, diagnosis, and treatment, relies heavily on our ability to detect and understand biomarkers. Biomarkers are defined as specific traits that are used as indicators of normal or pathogenic biological processes, or the body’s response to an exposure or intervention including treatments.^{1, 2} In 2016, the National Institutes of Health (NIH) and the Food and Drug Administration (FDA) published the BEST guideline defining seven categories of biomarkers by the function they serve, including prognostic, diagnostic, and monitoring biomarkers among others.^{2, 3} The process of discovering and validating biomarkers is challenging and requires meticulous testing of candidate biomarkers to measure their accuracy and reliability.⁴ As personalized medicine gains increasing prominence, the identification of biomarkers becomes paramount for early disease detection and potential treatment strategies,^{5, 6} guiding clinical decision-making from diagnostics to therapeutic monitoring.

Among the vast repertoire of biomolecules, glycoproteins stand out for their pivotal roles in numerous physiological and pathological processes. Glycoproteins are proteins that have carbohydrate groups attached to the polypeptide chain. This glycosylation, the addition of sugar moieties to proteins, is not merely a post-translational modification but a critical determinant of a protein’s structure, stability, distribution, and function.^7–9 The two primary forms of glycosylation are N-linked and O-linked. In N-linked glycosylation, glycans are attached to the amino acid asparagine (N) within the sequence motif N-X-S/T, where ‘X’ represents any amino acid other than proline.^{10, 11} In contrast, O-linked glycosylation involves the attachment of glycans to either serine (S) or threonine (T), and this type does not follow a specific sequence motif.^{12, 13} The glycans are mainly composed of the following monosaccharides; N-acetylglucosamine (GlcNAc), N-acetylgalactosamine (GalNAc), mannose (Man), galactose (Gal), glucose (Glc), fucose (Fuc), sialic acid (Sia, Neu5Ac), xylose (Xyl), and glucuronic acid (GlcAc). The dynamic nature of glycosylation, changing in response to biological conditions and stages of disease, positions glycoproteins as prime candidates for biomarkers. They play integral roles in cell-cell communication,^{14, 15} cell adhesion,^{16, 17} immune response,^{18, 19} and the progression of many diseases, including cancer,^20–22 cardiovascular diseases,²³ neurodegenerative diseases,^24–26 and infectious diseases,^{27, 28} making them invaluable for diagnosing diseases and monitoring therapeutic responses. However, the analysis of glycoproteins and glycopeptides presents significant challenges due to the high dynamic range of the human proteome making the detection of low-abundance glycoproteins challenging.²⁹

The inherent complexity and heterogeneity of glycoprotein glycosylation, arising from the diverse structures of carbohydrate chains and their various attachment sites on proteins resulting in a variety of glycoforms complicate their identification and quantification.³⁰ Glycopeptides exhibit extensive microheterogeneity due to the presence of different glycan structures attached to the same peptide backbone.³¹ This results in a wide variety of glycoforms, each differing by the composition and linkage of sugars. Glycans can be highly branched with various types of glycosidic linkages, making the determination of the exact structure challenging.³² O-glycosylation presents additional challenges because a single peptide backbone often has multiple glycosylation sites, each of which can be occupied by similar or different O-glycans. ^{33, 34}

Over the past decade, mass spectrometry has established itself as the optimal analytical technique for glycoprotein analysis, owing to its exceptional sensitivity and capacity to elucidate detailed structural information.^35–38 Although MS instruments are expensive and require expertise in instrumentation and data analysis, they remain the most widely utilized technique for glycoproteomics analysis. The most common MS-based techniques use matrix-assisted laser desorption (MALDI) and electrospray ionization (ESI) to introduce the glycopeptide sample into the MS system.^38–41 Together, the glycoprotein enzymatic digestion and the glycopeptide enrichment are the most important steps for achieving a robust glycoproteomics analysis. For enzymatic digestion, enzymes such as trypsin, Glu-C, or chymotrypsin are commonly used in single or combined approaches. Glycopeptide enrichment is a pre-analytical process that isolates glycopeptides from a mixture of peptides and glycopeptides obtained in the enzymatic digestion of the sample proteome. This selective isolation improves the detection and the identification of glycopeptides providing the N- and O-site glycoprotein structural information. The majority of the enrichment protocols focused on the separation of N-glycopeptides; the large hydrophilicity provided for the N-glycan moiety facilitates the application of selective strategies targeting its separation from the peptidome. Conversely, the low size of most of the O-glycans types attached to the glycoproteins does not provide a strong affinity for the conventional materials used for glycopeptide enrichment. Therefore, the O-glycopeptide enrichment should be addressed using different strategies, for example, applying a PNGase F digestion of the N-glycans prior to the O-glycopeptide enrichment, ^{42, 43} other strategies used O-glycans-selective novel material.⁴⁴ Various glycopeptide enrichment techniques such as hydrophilic interaction liquid chromatography (HILIC) and electrostatic repulsion liquid chromatography (ERLIC) have been developed to selectively enrich glycopeptides.^{41, 45, 46} HILIC enrichment exploits the polar nature of the glycan moieties to selectively retain glycopeptides over tryptic peptides.⁴⁵ Whereas, in ERLIC enrichment, a positively charged polyethyleneimine repels the similarly charged peptides and attracts negatively charged glycans of the glycopeptide.^{45, 47, 48} Other glycopeptide enrichment techniques include lectin affinity chromatography (LAC)^{49, 50} and hydrazide enrichment.^{51, 52} The comprehensive overview of the workflow used for glycoproteomics analysis and the validation of glycosignatures is shown in Figure 1.

Figure 1.

Analytical workflow for glycoproteomics analysis in biological samples. Sample digestion and glycopeptide enrichment is followed by LC-MS/MS analysis identification and characterization of glycopeptides. Targeted glycoproteomics analysis using an LC-PRM-MS approach is used for the validation of observed glycosignature changes.

This review aims to spotlight the critical area of glycopeptide enrichment techniques in the landscape of biomarker discovery. Given the challenges associated with glycoprotein analysis, the development and optimization of methods for glycopeptide enrichment are paramount. These techniques aim to separate and enrich glycosylated peptides from complex biological samples, thereby facilitating their identification, quantification, and characterization. By enhancing our ability to analyze glycoproteins with precision and depth, glycopeptide enrichment techniques stand at the forefront of advancing biomarker discovery. They offer the promise of uncovering novel biomarkers that could revolutionize our approach to diagnosing and treating diseases, marking a significant stride toward the realization of personalized medicine. This review will delve into the latest advancements in glycopeptide enrichment methods, evaluating their effectiveness, challenges, and potential in clinical biomarker discovery, with a particular focus on their application in understanding disease mechanisms and improving patient care.

2. Glycopeptide Enrichment Techniques: A Comparative Overview

2.1. Glycopeptide Enrichment Techniques

Glycopeptide enrichment is a crucial step in glycoproteomics, designed to separate glycosylated peptides from non-glycosylated peptides. This process is essential for simplifying sample complexity, enhancing the detection of glycopeptides, and facilitating subsequent LC-MS analysis. The general strategies for glycopeptide enrichment exploit the structural properties of glycopeptides including their glycan moiety together with their physicochemical characteristics. Over recent decades, a variety of enrichment methods have been devised and utilized in the field of glycoproteomics. The working principles of the common glycopeptides/glycoproteins enrichment techniques are outlined in Figure 2.

Figure 2.

Workflow schematics describing the basic working principle of commonly employed glycopeptide enrichment techniques followed by LC-MS/MS qualitative and quantitative characterization. Reused with permission from Song et al.³⁷⁴

2.1.1. Lectin Affinity Chromatography (LAC) Enrichment

Lectins are proteins that specifically identify and attach to particular carbohydrate structures found on biomolecules.^{53, 54} Therefore, different lectins have affinities for different sugar moieties conferring on them the ability to selectively isolate certain glycopeptides from complex biological samples. For instance, concanavalin A (ConA) specifically attaches to glucose and mannose moieties, while wheat germ agglutinin (WGA) binds to GlcNAc and sialic acid residues.^55–57 In lectin affinity enrichment, a lectin is immobilized on a solid support, such as magnetic beads, monolithic resins, microarrays, or in the stationary phase of a chromatography column.⁵⁸ The functionalized lectins bind to the sugar moieties on the sample glycopeptides effectively capturing them while non-glycosylated peptides pass through (Figure 2). Afterward, the bound glycopeptides are eluted by modifying the strength of the lectin-carbohydrate interactions by changing the pH of the elution solution.⁵⁹ Lectins are often engineered to target specific sugars in the glycan structure. For example, lentil lectin (LcH) is selective for core fucose,⁶⁰ whereas Ricinus comminus agglutinin (RCA) and peanut agglutinin (PNA) selectively bind galactose.⁶¹ Another commonly used lectin Jacalin (AIL) has been widely used for its affinity towards the core GalNAc in O-glycopeptide structures⁶² and the selectivity of elderberry lectin (SNA) for sialic acid has been shown.⁶³ Lectin affinity chromatography has been widely utilized in various studies to selectively enrich N- and O-glycopeptides from intricate biological systems.^64–67 Moreover, lectins can be used for the selective enrichment of different type of glycosylated peptides. Hutte et al.⁶⁸ used the lectin from Burkholderia cenocepacia (BC2L-A) as a selective enrichment of C- and O-mannosylated peptides, outperforming ConA in mannosylated glycopeptide retention and selectivity. However, the high selectivity possessed by these lectins also limits the type of glycopeptides that may be enriched by a specific lectin. This issue can be mitigated by the incorporation of multiple lectins into the workflow to bind a wider range of glycopeptide structures.^{62, 69, 70}

2.1.2. HILIC Enrichment

Hydrophilic Interaction Liquid Chromatography (HILIC) remains the most used method for enriching and characterizing the glycoproteome, a practice well-established and widely utilized within the field.^{41, 71–74} HILIC provides excellent selectivity for glycopeptides due to its ability to preferentially retain compounds with hydrophilic glycan structures.⁵⁸ This selectivity is beneficial in complex biological samples where non-glycosylated peptides may overwhelm glycopeptides. HILIC facilitates the retention of intact glycan structures on peptides, which is crucial for conducting detailed structural analysis and determining glycosylation sites (Figure 2). HILIC utilizes a stationary phase that is highly hydrophilic. Common materials include silica particles modified with polar groups such as amide, amino, or cyano groups.^{75, 76} These modifications help create a strong affinity for the polar molecules such as glycans. In HILIC a binding solution rich in acetonitrile with a small percentage of water will reduce the solubility of hydrophilic molecules, thereby promoting their interaction with the hydrophilic stationary phase facilitated by the formed water layer ⁷⁵ The glycopeptide interaction results in longer retention and separation from non-glycosylated peptide species. HILIC enrichment techniques have been utilized to enrich N-glycopeptides from a variety of biological samples derived from different disease conditions.^{74, 77} One major limitation of HILIC enrichment is that hydrophilic non-glycosylated peptides can coelute with the glycosylated peptides.⁷⁸ In addition, HILIC enrichment tends to have biases towards O-glycopeptides with small glycan structures.^79–84

Glycopeptides are typically polar and hydrophilic due to their carbohydrate moieties. Mobile phases must be carefully selected to match these properties, ensuring that glycopeptides interact appropriately with the stationary phase for effective separation. Kolarich and coworkers have investigated the impact of mobile phase composition on ZIC-HILIC glycopeptide enrichment.⁸⁵ They compared the efficiency of the four most commonly used MS organic solvents including acetonitrile, methanol, ethanol, and isopropanol for enriching a wide range of samples. Their results show that acetonitrile was best for retaining both hydrophilic and hydrophobic glycopeptides whereas methanol was unsuitable due to hydrogen bonding interfering with hydrophilic partitioning. Ethanol and isopropanol showed better suitability for enriching larger hydrophobic glycopeptides but co-enriched unmodified peptides. Glycopeptide enrichment efficiency depends significantly on the chosen mobile phase, with acetonitrile generally being the most effective.

2.1.3. Electrostatic Repulsion-Hydrophilic Interaction Chromatography

Electrostatic Repulsion Hydrophilic Interaction Chromatography (ERLIC) combines the principles of HILIC with an additional layer of complexity by utilizing a charged stationary phase. ^{86, 87} When the analyte and stationary phase have the same charge, electrostatic repulsion occurs, potentially pushing the analyte away from the stationary phase. However, the presence of organic solvent in the mobile phase weakens this repulsion, allowing hydrophilic interactions to dominate at sufficient organic solvent concentrations. This interplay between repulsion and interaction enables fine-tuning of analyte retention based on their charge. ERLIC can be used to enrich glycopeptides and phosphopeptides simultaneously through the optimization of several parameters, including the organic phase ratio, the use of ion-pairing reagents, pH adjustment, and salt concentration.⁸⁸ ERLIC has been successfully applied in various glycoproteomic studies to characterize the overall N- and O-glycosylation profiles of complex biological samples, resulting in the identification of a larger number of glycosylation sites when compared with HILIC enrichment.^{45, 62, 89} A study by Totten et al. ⁹⁰ combined strong anion exchange (SAX) with ERLIC (SAX-ERLIC) for enriching N-glycopeptides.⁹⁰ This approach yielded a high number of identified N-glycopeptides compared to traditional methods. ERLIC requires careful optimization of separation conditions for specific sample types. The interplay between electrostatic repulsion, hydrophilic interaction, and glycan properties necessitates adjustments to mobile phase composition, pH, and ionic strength for optimal results.

2.1.4. Ion Exchange Chromatography

The fundamental principle of Ion Exchange Chromatography (IEC) lies in the separation of molecules based on their charge. Glycopeptides, which are unique biomolecules containing both protein and glycan moieties, exhibit a net charge influenced by the protein sequence and glycan structure. ^{91, 92} The amino acid composition of a peptide dictates its pI, which directly impacts its charge at a specific pH, while the presence of sialic acids and other acidic functional groups within the glycan contributes negative charges, influencing the overall charge of the glycopeptide.^{93, 94} IEC is a powerful technique used in glycopeptide enrichment, which operates primarily in two modes: cation exchange chromatography (CEX) and anion exchange chromatography (AEX).^{95, 96} In CEX, glycopeptides that are positively charged are targeted. These glycopeptides usually have an isoelectric point (pI) that exceeds the pH at which the chromatography is conducted. They interact with a negatively charged stationary phase. To elute these glycopeptides from the column, the ionic strength of the mobile phase is increased, or the pH of the mobile phase is adjusted. This change in conditions disrupts the interactions between the positively charged glycopeptides and the negatively charged stationary phase, allowing the glycopeptides to be washed out of the column.^{97, 98} Conversely, in AEX, the focus is on glycopeptides that carry a negative charge. These often include glycopeptides with sialylated or sulfated glycans. In this method, these negatively charged glycopeptides bind to a positively charged stationary phase. Similar to CEX, the elution of glycopeptides in AEX is facilitated either by increasing the ionic strength of the mobile phase or by altering its pH, which weakens the electrostatic bonds between the negatively charged glycopeptides and the positively charged stationary phase. Both methods utilize the principle of electrostatic interactions between charged glycopeptides and oppositely charged stationary phases, but they target different types of glycopeptides based on their specific chemical characteristics. Several works have utilized IEC for glycopeptide enrichment in complex biological samples using different modes and materials.^{95, 99, 100} Recently, Zhong et al. ⁹² developed an efficient method for enriching sialylated glycopeptides using a derivatization approach known as Derivatization of Sialylated Glycopeptides Plus (DOSG+). This method introduces permanent positive charges on sialoglycopeptides, allowing for their effective enrichment through weak cation exchange (WCX) chromatography. Also, it is capable of identifying linkage isomers of sialic acids by introducing mass difference. It has been proven to be efficient for both N- and O-linked sialoglycopeptides.

2.1.5. Immunoaffinity Capture

This technique leverages the strong and selective binding affinity of antibodies towards specific epitopes or glycan structures on glycopeptides.^101–103 These antibodies can be raised against known glycan structures allowing for targeted enrichment. Antibodies are commonly immobilized on a solid support, such as beads in a column or on a microplate.^{104, 105} The sample containing a mixture of proteins/peptides, including glycoproteins/glycopeptides, is introduced to the immobilized antibodies. Only the glycoproteins/glycopeptides with the specific glycan structures or epitopes recognized by the antibodies will bind to the column, while other proteins/peptides will not be retained and can be washed away (Figure 2). The bound glycopeptides are then eluted from the antibodies using an elution buffer. This buffer might contain a high concentration of a competitive ligand or altered pH or ionic strength to disrupt the antibody-glycopeptide interaction, releasing the glycopeptides for further analysis.

In a work by Teo and coworkers,¹⁰² they successfully created a new set of monoclonal antibodies against O-GlcNAc using a three-component immunogen methodology. These antibodies demonstrated unique binding properties and enabled the identification of over 200 mammalian proteins modified by O-GlcNAc, many of which were previously unknown. The study highlighted the role of O-GlcNAc in responding to cellular stress and its involvement in various biological functions, including transcriptional regulation and signal transduction. Their findings suggest that these antibodies can be crucial tools for studying O-GlcNAc modifications, with potential implications for understanding cellular processes and disease mechanisms related to this modification. Furthermore, diverse studies have reported different anti-glycan antibodies against glycans and glycoprotein epitopes such as O-GlcNAc,¹⁰⁶ O-GalNAc,¹⁰⁷ Sialyl Lewis X antigen,¹⁰⁸ and fucosylated N-glycosylation epitopes.^{109, 110} Efforts in developing antibodies against glycans and glycoprotein epitopes are crucial for deepening our understanding of the pivotal roles of glycosylation in health and disease, potentially leading to novel biomarker discoveries and therapeutic targets.

2.1.6. Metabolic labeling

Metabolic labeling is a powerful technique used in glycoproteomics to study glycosylation patterns and dynamics. This method entails the in vivo introduction of identifiable markers to glycoproteins, enabling detailed analysis of glycopeptides through mass spectrometry. The approach incorporates modified glycan precursors, such as azide sugars into cell lines or tissue samples. These modified sugars are then metabolically assimilated into the cell or tissue and incorporated into glycoproteins during biosynthesis.¹¹¹This approach offers the benefits of selectively enriching and subsequently detecting glycopeptides. In contrast to certain enrichment techniques like HILIC, which aim to capture the entire spectrum of glycopeptides for a comprehensive understanding of the glycoproteome. This method in combination with selective techniques can be used to target a specific subset of glycopeptides carrying particular glycan structures.¹¹² In a study conducted by Hang et al.,¹¹³ they treated cells with N-azidoacetyl galactosamine tetra acylated analog, leading to the formation of glycoproteins containing Azides. Functional groups can be specifically tagged using phosphine probes through a process known as azide-phosphine conjugation.¹¹³

Debets et al.^103,105 demonstrated precise tagging of O-GalNAcylated proteins. They achieved this by introducing a membrane-permeable caged N-(S)-azidopropionylgalactosamine (GalNAzMe)-1-phosphate probes into cells that had been genetically modified to express a mutant form of the pyrophosphorylase AGX1. This modification allowed the synthesis of uridine diphosphate (UDP)-GalNAzMe donor molecules, facilitating the targeted labeling of O-glycoproteins. Coupled with mass spectrometry, metabolic labeling facilitates the enrichment and examination of labeled glycoproteins, aiding in the analysis of glycosylation patterns and structures. Nonetheless, it’s crucial to acknowledge potential drawbacks, including off-target labeling by certain metabolic agents and the risk of perturbing the normal biological system, especially with high concentrations of artificial metabolic precursors.

2.1.7. Fractionation

Sample fractionation enhances glycoproteomics analysis by facilitating subsequent glycopeptide enrichment steps, leading to a less complex sample for mass spectrometry analysis. This process takes advantage of the different physicochemical properties of the sample components. For example, protein depletion protocols are commonly used to simplify the sample complexity eliminating the high-abundance proteins from the sample such as albumin, facilitating further glycoproteomics analysis of the remaining proteins.¹¹¹ One effective method for fractionation involves the use of C18 fractionation in combination with titanium dioxide (TiO₂) chromatography.¹¹⁴ C18 fractionation enables the separation of peptides and glycopeptides based on their hydrophobicity. This method utilizes the hydrophobic properties of C18-RP to retain non-glycosylated peptides while allowing glycopeptides to pass through. This selective retention facilitates the enrichment of glycopeptides, as non-glycosylated peptides are effectively removed from the sample. Titanium dioxide (TiO₂) chromatography is employed in the subsequent enrichment step to selectively capture glycopeptides. TiO₂ has a high affinity for phosphorylated and glycosylated peptides, enabling the specific binding of glycopeptides to the TiO₂ resin. This step further enriches the sample for glycopeptides by isolating them from non-glycosylated peptides. The combined use of C18 fractionation and titanium dioxide chromatography allows for the selective and efficient enrichment of glycopeptides from complex samples. The selective fractionation of glycopeptides not only improves the detection sensitivity of mass spectrometry but also enables the exploration of the diverse glycan forms and sequence lengths of N-linked glycopeptides, providing valuable insights into the complex landscape of protein glycosylation. ¹¹⁵ This method holds great potential for advancing our understanding of the roles of glycosylation in diverse biological processes. Recently, Zurawska et al.¹¹⁶ introduced a method based on high-pH RP-LC fractionation of peptides using micro-flow rates, the pH was adjusted using ammonium bicarbonate (ABC) buffer. The application of this approach prompted the detection of 138,417 peptides derived from 8,896 proteins in comparison to the 23,093 peptides derived from 3,344 proteins detected in the non-fractionated samples.

2.1.8. Chemical Conjugation

Chemical conjugation in glycopeptides enrichment involves the covalent attachment of specific tags or analogs to glycopeptides, enabling their selective enrichment and subsequent analysis using mass spectrometry. By covalently linking these molecules to glycopeptides, targeted isolation and analysis of glycoproteins becomes feasible, leading to a comprehensive characterization of glycosylation patterns. The protein glycosylations and even the peptide amino acids can serve as linkers for the selective immobilization of conjugated chemicals. Sialic acid units of sialylated glycans can be oxidized and used as targets of hydrazide groups immobilized on a solid phase facilitating the enrichment process. This reaction is pH reversible, facilitating the recovery of the enriched glycopeptides.^{35, 111} For example, poly-NIPAM-co-hydrazide has been employed to aid in the enrichment of glycoproteins and N-glycopeptides.³⁵ The protocol was applied for the N-glycopeptide enrichment from tryptic-digested plasma samples. The results showed the identification of 329 N-glycosylation sites distributed across 180 glycoproteins. In a similar study conducted by Sun et al.,¹¹⁷ hydrazide beads were used for the capture and subsequent dimethyl isotopic labeling of N-glycopeptides. The protocol was used for the enrichment and isotopic labeling of N-glycopeptides derived from serum samples of patients with hepatocellular carcinoma (HCC). The results demonstrated an enhancement of the recovery of 330%, as well as an increase in sensitivity due to the peptide demethylation.¹¹⁸ Hydrazide chemistry also allows the enrichment and quantitation of O-glycopeptides. ¹¹¹ Chen et al.⁴⁴ introduced the CHO-GlcNAc method, combining enzymatic labeling, chemical oxidation, and reversible hydrazide chemistry for efficient O-GlcNAc glycopeptide enrichment, identifying 829 O-GlcNAcylation sites, on 274 glycoproteins in HeLa cell nuclear proteins, including novel modifications on m6A readers YTHDF1 and YTHDF3.

Another approach for glycopeptide enrichment via chemical conjugation is Boronate Affinity. Boronic acid materials interact with the diol groups commonly observed in the glycan moiety of the glycopeptides during enrichment protocols. Unlike hydrazide-based materials, the interaction between boronic acid and glycans can be disrupted by adding sugars with higher affinity for the glycans, such as glucose.³⁵ Chen et al.¹¹⁹ identified 816 N-glycosylation sites derived from 332 yeast proteins using a boronic acid enrichment coupled with PNGase F treatment in heavy-oxygen water. Hua et al.¹²⁰ developed a boronic acid rich carbonaceous sphere (BCS) used for the enrichment of N-glycopeptides derived from serum samples of patients normal pregnancy and preeclampsia patients. The results showed the identification of 235 N-linked glycopeptides from 166 glycoproteins. The carbonaceous sphere showed good stability after 10 cycles, and a sensitivity of 0.1 fmol mL⁻¹. The Saeed group¹²¹ used boronic acid to modify a hydrophilic MOF, UiO-66-NH2 to form GO@UiO-66-PBA. The modification improved the thermal and chemical stability of the MOF leading to a good binding capacity and recovery rate of 86.5% in N-glycopeptide enrichment. The results show the identification of 372 N-glycopeptides from one microliter of tryptic-digested human serum. The GO@UiO-66-PBA also showed a specificity (1:200) from the IgG N-glycopeptides against the serum albumin peptides. Xue et al.¹²² designed a combined protocol using boronic acid alongside the common enrichment methods HILIC and PGC. The results showed a significant improvement in the enrichment efficiency for the high mannose glycan compared to the stand-alone HILIC and PGC strategies. The pH and buffer effect on the boronic acid enrichment method was also investigated resulting in a binding buffer of 50 mM ammonium bicarbonate at pH 10 as the optimal conditions. Chen et al.¹²³ used a thiol-ene click chemistry technique to synthesize a phenylboronic acid bound to SiO₂ microspheres (PBA). The hydrophilic and affinity interactions of the HILIC and Borane components prompt the enrichment of neutral and acidic N-glycopeptides. Using this novel material the authors identified 101 unique glycosylation sites from 71 glycoproteins in one microliter of human serum. Kong et al.¹²⁴ developed GO@mSiO₂-GLYMO-APB, a novel boronate affinity material for glycopeptide enrichment, showcasing superior performance in simultaneous capture of N- and O-linked glycopeptides with enhanced selectivity boasting high surface area, size exclusion, hydrophilic interaction, and reversible covalent binding properties. The material showed uniform pore size, high total pore volume, and large BET surface area, identifying 724 intact N-glycopeptides and 152 O-glycopeptides in standard glycoproteins and human serum. Consequently, a number of research teams have recently reported on cutting-edge methods for using boronic acids to enhance glycopeptide identification.¹²⁵

2.2. Comparative Overview

Enrichment techniques for glycoproteomics analysis are critical, especially in finding novel candidate glycopeptide biomarkers in disease conditions and understanding post-translational modifications like glycosylation. Different techniques vary widely in their specificity, sensitivity, scalability, and suitability for particular types of samples. In this section, we provide a comparative overview of several key enrichment techniques used in the analysis of glycopeptides in biological samples, Table 1.

Table 1.

Comparative overview of the most common glycopeptide enrichment techniques used for the analysis of glycoproteins.

Techniques	Strengths	Weaknesses	Suitability
LAC	High specificity.126 Combination of lectins for improved enrichment.127 Mild elution conditions.128	Potential for non-specific binding.127 Unavailability of lectins specific for some glycan structures.129 Possible co-purification of non-glycosylated proteins.128	Selective enrichment of glycopeptide or a wide variety of glycopeptides via a combination of multiple lectins.130 Enrich both N- and O-glycopeptides.131
HILIC	Efficient enrichment of glycans and glycopeptides.132–134 Can be combined with reverse-phase liquid chromatography.135Analysis of other polar biomolecules.86, 136	Requires complex mobile phase optimization.134 Peak broadening.136	Suitable for the enrichment and fractionation of polar and hydrophilic molecules. 134, 136 Suitable for enriching N-glycopeptide.131
ERLIC	Effective for the enrichment of charged or polar residues.87 Combines electrostatic interactions and hydrophilic partitioning mechanisms87 Ability to separate and enrich different types of post-translational modifications88	Complex method development and optimization 87 Potential for analyte loss due to the multiple steps involved. Lack of proper enrichment of fucosylated glycans.137	Suitable for the enrichment of charged and polar residues. 87 Enrich both N- and O-glycopeptides.131
IEC	Efficient for the separation and enrichment of charged analytes.138–140 Able to achieve high purification and concentration factors. Versatile for a wide range of sample types and analytes.141	Harsh elution conditions. Require optimized buffer conditions.138 May require a desalting step before MS analysis.	Suitable for the enrichment and purification of charged biomolecules. 138, 139, 141 Enrich both N- and O-glycopeptides.42
Immunoaffinity Capture	High selectivity. Ability to capture low-abundance proteins and peptides from complex samples. Mild elution conditions.142	Limited availability of antibodies. Potential for non-specific binding. Antibodies may not be reusable.	Specific enrichment of proteins, peptides and glycopeptides. Enrich both N- and O-glycopeptides.131
Boronic Acid Enrichment	Selective for the capture of glycosylated proteins and glycopeptides.143 Capures a wider range of glycan structures.144 Relatively mild elution conditions.143	Potential for non-specific binding.145 Possible interference from other diol-containing compounds. 146	Selective enrichment and purification of glycoproteins and glycopeptides from complex biological samples.147, 148 Complementing lectin affinity approaches.149 Enrich both N- and O-glycopeptides.131
Metabolic Labeling	Accurate quantification of glycopeptides.150 Enhance detection and identification of glycopeptides. Enable glycosylation study in live cells. 151	Can alter cell physiology.152 Complex labeling process. Generally requires actively proliferating cells. Complex data analysis. 153	Quantitation of glycans and glycopeptides in live cells.151 Enrich both N- and O-glycopeptides.131

The field of glycopeptide enrichment is evolving rapidly, with each technique offering unique advantages and challenges. The choice of enrichment method depends on the specific objectives of the study, including the complexity of the sample, the type of glycosylation of interest, and the analytical platform used. Recent advancements are expanding the capabilities of glycopeptide enrichment, promising to unveil new layers of glycoproteome complexity and drive forward the discovery of novel biomarkers for clinical diagnostics and therapeutic monitoring.

2.3. Recent Advances in Glycopeptide Enrichment

2.3.1. Metal-Organic Frameworks (MOFs)

Metal-organic frameworks (MOFs) pioneered by Yaghi and coworkers in 1999 are porous materials made through the self-assembly of metal ions or clusters with organic ligands.¹⁵⁴ Their distinct structural and chemical properties render them highly versatile and effective for glycopeptide enrichment. MOFs are known for their exceptionally high porosity and surface area, which provide ample sites for the adsorption of glycopeptides.^155–157 This increases the interaction between the MOF and glycopeptides, leading to the efficient capture of these molecules from complex mixtures. The pore sizes and the functional groups within MOFs can be tailored during synthesis. This allows for the selective enrichment of glycopeptides by optimizing the interaction between the glycopeptide’s glycan groups and the MOF’s functional groups. MOFs can be functionalized with various chemical groups that can form multiple types of interactions (like hydrogen bonding, ionic interactions, and hydrophobic interactions) with the glycopeptides.^{134, 155} This selective binding is crucial for differentiating glycopeptides from other peptides enabling their enrichment from biological samples such as blood, tissue extracts, or cell cultures. This is particularly useful in proteomics, where the detection and analysis of glycopeptides can provide insights into disease mechanisms and biomarker discovery. MOFs can also be used as stationary phases in chromatography setups specifically designed for the separation of glycopeptides.^{158, 159} Their structural and chemical diversity allows for fine-tuning the separation process according to the specific characteristics of the glycopeptides. Deng And co-workers designed and synthesized a mercaptosuccinic acid functionalized hydrophilic magnetic metal-organic framework nanocomposite known as mMOF@Au-MSA.¹⁶⁰ They reported a high sensitivity of the material for glycopeptide enrichment even after decreasing tryptic HRP digest to a final concentration of 0.5 fmol μL⁻¹. In addition, it possessed a good glycopeptide selectivity by enriching N-glycopeptides from a mixture of 1:100 HRP digests to BSA digests. Figure 3a briefly describes the synthesis of mMOF@Au-MSA and the procedure for glycopeptide enrichment prior to MS analysis is shown in Figure 3b. They observed increased detection of glycopeptides from 250 fmol μL⁻¹ HRP digest after enrichment with mMOF@Au-MSA (Figure 3c and d). In another experiment using 300 fmol μL⁻¹, They observed an increased detection of glycopeptides after enriching the samples highlighting the efficiency of the enrichment technique (Figure 3e and f). In a more recent work, Ba et al.¹⁶¹ prepared a dual-hydrophilic hierarchical porous MOF nanospheres by metal organic assembling (MOA). In this novel material the porous structure improved the diffusion and binding affinity of the N-glycopeptides. Therefore, facilitating their HILIC-based enrichment with the MOF nanospheres. The strategy showed a selectivity of 1/500 for the N-glycopeptides derived from human serum immunoglobulin G and combined with the peptidome of bovine serum albumin (m/m). The application of this method in biological samples helped the identification of 550 N-glycopeptides in rat liver samples. Zhou et al.¹⁶² developed the high hydrophilic UIO-PBA&FDP zirconium-based MOF for the enrichment of N-glycopeptides. The results showed 359 N-glycopeptides from tryptic-digested human serum. Using MALDI-TOF MS, three glycopeptides were identified from HRP digests before enrichment and 39 after. Additionally, the authors reported 44 glycopeptides from IgG following enrichment with Zr-Fc MOF@Au@GC. According to the authors, these are the highest numbers of glycopeptides detected in HRP and IgG digest demonstrating the outstanding efficiency of this material for glycopeptide enrichment. The material was also capable of glycopeptide enrichment in a mixture of HRP to BSA tryptic digest ratio of 1:2000 demonstrating excellent selectivity. They were able to detect 17 glycopeptides at this mass ratio. In addition, this material is reported to demonstrate an increased glycopeptide binding capability of 200 mg/g, satisfactory reusability after five cycles, and a long-term storage capacity. The efficiency of enrichment remained constant after five cycles of use and after three months of storage at room temperature. Using this novel material for glycopeptide enrichment, they identified 655 N-glycopeptides corresponding to 366 glycoproteins from human serum samples utilizing a nano-LC-MS/MS technique. Several other MOFs have also been recently described for the enrichment of glycopeptides demonstrating their capabilities for identifying N- and O-glycopeptide biomarkers in biological samples.^{157, 163, 164} Despite their advantages, the use of MOFs in glycopeptide enrichment faces several challenges. Some MOFs may degrade or lose their functional integrity in the presence of water or biological fluids, affecting their performance. In addition, manufacturing MOFs with consistent quality and in large quantities can be challenging, which might limit their widespread application in routine analytical labs and glycopeptide enrichment.

Figure 3.

Synthesis and application of mMOF@Au-MSA for glycopeptide enrichment. (a) A scheme describing the synthetic procedure for mMOF@Au-MSA (b) Flowchart showing procedure for enrichment of glycopetides. MALDI-TOF mass spectra for the glycopeptide enrichment from 250 fmol μL⁻¹ HRP tryptic digest: (c) before enrichment, and (d) after treatment; from 300 fmol μL⁻¹ IgG: (e) before enrichment, and (f) after treatment. Peaks of glycopeptides are marked with red diamonds. Modified and reused with permission from Hu et al.¹⁶⁰

2.3.2. Covalent Organic Frameworks (COFs)

Covalent-Organic Frameworks (COFs) are another class of porous materials, like Metal-Organic Frameworks (MOFs), that have found significant utility in the field of analytical chemistry, particularly for the enrichment and analysis of biomolecules such as glycopeptides.^165–167 COFs consist of lighter elements such as hydrogen, boron, carbon, nitrogen, and oxygen, which are connected through robust covalent bonds to form frameworks in either two or three dimensions.^{168, 169} COFs can be precisely customized to allow their preferential interaction with glycopeptides based on size, shape, and functional groups. Like MOFs, COFs have a high surface area, which facilitates a high loading capacity for glycopeptides, enhancing the sensitivity and efficiency of enrichment processes.^{166, 170} COFs can be engineered to selectively bind glycopeptides from complex biological samples. This selectivity can be achieved by functionalizing the COF with specific chemical groups that interact favorably with the carbohydrate moieties of glycopeptides, thereby improving the specificity of binding. COFs can be utilized as stationary phases in chromatographic applications or as sorbents in solid-phase extraction (SPE) setups.¹⁷¹ Their selective binding properties and high surface areas make them particularly useful for separating glycopeptides from other types of peptides and proteins. Wu and coworkers¹⁷² developed a novel glutathione-functionalized magnetic covalent organic framework microsphere termed MCNC@COF@GSH for glycopeptide enrichment.¹⁷² They demonstrated the performance of this material with a low detection limit of 0.01 fmol/μL, high selectivity, reusability, and an excellent size exclusion effect. Five N-glycopeptides were identified at a decreased concentration of 0.01 fmol/μL, demonstrating the outstanding sensitivity of the enrichment technique. The material was also capable of glycopeptide enrichment in a mixture of IgG to BSA tryptic digest ratio of 1:5000 demonstrating an excellent selectivity. They were able to detect 18 glycopeptides at this mass ratio using MALDI-TOF MS analysis. Figure 4a briefly describes the synthesis of mMOF@Au-MSA and the procedure for N-glycopeptide enrichment prior to MS analysis is shown in Figure 4b. They observed an increased detection of glycopeptides from human IgG digest after enrichment using MCNC@COF@GSH microspheres (Figure 4c and d). Recently, Wang et al. ¹⁷³ reported the use of a polymer brush-functionalized porphyrin-based COF (TAPBB@GMA@AMA@Cys) for enriching N-glycopeptides. The N₂ adsorption-desorption experiment, the Brunauer-Emmett-Teller (BET) surface area of the material 63 m²/g, and a water contact angle of 59° indicates increased hydrophilic binding towards N-glycopeptides. The enrichment performance of the synthesized COF was tested on HRP tryptic digests using MALDI-MS. For the non-enriched tryptic digest, only non-glycopeptides were identified. However, after enrichment using TAPBB@GMA@AMA@Cys, 20 N- glycopeptides were identified. Further assessment of the material for N-glycopeptide enrichment at lower concentrations (20, 2, and 0.2 fmol/μL) shows the material’s potential for enriching N-glycopeptides with low abundance. In another recent study, Lin and colleagues¹⁷⁴ developed a 2D COF nanosheet containing sulfonic groups (NUS-9) using an oil-way biphase strategy for N-glycopeptide enrichment. Their experiment showed a glycopeptide recovery of 92 ± 2% after enrichment of tryptic-digested HRP using NUS-9. The material was also capable of N-glycopeptide enrichment in a mixture of HRP to BSA tryptic digest ratio of 1:1500 demonstrating an outstanding enrichment capability. Using NUS-9, they enriched and identified 631 N-glycopeptides from the human saliva sample. A boronate magnetic COF synthesized through a multiligand strategy has recently been shown to have an enhanced selectivity for N-glycopeptide enrichment in an IgG to BSA tryptic digest mixture with a ratio of 1:2000. Using this magnetic COF, they identified 1921 N-glycopeptides corresponding to 1154 glycoproteins from rat liver.¹⁷⁵ Other COFs have also been reported for N-glycopeptide enrichment including boronate affinity COFs,^176–179 and amide-linked COFs.^{180, 181} While COFs offer many advantages for glycopeptide enrichment, several challenges remain. Synthesizing COFs with consistent properties on a large scale can be challenging, which affects the reproducibility of analytical results.¹⁸² The strong covalent bonds that provide stability can also make it difficult to fully recover adsorbed glycopeptides without altering their structure or function.

Figure 4.

Synthesis and application of MCNC@COF@GSH for glycopeptide enrichment. Schematic Representation of the (a) Synthetic Route and (b) Workflow of N-Linked Glycopeptide Enrichment by the MCNC@COF Microspheres. MALDI-TOF-MS analysis of (c) human IgG digest (10⁻⁷ M) and (d) after enrichment by the MCNC@COF@GSH microspheres. N-linked glycopeptides are marked with solid red circles. Modified and reused with permission from Luo et al.¹⁷²

2.3.3. Functional Nanomaterials

Functional nanomaterials offer cutting-edge opportunities for the enrichment of glycopeptides, combining nanoscale engineering with specific biochemical interactions. These materials are crafted to interact with the unique structures of glycopeptides, which include both peptide backbones and attached glycan chains. Their design, functionalization, and application in glycopeptide enrichment leverage several advantages such as high surface area, selectivity, and the ability to integrate with other analytical techniques. Magnetic nanoparticles are often coated with various functional groups or layers that can specifically bind to glycopeptides.¹⁸³ The magnetic properties allow for easy separation from complex mixtures using a magnetic field, which simplifies the enrichment process and reduces sample handling.¹⁸⁴ Amine-functionalized magnetic nanoparticles have been demonstrated to be efficient for rapid glycopeptide enrichment in biological samples.¹⁸⁵ Gold nanoparticles can be functionalized with thiol groups that form strong bonds with gold, allowing for the attachment of capture agents such as lectins or antibodies specific to glycopeptide structures.^{186, 187} Functionalized silica nanoparticles provide a robust platform for the covalent attachment of various chemical groups or biomolecules that can selectively interact with glycopeptides.^{157, 188} Their stability and ease of functionalization make them particularly useful in chromatographic applications. Carbon-based nanomaterials including graphene, carbon nanotubes, and fullerenes, these materials are known for their electrical conductivity and high surface area.¹⁸⁹ They can be functionalized to interact with glycopeptides through hydrophobic interactions, π-π stacking, or electrostatic interactions.^{190, 191} Mechref and coworkers have demonstrated the use of magnetic carbon nanoparticles for the efficient characterization of glycans and their isomers.^{192, 193} Chen and colleagues introduced a highly effective affinity nanoprobe made of zwitterionic n-dodecyl-phosphocholine functionalized magnetic nanoparticles (ZIC-cHILIC@MNPs) for large-scale N- and O-glycopeptide enrichment.¹⁹⁴ For the non-enriched tryptic digest, only 1 glycopeptide was identified from fetuin and 2 from HRP (Figures 5a and 5b respectively). However, after enrichment using ZIC-cHILIC@MNPs, 21 N-glycopeptides were identified from fetuin and 16 from HRP (Figures 5c and 5d respectively). They demonstrate a specificity between 80–91% for the sialylated glycopeptides using the standard glycoproteins fetuin and HRP to test the novel strategy.

Figure 5.

Detection of intact glycopeptides from 2 μg fetuin and HRP by MALDI-TOF mass spectrometry analysis. MALDI-TOF spectra of standard fetuin and HRP before enrichment (a,b) and after enrichment (c,d) using ZIC-cHILIC@MNPs, respectively. Modified and reused with permission from Pradita et al.¹⁹⁴

2.3.4. pH Response Polymers

pH-Responsive polymers represent an innovative class of materials utilized in the selective enrichment of glycopeptides.^195–197 These polymers are designed to alter their chemical properties in response to changes in pH, making them especially useful for targeting glycopeptides in complex biological samples. These polymers possess functional groups capable of either accepting or donating protons in response to pH changes, leading to alterations in their solubility, charge, or conformation.^{198, 199} Common functional groups include carboxylic acids, amines, and sulfonamides, which respond dynamically to the pH variations typically found in biological environments. For glycopeptide enrichment, pH-responsive materials operate using different mechanisms. One is solubility switching where the polymer may become insoluble at certain pH levels, precipitating out of solution and capturing specific glycopeptides that interact with it.¹⁹⁵ Conversely, adjusting the pH can redissolve the polymer, releasing the enriched glycopeptides for analysis. In another mechanism, pH changes can alter the charge of the polymer.¹⁹⁶ For example, a polymer might exhibit a positive charge at low pH by protonating amine groups, binding negatively charged glycopeptides. At higher pH levels, the polymer becomes neutral or negatively charged, releasing the bound glycopeptides. Qin and colleagues reported a pH-responsive polymer system denoted poly-(AA-co-hydrazide) for the enrichment of glycopeptides derived from biological samples.¹⁹⁵ With their unique polymer, they were able to get rid of nonglycosylated peptides and observed increased intensity of glycosylated peptides from 100 fmol asialofetuin (Figure 6a and b). Interestingly, they were still able to detect glycopeptides at 1 fmol of asialofetuin after enrichment using poly-(AA-co-hydrazide) as shown in Figure 6c. Furthermore, they conducted a comparative analysis of their unique polymer for enrichment to HILIC, agarose-hydrazide beads, and lectin (WGA) beads and they obsersed a significant increase in the intensities of glycopeptides in method compared to the establish techniques (Figure 6d).

Figure 6.

MALDI-TOF-MS analysis conducted on tryptic digests of a mixture of asialofetuin and BSA at a 1:100 molar ratio. Asialofetuin (100 fmol) analyzed before (a) and after (b) enrichment with poly-(AA-co-hydrazide). Asialofetuin at a concentration of 1 fmol was examined after enrichment (c). Signal intensities of the enriched N-glycopeptides from asialofetuin, obtained through various enrichment techniques, were compared (d). These results are presented as the average of three independent measurements ± standard deviation (SD). Sequence and modification site of the identified N-glycopeptides: m/z = 1741.8 LCPDCPLLAPLn*DSR, m/z = 3017.5 VVHAVEVALATFNAESn*GSYLQLVEISR, m/z = 3672.7 RPTGEVYDIEIDTLETTCHVLDPTPLAn*CSVR. For the non-glycopeptides, the sum of the intensity of all the non-glycosylated peptide residues in the MS spectrum was used. Modified and reused with permission from Bai et al. ¹⁹⁵

pH-responsive polymers are a powerful tool in the enrichment of glycopeptides, offering tailored interactions based on controllable, pH-driven properties. Their development continues to advance, driven by the need for more selective and efficient materials in proteomics and biotechnology research. As these materials are further refined, they are expected to play a pivotal role in the analysis and study of glycoproteins in various scientific and clinical settings.

2.3.5. Other recent materials

Other than the described novel materials for the enrichment of glycopeptides from biological samples, some metal-based materials such as dendrimers and titanium/zirconia-based approaches have been described. Kayili et al.²⁰⁰ have described a titania-based sol-gel material for the rapid enrichment of glycopeptides using a pipette tip strategy. The novel material was compared with titanium dioxide (TiO₂). The titania-based sol-gel material strategy identified 29 N-glycopeptides, significantly larger than the N-glycopeptides identified using TiO₂ as enrichment material for IgG tryptic digests. In another work, Li et al. ²⁰¹ have employed an immobilized Titanium (IV) ion affinity chromatography material (Ti⁴⁺-IMAC) for the simultaneous enrichment of O-glycopeptides and phosphopeptides. They identified 32 O-glycopeptides and 10 phosphopeptides. Before enrichment, the signal of these biomolecules was not detected in the sample. Comparable numbers of O-glycopeptides and phosphopeptides were identified after 500 mass folds of BSA using Ti⁴⁺-IMAC. Yang et al. ¹⁶³ have also reported a 2D titanium-based MOF nanosheets for specific enrichment of glycopeptides from complex biological sample. Dendrimers, with their highly branched, tree-like structure and abundant functional groups, present an effective tool for enriching glycopeptides. Kayili et al.²⁰² have shown the efficiency of Poly(amidoamine) dendrimer-coated magnetic nanoparticles for the enrichment of N-glycopeptides and N-glycans.²⁰² Another work by Zhang and coworker ²⁰³ have used a dual functionalized hydrophilic dendrimer-modified MOF for the selective enrichment of neutral N-glycopeptides. This material showed superior N-glycopeptide enrichment capability by identifying 15 neutral N-glycopeptides from IgG tryptic digest compared to 10 identified by commercially available HILIC amide material. In addition to these materials, some polymeric materials including cotton and silk have been reported for enriching glycopeptides from biological samples. Xi et al.²⁰⁴ stuffed a pipette tip with adsorbent cotton termed cotton HILIC for the enrichment of N- and O-glycopeptides from mouse brain and seminal plasma. They compared the enrichment efficiency of their cotton HILIC to the widely used Venusil HILIC sorbent and mixed anion exchange chromatography (MAX). Using LC-MS/MS, their HILIC cotton material identified 2720 unique N-glycopeptides from 528 glycoproteins, while Venusil HILIC and MAX identified 1705, and 1844 unique N-glycopeptides from 338, and 415 glycoproteins respectively. Additionally, HILIC cotton material also showed a higher identification of N-glycopeptides in plasma samples compared to Venusil HILIC and MAX. Although the material showed greater efficiency for enriching O-glycopeptides from both mouse brain and plasma than other materials tested, it showed bias for enriching N-glycopeptides. Salih and his team ²⁰⁵ used a micropipette tip packed with silk as an effective solid-phase extraction tool for the quick purification of glycans and glycopeptides. This technique was capable of removing salts and detergents from samples and enriching N-glycopeptides.

2.4. Online Enrichment of Glycopeptides

While there have been great advances in the off-line enrichment of N- and O-glycopeptides, they often involve adding multiple steps to the protocol, limiting the throughput and scalability of analysis. In recent years, efforts have been made toward the development of online enrichment technologies that would allow for efficient enrichment in less time. Recently, Hu et al.²⁰⁶ developed a system titled AutoGP for online enrichment and separation of N-glycopeptides. The setup involved the use of a HILIC column, made with ZIC-HILIC material, for the preconcentration of glycopeptides, while proteins were washed out. The retained glycopeptides were then injected into a C18 column interfaced with an Orbitrap Eclipse mass spectrometer. They noted an average of 2084 glycopeptide-spectrum matches from 267 glycoproteins using the AutoGP-MS analysis method, which was approximately 1.5 times higher than manual off-line HILIC enrichment methods. Similarly, Yang et al.²⁰⁷ developed an online proteolysis and glycopeptide enrichment platform using serial thermoresponsive porous polymer membrane reactors (TPPMRs). In this platform, trypsin was immobilized on the first TPPMR for proteolysis, while the second TPPMR was immobilized with ConA and WGA in a 3:1 ratio for the enrichment of glycopeptides. A total of 262 N-glycopeptides from 155 glycoproteins were idemtified. Although this number of glycopeptides was lower than the 424 identified using the solid-phase HILIC enrichment method (HILIC-SPE), they identified a higher number of glycoproteins (140 total glycoproteins via HILIC-SPE). Additionally, the online method allows for automated proteolysis and enrichment, along with the ability to modify the membrane reactors with a variety of enzymes/lectins for desired affinity.

3. Mass Spectrometry in Glycoproteomics Analysis

3.1. MS-Based Glycoproteomics Strategies

Mass spectrometry (MS) has become an essential tool for glycoprotein analysis, offering exceptional sensitivity and specificity, along with the capability to manage complex biological samples.^{153, 208–211} Techniques based on MS facilitate the comprehensive characterization of glycoprotein structures, allowing for the identification of glycosylation sites and the examination of glycan components linked to peptides. Before the introduction of the glycopeptides into the MS system, a suitable chromatographic separation must be performed to improve the identification and characterization of the extremely heterogeneous glycopeptide structures. The most commonly utilized separation techniques in MS are LC-MS/MS,^212–215 capillary electrophoresis (CE-MS),²¹⁶ matrix-assisted laser desorption ionization (MALDI),²¹⁷ and ion mobility mass spectrometry (IMS). These methods improve the sensitivity and selectivity of glycopeptide detection by reducing the number of species ionized at the same time. LC approaches include the use of reverse phase (RP),^{218, 219} HILIC,³⁷ and PGC²¹⁵ stationary phases. Glycopeptide separation using C18 phases is the most common RP technique. In this method, the two primary moieties of the glycopeptide influence the retention time. First, the peptide backbone interacts directly with the C18 material, then the glycan structure which can show large heterogeneity dramatically altering the retention times of the glycopeptides. An increase in N-acetylhexosamine residues in the glycopeptide reduces the retention, while more neuraminic acid residues lead to increased retention.^{213, 220} Neutral monosaccharides such as mannose and galactose also increase retention, whereas fucose units have minimal impact on the elution time of glycopeptides. Temperature and column length are additional chromatographic parameters that can be further adjusted to allow for the separation of glycopeptide isomers, providing additional dimensions of molecular microchanges, and enhancing the differentiation between the studied groups.^{72, 215, 220} Gutierrez Reyes et al.²²⁰ used a 50 cm long C18 column to separate glycopeptide isomers from human serum haptoglobin samples obtained from cirrhosis and hepatocellular carcinoma (HCC) patients. The results showed a set of isomeric N-glycopeptides with better capabilities differentiating the cirrhosis and HCC evaluated groups. For example, the DOSG protocol described by Zhong et al. ⁹² focused on the differentiation of the isomeric α2,3 and α2,6 sialic acid linkages using a 60-minute long gradient. HILIC phases have different selectivity for which the interaction of the polar stationary phase and the glycopeptides depends considerably on the glycan moiety hydrophilicity. ^{221, 222} Huang et al. ³⁷ used a Halo Penta-HILIC column to differentiate the α2,3 and α2,6 sialic acid linkage of tryptic-digested N-glycopeptides obtained from fetuin and IgG. Other HILIC approaches are described in the next references.^223–225 PGC, on the other hand, has two unique mechanisms; the polar glycan moieties of the glycopeptides exhibit strong retention in the stationary phase, whereas the retention of peptides is dependent on the relative size of the backbone and the type of side chains present.²²⁶ Zhu et al.²¹⁵ investigated the effect of the pH and temperature on the retention of sialylated N-glycopeptides derived from a variety of standard glycoproteins. They observed that a pH of 9.9 and a temperature of 75 °C are the best conditions for a reproducible separation of the isomeric N-glycopeptides. More recently, Daramola et al. ²²⁷ demonstrated the efficiency of mesoporous graphitized carbon (MGC) material for separating N- and O-glycopeptides using LC-MS/MS. The technique was validated by applying it to a wide range of glycoprotein sources, such as human blood serum, α1-glycoprotein, asialofetuin, and bovine fetuin.

While LC-MS/MS has become a standard in the analysis of glycopeptides, other techniques such as capillary electrophoresis (CE) can be employed alongside or in combination with the LC system to provide complementary information regarding the glycopeptide sample. In the past, capillary electrophoresis has suffered from poor sensitivity due to the low injection volume and limited optical path length for optical detectors.²²⁸ The development of cutting-edge CE-MS interface technologies, CE has seen a significant increase in sensitivity, separation efficiency, and usage in proteomic/glycoproteomic workflows, allowing for the analysis of much smaller sample volumes on a much shorter time scale.^{229, 230} Furthermore, capillaries may be modified with micro cartridges containing adsorbents allowing for online solid-phase extraction (SPE) enrichment of glycopeptides or other analytes of interest based on the adsorbents.^{216, 231} Mancera-Arteu et al. ²³² developed an online titanium dioxide SPE-CE-MS platform for the selective enrichment of N- and O-glycopeptides derived from recombinant human erythropoietin (rhEPO) – a commonly used glycoprotein hormone for the treatment of cancer and kidney diseases. They reported up to a 100 times lower limit of detection (LOD) for rhEPO-derived N- and O-glycopeptides compared to traditional CE-MS. More recently, Mancera-Arteu et al. ²¹⁶ pioneered another online SPE-CE-MS method for the enrichment tryptic digests of rhEPO using adsorbed phenylboronic acid, achieving LODs up to 200 times lower than titanium dioxide SPE-CE-MS for N-glycopeptides.

Regarding the MS system, electrospray-ionization (ESI-MS)^{32, 233} and MALDI-MS²¹⁷ are the most common ionization techniques used for sample introduction. ESI-MS, frequently coupled with liquid chromatography (LC-ESI-MS), has become a staple in glycopeptide research due to its ability to generate multiple charged ions The advent of nano-ESI technology has further refined this approach, dramatically improving sensitivity and enabling the detection of glycopeptides at lower concentrations. ²³⁴ In contrast, MALDI-MS has carved its niche in glycopeptide analysis, prized for its straightforward methodology and resilience to sample impurities. The predominant production of singly charged ions in MALDI simplifies spectrum interpretation, a significant advantage when dealing with complex glycopeptide mixtures. Recent innovations in matrix formulations have addressed previous limitations, particularly enhancing the analysis of sialylated glycopeptides, which were historically challenging to detect using this method ^{235, 236}. While MALDI-MS excels in the analysis of isolated glycans, LC-MS/MS has established itself as the preferred approach for comprehensive glycoproteomics studies. This preference stems from the ability of LC-MS/MS to provide detailed, site-specific glycosylation information, a critical aspect in quantitative glycosylation studies.²³⁷

IMS is a gas-phase approach recently gaining attention for its application in the separation of different biomolecules. This technique has the ability to separate glycopeptide isomers and isobars due to the differences in the structural conformations of the glycopeptides.^238–240 Differences in shapes and sizes of the glycopeptide isomers permit separation in the gas phase using ion mobility separation technique. Lu and co-workers²⁴¹ have employed the IMS technique to identify and quantify sialic acid linkage isomers of intact N-glycopeptides in HCC. They were able to show aberrant sialylation of haptoglobin in this disease condition. Other works have also shown the site-specific characterization of sialic linkage N-glycopeptide isomers using IMS.^{242, 243} In addition, O-glycopeptide isomers have been separated using IMS further demonstrating the capabilities of this technique for glycoproteomics analysis.²⁴⁴ A notable advancement in this field is field asymmetric waveform ion mobility spectrometry (FAIMS). Recent studies have demonstrated the ability of FAIMS to differentiate O-glycopeptide isomers from mucins, as well as α- and β-GalNAc anomeric glycopeptides, however, isomeric-free glycans were only partially distinguishable.²⁴⁵ The integration of ion mobility separation with LC and mass spectrometry creates a multi-dimensional analytical platform that operates on distinct time scales. LC separation, which typically occurs over seconds to minutes, differentiates analytes based on their hydrophobicity. ²⁴⁶ Subsequently, these separated analytes undergo ion mobility separation, which takes place on a millisecond timescale and distinguishes ions based on their collision cross-section (CCS). Finally, time-of-flight (TOF) detection occurs within microseconds, allowing for the acquisition of several mass spectra during each IMS separation event. ^{247, 248} This multi-tiered approach offers several advantages over traditional analytical methods. It enhances the speed of analysis while maintaining high resolution, significantly improves peak capacity, and increases overall selectivity. These benefits make IMS an invaluable tool in the characterization of complex biological samples, where the differentiation of structurally similar compounds is crucial. ²⁴⁹ The compatibility of TOF mass spectrometers with IMS is attributed to their rapid scan speeds, which align well with the fast separation times of ion mobility. An advanced iteration of this technology is the trapped ion mobility spectrometry (TIMS) device, which further leverages the resolving power of IMS and the acquisition speed of TOF instruments.²⁵⁰ Another notable approach that has gained prominence is MALDI coupled with TOF mass analyzers. This combination has emerged as a powerful tool in clinical glycomics analysis, offering several advantages over traditional methods.²³⁶ The popularity of the technique in clinical settings can be attributed to its high sensitivity, ease of use, and ability to generate easily interpretable spectra.²⁵¹ A recent study by Hu et al. ²⁵² has shown the potential of novel materials in enhancing glycopeptide analysis using MALDI-TOF MS. The researchers developed a unique enrichment method utilizing Knoevenagel copolymers modified with polydopamine-adenosine (PDA-ADE@KCP).²⁵² This innovative approach was applied to analyze intact glycopeptides from human IgG, a crucial class of glycoproteins in the immune system.²⁵² The results of this study were particularly noteworthy, as the PDA-ADE@KCP material demonstrated superior performance in capturing intact N-glycopeptides compared to many recently reported HILIC-based materials, which include the study of Sajid et al. ²⁵³ and Zhou et al.^{253 162}

Considering the recent advances in MS techniques, system automation, and high-throughput screening methods are the major focus in the field. This allows for the rapid processing of multiple samples simultaneously, significantly reducing the time required for analysis. Chen et al.²⁵⁴ described a glycopeptide, glycan, and glycosite analysis that fully screen the glycoprotein structural information. The research group combined C4-tip and C18/MAX-tip workflows to reduce the sample preparation considerably reducing analysis time. Using the proposed protocol, the intact glycopeptides can be enriched in a high-throughput system. After N-glycopeptide enrichment, the users can continue with LC-MS glycoproteomics analysis or process the intact glycopeptides by PNGaseF into de-N-glycosylated peptides and glycans. Subsequently, the glycans can be separated from the peptides using reverse-phase C18. The developed format offers the flexibility to process the samples using a single pipet, a multichannel pipet, or liquid handling systems. The methodology was tested on standard glycoproteins, and bodily fluids, and the C18/MAX tips showed a reusability of 5 cycles. In another work, Liu et al.²⁵⁵ developed an automated method for the enrichment of blood serum N-glycopeptides using an HPLC system with an enrichment time of 20 min. The automated enrichment method showed a specificity of 84% compared to the 36% observed using the common tip method. The number of N-glycopeptide identifications was 1198 and 745, respectively. The method was used for the identification of N-glycosylation alterations in the blood serum of pancreatic cancer (PC) patients. Results showed a significant decrease in the IgG1 monofucosylated glycans of the PC cohort.

MS glycoproteomics analysis offers the ability to elucidate structural information using MSⁿ analysis with different fragmentation techniques such as electron transfer dissociation (ETD), high-energy collision dissociation (HCD), ²⁵⁶ ultra violet photon dissociation (UVPD), ²⁵⁷ and combined strategies like EThcD (ETD + HCD) ^{258, 259} which allows a selective fragmentation of both, the glycan and peptide moieties. A study by Riley et al.²⁶⁰ compared various fragmentation methods for intact glycopeptide analysis, focusing on their applicability to N- and O-glycoproteomics. For N-glycopeptides, higher-energy collisional dissociation (HCD) and stepped HCD (sHCD) proved equally effective in terms of identification numbers. However, sHCD generally produced higher quality spectra using sHCD with 30 ± 10 of normalized collision energy (NCE) values. While HCD-based methods outperformed electron-transfer dissociation (ETD) techniques in N-glycopeptide identification, its combination with electron transfer dissociation (ETD), particularly EThcD, proved to be crucial for site-specific analysis of O-glycopeptides. This finding underscores the importance of incorporating ETD-based dissociation techniques in glycoproteomics methodologies aimed at comprehensive O-glycopeptide characterization.²⁶⁰

Ultimately, the combination of a suitable separation technique and a sensitive detection system is essential for accurate and precise glycoprotein analysis. Selecting complementary techniques can further enhance the sensitivity of the analytical system. In glycopeptide derivatization, either the glycan or the peptide fractions of the glycopeptide may be derivatized for improved separation in a given system. Tandem mass tag (TMT) has been used for the derivatization of the peptide backbone.^{261, 262} 2AB²⁶³ labeling and the DOSG ⁹² sialic acid derivatization are two examples of glycan derivatization. All the strategies described have displayed an increase in the sensitivity of glycopeptide detection and also the ability to achieve isomeric separation. After data acquisition, the compiled data should be normalized prior to the application of statistical tests. Relative abundance using the total glycoproteome signal, the addition of known peptides or glycopeptides as internal standards are common techniques used for glycoproteomics data normalization. Recent glycoproteomics approaches use isobaric tags such as TMT, iTRAQ, and DiLeu for the absolute quantitation of glycopeptides.^{35, 211, 261, 264–266} Undoubtedly, the quantitation of glycopeptides in clinical tests should be assessed by absolute quantitative analysis. Hence, it is crucial to employ MS-targeted methods that offer an accurate and adaptable quantitative approach, which can be replicated in many laboratories and applied to extensive clinical cohorts.^{267, 268} MS-targeted techniques provide a precise and sensitive methods designed specifically for the detection and quantitation of specific analytes. This is a critical step in the process of discovering and validating biomarkers.²⁶⁷ Initially, targeted analysis using MRM on triple quadrupole mass spectrometers was the preferred method for targeted studies. However, the low resolution of triple quadrupole and the complex process of product ion selection present significant challenges for this approach. Parallel reaction monitoring (PRM) is an alternate approach that utilize the high resolution and accuracy of orbitrap mass spectrometers to simultaneously detect all quantitative fragment ions of a targeted analyte with increased specificity and sensitivity.^269–274

3.2. Data Analysis and Interpretation

Despite the advancements in MS technology, the analysis and interpretation of MS data for glycopeptide analysis remains challenging due to the complexity of glycosylation and the vast amount of data generated. Glycopeptides exhibit heterogeneity in glycan structures and glycosites, complicating their identification and quantification. Addressing these challenges requires sophisticated bioinformatics tools and algorithms capable of processing and interpreting MS data. Several bioinformatics tools have been developed for glycopeptide identification and quantification, including software that can predict glycosylation sites, analyze glycan compositions, and match MS/MS spectra to glycopeptide sequences. Byonic^™ is one of the most utilized software for glycoproteomics analysis and one of its important features is the simultaneous searching of modifications such as oxidations, deaminations, phosphorylations, and glycosylations.²⁷⁵ In the past years, several tools have been developed including GlycopeptiudeGraph,²⁷⁶ GPSeeker,²⁷⁷ MSfragger-Glyco,²⁷⁸ pGlyco 2.0,²⁷⁹ O-PairSearch,²⁸⁰ Glycodecipher,²⁸¹ GlycoHybridSeq,²⁸² GlycReSoft,²⁸³ and Protein Prospector.²⁸⁴ Additional information on these and other interesting tools are described in the next references.^{35, 285} Although some of the listed software offers the ability to process and quantify large glycopeptide data sets, the reported results still produce large false positive signals. Therefore, a thoughtful selection of the identification and quantitation software is mandatory. For example, MSFragger-Glyco^{286, 287} employs an innovative approach, utilizing open (mass-tolerant) and mass offset search algorithms in a single pass, coupled with fragment ion indexing. This method significantly enhances computational efficiency, making it particularly suitable for large-scale glycoproteomic studies. However, its limitation in localizing only one glycosylation site may pose challenges when analyzing heavily altered O-glycoproteins, such as those with mucin domains.^{288, 289} In contrast, O-Pair Search, integrated into the MetaMorpheus suite, is specifically tailored for O-glycoproteomic data analysis. It employs a two-step strategy, first conducting an open search of HCD spectra, followed by the utilization of EThcD spectra for precise O-glycan localization ²⁹⁰. The pGlyco software utilizes a glycan-first search method and glycan ion-indexing technique. This methodology accelerates the speed at which glycopeptides are searched and also accommodates modified saccharide units.²⁹¹ The ability of pGlyco to estimate both glycan and peptide false discovery rates and localize glycopeptides by sHCD and ETD spectra, enhances its versatility, ²⁹² while GlycReSoft adopts a more traditional proteomics approach, treating glycans as variable modifications. The distinguishing feature of this software is its ability to study both glycoproteomics and glycomics data, which sets it apart from other tools in this comparison. Its flexibility in building custom glycan databases based on user-defined parameters offers researchers considerable control over their analyses.²⁸³ Notably, Protein Prospector, which was originally developed for phosphoproteomics, has successfully been adapted for the quantitation of O-glycopeptides. Its comprehensive approach to modification analysis, including consideration of neutral losses and comparative analysis of various allocations of alteration sites, has positioned it as a high-performing tool for O-glycoprotein analysis. ²⁹³ Lastly Byonic software, is widely regarded as the gold standard in glycoproteomics due to its flexibility in defining variable modification types. However, recent studies have highlighted variability in results from Byonic from different research groups using identical raw data, underscoring the importance of standardized analytical protocols.²⁸⁹ Each of these tools exhibits unique strengths in handling different spectrum types and data processing approaches. While MSFragger-Glyco and O-Pair Search focus primarily on HCD and EThcD spectra, respectively, pGlyco demonstrates versatility in merging various spectral types. GlycReSoft, Protein Prospector, and Byonic offer broader flexibility, adapting to diverse experimental setups. Additional information on these and other interesting tools are described in the next references.^{35, 285} The selection of an appropriate software tool depends on various factors, including the specific type of glycosylation under study, the required depth of analysis, available spectral data, and computational resources. Given the complementary strengths of these tools, researchers may benefit from employing multiple software packages to achieve a more comprehensive analysis of glycopeptide data, particularly when dealing with complex samples or when high-confidence identifications are crucial. Although some of the listed software offers the ability to process and quantify large glycopeptide data sets, the observed results still produce large false positive signals, therefore, a manual revision is still needed.

4. Advancing Clinical Diagnostics: Glycoproteomics Analysis and the Role of Glycopeptide Enrichment

Glycosylation is crucial in biological functions and the pathogenesis of diseases. Their structural complexity and diversity pose analytical challenges, necessitating efficient enrichment techniques for their study. This selective isolation is crucial for reducing sample complexity, improving detection sensitivity, and facilitating the detailed analysis of protein N- and O-site glycosylation. Glycoproteomics analysis has led to significant advancements in the identification of disease biomarkers in clinical research. These biomarkers hold the potential to revolutionize diagnostics, prognostics, and therapeutic strategies across a spectrum of diseases.

4.1. Glycoproteomics Contributions to Cancer Research

Cancer is characterized by aberrant changes in glycosylation, which influences tumor progression, metastasis, and immune evasion. Through glycoproteomics analysis, researchers can elucidate the glycosylation alterations in tumor cells, contributing to the development of cancer biomarkers and targeted therapies. This detailed N- and O-site glycan analysis enables the identification of tumor-associated antigens and the exploration of cancer immunotherapy strategies. HILIC materials are commonly employed to enrich glycopeptides derived from several types of cancer-related samples including hepatocellular carcinoma,^{220, 294, 295} lung cancer,^{296, 297} and colorectal cancer. ²⁹⁸ Lectin affinity enrichment has also found application in a wide range of cancer biomarker discovery research including lung cancer,²⁹⁹ brain cancer, prostate cancer,⁷⁰ breast cancer,^{300, 301} and hepatocellular carcinoma.³⁰² Some examples of the application of hydrazide chemistry-based enrichment in cancer include lung cancer,^{303, 304} breast cancer,^{305, 306} prostate cancer,³⁰⁷ hepatocellular carcinoma.³⁰⁸ Hydrazide chemistry has also been employed in combination with HILIC for enriching glycopeptides in human metastatic hepatocellular carcinoma.³⁰⁹ Song et al. used hydrazide chemistry in combination with lectin affinity chromatography to enrich glycopeptides in esophageal adenocarcinoma.³¹⁰

Wang et al.³¹¹ successfully enriched N-glycopeptides from the tryptic digests of the proteome obtained from urine samples of patients with bladder cancer. They used the novel thickness-controlled hydrophilic Mg-MOF coating-functionalized magnetic graphene composite (MagG@Mg-MOFs-1C), and the result showed the identification of 406 N-glycopetides from 185 glycoproteins. Among the identified glycoproteins, α−2-macroglobulin, complement C4-B, and α−1-antitrypsin showed N-glycopeptides with the potential for differentiating the bladder cancer cohort from the healthy control samples. Chu et al. ³¹² functionalized a hydrophilic tripeptide in a magnetic metal oxide composite. The novel material was used in the identification of potential N-glycopeptide biomarkers in cervical cancer. The statistically significant N-glycopeptides between the studied cohorts derived from the glycoproteins include heat shock protein beta-1, DNA topoisomerase 2-alpha, proliferation marker protein Ki-67, and 60 KDa heat shock protein. Zacharias et al.³¹³ studied the N-glycosylation of breast and brain cancer cells using HILIC and ERLIC enrichment, a total of 497 N-glycopeptides were characterized. The IPA (QIAGEN^™) analysis showed a strong association of the cell glycoproteome with the patways associate with brain cancer metastasis. In another work, Du et al.,³¹⁴ combined HILIC and TiO₂ microparticles for the enrichment of N-glycopeptides and phosphopeptides derived from lung cancer and normal tissue samples. The results showed a significant dysregulation in 249 N-glycopepetides and 486 phosphopepetides. Pickering et al.³¹⁵ completed glycoproteomics analysis in plasma samples derived from an immune checkpoint inhibitors (ICI) therapy. The findings indicated that 143 N-glycopeptide biomarkers were differentially expressed between patients who passed away within six months after beginning ICI treatment and those who did not show disease progression for three years. In another study, Shah et al.³¹⁶ conducted an in-depth proteomics and glycoproteomics analysis to compare the LNCaP and PC3 prostate cancer cell lines, which are commonly used to model androgen-dependent and androgen-independent prostate cancer, respectively. These cell lines are known for their differing metastatic capabilities. The approach included a peptide and glycopeptide labeling strategy using iTRAQ. The comparison revealed significant differences between the two cell lines, with numerous glycoproteins and proteins being differently expressed. Hubbard et al.³¹⁷ reported a selective labeling approach focused on the identification of cell-surface glycoproteins from prostate cancer tissues and PC-3 cells. The samples were treated with the peracetylated N-azidoacetylgalactosamine azidosugar. Using this approach, 70 glycoproteins were identified including CD146 and integrin beta-4 known to facilitate metastatic behavior.

Recently, Tian and colleagues ³¹⁸ developed a chemical strategy involving a trifunctional probe to capture, crosslink, and enrich carcinoembryonic antigen (CAE) from plasma samples of cancer patients. Intact N-glycopeptides were generated from the glycoprotein using multiple enzymes and subsequently enriched through HILIC. Using this innovative method, 26 N-glycosylation sites were identified, which is the highest number of N-sites reported for CAE to date. In patients with lung cancer, glycosylation at N274 and N580 was increased, while glycosylation at N104, N152, N182, N360, N466, N529, and N665 was elevated in patients with colorectal cancer. Li et al. ³¹⁹ employed ZIC-HILIC for the enrichment of N-glycopeptides from serum samples of locally advanced cervical cancer (LACC) patients. They were able to identify six glycopeptide biomarkers derived from the glycoproteins MASP1, LUM, ATRN, CO8A, CO8B, and CO6 that can predict the efficacy of neoadjuvant chemotherapy for treating LACC by measuring their expression levels before and after the chemotherapy. These tentative markers showed increased abundance before neoadjuvant chemotherapy and a significant decrease in abundance after.

4.2. Glycoproteomics Contributions to Neurodegenerative Diseases

In neurodegenerative diseases such as Alzheimer’s disease (AD) and Parkinson’s disease (PD), abnormal glycosylation patterns have been implicated in disease pathogenesis. Moreover, glycoproteomics analysis aids in understanding the role of glycosylation in protein aggregation and neurodegeneration, providing a molecular basis for disease progression and treatment response monitoring. Recent studies examining the role of glycoproteins in neurodegenerative diseases such as AD^320–323, PD^{324, 325}, and mild cognitive impairment (MCI)^{326, 327} have employed these techniques, either individually or in combination. The glycosylation of the tau protein is an initial aberration that could potentially aid in hyperphosphorylation, which is a pathogenic characteristic of the brain in Alzheimer’s disease. ³²⁸ Tau is involved in stabilizing axonal microtubules in healthy brains and does not undergo glycosylation. In contrast, Tau experiences abberant post-translational modifications and aggregates into harmful structures in AD brain.³²⁸ Both phosphorylation and glycosylation are PTMs that have been reported to modulate the aggregation of Tau, thereby impacting Tau pathology.^{329, 330} Zhang et al. ³³¹ employed ZIC HILIC to enrich glycopeptides derived from human brains in a large-scale AD study that provided new insights into the molecular alterations in N-glycosylation that are involved in AD pathogenesis.³³¹ Pan et al. ³³² have employed HILIC enrichment and fractionation technique coupled with nanoLC-HCD-PRM-MS analysis for the identification and quantification of potential serum N-glycopeptide biomarkers for early-stage detection of AD. ³³² Another study has employed a combination of multi-lectin affinity with HILIC for selective enrichment of N-glycopeptides before LC-MS/MS analysis with the potential to identify diagnostic biomarkers and or therapeutic targets in AD. ³³³ A similar sequential approach of N-glycopeptide enrichment using HILIC and boronic acid was also used to investigate AD glycosylation signatures.³³⁴ In addition, a recent study conducted by Gutierrez Reyes et al. ⁷² utilized the HILIC enrichment technique to identify potential N-glycopeptide biomarkers derived from low abundant-serum proteins of MCI patients. Identification of O-glycopeptide biomarkers has also been facilitated by several enrichment techniques. Chen et al. ³²⁶ have utilized boronic acid enrichment in combination with high-pH fractionation to study O-glycosylation alterations in MCI and AD. Cummings et al. ³²⁰ employed a combination of multi-lectin and HILIC enrichment during the glycoproteomics analysis of normal, asymptomatic, and symptomatic AD brain tissues. They used a combination of six lectins incuding ConA, RCA, SNA, WGA, AAL, and VVA for the enrichment of glycoproteins from brain lysates, followed by tryptic digestion of the proteins. Then, the tryptic digested glycopeptides were subjected to enrichment using HILIC SPE columns. The results showed a total of 2035 glycopeptides composing a total of 303 glycoproteins. They found out a glycome with a large percentage of fucosylated glycans (44%), followed by high mannose (40%), and sialylated (15%).

In an aim to identify potential glycopeptide biomarkers in Parkinson’s disease, several studies have employed glycoproteomics analysis using different enrichment strategies to achieve this goal. ^{335, 336} Yang et al. ³²⁴ utilized lectin affinity chromatography and HILIC combined with LC-MS to identify the glycoproteome changes in the serum of patients with PD. They employed three specific lectins: AAL, which preferably binds fucose; SNA, which targets α2,6-linked sialic acids; and MAL II which binds α2,3-linked sialic acids. This was followed by tryptic digestion and further HILIC enrichment. In their study, they identified 1360 N-glycopeptides, with 116 glycopeptides being exclusively found in the serum of patients with PD. They noted unique glycosylation changes in 32 glycoproteins and 10 glycan structures, involving core fucosylation, sialylation, and bisecting GlcNAc. Additionally, unique sialylated biantennary/triantennary structures in eleven proteins and an increase in the core fucosylation and sialylation in ten proteins were identified. Suggesting that the levels of specific glycosyltransferases might be altered in PD brain.

4.3. Glycoproteomics Contributions to Metabolic Disorders and Other Diseases.

Glycoproteins analysis has also been implicated in other biological conditions including metabolic disorders and infectious diseases. Specific glycoproteins, including fibronectin,^{337, 338} integrin,³³⁷ apolipoprotein C3,³³⁹ plasminogen,³⁴⁰ and complement C8α,³⁴¹ have been associated with diabetes-related complications such as diabetic retinopathy and inflammation.³⁴² Furthermore, specific diabetes-related glycoproteins have also been implicated in the worsening of age-related conditions such as osteoarthritis. Recently, Luo et al. ³⁴¹ employed the use of three lectins including ConA, WGA, and RCA 120 to enrich glycopeptides derived from the knee cartilage of 10 patients diagnosed with primary knee degenerative osteoarthritis (OA), with five of the ten patients also having a type 2 diabetes (T2D) diagnosis (DMOA). In the study, a total of 729 N-glycopeptides were identified derived from 374 glycoproteins, out of which 444 N-glycopeptides and 257 glycoproteins were quantified. A significant increase in N-glycosylation of complement C8α chain at glycosylation site N437 was noted in patients with DMOA compared to OA. The N-glycosylation of N437 plays a critical role in complement activation, a process that has been shown to contribute to the development and progression of osteoarthritis.^{343, 344} This connection suggests a link between T2D and osteoarthritis. Li and colleagues³⁴⁵ recently conducted a study utilizing a novel MOF named U6N/Pv@Glc, which was synthesized using an amide and multihydroxyl strategy. By employing their enrichment technique, they identified 379 glycopeptides and 347 glycosylation sites originating from 165 glycoproteins. This MOF was employed for the enrichment of glycopeptides in human serum samples from patients with T2D. By employing their enrichment technique, they identified 379 N-glycopeptides and 347 glycosylation sites originating from 165 glycoproteins. They reported 14 glycoproteins to be correlated with T2D some of which included LPA, SERPINA3, C2, CPS1, SELENOP, APOM, and HP.

Glycosylation also plays an essential role in adaptive and innate immune responses, particularly in combating bacterial, viral, and fungal infections. N- and O-Glycosylation of B-cell and T-cell plays a role in receptor trafficking and signal transduction.³⁴⁶ Furthermore, glycosylation of membrane receptors also plays a role in pathogen recognition by recognizing specific glycans signatures on the pathogen surface.³⁴⁷ In 2023, Willems et al. ³⁴⁸ used cross-linked agarose gel material (Sepharose CL-4B beads) to enrich glycopeptides derived from plasma samples obtained from 42 controls and 91 patients. Of the 91 patients, 53 were diagnosed with bacterial infections, while 38 were diagnosed with viral infections. A total of 463 N-glycopeptides derived from highly abundant plasma glycoproteins were identified. A significant alteration in glycopeptides originating from haptoglobin, α−1-acid glycoprotein 1, complement C3, and hemopexin was observed between healthy individuals and infected (viral or bacterial) patients.³⁴⁸ Glycoproteomics biomarkers have also been identified in autoimmune disorders such as narcolepsy,⁴¹ multiple sclerosis,³⁴⁹ and rheumatoid arthritis.³⁵⁰

4.4. Glycoprotein Biomarkers for Clinical Applications

Glycoproteins play a crucial role in various biological processes and have become essential in clinical diagnostics, especially for diseases such as cancer. It is one among the currently accepted biomarkers for a range of disorders ³⁵¹. Glycoproteomic analysis has greatly progressed, providing potential techniques for early cancer detection and predicting therapy effectiveness, including responses to immunotherapies ³⁵². The process of biomarker development encompasses multiple crucial stages. Promising biomarkers are found and confirmed using small sample volumes. These are subjected to two primary forms of validation: analytical validation, which assesses the accuracy and reproducibility of measuring the desired analyte within the samples collected from patient, and the clinical validation evaluates the reliability of the test result and its association with the clinical phenotype.³⁵² While several biomarkers are still in the process of being discovered, others have progressed to the stage of receiving approval from the FDA. Nevertheless, existing biomarkers encounter difficulties, including problems with selectivity and specificity, especially when it comes to detecting diseases at an early stage.³⁵³ This underscores the continuous requirement for research in this area of study. According to Suttapitugsakul et al.,³⁵³ the glycoproteins human epidermal growth factor receptor 2 (HER2/neu), human epididymis protein 4 (HE4), thyroglobulin (TG), alpha-fetoprotein (AFP), cancer antigen 15–3 (CA15–3, MUC1), prostate-specific antigen (PSA), carcinoembryonic antigen (CEA) and cancer antigen 125 (CA125, MUC 16), are among the cancer biomarkers that have been approved by the FDA. In addition, He et al.,³⁵² made a list of glycoprotein biomarkers unique to various cancers. These include α-Fetoprotein (AFP) for Liver cancer, β-Human chorionic gonadotropin (β-hCG) for Testicular and ovarian cancer, Carbohydrate antigen 19–9 (CA19–9) used for pancreatic, ovarian, and gastrointestinal cancer, Mucin1 (Cancer antigen 15–3/27–29) (MUC1 (CA15–3 /CA27–29)) for Breast cancer, Carcinoembryonic antigen-related cell adhesion molecules (CEA and various CEACAMs) are used for colorectal, pancreatic, and lung cancer, Mucin16 (Cancer antigen 125) (MUC16 (CA125)) used for ovarian cancer. In another study by Nie et al.,³⁵⁴ glycoprotein biomarkers were found to distinguish pancreatic cancer from other pancreatic disorders such as cysts, diabetes, obstructive jaundice, chronic pancreatitis, and healthy individuals. The trio of α−1-antichymotrypsin (AACT), haptoglobin (HPT), and thrombospondin-1 (THBS1) exhibited superior performance compared to CA 19–9 in differentiating pancreatic cancer from healthy controls with AUC of 0.95, diabetes with AUC of 0.89, cysts with AUC of 0.82, and chronic pancreatitis with AUC of 0.90. The study determined that a potential glycoprotein biomarker panel for the early identification of pancreatic cancer was discovered by combining lectin-array assays and serum quantitative proteome analysis.

Glycoproteins are essential in the diagnosis and monitoring of diseases, particularly in the field of cancer. Although there has been advancement and acceptance of several glycoprotein biomarkers, there are still obstacles to overcome in order to enhance their selectivity and specificity for the early diagnosis of diseases. Continuous progress in glycoproteomic technology and additional research are necessary to improve the therapeutic usefulness of these biomarkers.

5. Challenges and Future Directions

The enrichment of glycopeptides, a crucial step in the analysis of glycoproteins, faces several analytical challenges. The complexity of glycosylation patterns and the heterogeneity of glycopeptide structures necessitate highly selective and efficient enrichment techniques. Current methods, including LAC, HILIC, ERLIC, and Immunoaffinity Capture have limitations in terms of specificity, recovery rate, availability, and potential biases towards certain glycopeptide forms. For instance, HILIC often favors N-glycopeptides with large glycan structures but not small of O-glycopeptides, and LAC may not capture glycopeptides with less common glycan motifs effectively such as the hybrid structures. These limitations highlight the need for advancements in enrichment technologies that can handle the diverse and complex nature of glycopeptides. The incorporation of novel materials, such as MOFs and COFs, offers new possibilities for glycopeptide enrichment. These materials are noted for their high porosity and customizable surfaces, which can be tailored to enhance interactions with specific glycopeptide structures. Future research could focus on enhancing the stability and robustness of these materials in biological matrices, ensuring consistent performance and preventing degradation in the presence of complex biological fluids. Functional nanomaterials, especially those with magnetic properties or specific functional groups, also present opportunities for rapid and efficient enrichment directly from bodily fluids or tissue extracts. The enormous amount of data generated in the LC-MS/MS glycoproteomics analysis makes data analysis and interpretation challenging. Enhanced bioinformatics tools specifically designed for glycoproteomics are necessary. Database searching tools such as Byonic or MSFragger have been predominantly used in the past for spectral matching. ^{290, 355–357} However, they encounter several limitations, particularly when dealing with the complexity and variability of glycoproteomics data. The integration of artificial intelligence (AI) and machine learning (ML) can address some of these limitations and enhance the analysis significantly.

AI can assist in pattern recognition and prediction of glycosylation sites, improving the identification and quantification of glycopeptides. They can automate and streamline data processing, handling large volumes of data more efficiently than traditional methods. This capability is crucial for glycoproteomics, where datasets are extensive and complex. Deep learning, a subdivision of artificial intelligence, consists of layers of interconnected nodes that mimic neurons in a human brain. The weight of connections between these nodes is adjusted by training the model using sample data to minimize the difference between the predicted and actual outcomes. ³⁵⁸ The broad availability of open-access datasets further supports the extensive training of these neural networks. ^{359, 360} Deep learning models such as pDEEP, MS2CNN, and DeepGlyco have been developed, leveraging neural networks trained on sample data to predict MS/MS spectra. ^361–364 Moreover, DL models have been used for other aspects of LC-MS/MS data prediction such as retention time prediction and post-translational modification prediction. ^{362, 365, 366} Machine learning models can learn from existing glycoproteomic data to predict glycan structures, glycosylation sites, and even the biological function of glycopeptides. These models can be trained to distinguish between disease-specific glycosylation patterns, facilitating the discovery of biomarkers for diseases such as cancer, neurodegenerative disorders, and inflammatory conditions. Simpler machine-learning models, including, received operator characteristic (ROC) analysis,³⁶⁷ support vector machines, principal component analysis (PCA),³⁶⁷ decision trees,³⁶⁸ random forest,^{369, 370} and binary logistic regression models have long been prevalent in scientific laboratories and clinical settings for cohort classification tasks and biomarker validity testing. ROC analysis is by far the most commonly used ML algorithm used for measuring the accuracy of a diagnostic test. It relies on measuring the true positive rate (sensitivity) vs the false positive rate (specificity) at different thresholds of classification.³⁷¹ The accuracy is measured by the area under curve (AUC) score of a sensitivity vs specificity plot, with a value closer to 1 indicating higher accuracy of prediction.

The integration of AI with glycoproteomic data could lead to more precise diagnostic tools and targeted therapies, paving the way for personalized medicine approaches. Figure 7 describes the integration of AI and ML in analyzing MS-generated glyocoproteomics data to facilitate data analysis, pathway prediction, identification of novel biomarkers, therapeutic targets, and ultimately influencing precision medicine. To translate glycoproteomic discoveries into clinical applications, it is crucial to develop standardized protocols and validation procedures for glycopeptide enrichment and analysis. Improved techniques for the selective enrichment of glycopeptides will enable more detailed and accurate profiling of glycosylation in various diseases. Collaborations between academia, industry, and clinical laboratories are essential to drive innovations from the laboratory bench to clinical implementation. The development of portable or miniaturized devices for glycopeptide analysis could also enhance the applicability of glycoproteomic studies in clinical settings, offering rapid diagnostic capabilities.

Figure 7.

Integrative workflow from mass spectrometry data acquisition to artificial intelligence/machine learning-assisted data analysis, disease associated pathway prediction, identification of disease biomarkers and therapeutic targets, and precision medicine.

6. Conclusion

In conclusion, this comprehensive review underscores the pivotal role of advanced glycopeptide enrichment techniques and mass spectrometry in the burgeoning field of clinical glycoproteomics. The detailed evaluation of various enrichment strategies and novel material-based methods like MOFs and COFs, highlights significant advancements in the selective isolation and analysis of glycopeptides. These techniques have proven instrumental in enhancing the sensitivity and specificity of biomarker identification, which is crucial for the early detection and therapeutic monitoring of diseases. Despite these advancements, the review also illuminates persistent challenges within the field. Specificity and efficiency in glycopeptide enrichment remain hindered by the complex and diverse nature of glycosylation patterns. The scalability of these techniques in clinical settings also presents a significant hurdle, necessitating ongoing innovation and optimization. Looking forward, the integration of cutting-edge computational tools, such as AI and ML, promises to revolutionize glycoproteomic analyses. These technologies offer the potential to refine data processing and enhance the predictive accuracy of glycopeptide-based diagnostics and therapeutics. Furthermore, the development of robust, high-throughput platforms that can seamlessly integrate with clinical workflows will be essential to translate glycoproteomic discoveries into practical medical applications. While current glycopeptide enrichment techniques and analytical platforms lay a solid foundation for understanding glycoprotein functions in health and disease, significant opportunities exist to address the challenges highlighted. Future research should aim to develop more refined and scalable approaches that can accommodate the complexity of glycoproteomics. Such advancements will undoubtedly enrich our understanding of disease mechanisms and foster the development of personalized medicine, ultimately improving patient outcomes.

7. Expert Opinion

Glycoproteomics has proven instrumental in the discovery of biomarkers across various diseases. The glycans bound to proteins influence how these proteins fold, their stability, and function. Consequently, glycopeptides, with their critical roles in biological processes and disease states, hold substantial promise for utilization as biomarkers. Advances in glycopeptide enrichment techniques, as discussed in the review, have profound implications for clinical diagnostics and therapeutic interventions. Improved glycopeptide enrichment, particularly through techniques such as LAC, HILIC, and the utilization of novel materials like MOFs, COFs, and nanomaterials enhance the detection and characterization of disease biomarkers. These advancements could lead to more accurate and earlier diagnosis and better identification of drug targets impacting healthcare positively. However, the integration of these technologies into clinical practice faces several barriers. The main challenges include the complexity and cost of the equipment and procedures, the need for highly trained personnel to perform and interpret complex analyses, and the scalability of these techniques in routine clinical settings. Another major challenge limiting the application of glycopeptide biomarkers in clinical settings is the size of the sample cohorts involved in the studies. In many cases, the sample sizes are limited generating insufficient clinical data to verify the use of glycopeptides as potential biomarkers.

The current limitations in glycopeptide enrichment include issues with specificity, sensitivity, efficiency, and the ability to scale these techniques for widespread clinical use. The specificity and efficiency can be further enhanced by continuing to develop and integrate innovative materials that provide more selective and robust enrichment platforms. Additionally, integrating computational advancements such as artificial intelligence and machine learning could improve the processing and interpretation of complex data, boosting the predictive accuracy of biomarker discovery. Technical and technological enhancements are required to make these methods more accessible and cost-effective for routine use in clinical diagnostics. This includes the development of more durable materials that do not degrade in biological samples and the simplification of analytical equipment.

Further research in glycopeptide enrichment has significant potential to revolutionize personalized medicine. There is no definitive endpoint in sight for this research area, as the complexity of diseases and the continuous evolution of technology will always drive further innovation. Future research could focus on refining the sensitivity and selectivity of enrichment techniques, developing portable or miniaturized devices for clinical settings, and enhancing the integration of glycopeptide data with clinical outcomes to better guide treatment decisions. Most research on glycopeptide biomarkers has centered on cancer and neurodegenerative disorders. However, there is a significant opportunity to investigate these biomarkers in other diseases to facilitate early diagnosis. The application of innovative materials like MOFs, COFs, and nanomaterials to enrich glycopeptides in various disease contexts remains underexplored. Expanding the use of these effective materials could enhance the identification of glycopeptide biomarkers across a broader range of health conditions.

While the future of study in glycopeptide enrichment is promising due to its critical role in personalized medicine and biomarker discovery, other areas such as direct biomolecule sequencing and real-time monitoring technologies also hold significant potential. However, the unique ability of glycopeptide enrichment to provide detailed insights into protein modification and function makes it an indispensable area of study in the foreseeable future.

The field of glycoproteomics and glycopeptide enrichment is expected to evolve significantly, driven by advancements in materials science, bioinformatics, and nanotechnology. As new materials and computational tools become available, the precision and efficacy of glycopeptide analyses will improve, making these techniques more robust and easier to integrate into clinical workflows. This evolution will likely lead to a broader acceptance and adoption of glycopeptide-based diagnostics and therapeutics, shaping the future of disease diagnosis and personalized medicine.

In the next five years, the field of glycopeptide enrichment techniques for the identification of clinical biomarkers is poised for transformative advances. We can expect these changes to be driven by a convergence of technological innovations, increased understanding of disease biomarkers, and enhanced integration of computational tools in biomedicine. The application of aforementioned novel materials in glycopeptide enrichment is likely to advance significantly. These materials will see improvements in stability, biocompatibility, and functional customization. This will allow for more selective and efficient capture of glycopeptides, even from highly complex biological samples like blood or tissue fluids. Another anticipated innovation includes the development of dynamically responsive materials that can adjust their binding properties in real-time, driven by changes in external factors such as pH, temperature, or light. This could greatly enhance the specificity and yield of glycopeptide enrichment by allowing the tuning of material properties during the enrichment process. In addition, the integration of nanotechnology will play a crucial role in enhancing the sensitivity and efficiency of glycopeptide enrichment techniques. Nanostructured materials, due to their high surface area-to-volume ratios and customizable surface chemistry, will provide unprecedented control over the molecular interactions required for effective glycopeptide isolation. Over the next five years, glycopeptide enrichment techniques will experience great advances. The development of liquid chromatography (LC) columns and capillaries with novel materials will enhance online proteolysis and glycopeptide enrichment, leading to greater automation and a substantial increase in the throughput and scalability of glycoproteomic workflows.^{206, 207, 229} Furthermore, with a greater understanding of glycosylation and its role in neurodegeneration, cancer, metabolic disorders, and infectious diseases, glycopeptide enrichment techniques are expected to find applications in clinical settings.

The use of AI and ML will become increasingly central in glycoproteomics research. These computational tools will not only enhance the ability to predict and interpret complex patterns in glycoproteomics data but also streamline the design and optimization of enrichment processes. AI could be employed to simulate and predict the outcomes of experimental enrichment setups, thereby reducing the need for extensive physical experimentation. Machine learning models might be developed to learn from vast datasets of glycoproteomics, enabling the identification of subtle biomarker signatures that may be undetectable by human analysts. As these technologies mature, their integration into clinical diagnostics platforms is likely to increase. This will facilitate a shift towards more personalized medicine approaches, where glycopeptide biomarkers are used to tailor therapeutic strategies to individual patients. Recent advances in deep learning models are poised to revolutionize glycopeptide search algorithms, significantly enhancing both the number of glycopeptides identified and the accuracy of their identification. Models such as Graph Neural Networks (GNNs) are already being used for in-silico prediction of glycopeptide fragments.^{361, 372} The ongoing advancement of these models will facilitate the creation of in-silico predicted spectral libraries for Data-Independent Acquisition (DIA) glycoproteomics experiments, allowing for the identification and quantification of novel glycopeptides. Convolutional neural networks (CNNs) have recently been proposed for the detection of Alzheimer’s disease using MRI data demonstrating classification accuracies up to 99.43%.³⁷³ Similar use of CNN models for classification between healthy and disease cohorts using glycoproteomics data is expected to gain significant traction within the next five years. This development would be especially beneficial in clinical settings, allowing for faster diagnoses.

The near future will likely witness significant advancements in glycopeptide enrichment techniques, driven by breakthroughs in materials science, nanotechnology, and computational biology. These developments will enhance our ability to identify and utilize biomarkers for disease diagnosis and treatment, potentially revolutionizing the field of clinical diagnostics and personalized medicine.

Article Highlights.

• Complexity of glycosylation necessitates enrichment before mass spectrometry analysis

• Review of common strategies and novel materials for glycopeptide enrichment

• Artificial intelligence and machine learning will advance glycoproteomics analysis

• A robust platform will facilitate biomarker discovery in clinical glycoproteomics

Abbreviations:

AEX

Anion Exchange Chromatography

ConA

concanavalin A

CEX

Cation Exchange Chromatography

COF

Covalent Organic Framework

DOSG+

Derivatization of Sialylated Glycopeptides Plus

ERLIC

Electrostatic Repulsion Hydrophilic Interaction Chromatography

HILIC

Hydrophilic Interaction Liquid Chromatography

IEC

Ion Exchange Chromatography

IsoTAG

Isotope Target Glycoproteomics

LAC

Lectin Affinity Chromatography

LcH

lentil lectin

MOF

Metal Organic Framework

PNA

Peanut Agglutinin

RCA

Ricinus Communis Agglutinin

SAX

Strong Anion Exchange Chromatography

SNA

elderberry lectin

WCX

Weak Cation Exchange Chromatography

WGA

Wheat Germ Agglutinin

AIL

Jacalin lectin