Isolation and Characterization of Glycosylated Neuropeptides

Abstract

Glycosylation is one of the most ubiquitous and complex post-translational modifications (PTMs). It plays pivotal roles in various biological processes. Studies at the glycopeptide level are typically considered as a downstream work resulting from enzymatic digested glycoproteins. Less attention has been focused on glycosylated endogenous signaling peptides due to their low abundance, structural heterogeneity and the lack of enabling analytical tools. Here, protocols are presented to isolate and characterize glycosylated neuropeptides utilizing nanoflow liquid chromatography coupled with mass spectrometry (LC-MS). We first demonstrate how to extract neuropeptides from raw tissues and perform further separation/cleanup before MS analysis. Then we describe hybrid MS methods for glycosylated neuropeptides profiling and site-specific analysis. We also include recommendations for data analysis to identify glycosylated neuropeptides in crustaceans where a complete neuropeptide database is still lacking. Other strategies and future directions are discussed to provide readers with alternative approaches and further unravel biological complexity rendered by glycosylation.

1. Introduction

Proteins and peptides are usually released to the circulating system after cellular enzymatic processes from inactive precursors and multiple post-translational modifications (PTMs) (Hook et al., 2008; Hook, Lietz, Podvin, Cajka, & Fiehn, 2018; McClure, Walls, & Grinnell, 1992). PTMs are of great significance in determining or altering peptide structures and biological behaviors (Duan & Walther, 2015; Seo & Lee, 2004). Glycosylation is one of the most common yet the most challenging PTMs, due to the complexity of attached glycans and the site occupancy on the backbones. It is broadly related to various functional roles in a wide range of biological processes, including signal transduction, ion homeostasis, protein folding stability and kinetics, cell-cell adhesion, protein-protein interaction, and immune defense (Arey, 2012; Arnold, Wormald, Sim, Rudd, & Dwek, 2007; Helenius & Aebi, 2001, 2004; Jayaprakash & Surolia, 2017; Khidekel, Ficarro, Peters, & Hsieh-Wilson, 2004; Ohtsubo & Marth, 2006; Rudd, Elliott, Cresswell, Wilson, & Dwek, 2001). The distribution of glycosylation covers all three domains of life, eukarya, bacteria, and archaea, and especially in eukarya, more than half of proteins are modified by various attached glycans (Apweiler, Hermjakob, & Sharon, 1999; Dell, Galadari, Sastre, & Hitchen, 2010).

The glycans appear as short carbohydrate chains consisting of a single monosaccharide or polysaccharides, exhibiting broad diversities in structure, linkage and branching patterns. The complexity of glycan moieties results in a variety of glycosylated modifications. Among different types of glycosylation, there are two categories that are widely studied, namely N- and O-linked glycosylation. Specifically, an oligosaccharide attached to the side chain amide nitrogen of asparagine (Asn) is recognized as N-linked glycans with the Asn-X-Ser/Thr sequon, where X could be any amino acid except proline. O-linked glycosylation usually occurs on the hydroxyl group of serine and threonine when the carbohydrate is attached to the oxygen. Unlike the N-linked structures, O-linked glycans are found to be associated with no consensus sequence yet. Besides the different residues that glycans are attached to, N- and O-linked glycans also have distinct structures, due to their unique enzymatic pathway during biosynthesis (Lodish et al., 2000). N-glycans start with a conserved chitobiose core, (GlcNAc)₂Man₃. With the extension of other monosaccharides from the core unit, this group will be further divided into high-mannose, complex and hybrid types. However, O-linked carbohydrates do not share a common core structure, only featured by eight common formations (Alley, Mann, & Novotny, 2013). Challenges of glycosylation studies mostly stem from the sophisticated nature of glycans and their site occupancy throughout the primary amino acid sequences (Kolarich, Jensen, Altmann, & Packer, 2012; Stavenhagen et al., 2013). On a single glycosylated site in a peptide backbone, there is often a set of modifications with multiple glycan isoforms at different length and branch patterns. Throughout the full peptide sequence, there can be multiple glycosylation sites, which are completely, partially or scarcely occupied. Extensive research efforts have been devoted to the development of analytical tools applied in glycomics and glycoproteomics (Z. Chen, Huang, & Li, 2018; Ruhaak, Xu, Li, Goonatilleke, & Lebrilla, 2018; Xiao, Suttapitugsakul, Sun, & Wu, 2018; A. Yu et al., 2018). However, limited studies have focused on the endogenous glycosylated signaling peptides.

Neuropeptides and peptide hormones are main classes of signaling molecules in nervous and endocrine systems that regulate physiological processes and behaviors (Buchberger, Yu, & Li, 2015; Elphick, Mirabeau, & Larhammar, 2018; Heck et al., 2018; Matsubayashi, 2018). They are classified into different families based on the shared sequence motifs (Christie, Stemmler, & Dickinson, 2010; Enman, Sabban, McGonigle, & Van Bockstaele, 2015; Hökfelt et al., 2000). Although isoforms in the same family often share similar sequences, and even with identical masses, they may still have distinct physiological roles (Chung & Zmora, 2008; Nusbaum & Blitz, 2012). Therefore, the similarity in chemical and physical properties does not necessarily imply identical function. Neuropeptides are typically short amino acid chains, however, some peptide families could consist of up to 70 residues (Chung, Zmora, Katayama, & Tsutsui, 2010; Webster, Keller, & Dircksen, 2012). With various PTMs, the structural diversity presents significant challenges in global characterization. Furthermore, the concentration of neuropeptides and peptide hormones are oftentimes at nM-pM level (Adrian et al., 1983; Q. Li, Zubieta, & Kennedy, 2009), which would also cause great difficulties in analysis.

The glycosylated endogenous peptides undertake essential and non-negligible functional roles in many biological processes, although they exist at a further lower concentration level compared to the non-glycosylated counterparts (Halim et al., 2011). For example, glycosylation of opioid peptides (cyclic/linear enkephalin, Met-/Leu-enkephalin, endomorphin, deltorphin) improves the transportation of these pharmacologically active neuropeptides into the brain (Egleton et al., 2000, 2001; Masand et al., 2006; Varamini et al., 2012; Witt & Davis, 2006) by crossing the blood-brain barrier (BBB), a highly selective barrier that prohibits most compounds from penetrating into the brain, the central nervous system (CNS) (Serrano, Ribeiro, & Castanho, 2012; Varamini et al., 2012). Furthermore, the attachments of carbohydrates to peptides would increase the resistance to enzymatic cleavage and degradation (Semenov et al., 2009; T. Yamamoto et al., 2009), lengthen the half-life time in serum (Polt et al., 1994) and enhance the receptor binding (Egleton et al., 2000). The improved factors make the glycosylated analogues a more effective methodology as peptide drugs. For example, studies showed that the additional sugar moieties boost the analgesic activity (Masand et al., 2006; Polt et al., 1994), an antinociceptive effect similar to morphine (Bilsky et al., 2000), both in maximal analgesic level and the analgesia duration (Egleton et al., 2001). On another aspect, radiolabeled glycopeptides affect the biodistribution in the application to cancer cell imaging (Schweinsberg et al., 2008; Watanabe et al., 2012), including less liver and renal accumulation and improvement in tumor uptake, leading to a minimum effective dosage and an increased tumor-to-background ratio in imaging (Haubner et al., 2001; Schottelius, Wester, Reubi, Senekowitsch-Schmidtke, & Schwaiger, 2002). Besides functioning as potential therapeutic drugs, O-glycosylated peptide toxins, Vespulakinins, isolated from yellowjacket venom sacs were the first reported vasoactive glycopeptides and naturally occurring glycosylated derivatives of bradykinin, with an identical C-terminus sequence (Yoshida, Geller, & Pisano, 1976). In the cone snail, two other O-glycopeptides, κA-conotoxin SIVA and Contulakin-G, were found in the venom, as the first evidence of biologically active glycosylated peptides in this model. They were demonstrated to have higher activity compared to nonglycosylated counterparts when administered in vivo (A. G. Craig et al., 1998; A. Grey Craig, Bandyopadhyay, & Olivera, 1999). Notably, Contulakin-G was the first known neurotensin family peptide from a nonvertebrate model. The C-terminus of its glycosylated isoform, due to the high homology to mammalian neuropeptides, was detected to target the mammalian neurotransmitter receptors (A. Grey Craig, Norberg, et al., 1999). Recently, more studies have discovered that the glycosylated endogenous peptides are associated with many diseases, such as diabetes, heart failure and neurodegenerative diseases (Brinkmalm et al., 2012; Halfinger et al., 2017; Halim et al., 2011; Semenov et al., 2009; Q. Yu, Canales, et al., 2017).

To deal with the high complexity and low abundance of the endogenous glycosylated peptides, mass spectrometry (MS)-based approaches have emerged as powerful tools, benefiting from the high sensitivity and high resolution. With the instrumentation advances in ionization sources, mass analyzers and especially, the fragmentation modes, significant advancements in glycopeptide analysis enable revealing peptide backbone structures, site-specific information and glycan types with less instrument time. Meanwhile, the extraction and enrichment of these endogenous peptides from biological samples are also the crucial steps in the entire workflow. Extraction methods may vary depending on the specific properties of the samples and the targeted peptides. Prior to rapid and highly accurate MS detection, glycopeptide enrichment and multidimensional chromatographic separation would facilitate the qualitative and quantitative analysis of glycosylated peptides at a lower detection limit with a larger dynamic range. In this chapter, we summarize the current state of endogenous glycopeptide studies and highlight recent advances in various techniques benefiting glycosylated peptide analysis (Fig. 1). Specifically, we present detailed workflows ranging from tissue collection to MS analysis that enable discovery and identification of glycosylated neuropeptides in crustacean model organisms. Protocols of other strategies are also discussed, including peptide extraction, glycopeptide enrichment and mass spectrometric analysis to provide the readers sufficient details so that effective experiments can be designed and performed according to different sample types, target analytes and available instrument platforms.

Figure 1.

A typical workflow to identify glycopeptides via multifaceted MS approaches in complex biological samples.

2. Neuropeptide Extraction from Crustacean Neuronal Tissues

Besides endogenous peptides, biological samples usually also contain lipids, salts and large proteins (Aebersold & Mann, 2016; Arnaud et al., 2005; Cajka & Fiehn, 2016; Paglia, Kliman, Claude, Geromanos, & Astarita, 2015). They will interfere the ionization efficiency, contaminate the LC platform and suppress the signal intensity of target peptides. Reduction of sample complexity is the primary step in peptidome analysis. Commonly used peptide extraction methods will be introduced here, while the purification approaches specific to glycopeptides will be described. The crustacean nervous system contains rich contents of neuropeptides. Acid extraction is a classical method to easily purify neuropeptides. Here, we first describe how to collect crustacean neuronal tissues and then a detailed workflow to extract neuropeptides. The whole experiment can take 1 day to 2 days, depending on the tissue amount involved.

2.1. Neuronal Tissue Collection

2.1.1. Materials and Buffer Recipes

1. Physiological saline buffer stored at 4 °C: 440 mM NaCl, 11 mM KCl, 26 mM MgCl₂, 13 mM CaCl₂, 11 mM Trizma base, 5 mM maleic acid

2. Ice-cold acidified methanol (methanol/water/glacial acetic acid = 90/9/1, v/v/v)

3. Dry ice in a sealed box

4. Dissection tools, including petri dish, spatula, scissors, tweezers, forceps and needles

5. Microscope

2.1.2. Tissue Harvest via Animal Sacrifice

1. Anesthetize a crab by burying it in ice for 15 min.

2. Use forceps to open crab shell, isolate large tissues where neuronal tissues are located and transfer them to a petri dish, add physiological saline buffer and use needles to pin down specific tissues for micro-dissection.

3. Use scissors and tweezers to collect desired neuronal tissues under a microscope.

4. Transfer isolated neuronal tissues into centrifuge tubes filled with 100 μL ice-cold acidified methanol and immediately put tubes in dry ice. Tissues can be used for immediate handling or stored in −80 °C freezer until further processing. Repeat dissection and tissue collection until enough samples are collected.

2.2. Acid Extraction

2.2.1. Materials and Buffer Recipes

1. Ice-cold acidified methanol (methanol/water/glacial acetic acid = 90/9/1, v/v/v)

2. Probe dismembrator

3. 30 kDa molecular weight cutoff filter

4. ACN/MeOH/H₂O = 20/30/50, v/v/v

5. Ice

6. Speed-vacuum sample dry system

2.2.2. Neuropeptide Extraction

1. Prerinse 30 kDa molecular weight cutoff filter with 80 μL ACN/MeOH/H₂O (20/30/50, v/v/v) and spin at 4 °C for 15 min at 14,000g. Repeat once.

2. Put centrifuge tubes containing neuronal tissues onto ice-water mixture. Use a probe dismemberator in a 4 °C cold room to break large tissues in acidified methanol for efficient extraction.

3. Centrifuge the tissue content at 4 °C for 5 min at 16,100g.

4. Transfer supernatant onto the prerinsed 30 kDa molecular weight cutoff filter and centrifuge at 4 °C for 20 min at 14,000g.

5. Collect the supernatant that has spun through the filter. Repeat step 4 until no more flow-through is observed.

6. Transfer 50 μL acidified methanol onto the filter and further centrifuge at 4 °C for 20 min at 14,000g. Combine all the supernatants.

7. Evaporate the solvent under vacuum and store the dried sample at −80 °C for future analysis.

2.3. Neuropeptide Reduction and Alkylation

2.3.1. Materials and Buffer Recipes

1. 50 mM ammonium bicarbonate (ABC)

2. 100 mM DTT

3. 200 mM IAA

4. Speed-vacuum sample dry system

2.3.2. DTT and IAA Reactions

1. Resuspend neuropeptide extract in a centrifuge tube with 100 μL 50 mM ABC. Vortex to allow sufficient mixing

2. Add 5.26 μL DTT into the centrifuge tube and incubate at room temperature for 1 hr.

3. Add 8.53 μL IAA into the centrifuge tube and incubate in dark at room temperature for 30 min.

4. Add 5.26 μL DTT into the centrifuge tube and incubate at room temperature for 5 min.

5. Dry down sample solvent under vacuum and store the dried sample at −80 °C for future analysis.

Our method incorporates both organic solvent precipitation and acid extraction to remove large proteins from neuropeptides in tissue samples. Organic solvent precipitation is one of the most common methods applied to isolate peptides from insoluble proteins (Patel et al., 2018; Simpson & Beynon, 2010; Williams et al., 2010). The addition of organic solvents disrupt the ordered water molecules surrounding the proteins and the hydrate layer around are decreased. Polar interactions between aqueous solvent and the protein are minimized, leading the protein to aggregate and precipitate due to the electrostatic interaction. The reduced solubility can effectively clean out the abundant large proteins, making peptides and small proteins in the solution accessible to analysis in the following steps. Popular choices for precipitants are various alcohols and acetone, among which, acetone is considered superior because of the minimal diluted volume and more dense precipitation (Chertov et al., 2004; Polson, Sarkar, Incledon, Raguvaran, & Grant, 2003).

Adding acid is also a widely adopted approach in precipitating proteins and extracting peptides (Kirkpatrick et al., 2017; S.-I. Park, Kim, & Yoe, 2015; Robert et al., 2015). Proteins, positively charged at pHs below their pI, interact with the acidic reagents and form insoluble salts. Besides inducing precipitant, acids also play another essential role in the preparation of biological samples containing active proteases. The acidic environment causes protein denature, thus curtailing the protease activity (Fink, Calciano, Goto, Kurotsu, & Palleros, 1994) to preserve fast-degrading peptides.

In peptide extraction, the interference of large proteins is not only their high abundance, but also the binding effects with some small peptides. The peptide-protein interaction is fundamental in cell signaling and its prevalence (London, Movshovitz-Attias, & Schueler-Furman, 2010; Schulze & Mann, 2004; M. Yang, Wu, & Fields, 1995) would lead to huge sample loss if peptides are not unbound from the proteins before being precipitated. Organic solvents, together with acidic reagents will dissociate the peptides from binding sites of large proteins, improving the extraction efficiency (Chertov et al., 2004). Another effective method to extract low molecular weight peptides would be differential solubilization (Kawashima et al., 2010). Sample was first mixed with denaturing solutions and then added ice-cold acetone, resulting in precipitation of all proteins and peptides. The precipitate was further washed with 70% ACN containing 12mM HCl, in which peptides were easily dissolved and separated from other proteins. It has been shown that compared to other extraction methods, differential solubilization gave higher yield and better reproducibility, being independent of the composition of the proteins/peptides from the samples.

Our experimental protocol also utilizes ultrafiltration enabled by molecular weight cutoff filter to further separate neuropeptides with soluble proteins. Ultrafiltration is also a frequently used approach in peptide extraction and purification, which retains the high molecular weight solutes while allowing the low molecular weight to pass through the semi-permeable membrane (Greening & Simpson, 2010; Luque-Garcia & Neubert, 2007). Various molecular weight cut-off of the membranes define the boundary between filtrate and retentate, providing more choices to different sample components and separation purposes (Chang, Ismail, Yanagita, Mohd Esa, & Baharuldin, 2015; Kirkpatrick et al., 2017; Zheng, Baker, & Hancock, 2006). The successful separation of centrifugation ultrafiltration relies on the low centrifugation speed and the low sample concentration (Luque-Garcia & Neubert, 2007). Although the centrifugal ultrafiltration method is facile, rapid and cost-effective in peptidome analysis (Finoulst, Pinkse, Van Dongen, & Verhaert, 2011; Greening & Simpson, 2010), it still lacks sharp separation indicated by the molecular weight cut-off membranes (Georgiou, Rice, & Baker, 2001; Picot et al., 2010). The large molecules stuck on the membrane can easily block the pores, affecting the separation efficiency. Electrophoresis is widely known as a high-resolution purification approach. A hybrid method of electrodialysis with insertion of ultrafiltration membrane was developed to enhance the peptide extraction and purification performance (Dlask & Václavíková, 2018; Picot et al., 2010). In electrodialysis ultrafiltration, the electric potential differences act as driving force and the variation in pH can selectively separate acidic, basic and neutral peptides, reducing the accumulation of molecules on membrane surface and clogs of the porous membranes (Dlask & Václavíková, 2018; Poulin, Amiot, & Bazinet, 2006).

To manually operate the protein precipitation would be quite time-consuming when being applied to large sample sets. The development of membrane-based protein precipitation filter plates helps to avoid the tedious workflow. By using a 96-well filter block, the separation of peptides and large proteins become more effective (Biddlecombe & Pleasance, 1999; Kitchen, Wang, Musson, Yang, & Fisher, 2003), with higher throughput and more simplified procedures, due to the elimination of centrifugation and supernatant transfer steps (Watt, Morrison, Locker, & Evans, 2000). In this method, filtrate containing low-molecular-weight proteins and peptides can be easily collected, suffering from less cross contamination, and being compatible to downstream analysis using MS at high accuracy and reproducibility (Walter, Cramer, & Tse, 2001).

2.4. Alternative Strategy - Restricted Access Materials

Restricted access materials (RAMs) are solid sorbents that exclude the access of macromolecules to the surface binding while the low molecular weight particles are selectively retained (Kole, Venkatesh, Kotecha, & Sheshala, 2011). The first silica-based RAM model was proposed in 1986, named as internal surface reverse-phase (ISRP) (Cook & Pinkerton, 1986). In 1991, the modifications on the surface of silica particles were successfully carried out and the concept of RAM was first introduced (Desilets, Rounds, & Regnier, 1991). From then on, various types of RAM have been developed, which can be classified based on the mechanisms how the large proteins are excluded from extraction (de Faria, Abrão, Santos, Barbosa, & Figueiredo, 2017; Souverain, Rudaz, & Veuthey, 2004). These special extraction methods allow automating sample preparation and on-line extraction followed by MS analysis using a RAM column. It enables simplified operation, high throughput and reproducible identification of peptides. Around 400 peptides were identified in 2μl human serum using one-dimensional separation and with the help of a SCX/SEC/RP-MS platform (Fig. 2); identifications of 1286 peptides were achieved by the on-line processing of 20μl of human serum (L. Hu, Boos, Ye, Wu, & Zou, 2009). The same group later exploited a more simplified system, constructed with one switching valve, two gradient pumps and one UV detector, to realize the on-line extraction, desalting and fractionation of peptides in human serum (L. Hu, Boos, Ye, & Zou, 2014).

Figure 2.

Schematic representation of the set up for the on-line extraction system and chromatographic principles of the bifunctional SCX/SEC-RAM for selective extraction. Solid line represents position for loading sample and dashed line represents position for separation. Adapted from (L. Hu et al., 2009) with permission.

2.5. Alternative Strategy - Magnetic beads

Magnetic bead-based technique is used in peptidome analysis due to its operation simplicity and high selectivity in peptide extraction (Safarik & Safarikova, 2004; Villanueva et al., 2004), especially when conjugated with antibodies (Whiteaker et al., 2007). Magnetic particles usually have affinity ligands or ion-exchange groups which interacts with isolated analytes. They can be directly mixed with crude samples followed by a period of incubation time for targeted compounds to bind with magnetic particles. The complex can be isolated from the sample using magnetic separators and finally the targeted molecules are eluted for further analysis. This effective peptide extraction method was also applied in tissue imaging using mass spectrometry, indicating that more peptides were detected with no compromise in localization information (T. Andrews, B. Skube, & B. Hummon, 2018).

Among many different peptide extraction approaches, each has its pros and cons. In general, non-specific methods, like organic precipitation and ultrafiltration, recovering peptides with no bias, are suitable for comprehensive studies. Methods with selectivity over certain structure or distinct characteristics of target analyte results in more efficient peptide extraction. However, there is no overall gold standard for method selection. Decisions on a single extraction method or a combination of multiple methods mainly depend on the aim of the analysis.

3. Sample Preparation — Glycopeptide Purification and Enrichment

Endogenous glycosylated neuropeptides are often present at low abundance compared to their non-glycosylated counterparts and exhibit a wide dynamic range in biological samples. In our protocol, an enrichment-free method is utilized, which is slightly different from other approaches where glycopeptides are often isolated and enriched via molecular interactions with other materials (Fanayan, Hincapie, & Hancock, 2012; Palmisano, Lendal, & Larsen, 2011; Xu et al., 2009; L. Yu, Li, Guo, Zhang, & Liang, 2009). Those methods can be generally catalogued based on how the glycosylated peptides are recognized and bound, such as interactions based on physicochemical properties of peptides or the specific recognition of the glycans attached to peptide backbones (C.-C. Chen et al., 2014). We first describe high pH fractionation and C18 cleanup workflow for upcoming glycosylated neuropeptide analysis with hybrid MS method and then discuss various efficient enrichment strategies for readers’ reference to select appropriate methods and address their specific experimental needs accordingly.

3.1. Purification with High pH Offline Fractionation

Glycopeptide samples from extraction and enrichment steps usually contain various salts, which requires further cleaning procedures. Common desalting steps performed on pipette tips, cartridges or HPLC column, depending on the loading amount and the consideration of sample loss. When handling relatively large tissue amount (>10 pieces) neuropeptide extracts, it is beneficial to first use offline LC to further separate neuropeptides and reduce sample complexity before MS analysis. As this separation step also helps desalt, the collected fractions are ready for MS characterization.

3.1.1. Materials and Mobile Phase Recipes

1. Phenomenex Kinetex 5u EVO C18 column (150 mm × 2.1 mm, 100 A, part number 00F-4633-AN)

2. Mobile phase A: 10 mM ammonium formate in H₂O, pH = 10

3. Mobile phase B: 10 mM ammonium formate in ACN/H₂O (90/10, v/v), pH =10

4. Fraction collector

5. Speed-vacuum sample dry system

3.1.2. Neuropeptide Separation and Purification

1. Attach a C18 column to HPLC system. Set up column temperature (30 °C for the column we used). Once temperature is stable, set up the flow rate (0.2 mL/min for the column we used). Allow sufficient equilibration of the column (10 times of column volume with mobile phase A).

2. Set up the sequence. Run a blank sample before real sample analysis to ensure instrument and column are functioning properly.

3. Resuspend the neuropeptide sample from Section 2.3.2 in 100 μL mobile phase A and inject onto the LC system.

4. Run the following gradient: 0–3 min, 1% mobile phase B; 3–50 min, 1–35% mobile phase B; 54–58 min, 60–70% mobile phase B; 59–74 min, 100% mobile phase B; 74.5–90 min, 1% mobile phase B. A typical elution profile is shown in Fig. 3.

5. Set up the fraction collector for sample collection in 2 min intervals from 6 to 62 min.

6. Combine the collected fractions into 4~5 centrifuge tubes and dry down under vacuum for solvent removal. Store the dried samples in a −80 °C freezer until MS analysis.

7. For LC-ESI-MS analysis, resuspend sample into 10 μL 0.1% FA H₂O.

Figure 3.

Chromatogram of crustacean neuropeptides separated by offline high pH fractionation.

3.2. Purification with C18 Ziptip Desalting

While high pH fractionation reduces sample complexity, it also causes sample loss during separation. For smaller amount (<10 pieces) of crustacean tissue neuropeptide extract, we prefer using C18 Ziptip to remove salt before MS analysis with no need of further separation, as sample complexity is not the major concern.

3.2.1. Materials and Buffer Recipes

1. 0.1% trifluoroacetic acid (TFA)

2. Acetonitrile (ACN)

3. Elution buffer 1: ACN/H₂O, 50/50, v/v

4. Elution buffer 2: ACN/H₂O, 75/25, v/v

5. Agilent C18 Ziptip, 100 μL volume

6. Speed-vacuum sample dry system

3.2.2. Neuropeptide Desalting

1. Aspirate the Ziptip with 100 μL ACN and dispense to waste. Repeat 2 times.

2. Aspirate the Ziptip with 100 μL 0.1% TFA and dispense to waste. Repeat 3 times.

3. Resuspend neuropeptide sample from Section 2.3.2 in 100 μL 0.1% TFA. Load the sample onto C18 Ziptip binding material by aspirating the resuspended extract for 10 times.

4. Load 100 μL 0.1% TFA into Ziptip, aspirate and dispense to waste. Repeat 3 times.

5. Elute neuropeptides with 100 μL ACN/H₂O (50/50, v/v) by aspirating for 10 times

6. Elute neuropeptides with 100 μL ACN/H₂O (75/25, v/v) by aspirating for 10 times

7. Combine the elution and dry down sample under vacuum.

8. For LC-ESI-MS analysis, resuspend sample into 10 μL 0.1% FA H₂O.

3.3. Enrichment with Hydrophilic Interaction Chromatography (HILIC)

The term of hydrophilic interaction chromatography (HILIC), as a variation of normal phase-HPLC, was first proposed and described in 1990 (Alpert, 1990). In HILIC separation, the higher polarity of stationary phase strongly retains hydrophilic analytes and the less polar mobile phase, which usually consists of 50–70% organic solvent, elutes the retained molecules in an order from least hydrophilic to the most hydrophilic (Buszewski & Noga, 2012). Due to the sugar side chains attached, glycopeptides usually exhibit higher hydrophilicity than their counterparts without glycosylation, therefore, they can be fractionated and collected at different retention times (P. G. Wang, He, & He, 2011). The hydrophilicity difference between glycosylated and non-glycosylated peptides can be exaggerated and the overlap in chromatography will diminish with the addition of ion pairing reagents (Ding, Hill, & Kelly, 2007; Furuki & Toyo’oka, 2017; Palmisano et al., 2010). Different types of HILIC methods have been reported using a wide range of solid phase materials, including silica particles (Wan et al., 2011), cellulose (Ohta, Kameda, Matsumoto, & Kawasaki, 2017; Snovida, Bodnar, Viner, Saba, & Perreault, 2010), polysaccharide-based (Xiong et al., 2013), functional group based, such as amide (Palmisano et al., 2012) and amine (Kuo, Wu, Hsiao, & Khoo, 2012), even just simple cotton (Selman, Hemayatkar, Deelder, & Wuhrer, 2011). A recently emerged and largely adopted method is ZIC-HILIC, functioned by sulfobetaine (Dedvisitsakul et al., 2014; Pohlentz, Marx, & Mormann, 2016; Zhao et al., 2017). It has been demonstrated to selectively separate sialylated N-glycan glycopeptides (Takegawa et al., 2006). The composition of mobile phase can largely alter the separation mechanism and selectivity as well (Alagesan, Khilji, & Kolarich, 2017). Electrostatic repulsion hydrophilic interaction chromatography (ERLIC), a combination of HILIC and ion-exchange chromatography, was proposed in 2008 (Alpert, 2008). The high portion organic solvents and the ion-exchange stationary phase enable the retention of both the same and opposite charged analytes as the ion-exchange resins. Glycopeptides with noncharged glycans are mainly subject to hydrophilic interaction, while for those charged glycans with sialic acid, the retention also depends on the charge-based repulsion forces. The sequential HILIC-ERLIC enrichment provided complementary identification of glycopeptides to HILIC-only method (Zacharias et al., 2016). In addition to the enrichment of glycopeptides from their non-glycopeptide counterparts, the HILIC approach is also capable to differentiate the isomeric glycopeptides, with the same backbone but various sugar moieties (Y. Huang, Nie, Boyes, & Orlando, 2016; Takegawa et al., 2006). Improvements in HILIC technique include the combination with other enrichment methods, such as lectin affinity capture (C. Ma et al., 2013) and titanium dioxide (Palmisano et al., 2010) to increase selectivity and manufacture of microcolumns (Boersema, Mohammed, & Heck, 2008; Dedvisitsakul et al., 2014) and microtips (J. Jiang et al., 2014; Selman et al., 2011; Vreeker et al., 2018) to decrease sample loss, especially for low abundance endogenous peptides (Kawahara, Saad, Angeli, & Palmisano, 2016). Most recently, HILIC probes incorporated with magnetic mesoporous materials were successfully applied to capturing endogenous glycopeptides. A zwitterionic hydrophilic mesoporous silica material (FSAu@mSiO₂@_L-Cys) was synthesized and the pore size was designed to enrich endogenous glycopeptides with large-size proteins excluded (Z. Wang, Wu, Chen, Sun, & Deng, 2018). 40 endogenous glycopeptides were identified from 10 μl human saliva, which was for the first time used as biological sample for endogenous glycopeptidome analysis. Shortly after, the same group developed another hydrophilic probe in mesoporous pore (FE₃O₄@mSiO₂@G6P) with an even smaller pore size. From 10 μl healthy and gastric carcinoma human saliva, a total of 39 and 25 endogenous glycopeptides were detected (Y. Li, Deng, & Sun, 2018). The total number of endogenous glycopeptides found in both studies were comparable (40 and 39 from the same volume of healthy saliva sample), however, there was only one peptide appeared to be overlapped. This could probably be explained by the dynamic change of peptide expression or the different affinity property of each mesoporous material. However, it also reveals that poor reproducibility could present significant barrier in glycosylated neuropeptides studies.

3.3.1. Materials and Buffer Recipes

1. ZIC-HILIC cartridge.

2. 80% ACN (v/v) 1% TFA (v/v).

3. 0.1% TFA.

4. Speed-vacuum sample dry system.

3.3.2. Peptide Extraction-ZIC-HILIC Cartridge

1. Add 1ml 1% TFA in water to balance the cartridge. Repeat for 3 times.

2. Add 1ml 80% ACN 1%TFA (v/v) to equilibrate the cartridge and repeat for 3 times.

3. Dissolve sample in 80% ACN 1% TFA (v/v).

4. Load sample to cartridge and wash with 1ml 80% ACN (v/v) 1% TFA (v/v) for 3 times.

5. Elute glycopeptide into 0.1% TFA.

6. Evaporate the solvent using speedvac and store the dried sample at −80 °C.

3.4. Enrichment with Lectin Affinity Chromatography (LAC)

Lectin affinity chromatography (LAC), is another extensively used method in glycopeptide enrichment (F. Zhu, Clemmer, & Trinidad, 2016). Different from HILIC isolating glycopeptide based on their hydrophobicity, lectins are proteins which specifically bind with carbohydrate groups. They recognize and bind to glycans through a combination of hydrogen binding to sugar hydroxyl groups and the van der Waals interaction between sugars and amino acid side chains (Weis & Drickamer, 1996). Therefore, the binding selectivity of lectins to glycopeptides is highly diverse and depends on the structures of both lectin and the glycans. There are at least 160 known species of lectins and more than 60 of them are commercially available. Concanavalin A (ConA), the most widely used lectin, has a high affinity in binding the high mannose core in N-glycans (Y. Liu et al., 2016; Ohyama, Kasai, Nomoto, & Inoue, 1985). Wheat germ agglutinin (WGA) tend to specifically binds to sialic acid and N-acetyl-glucosamine (Bhavanandan & Katlic, 1979; Nagata & Burger, 1974; K. Yamamoto, Tsuji, Matsumoto, & Osawa, 1981). Aleuria aurantia lectin (AAL) is used to recognize α1,6-fucosyl residues (Matsumura et al., 2007; Yamashita, Kochibe, Ohkura, Ueda, & Kobata, 1985). However, some lectins have relative narrow specificity and weak affinity to unique glycans. _L-phytohemagglytinin (_L-PHA) usually targets the structure of β−1,6-GlcNAc moiety of N-glycans (Abbott et al., 2008; Ahn et al., 2009). Jacalin (JAC) is a seed lectin from jackfruits, which specifically binds with Galβ1–3GalNAc and O-linked glycans (Bourne et al., 2002; Hortin, 1990; Tachibana et al., 2006). These lectins with less robust binding with targeted sugar moieties may result in low recovery and sample loss. A serial lectin affinity chromatography (SLAC) was devised to avoid the coverage limit given by a single lectin (Cummings & Kornfeld, 1982). By applying several lectin enrichment strategies in succession, different glycosylation patterns can be better covered, at the same time, effectively separated (Harada et al., 1987; Lehoux & Ju, 2017). SLAC offers more flexibility in the design of enrichment workflow with some specific goals. For example, in order to identify peptides with O-glycosylation in human serum, N-linked high mannose glycopeptides were removed in the first step by ConA. The “cleaned” sample was then applied to the Jacalin column for exclusive binding to O-glycosylated peptides, which yield an improved selection of broad identification of O-linked glycopeptides compared to that result from Jacalin enrichment alone (Durham & Regnier, 2006). In 2004, a more powerful enrichment method was designed as multi-lectin affinity column (MLAC), which contains several different lectins in the same column (Z. Yang & Hancock, 2004). It has been demonstrated that MLAC can produce reproducible results, higher dynamic ranges and enable more comprehensive capture of glycopeptides than a single-lectin column (Totten et al., 2018; Totten, Kullolli, & Pitteri, 2017; Z. Yang, Harris, Palmer-Toy, & Hancock, 2006). This method was optimized to a high-performance multi-lectin affinity chromatography (HP-MLAC) by incorporating commercial ConA, WGA and JAC to a HPLC column (Kullolli, Hancock, & Hincapie, 2008). The automated HPLC platform integrated with inline MLAC enrichment was shown to generate high recovery, high throughput for clinical studies and disease related biomarker analysis (Kullolli, Hancock, & Hincapie, 2010; Z. Zeng et al., 2011).

3.4.1. Materials and Buffer Recipes

1. ConA cartridge.

2. Äkta Purifier (Amersham Biosciences, Piscattaway, NJ) equipped with a UV detector set at 280 nm.

3. Loading buffer: 10mM TRIS buffer, 500 mM NaCl, 1mM CaCl₂, 1mM MnCl₂, 0.02% sodium azide adjust pH to 7.4 using HCl.

4. Elution Buffer: 10mM TRIS buffer, 500 mM NaCl, 1mM CaCl₂, 1mM MnCl₂, 0.02% sodium azide, 100 mM Methylmannose, adjust pH to 7.4 using HCl.

5. Speed-vacuum sample dry system.

3.4.2. Peptide Extraction-ConA Microcolumn

1. Equilibrate the cartridge with loading buffer at 100μL/min with 10 column volumes.

2. Reduce flow rate to 50 μL/min.

3. Dissolve 50 μg sample in loading buffer (injection volume depends on the volume held by the sample loop).

4. Inject sample into the sample loop and load onto the microcolumn.

5. Wash the microcolumn with 5–10 column volumes or until the A280 returns to baseline value.

6. Elute peptides using elution buffer at step-gradient 0–100% over 10 column volumes.

7. Evaporate the solvent using Speedvac and store the dried sample at −80 °C.

3.5. Enrichment with Metal Ion Affinity

3.5.1. Titanium

Enrichment and separation based on the metal ion affinity are extensively adopted in glycopeptide studies, benefiting from the critical interaction between the spare pair electrons of oxygen atoms in carboxylate or hydroxyl groups and the unoccupied orbitals. The most representative example is titanium oxide (TiO₂), which is also a tool for phosphopeptide enrichment (Thingholm, Jørgensen, Jensen, & Larsen, 2006; Thingholm & Larsen, 2016; B. Zhang et al., 2011). It is an ideal choice when simultaneous analysis of phosphopeptides and glycopeptides are desired (Melo-Braga, Ibáñez-Vea, Larsen, & Kulej, 2015; Palmisano et al., 2012). For studies focusing on glycosylation, the primary interference would be the competitive binding with phosphopeptides. Therefore, the phosphatase pretreatment is necessary prior to the glycopeptide purification (Larsen, Jensen, Jakobsen, & Heegaard, 2007; Palmisano et al., 2010). TiO₂ is usually considered to bind sialic acid-containing glycopeptides preferentially (Larsen et al., 2007; Palmisano et al., 2010); however, it also interacts with peptides containing neutral glycans (Yan et al., 2010) based on the understanding of different enrichment mechanisms of TiO₂ towards glycopeptides with different solvent compositions (Sheng et al., 2013).

3.5.1.1. Materials and Buffer Recipes

1. Alkaline phosphatase.

2. Denaturing buffer: 6M urea, 2M thiourea and 50 mM NH₄HCO₃.

3. TiO2 beads.

4. TiO2 loading buffer: 1 M glycolic acid in ACN/H₂O, 80/20, v/v; 5% TFA, v/v.

5. TiO2 washing buffer 1: ACN/H₂O, 80/20, v/v; 1% TFA, v/v.

6. TiO₂ washing buffer 2: ACN/H₂O, 20/80, v/v; 0.1% TFA, v/v.

7. TiO₂ elution buffer: NH4OH/H₂O, 25/75, v/v, pH 11.3.

8. ACN/H₂O, 50/50, v/v.

9. Vortex mixture.

10. Centrifuge.

11. Speed-vacuum sample dry system.

3.5.1.2. Peptide extraction-Titanium Dioxide

1. Dissolve extracted peptides in 100 μL denaturing buffer.

2. Treat the peptides with 20 U alkaline phosphatase at 30 °C for 2 h to avoid the co-purification of phosphorylated peptides in the following steps.

3. Prepare TiO₂ beads (~10mg) in 100 μL ACN.

4. Mix 0.6 mg of TiO₂ beads with per 100 μg peptides in TiO₂ loading buffer. Then incubate 30 min on a mixer.

5. Centrifuge the mixture for 1 min × 1000g at room temperature and remove supernatant.

6. Add 500 μL TiO₂ loading buffer to the beads and incubate on the mixer for 1 min.

7. Centrifuge the mixture for 1 min × 1000g at room temperature and remove supernatant.

8. Add 500 μL TiO₂ washing buffer 1 to the beads, vortex 15 s.

9. Centrifuge for 1 min × 1000g and remove supernatant.

10. Add 500 μL TiO₂ washing buffer 2 to the beads.

11. Dry the mixture for 5–10 min in SpeedVac.

12. Incubate the beads with 200 μL TiO₂ elution buffer for 20 min on the mixer to elute the sialoglycopeptides.

13. Remove the beads and wash with 10 μL 50/50 ACN/H₂O and repeat twice.

14. Combine supernatant from steps 12 and 13for future analysis.

3.5.2. Zirconium

Zirconium dioxide (ZrO₂) is also commonly used in phosphopeptide enrichment (Dong, Zhou, Wu, Ye, & Zou, 2007; Zhou et al., 2007). Zr belongs to the same group in periodical table as titanium, therefore it has similar electron configuration; thus, similar chemical properties. It is inevitable to explore the advantage of zirconia in glycopeptide enrichment. A method of using zirconia layer coated mesoporous silica microspheres (Fig. 4) was proposed to specifically enrich glycopeptides (Wan et al., 2011). It has been tested on trypsin digested IgG, RNase B and α-casein. Under different solvent compositions, nonglycopeptides and glycopeptides were eluted in different fractions to minimize signal suppression, meanwhile, the phosphopeptides are strongly bound by the Lewis acid-basic action without being eluted to the same fraction as glycopeptides. This method resulted in better site-specific identification and higher selectivity of glycopeptides out of nonglycopeptides, compared to commercial sepharose and silica microspheres. In a more recent study, Zr-magnetic metal-organic framework (MOF) was modified onto the polydopamine (PDA)-coated magnetic microsphere (Fe₃O₄@PDA@UiO-66-NH₂) to build up a “one for two” hydrophilic magnetic amino-functionalized metal-organic framework (Xie & Deng, 2017). Owing to the large surface area of MOF, the strong magnetic responsiveness given by the Fe₃O₄ core, the hydrophilic interaction of Zr towards glycopeptides and its binding with phosphopeptides, this route exhibited excellent performance in identification of glyco- and phosphopeptides with high sensitivity and selectivity.

Figure 4.

Illustration of glycopeptide enrichment ability with ZrO2/MCF mesoporous microspheres, Adapted from (Wan et al., 2011) with permission.

3.5.3. Silver

Silver nanoparticles (Ag NPs) have been reported to be applied to glycopeptide separation and enrichment (B. Jiang, Wu, Zhang, & Zhang, 2017; W.-F. Ma et al., 2012; Y. Ma, Wang, Zhou, & Zhang, 2019) (Fig. 5). They proposed that the strong affinity between glycopeptides and Ag NPs was the multivalent interactions of hydroxyl groups and the silver surface. The affinity of glycopeptides to Ag NPs is comparable to that of water molecules. Therefore, the binding and elution in glycopeptide enrichment process were realized by adjusting the water content in the buffer. Ag NPs were shown to enable trace level glycopeptide enrichment at high recovery rate.

Figure 5.

The procedure for the post-synthetic modification of TpPa-1 and the enrichment and detection of the N-linked glycopeptides based on TpPa-1@Ag@GSH. Adapted from (Y. Ma et al., 2019) with permission.

3.6. Covalent Bond Interaction

3.6.1. Boronic acid chemistry

Boronic acid can enrich and isolate a wide range of N- and O-linked glycopeptides with no bias because it reacts with cis-diol-containing saccharides, like mannose, galactose and glucose and forming stable heterocyclic diesters (Sparbier, Koch, Kessler, Wenzel, & Kostrzewa, 2005). The covalent bonding between boronic acid and cis-diol group usually occur under basic aqueous or non-aqueous condition. The covalent linkage is reversible and the cyclic esters dissociate in acidic environment with no alteration to the glycan structures (X. Wang, Xia, & Liu, 2013). In addition to the advantages of high tolerance to glycan structures (branches or linear), saccharide types, as well as monosaccharide modifications, boronic acid can be easily incorporated onto various solid supports. Diboronic acid was grafted onto the surface of ordered mesoporous silica matrix (Xu et al., 2009). The universal covalent bonding of boronic acid and the high surface area of mesoporous silica enabled 2-orders of magnitude improvement in glycopeptide detection limit. Boronic acid-conjugated nanoparticles were designed and have been reported to enrich trace level glycopeptide from complex biological samples rapidly and efficiently (D. Li, Xia, & Wang, 2018; Tang et al., 2009; X. Zhang, Wang, He, Chen, & Zhang, 2015). To enable micro-scale analysis of glycopeptides, boronate affinity monolith columns were developed (M. Chen, Lu, Ma, Guo, & Feng, 2009; Jin, Zhang, Yang, Dai, & Zhou, 2018; H. Li & Liu, 2012). For example, phenylboronic acid (PBA) bound to SiO₂ microsphere was synthesized and packed into columns to enrich glycopeptides in HILIC mode (Jianying Chen et al., 2017). This method exhibited high selectivity for both neutral and acidic glycopeptides, with high resistance of interference from 100-fold (in molar) excess of bovine serum albumin digests. MOF with abundant boronic acid sites was designed (S. Li et al., 2018) and has been reported to enable detection of 209 N-glycosylated peptides from human serum digests with enhanced improvement (0.5 fmol/μl) and selectivity (1:100) (Xie, Liu, Li, & Deng, 2018). For endogenous peptidome study, mesoporous silica MCM-41, with pore size as small as 2–3nm, was functionalized with boronic acid groups at inner surface, so that only peptides with molecular weights smaller than 12 kDa were allowed to enter and then glycopeptides were selectively captured by the grafted boronic acid groups (L. Liu et al., 2012). As a result, fifteen unique glycosylation sites mapped to 15 different endogenous glycopeptides were identified from the rat serum.

3.6.1.1. Materials and buffer recipes

1. Boronic acid conjugated magnetic beads.

2. Binding buffer: 200mM ammomium acetate buffer (pH=10).

3. Elution buffer: ACN: H₂O: TFA = 50:49:1.

4. Incubation shaker.

5. Centrifuge.

3.6.1.2. Peptide extraction-Boronic acid magnetic beads

1. Store boronic acid (BA) beads in ethanol (6mM).

2. Wash BA beads with 2ml of binding buffer. Repeat 3 times.

3. Resuspend BA beads in 500 μL of binding buffer.

4. Transfer the beads into sample tubes.

5. Incubate at 37 °C for 1 h with appropriate shaking.

6. Centrifuge the mixture at 900g for 3min.

7. Collect the supernatant.

8. Add 500 μL of binding buffer to the remaining bead.

9. Repeat step 6–8 to wash the beads for 5 times.

10. Elute the glycopeptide using 500 μL elution buffer. Repeat 3 times and combine the elute.

3.6.2. Hydrazine capture chemistry

Hydrazine capture of glycopeptides, relying on the covalent bond formation as well, is a selective isolation and enrichment method first introduced in 2003 (H. Zhang, Li, Martin, & Aebersold, 2003). The cis-diol groups in carbohydrates were oxidized to aldehydes by NaIO₄ and further coupled with hydrazide groups immobilized on solid support materials via covalent bonds (Fig. 6). Nonglycosylated peptides were removed and the N-linked glycopeptides were released by PNGase F treatment. The choice of solid support can affect the enrichment efficiency and preference (Sajid, Jabeen, Hussain, Ashiq, & Najam-Ul-Haq, 2017). Soluble nanopolymers was modified by hydrazide and followed by filter-aided sample preparation (FASP), resulting in 158 unique glycopeptides mapped to 60 glycoproteins from 5μl human serum (L. Zhang et al., 2014). Magnetic nanoparticles functionalized with multiple hydrazide groups offer a large capacity of binding sites, hence, enabled enrichment and isolation of glycopeptides from non-glycosylated counterparts at even 100-fold greater abundance. The application of this method in analysis of mouse liver concluded in 511 unique glycopeptides from 372 different glycoproteins (Qichen Cao et al., 2013). Hydrazide resin has also been incorporated with pipet tips to achieve rapid, automated, high throughput isolation of N-linked glycopeptides (Jing Chen, Shah, & Zhang, 2013). Although the covalent bonding between hydrazide and carbohydrates are of no discrimination, the recovery of glycopeptides highly depends on the enzymatic release step. For example, the PNGase F detachment efficiency in the original protocol described above is limited by the inherent steric hindrance effect. A hydroxylamine assisted PNGase F deglycosylation (HAPD) method was developed to release glycopeptides through cleavage at hydrazine bonds by transamination followed by the PNGase F treatment (J. Huang et al., 2015). Using the HAPD strategy, the detection of N-glycosylation sites was increased by 27% and the identification of N-terminus glycosylated peptides was improved by 5-fold compared to the conventional methods. Another concern in the application of hydrazide capture in O-linked glycopeptide isolation is limited by the lack of an enzyme to specifically cleave O-linked glycans and release the peptides. Also, according to the original method, glycan structure information is not available since the glycans remained on hydrazide resin. These concerns were partially solved by a method proposed to enrich specifically sialic acid-containing peptides, involving both N- and O-linked glycosylation (Nilsson et al., 2009). Sialic acids were selectively oxidized by mild periodate oxidation, followed by hydrazide capture. After the non-glycosylated peptides being removed, the captured glycopeptides were released by acid hydrolysis rather than PNGase F; because the terminal sialic acid and the penultimate monosaccharide are vulnerable to mild acid hydrolysis and the O-linked glycopeptides can be collected in this way. They optimized this method by applying PNGase F to reduce the interference from N-linked glycopeptides prior to the sialic acid-capture-and-release protocol so that selective characterization of O-glycopeptides was largely improved in the analysis of human cerebrospinal fluid (Halim, Rüetschi, Larson, & Nilsson, 2013). They also detected endogenous neuropeptides containing three different glycopeptide stretches from ProSAAS which were cleaved from the proprotein by convertase. The identical endogenous neuropeptides have also been investigated in previous neuropeptidome study (Gupta et al., 2010) and were found to carry sialylated O-glycans as PTMs in a CSF peptidedome study (Zougman et al., 2008). However, this strategy is still subject to the limitations that information of sialylation and nonsialylation peptides is absent. It was reported that the sialylation information can be preserved by using ice-cold 1M hydrochloride to release the sialic acid-containing peptide from hydrazide beads, leaving the sialic acid intact (Kurogochi et al., 2010).

Figure 6.

Graphical features in hydrazide-capturing method for glycoproteome enrichment. Adapted from (Ahn, Kim, & Yoo, 2015) with permission.

3.6.2.1. Materials and buffer recipes

1. Hydrazine resin.

2. Oxidation buffer (100 mM NaAc, 150 mM NaCl, pH = 5.5).

3. Methanol.

4. NaIO₄.

5. Na₂SO₄.

6. ACN/H₂O, 80/20. v/v.

7. NaCl, 1.5M.

8. NH₄HCO₃, 10mM, 100 mM.

9. PNGase F.

3.6.2.2. Peptide extraction-Hydrazine capture

1. Peptides from serum or tissue extract was dissolved in 100 μL of oxidation buffer.

2. Add NaIO₄ to final concentration of 15 mM.

3. Incubate the mixture at room temperature in dark for 1 hour.

4. Quench the reaction by adding Na₂SO₄ to consume excess NaIO₄.

5. Add hydrazine beads into the reaction mixture.

6. Incubate overnight at room temperature.

7. Wash glycopeptide-captured hydrazine beads with 80/20 ACN/H₂O, 1.5 M NaCl and 100% methanol sequentially for 3 times.

8. Add NH₄HCO₃ buffer containing 500U PNGase F enzyme.

9. Incubate at 37 °C overnight to release glycans from the glycopeptides to produce deglycosylated peptides. This process leaves a 0.9858 Da mass on the glycosylation site which makes it possible for site-specific identification.

10. Collect supernatant after enzyme digestion.

11. Wash the beads with 100 μL of 100 mM NH₄HCO₃. Repeat twice.

12. Combine supernatant for further analysis.

To select a proper isolation and enrichment method is the crucial step in a glycopeptidome experiment. The decision usually depends on the aim of the study. Method performing nondiscriminatory separation would be an optimal choice in experiments aiming to obtain comprehensive global glycopeptide identifications. For studies concerning a specific or a rare group of peptides, the lectin affinity-based strategies contribute more precise isolation by choosing the corresponding lectin species. It would be an efficient way to combine methods introduced above, taking advantages of multiple interactions in capturing glycopeptides with unique features. Nonetheless, there will be a trade-off between enrichment efficiency and the sample loss when the multi-step experiment design gets more complicated.

4. Glycopeptide Profiling by MS

Mass spectrometry has grown to be a powerful tool to address challenges in the analysis of complex and low-abundance endogenous glycopeptides. Typical MS strategies consist of various ionization methods and fragmentation methods that would contribute to the depth and the breadth of glycopeptide identification/characterization. Matrix-assisted laser desorption/ionization (MALDI) facilitates simple sample preparation, rapid instrument operation and more straightforward spectral interpretation, which offers benefits in high throughput analysis. Electrospray ionization (ESI), which is more compatible with LC separation system, usually produces multiply charged ions that can be further fragmented to generate tandem mass spectra, delivering more information in peptide backbone sequences, glycan structures and glycosylation sites. As the fragmentation methods continue to develop, MS/MS strategies present more choices and solutions to site-specific glycosylation characterization. Here, in addition to the common MS strategies, we also introduced a novel method, using HCD product ion-triggered EThcD fragmentation, specifically applied to tackle glycosylation occurring on crustacean neuropeptides.

4.1. MALDI-MS

4.1.1. Derivatization approaches

Glycosylated peptides are more hydrophilic due to the attached sugar moieties, which cause reduction in ionization efficiency. The signal suppression of high abundance non-glycosylated peptides will further exacerbate glycopeptide detection. The glycosidic bonds, especially oligosaccharides with sialic acid and fucose, are labile to in-source or post-source dissociation. The preferential loss of sugar moieties during mass measurement is a major problem for MALDI analysis (Nishikaze, 2017). In order to enhance the relative signal intensities of glycopeptides, derivatization methods, such as methyl esterification (Powell & Harvey, 1996), permethylation (Z. Wu et al., 2017) and perbenzolylation (P. Chen, Werner-Zwanziger, Wiesler, Pagel, & Novotny, 1999) have been routinely introduced into sample preparation process (Morelle & Michalski, 2007). The derivatization benefits MALDI MS analysis by stabilizing the labile sialic acid moieties, enabling detection of neutral and acidic glycans in the positive mode and improving the ionization efficiency of derivatives (X. Liu, Qiu, Lee, Chen, & Li, 2010). The stabilization can be accomplished by derivatization of the released glycans (Reiding, Blank, Kuijper, Deelder, & Wuhrer, 2014; Sekiya, Wada, & Tanaka, 2005; Thaysen-Andersen, Mysling, & Højrup, 2009). A microplate-based permethylation kit, which could process up to 96 samples in parallel, was developed to be robotics compatible, high-throughput, and capable to provide linkage-specific analysis of glycans (Shubhakar et al., 2016). Another pathway is to derivatize the glycans without detachment from the intact glycopeptides, allowing simultaneous interpretation for both site-specific glycosylation and peptide sequences. In addition to the derivatization of free glycans, the glycopeptide-level reaction requires more attention in the experimental condition control. For example, the carboxylic acids on the peptide backbones may participate in the derivatization, like ethyl or methyl esterification, causing internal or external reaction and producing heterogeneous products. A dimethylamidation protocol which induced mass difference between α2,3- and α2,6-linked sialic acids, was employed in linkage-specific analysis of sialic acids (de Haan et al., 2015). In addition to reveal accurate structures of the glycans, the informative fragmentation spectra of derivatized glycopeptides enable derivation of peptide backbone sequences as well. This method has been proven to be rapid, simple, reproducible and high-throughput compared to other strategies, like LC-MS, CE-MS and the usage of sialidases.

4.1.2. On-plate enrichment

There also have been efforts combining on-plate enrichment with MALDI MS analysis for fast and selective characterization of glycosylated peptides. The MALDI plate was directly fabricated with boronate functional particles or first modified with metal substrates (e.g. gold) then further functionalized by boronic acid-containing molecules to utilize the large surface area and affinity to glycans and to yield rapid and high-throughput analysis of low-abundance glycosylation in complex samples (Tang et al., 2009; Xu, Zhang, Lu, & Yang, 2010; Y.-C. Liu & Chen, 2017; Zhoufang Zeng, Wang, Guo, Wang, & Lu, 2013). These novel and facile strategies are proven to be effective and versatile with ease of use to profile glycopeptides in complex biological mixtures with minimum sample pre-treatment. When metal gold is used as supporting substrate on MALDI plate, it enhances UV absorption thus benefits desorption and ionization of enriched glycopeptides for improved MS sensitivity.

4.2. ESI-MS

As another soft ionization approach, ESI is a prevalent method to produce multiply charged ions before MS analysis. The coupling of LC and ESI-MS has enabled separation and site-specific characterization of glycosylated peptides simultaneously to unravel biological complexity. The advancements on mass spectrometry fragmentation techniques significantly facilitate the understandings toward glycosylation on peptides via systematic elucidation of peptide backbone, glycan composition and glycosylation site. We present various MS/MS fragmentation approaches utilized to identify glycopeptides, with an emphasis on employing hybrid MS/MS techniques to uncover less studied crustacean glycosylated neuropeptides.

4.2.1. Different MS/MS fragmentation techniques

In tandem MS experiments, multiple MS dissociation approaches (Reiding, Bondt, Franc, & Heck, 2018) can be employed to generate structural fragments of precursor ions (Fig. 7). Commonly applied MS/MS fragmentation processes include electron-induced dissociation (ExD), such as electron-transfer dissociation (ETD) and electron-capture dissociation (ECD), and collision-induced dissociation (CID) whose higher energy version is also known as high-energy collisional dissociation (HCD) in Orbitrap instruments. CID and HCD tandem MS spectra often contain peptide b/y-ion series and monosaccharide ions produced from glycosidic bond cleavage. While in ExD events, the major fragments (c/z-ions) are peptide backbone fragments with intact glycan preserved. Although these tandem MS strategies provide meaningful information of the molecular structures, a fast and thorough characterization of glycosylated peptides can be challenging to achieve due to a lack of full understanding of peptide primary structure, site of glycosylation and glycan elucidation (H. Hu, Khatri, Klein, Leymarie, & Zaia, 2016).

Figure 7.

Overview of contemporary glycopeptide fragmentation methods and their preferred sites of action. The displayed peptide sequence matches that of a tryptic N-glycopeptide from alpha-1-acid glycoprotein, whereas the O-glycan has been added for illustrative purposes. Locations of fragmentation are exemplary, and dissociation of glycosidic linkages and peptide bonds can also occur elsewhere on the molecule. The actual observed fragments will depend highly on both the glycan and the peptide in question and the particular energy deposited in the precursor ions. Adapted from (Reiding, Bondt, Franc, & Heck, 2018) with permission.

4.2.2. Integration of MS/MS fragmentation techniques

4.2.2.1. Combination of different MS/MS fragmentation techniques

As various tandem MS dissociation approaches provide different aspects of glycopeptide structure, complementary fragmentation techniques have been utilized to generate comprehensive information and served as effective tools to uncover intact glycopeptides. CID/HCD and ExD have been implemented sequentially into a single online LC-MS experiment (S.-L. Wu, Hühmer, Hao, & Karger, 2007) or in independent LC-MS experiments (Halim, Nilsson, Rüetschi, Hesse, & Larson, 2012; Halim et al., 2012; C. Ma et al., 2016) for the elucidation of glycosylated peptides. CID and HCD have also been integrated to map native glycopeptides through the sequencing of deglycosylated peptides and intact peptides, respectively. (R. Chen, Seebun, Ye, Zou, & Figeys, 2014)

4.2.2.2. Product ion-triggered MS/MS fragmentation techniques

Glycan diagnostic ions produced from HCD scans are beneficial to fast probing the presence of glycosylated peptides and improving instrument duty cycles. In the low m/z region, those diagnostic oxonium ions include HexNAc (m/z 204.09), HexNAcHex (m/z 366.14) and several HexNAc fragmented pieces (m/z 126.06, 138.06, 144.07, 168.07, and 186.08) (Halim et al., 2014). Taking advantage of those signature ions in HCD scans, subsequent MS/MS scans can be selectively triggered to further elucidate glycopeptide structures, with no need of prior knowledge of glycan structures or sample enrichment steps. An HCD-triggered ETD method has been applied to study N-linked glycosylation on ribonuclease B (Lys-C digested) and human immunoglobin G (Trypsin-digested) (Singh, Zampronio, Creese, & Cooper, 2012). This type of method is generally referred as HCD-product dependent-ETD (HCD-pd-ETD) that has been widely applied to elucidate glycopeptide compositions. (Saba, Dutta, Hemenway, & Viner, 2012; S.-W. Wu, Pu, Viner, & Khoo, 2014; Yin et al., 2013)

4.2.2.3. Hybrid MS/MS fragmentation techniques

Recently the advancement of modern MS technologies has enabled several hybrid MS/MS fragmentation approaches to acquire in-depth glycopeptide information in a single tandem MS spectrum. The most appealing strategies include stepping-energy CID, stepped collisional energy HCD, and electron-transfer/higher-energy collision dissociation (EThcD). For instance, stepping-energy CID has been employed to elucidate N-linked and O-linked glycosylation on human C1-inhibitor protein (Stavenhagen et al., 2018). As different HCD collisional energies produce various fragmentation patterns, the utilization of stepped collision energy HCD (normalized collisional energy set to 20%, 30% and 40%) offers a more comprehensive characterization of the N-linked glycopeptides after HILIC enrichment from human serum (H. Yang, Yang, & Sun, 2018). EThcD was first reported by the Heck group for the characterization of phosphosite by incorporating HCD and ETD fragments in one MS/MS spectrum revealing significantly larger amount of fragments. The employment of EThcD in glycosylation studies (Franc, Yang, & Heck, 2017; Khurana et al., 2019; Y. Zhang et al., 2018) has enabled simultaneous acquisition of glycan composition and peptide backbone fragments of the glycosylated peptide precursors. The Li group first reported the optimization of EThcD parameters including tuning the reaction time in ETD scans per precursor charges to improve spectral quality and produce more abundant structural information. The optimized method with improved duty cycle was applied to profile intact N-glycosylated peptides (Fig. 8) and increase our knowledge about in vivo glycoproteomes in various biological systems (Z. Chen, Huang, et al., 2018; Glover et al., 2018; Q. Yu, Wang, et al., 2017; J. Zhu et al., 2018). Subsequently, this method has been further adapted by Yu et al. for an oxonium ion-triggered HCD-pd-EThcD approach to achieve selective and targeted characterization of O-glycosylation on mouse insulin. An EThcD scan was only triggered when prior HCD spectrum detected oxonium ions, in order to save the ETD reaction time and to increase throughput (Q. Yu, Canales, et al., 2017). This approach enabled the discovery of first O-glycosylated insulin peptides in mouse and human pancreatic islets.

Figure 8.

MS/MS of 3+ charge state precursor ion at m/z 1577.9 of bovine fetuin triantennary N-glycopeptide KLCPDCPLLAPLNDSR (AA 126–141). Alternating between CID/ETD/EThcD resulted in different sets of ions. (a) CID and ETD spectra (inset). Asterisk (*) in the peptide sequence indicates carbamidomethylation. (b) EThcD spectrum. Starred peaks (*) in the spectra were deconvoluted and annotated in the inset. Adapted from (Yu, Wang, et al., 2017) with permission.

As prior evidence showed the glycosylation modification on a signaling peptide hormone, insulin, the HCD-pd-EThcD method has been recently applied to study N-linked and O-linked glycosylation of neuropeptides in crustacean nervous system without prior enrichment, to further extend our knowledge of neuromodulation and crustacean peptidome. In addition to all the benefits this HCD-pd-EThcD method offers, this approach is especially beneficial for studies of glycosylation occurring on neuropeptides in crustaceans whose complete neuropeptide database is still missing. Since the HCD scans also produce an overview of peptidome for the sample analyzed, an updated database with novel crustacean neuropeptides can be built and then used in the following data analysis procedures. This approach (Fig. 9) enables the discovery of glycosylation on novel neuropeptides in addition to known neuropeptides with limited amount of neuronal tissue samples (Qinjingwen Cao, Yu, Liu, Chen, & Li, submitted). The detailed MS workflow is shown below:

Figure 9.

Data analysis workflow for crustacean glyconeuropeptide identification and characterization.

1. Attach a C18 column (75 μm inner diameter × 15 cm, either home-packed or commercially available) onto the LC system coupled with MS. Mobile phase A is 0.1% FA in water and mobile phase B is 0.1% FA in acetonitrile. Set the flow rate to 0.3 μL/min. Allow sufficient time for column to equilibrate and reach stable pressure.

2. Inject 2 μL of the crustacean neuropeptide extract from Section 3.1 (starting tissues >10 pieces) or Section 3.2 (starting tissues < 10 pieces).

3. Set up the following gradient: 0–90 min, 3–30% B; 90.5–110.5 min, 30–75% B; 110–130 min, 75–95% B; and include another 15 min 3% B at the end for column equilibration.

4. Set up an HCD-pd-EThcD MS method:

   1. Acquire full MS scan at Top Speed mode (3s) from m/z 400–1800 at 120 K resolution.
   2. For data-dependent acquisition (DDA), use mode cycle time (3s) and dynamic exclusion 20 s.
   3. For HCD scans, use normalized collisional energy 35% with resolution 30 K.
   4. Include a targeted mass trigger containing m/z 138.0545, m/z 204.0867 and m/z 366.1396 and set up the following criteria: If any of the above three ions is within the top 30 most abundant peaks in the HCD spectra, trigger a subsequent EThcD MS/MS scan.
   5. For EThcD scans, precursors at different charges are exposed to distinct ETD reaction times followed by same amount of supplemental activation energy (33%) : z = 2, reaction time = 50 ms; z = 3–4, reaction time = 20 ms; z = 5, reaction time = 10 ms; z = 6–15, use instrument charge-dependent ETD parameters. All EThcD spectra are acquired at resolution 60 K.

4.2.3. Ion mobility MS

Ion mobility (IM) MS is a gas phase separation technique that allows differentiation of molecules based on their structural/conformational differences, measured by drift time analytes experienced inside the IM cell filled with buffer gas. IM-MS has drawn increasing attention in glycosylation studies (Z. Chen, Glover, & Li, 2018), because complete structural elucidation of glycopeptides remains to be challenging relying solely on masses obtained from tandem MS events. The challenges imposed mainly are contributed by glycan heterogeneity (Fig. 10), especially sugar structural isomerization, monosaccharide α/β-stereochemistry linkage at multiple potential sites, and anomeric α/β-linkage between amino acid residue and glycan. Often times there are only subtle structural variances among various glycan and glycoconjugate isomers thus it is highly demanding to develop novel techniques to assist differentiation of those molecules. A high field asymmetric wave ion mobility spectrometry (FAIMS) has been developed to separate isomeric O-linked glycopeptides with merely glycosylation site difference (Creese & Cooper, 2012). The two glycoconjugate isomers were indistinguishable in LC but the glycosylation site differences have potentially contributed to gas-phase intramolecular interactions resulting in the structural variance noticeable in drift times. IM MS has also been demonstrated as an efficient strategy resolving N-acetylneuraminic acid α2,3 and α2,6 linkages by measuring signature CID fragmentation ion drift times (Hinneburg et al., 2016). Utilization of IM-MS is anticipated to foster knowledge of glycosciences with the improved technologies and integration of other orthogonal techniques, including capillary electrophoresis (CE), LC and tandem MS.

Figure 10.

The isomerization of glycan and glycoconjugates. (a) The building blocks (monosaccharides) that compose larger glycans are structural isomers (hexose: galactose, glucose, mannose, N-acetylhexosamine: N-acetylgalactosamine, N-acetylglucosamine); monosaccharides can be connected either α-stereochemistry or β-stereochemistry at multiple potential linkage position; fucose could be either attached to N-glycan core or branches. (b) Epimeric glycoconjugates results from alternative configurations (α- or β-) at the anomeric linkages or the presence of epimeric glycan monomers (galactose or glucose), two isomeric N-glycopeptides differ in the site of N-glycan attachment. Adapted from (Z. Chen, Glover, & Li, 2018) with permission.

4.3. Data Analysis to Uncover Sequence and Structural Information about Intact Glycopeptides

There have been continuous efforts to develop software for MS data interpretation and large-scale glycopeptide identification. A plethora of search engines are available, including Byonic (Cooper, Gasteiger, & Packer, 2001), GlycoMaster DB (He, Xin, Shan, Lajoie, & Ma, 2014), GPQuest (Toghi Eshghi, Shah, Yang, Li, & Zhang, 2015), MAGIC (Lynn et al., 2015), Integrated GlycoProteome Analyzer (I-GPA) (G. W. Park et al., 2016), SweetNET (Nasir et al., 2016), pGlyco (W.-F. Zeng et al., 2016), pGlyco 2.0 (M.-Q. Liu et al., 2017). As glycosylated peptide standards are generally difficult to synthesize, the rich software selections have not only enabled deciphering of complicated glycopeptides but also provided various cross-validation opportunities to consolidate search results generated from different MS platforms. The MS data processing steps typically involve a match between MS/MS spectra and in silico peptide/protein sequences together with a glycan library. For the organisms and species missing a completely characterized genomic and peptide database, such as crustaceans, de novo sequencing can first be applied to the HCD scans in HCD-pd-EThcD spectra to generate a new peptide database to be used in data analysis to achieve better identification rate of intact glycosylated neuropeptides, including novel glycosylated neuropeptides.

A detailed procedure enabling discovery of crustacean glycosylated neuropeptides is provided below. Briefly, PEAKS Studio is first used to analyze the acquired data and generate de novo peptide sequences (typically >5K sequences); then those de novo sequences are combined with existing Li Lab crustacean neuropeptide database (containing ~650 neuropeptides) to generate a new peptide database; following this, the acquired MS data are searched against the new peptide database using Proteome Discoverer embedded with Byonic node. Finally, the identified glycosylated neuropeptide spectra are manually checked to ensure correct assignment.

1. Use PEAKS Studio to generate de novo sequences. The version we used is PEAKS 7 but there are updated versions available, such as PEAKS X. In data refinement, set full MS precursor tolerance to be 10 ppm and fragment tolerance ion to be 0.02 Da. Select “no enzyme” as no digestion has been involved. In the PTM types, select fixed modification to be Cys carbamidomethylation; for variable modifications, include C-terminal amidation, Met oxidation, pyro-Glu from E, pyro-Glu from Q, Tyr-sulfation, dehydration, methylation and glycan modifications consisting no more than 2 glycans. Import Li Lab crustacean neuropeptide database. After search is finished, go to “de novo only” tab and export all de novo sequences with average local confidence (ALC) > 50%.

2. Combine de novo sequences with custom-built crustacean neuropeptide database. Use Excel or customized scripts to remove any unnecessary columns and adapt the format of the sequences to be FASTA style. Save the new database into .fasta file. Note: name the Li Lab crustacean neuropeptides and de novo sequences in a distinguishable manner as the de novo sequences are not completely confirmed.

3. Use Proteome Discoverer with Byonic node to identify glycosylated neuropeptides. The software version used is Proteome Discoverer 2.1. Data are searched against mammalian glycan database and the new database from Step 2. Precursor tolerance is set to be 10 ppm and fragment tolerance ion to be ±0.01 Da. Include modifications used in Step 1.

4. Manually check the spectra in Byonic Viewer. A glycosylated neuropeptide should have oxonium ions in the low mass region with most of the fragments assigned. Fig. 11 illustrates a tandem MS spectrum annotated and assigned as a novel B-type allatostatin neuropeptide with O-linked HexNAc2Hex2Fuc2 discovered from blue crab pericardial organ.

Figure 11.

O-glycosylation occurring on a novel B-type allatostatin neuropeptide revealed by EThcD fragmentation.

5. Conclusions and Future Directions

Over the past decades, extensive efforts have been devoted to the large-scale characterization of glycopeptides and glycoproteins due to their important biological functions. The rapid technological innovations, ranging from sample enrichment strategies to MS/MS characterization techniques, have shed light on this complex PTM that is essential in a variety of biological processes, such as cell-cell communication and disease progression. The large-scale profiling of glycopeptidome or even glycoproteome with site-specific elucidation has become more feasible thanks to the development of hybrid MS/MS approaches on highly sensitive instrument platforms. However, to reveal complete glycopeptide information in a complex biological system, more selective and efficient enrichment methods are still in demand to preserve intact glycan structures to the best extent, especially for O-linked glycosylation. Furthermore, current MS/MS fragmentation approaches only enable complete characterization of peptides with relatively simple glycosylation structure. The isomer differentiation and high-order glycan complexity require more diverse tools (e.g. IM-MS and spectroscopy) to differentiate the subtle differences among structural isomers and characterize glycan microheterogeneity, respectively. Another aspect is to understand the roles of glycosylated signaling molecules, such as glycosylated neuropeptides, by incorporating quantitatively comparative studies and electrophysiological studies in the nervous system. The relative abundance changes of glycosylated neuropeptides in response to certain stimuli or perturbation and associated electrical potential alterations are expected to generate new insights into neuromodulation mechanism and signaling transmission. Future advancements should include sample enrichment improvement, streamlined hybrid characterization techniques, improved bioinformatics strategies for glycan sequencing and functional studies to enable more a comprehensive understanding the roles of these glycosylated peptides at the cellular and network levels.