Global site-specific analysis of glycoprotein N-glycan processing

Abstract

N-glycans contribute to the folding, stability and functions of the proteins they decorate. They are produced by transfer of the glycan precursor to the sequon Asn-X-Thr/Ser, followed by enzymatic trimming to a high mannose-type core and sequential addition of monosaccharides to generate complex-type and hybrid glycans. This process, mediated by the concerted action of multiple enzymes, produces a mixture of related glycoforms at each glycosite, making analysis of glycosylation difficult. To address this analytical challenge we developed a robust semi-quantitative MS-based method that determines the degree of glycan occupancy at each glycosite, and the proportion of N-glycans processed from high mannose type to complex type. It is applicable to virtually any glycoprotein, and a complete analysis can be conducted with 30 μg of protein or less. Here we provide a detailed description of the method that includes procedures for (1) proteolytic digestion of glycoprotein(s) with specific and non-specific proteases; (2) denaturation of proteases by heating; (3) sequential treatment of the glycopeptide mixture with two endoglycosidases, Endo H and PNGase F, to create unique mass signatures for the three glycosylation states; (4) LC-MS/MS analysis; (5) data analysis for identification and quantitation of peptides for the three glycosylation states. Full coverage of site-specific glycosylation of glycoproteins is achieved with up to thousands of high-confidence spectra hits for each glycosite. The protocol can be performed by an experienced technician or student/post-doc with basic skills for proteomics experiments and takes approximately 7 days to complete.

Introduction

Asparagine-linked-glycans (N-glycans) are among the most common and important post-translational modifications of proteins. They play critical roles in the folding, conformation and stability of the proteins themselves^1,2, and participate as ligands in intra- and inter-cellular recognition and host-pathogen interactions^3,4. Altered biosynthesis of N-glycans has been associated with many diseases, such as cancer^5,6, influenza^7–9, and acquired immunodeficiency syndrome (AIDS)^10–14. For example, increased levels of under-processed high mannose type glycans have been reported to occur during breast cancer progression in both mice and human⁶. Some broadly neutralizing antibodies (bnAbs) of the highly glycosylated human immunodeficiency virus (HIV) envelope glycoprotein include under-processed high mannose glycans in their epitopes, while others require fully processed complex type structures containing sialic acid^11,15. Since the structures of N-glycans impact the activity and pharmacodynamics of glycoprotein biotherapeutics¹⁶, careful choice of cell lines used to express proteins¹⁰, growth conditions¹⁷, and purification methods^18–20 to control the consistency of N-glycosylation is needed.

Given the importance of N-glycans to the structure and functions of glycoproteins, there is increasing need for robust methods for analysis of N-linked glycosylation that can be integrated with state of the art proteomics. Complicating the analysis is the inherent diversity of N-glycan structures present on any glycoprotein, which is a consequence of the non-template driven biosynthesis. It begins with the en bloc transfer of Glc₃Man₉GlcNAc₂ from a lipid linked glycosyl-donor to the nascent polypeptide by an oligosaccharyltransferase (OST) to a sequence defined glycosite, Asn-X-Thr/Ser (where X can be any amino acid residue except proline). Although none of the proteins analyzed in the present study contains an atypical glycosylation site (N-X-Cys/Val), these sites have been verified in previous studies by the presence of glycans on intact glycopeptides^21,22. As illustrated in Fig. 1, the glycan is then subject to trimming by removal of the glucose residues to a “high mannose” type glycan, then further trimming to the conserved Man₃GlcNAc₂ core, followed by addition of sugars by the sequential action of glycosyltransferases that produce a highly related set of “complex type” structures (Fig. 1). Moreover, for glycoproteins with more than one glycosite, processing at each site may differ based on the access of the glycans to the processing enzymes. Indeed, well documented examples of proteins that contain high mannose type glycans at one or more glycosites, and highly processed complex type glycans at other sites are IgM^23,24, influenza hemagglutinin^8,25,26, and the HIV envelope glycoprotein^10,27–29.

Figure 1.

N-linked glycan processing in the ER and Golgi apparatus. N-glycans that are Endo H-sensitive are shown in the red box, which include high mannose and hybrid glycans that has two terminal mannose residues required for recognition by Endo H (red oval). Fuc: fucose, Glc: glucose, Sia: sialic acid, Gal: galactose, GlcNAc: N-Acetylglucosamine, Man: mannose.

Analysis of N-glycans of glycoproteins

Over the last two decades, several strategies have emerged to characterize N-glycans of glycoproteins and identify the glycosites that are recognized and used by the cellular glycosylation machinery. Several methods employ endoglycosidases, such as PNGase F and PNGase A, to release the glycans from the protein followed by analysis of glycoforms using mass spectrometry (MS)^30–35 or high performance liquid chromatography (HPLC)^36,37 with or without prior derivatization³⁸. The MS based methods provide a composition for each molecular ion, which are annotated as high mannose or complex type glycans consistent with biosynthetic principles. Furthermore, tandem mass spectra of derivatized N-glycans generated by both matrix-assisted laser desorption/ionization (MALDI) and electrospray ionization (ESI) MS have been widely used for characterization of detailed structural information of N-glycans expressed in different biological systems^30,34. Freely available software tools such as GRITS Toolbox (http://www.grits-toolbox.org/) are able to automatically process, annotate, and archive glycomics data in a high throughput manner. The HPLC methods rely on retention times of N-glycan standards for identification of the glycan species. With both methods, supporting experiments using glycosidase digestions and/or permethylation analysis can provide additional support for structure assignments. Such methods find wide utility in characterizing glycosylation of highly studied glycoproteins like immunoglobulins (e.g. IgG), or detecting individual differences in N-glycans of serum glycoproteins^36,37,39. While these methods provide key information about the nature of the glycoforms present on a glycoprotein and their relative abundance, they do not reveal which glycosites are utilized or the extent to which each glycosite is occupied.

A number of proteomics based methods focus on identification of the glycosites that are utilized by the oligosaccharyltransferase and the degree to which the site is occupied by a glycan^40–44. The basic strategy is to immobilize glycopeptides using lectins, or by coupling periodate treated glycopeptides to hydrazide-activated beads. Then, peptides are released from the bound glycan with an endoglycosidase (e.g. PNGase F) that in the process converts the Asn-X-Thr/Ser sequence to Asp-X-Thr/Ser. When the reaction is done in H₂O¹⁸, it creates a mass difference from Asn to Asp of +3. Then, MS/MS analysis of the eluted peptides provides positive identification of the sites that are glycosylated. These methods are particularly useful for surveying complex biological systems, such as whole cells or tissues, to identify proteins that are glycosylated.

Recently there have been major advances in glyco-proteomics MS/MS methods that analyze intact glycopeptides, and establish glycoforms present at each glycosite based on the combined masses of the peptide and N-glycan^{8,28,29,45–48}. Major limitations of this approach are the relatively low abundance of glycopeptides in the protein digest mixture as well as the inherent low ionization efficiency of glycopeptides during MS analysis^40,41,49,50. For these reasons, glycopeptides are typically enriched from peptide digests prior to MS analysis with different purification methods, such as hydrophilic interaction chromatography (HILIC)^29,51,52 and hydrazide chemistry⁴⁰. Despite the use of different fragmentation methods, including collision-induced dissociation (CID), high-energy collision dissociation (HCD), and electron-transfer dissociation (ETD), the inherent heterogeneity of glycoforms at each glycosylation site as well as difficulties in getting good peptide backbone fragmentations for peptides with large glycans, impede comprehensive identification of intact glycopeptides^28,30. Annotation of LC-MS/MS data from such experiments is possible by using commercial algorithms such as Byonic™ (http://www.proteinmetrics.com/products/byonic/). However, quantitative assessment of the relative abundance of the glycan structures detected at each glycosite is still a significant challenge due to unknown ionization efficiency for peptides with each glycoform, especially for those with sialic acid-containing structures³⁸. Moreover, if the glycopeptides are enriched prior to analysis, peptides with no glycan are lost, and the degree of glycosite occupancy cannot be assessed.

Development of the protocol and applications of the method

The protocol described here arose from the need of the HIV vaccine effort for robust semi-quantitative glyco-proteomics methods that would: 1) establish the occupancy of N-glycans at each glycosite and 2) assess the degree to which glycans were processed from high mannose type to complex. The need stems from the fact that the primary candidate for a vaccine is the HIV envelope glycoprotein trimer (Env) that contains 75-90 N-glycans creating a glycan shield to protect against attack by the immune system^53,54. Despite the dense cover of N-glycans, broadly neutralizing antibodies (bnAbs) that bind to HIV Env do occur in 10-30% of HIV-1 infected patients^55,56. Importantly, some bnAbs show interactions with high-mannose glycans^57,58, while others show dependence on complex-type glycans^12,15,59,60. Thus, to support the rational design of an HIV Env vaccine, we developed a method that could assess the proportion of high mannose type and complex type N-glycans at each glycosite on the HIV Env¹⁰. However, as demonstrated here, the method is broadly applicable to any glycoprotein.

One of the key aspects of our method is the sequential use of two endoglycosidase treatments to introduce unique mass signatures for glycosites that carry no glycan, high mannose/hybrid type glycans, and complex type glycans, as illustrated in Fig. 2A. After protease digestion, Endoglycosidase H (Endo H) is used to remove all high mannose and hybrid type N-glycans that have at least 5 mannose residues (see Fig. 1). This enzyme leaves a GlcNAc-Asn residue that adds +203 to the peptide mass. Subsequently, the remaining complex N-glycans are removed with PNGase F in the presence of H₂O¹⁸, which both removes the glycan and converts Asn to Asp, with a +3 addition to the peptide mass. Peptides carrying glycosites (Asn-X-Thr/Ser) with Asn (unoccupied), GlcNAc-Asn (Endo H treated), or Asp (PNGase F treated), display a similar ionization efficiency during MS analysis^10,45, allowing us to use ion intensity peak area to quantify the relative distribution of the three glycosylation states at each glycosite detected^10,61. Another distinguishing feature of the method is using multiple proteases, which can generate up to thousands of spectra hits for each glycosite, resulting in >95% sequence coverage. This simple strategy effectively converts the glycoproteomics analysis to a proteomics analysis, allowing the use of robust proteomics software to analyze data in a high throughput manner.

Figure 2.

Schematic overview of the protocol. (a) Introduction of novel mass signatures for peptide glycosites that are not occupied, or occupied by high mannose/hybrid or complex type glycans. (b) The workflow of the protocol. Fig. 2a adapted with permission from ref. 10, Nature Publishing Group.

A major advantage of this method is that it provides a glimpse into the site occupancy and processing of N-glycans at each glycosylation site¹⁰. It provides a semi-quantitative analysis for the three glycosylation states, and a complete analysis can be conducted on only 30 μg of protein, even for complex glycoproteins like HIV Env comprising up to 30 glycosites per monomer. It should be noted that hybrid structures, which can potentially contain sialic acid (see Fig. 1), are included in the high mannose type category since they are cleaved by Endo H. However, hybrid structures are typically low abundance^29,62.

One major limitation of the method is that glycan structures are removed before LC-MS/MS, and the class of the glycan inferred by the specificities of the endoglycosidases. MS/MS methods based on analysis of intact glycopeptides provide more information on the spectrum of glycans at individual glycosites^{27- 29,63}, and are complementary to this protocol which provides semi-quantitative information on site occupancy and glycan processing. However, neither method provides the precise structure complete with glycosidic linkages between monosaccharides. Although this protocol is designed for site-specific glycosylation analysis of purified glycoproteins, it is in principle applicable to more complex protein samples, such as membrane enveloped viruses (e.g. HIV, influenza virus, coronavirus) that comprise only 10-12 proteins. However, as described in Experimental Procedure, modification of the protocol would be needed to survey glycosites in more complex samples such as cells or human serum.

Overview of the Procedure

The procedure for site-specific analysis of glycoprotein N-glycan processing is summarized in Fig. 2b, and consists of the following key stages.

• In the first stage, buffer exchange for glycoproteins is needed if they are dissolved in the buffers that contain non-volatile salts (steps 1-12). Glycoproteins are then denatured and alkylated at pH 6 (steps 13-17) to minimize non- enzymatic deamidation while retaining protein sequence coverage. Proteins are digested with several different protease treatments, including ‘triple digestion’, chymotrypsin, and a combination of trypsin and chymotrypsin, in order to maximize sequence coverage and increase confidence of detecting each glycosite (step 18).

• In the second stage (steps 19-25), proteases are denatured by heating to prevent incorporation of ¹⁸O-water into the C-termini of peptides during the subsequent PNGase F treatment.

• In the third stage (steps 26-36), sequential endoglycosidase treatment is employed to create unique mass signatures relative to the predicted amino acid sequence, for peptide glycosites that are not occupied (+0 Da), or occupied by high mannose/hybrid (+203 Da) or complex type glycans (+3 Da) (Fig. 2a).

• The resulting samples are subjected to LC-MS/MS analysis (step 37).

• Data processing for identification and quantitation of peptides for the three glycosylation status is done using software package Integrated Proteomics Pipeline (IP2) (stage 5, steps 41-53). The MS1 and MS2 data are extracted from MS raw files with RawConverter and processed using multiple components from software package IP2 (steps 38-53).

Alternative methods

Over the last decade there has been significant progress on site-specific glycosylation analysis of purified glycoproteins^{8,27–29,45,62–65}. Exemplary and perhaps the most relevant to this protocol is the work on HIV-1 Env, by the Desaire and Cripsin groups^27–29,63. The methods developed by both groups focus on characterizing intact glycopeptides with or without enrichment of glycopeptides prior to MS analysis, and thus can provide complementary information about individual glycoforms at each glycosite. However, milligram quantities of material may be required for their methods, which is attributed in part to the relatively low ionization efficiencies of glycopeptides, and because each glycopeptide is actually a mixture of many different glycoforms^28,29 Although the complexity is reduced by using only two proteases, chymotrypsin and trypsin, identification of all forms of a glycopeptide with multiple N-glycans is still a challenge even when using a combination of CID and ETD for fragmentation^27,28. Quantitative measurements are also challenging due to unknown ionization of glycopeptides with different N-glycan species during MS analysis. Moreover, there is limited information about site occupancy due to dramatically different ionization efficiency between peptides and corresponding glycopeptides, and/or enrichment of glycopeptides prior to MS analysis.

Other methods using endoglycosidases to identify glycosites have long been in used^40,41,44. However, typically glycopeptides are selectively captured using treatment with periodate followed by binding to hydrazide activated beads, or enriched using lectins. The peptides are then released using PNGase F, in which glycosylated asparagine is converted to aspartic acid (+1 Da mass shift), allowing indirect identification of glycosites in the released peptides.^40,41. Alternatively, for a purified protein, PNGase F can be applied directly, and site occupancy of glycosylation sites can be determined by measuring the ratio of peptides with glycosites containing asparagine and aspartic acid during ESI-MS analysis²⁸. By comparison with the protocol described here, addition of a second endoglycosidease, EndoH, provides additional information about the extent of glycan processing in addition to the degree of site occupancy.

Experimental design

Proteolysis with a combination of proteases

Glycoproteins are denatured and alkylated at pH 6 instead of the mildly alkaline pH to minimize the non-enzymatic deamidation of Asn to Asp that can complicate data analysis⁶⁶. To maximize sequence coverage, a number of different protease digestions are used, including chymotrypsin, a combination of trypsin and chymotrypsin, and “triple digestion” with combinations of Arg-C, trypsin, elastase and subtilisin⁶⁷ (step 18). Of note, a combination of all proteolytic digestions is essential for detecting all glycosites on heavily glycosylated proteins with large molecular weight, such as the HIV Env trimer and the spike glycoprotein of Middle East respiratory syndrome coronavirus (MERS-CoV S-2P protein). Triple digestion alone is able to identify all glycosites in glycoproteins containing only a few glycosites, while for most glycoproteins a combination of triple digestion and chymotrypsin is enough to generate detectable peptides that contain all the glycosylation sites¹⁰. Another benefit of the use of multiple proteases, including trypsin and non-specific proteases, is that it produces a sequence ladder that contains a series of peptides with variable numbers of amino acid residues on both sides of glycosite, allowing for highly confident identification for a given glycosite (Supplementary Table 1). For single glycoproteins or simple mixtures (e.g. viruses) the use of multiple proteases allows for much higher confident detection and semi-quantitative analysis of the three states of glycosylation for each glycosite. However, for more complex mixtures (e.g. cells, plasma etc.) use of specific proteases such as trypsin will result in reduced complexity and the identification of glycosites, while retaining the ability to detect the three different glycosylation states. Another key aspect of the protocol is that only volatile salts, such as ammonium bicarbonate and ammonium acetate, are employed so that no column purification is needed, resulting in high specificity of the protocol (requires ~30 μg of starting material).

Denaturation of proteases

In the next stage PNGase F treatment is employed for deglycosylation of glycoproteins, resulting in conversion of Asn to Asp on removal of the carbohydrate, and a mass change of +3 when carried out in O¹⁸-water (steps 19-25). Residual active trypsin and other proteases used for proteolysis can incorporate O¹⁸ into the C-termini of the peptides during the deglycosylation step⁶⁸, resulting in a high false positive identification of peptides that may have glycosites. To avoid this, we denature all proteases used in the protocol by heating at 100 °C prior to the deglycosylation steps.

Sequential endoglycosidase treatment

Sequential treatment of glycopeptides with Endo H followed by PNGase F is employed to generate different mass signatures for glycosites that contain no glycan (+0), high mannose/hybrid type glycan (+203), and complex type glycan (+3) (steps 26-36, Fig. 2a). The endoglycosidase digestions are highly efficient and go rapidly to completion, so the deglycosylation reactions are conducted for 1 h to minimize potential non-enzymatic deamidation occurring during this step. Removal of N-glycans increases the ionization efficiencies of peptides and allows us to localize glycosylation sites on peptides with multiple modifications, which is challenging for analysis of intact glycopeptides (Supplementary Fig. 1).

LC-MS/MS analysis

In principle, the deglycosylated peptides can then be analyzed by any type of LC-MS/MS (step 37). A high-resolution mass spectrometer, such as the Orbitrap Elite, provides satisfactory sequence coverage for most glycoproteins. An instrument with higher scan speed, such as an Orbitrap Fusion or Orbitrap Lumos, is more sensitive and generates several times more MS/MS spectra for a given sample than an Orbitrap Elite. Thus, they are able to provide site-specific glycan processing information for those sites that may be missed in the results generated by the Orbitrap Elite as a result of heavy glycosylation and the large molecular weights of glycoproteins. While single-dimension separation is sufficient for characterization of site-specific glycosylation of purified proteins, multidimensional protein identification technology (MudPIT), in which peptides are systematically separated based on the charge in the first dimension and hydrophobicity in the second, will accelerate measurement of site-specific N-glycan processing of glycoproteins in these complex protein samples⁶⁹.

Replicates and controls

Ideally, each glycoprotein is digested in at least two technical replicates and analyzed by the same MS instrument. Invertase produced by the yeast S. cerevisiae and alpha-1-acid glycoprotein from bovine serum are known to be occupied by high mannose type and complex type glycans respectively and can be used as controls to test completion of endoglycosidase treatment. As a check on the overall success of the protocol, non-enzymatic deamidation of Asn residues and/or O¹⁸-incorpation at the C-terminus should be seen in less than 5% of the total peptides.

Materials

Reagents

• Urea (MilliporeSigma, cat. no. U5128)

• Ammonium acetate (CH₃COONH₄, MilliporeSigma, cat. no. A1542)

• Dithiothreitol (DTT, Fisher BioReagents, cat. no. BP172-5)

• Iodoacetamide (IAA, MilliporeSigma, cat. no. I1149)

• Water (purified by the GenPure Pro water purification system, Thermo Fisher Scientific)

• Ammonium bicarbonate (NH₄HCO₃, MilliporeSigma, cat. no. A6141)

• Formic acid (MilliporeSigma, cat. no. F0507)

• Acetic acid (Fisher Scientific, cat. no. A38-212)

• Hydrochloric acid (HCl, 36.5-38% (vol/vol), A144-212) !CAUTION Concentrated hydrochloric acid is a corrosive acid and forms acidic mists. Both the mist and the solution have a corrosive effect on human tissue, with the potential to damage respiratory organs, eyes, skin, and intestines irreversibly. It should be handled in a hood with personal protective equipment, including lab coat, gloves, and safety glasses.

• Arg-C (Promega, cat. no. V1881)

• Trypsin (Promega, cat. no. V5111)

• Ethylenediaminetetraacetic acid disodium salt dihydrate (EDTA, MilliporeSigma, cat. no. E5134)

• Acetonitrile (ACN, VWR, cat. no. 200004-350) !CAUTION Acetonitrile has modest toxicity in small doses and should be handled in a hood using gloves.

• Tris Base (Fisher Scientific, cat. no. BP152-10)

• Calcium chloride dihydrate (CaCl₂·2H₂O, MilliporeSigma, cat. no. C3306)

• Elastase (Promega, cat. no. V1891)

• Subtilisin (MilliporeSigma, cat. no. P5380)

• Chymotrypsin (Promega, cat. no. V1061)

• Endo H (New England Biolabs, cat. no. P0702L)

• PNGase F (New England Biolabs, cat. no. P0705S)

• ¹⁸O-water (97 atom % of ¹⁸O, MilliporeSigma, cat. no. 329878)

  CRITICAL STEP Store the reagent in a desiccator.

• Fetuin (MilliporeSigma, cat. no. F3385)

• Invertase (MilliporeSigma, cat. no. I0408)

• IgG (MilliporeSigma, cat. no. I4506)

• IgM (MilliporeSigma, cat. no. I8260)

• AGP (MilliporeSigma, cat. no. G3643)

• Transferrin (MilliporeSigma, cat. no. T8158)

• Glycoprotein of interest: In this Protocol we use three example proteins: BG505 SOSIP.664 trimer (laboratory made, expressed in HEK 293-F cells as described in previous study¹⁰), MERS-CoV S-2P protein (laboratory made, expressed in HEK 293-F cells as described in previous study⁷⁰), and Influenza virus hemagglutinin (HA, laboratory made, expressed in HEK 293-F cells as described in previous study¹⁰)

Equipment

• Eppendorf Research Plus pipette (Eppendorf)

• Biotix Microcentrifuge Tubes-1.5 mL (VWR, cat. no. MTL-0150-BC)

• Low binding pipet tips-10 μl (Corning, cat. no. 4150)

• Low binding pipet tips-200 μl (Corning, cat. no. 4151)

• Centrifugal Filter-10kDa (MilliporeSigma, cat. no. MRCPRT010)

• Centrifugal Filter-30kDa (MilliporeSigma, cat. no. MRCF0R030)

• Incubator at 37 °C (Fisher Scientific, cat. no. 15-103-0514)

• Incubator at 56 °C (Fisher Scientific, cat. no. 15-103-0514)

• Incubator at 100 °C (Fisher Scientific, cat. no. 05-412-500)

• Water bath at 30 °C (Fisher Scientific, cat. no. 15-462-5Q)

• pH meter (MilliporeSigma, cat. no. Z283037)

• Eppendorf Microcentrifuge 5415R (MilliporeSigma, cat. no. Z605212)

• Microcentrifuge (Santa Cruz Biotechnology, cat. no. sc-358765)

• Lyophilizer (Labconco)

• Desiccator (HACH)

• Fisher Vortex Genie 2 (Fisher Scientific, cat. no. 12-812)

• Parafilm (VWR, cat. no. 52858-000)

• Kimtech Science Kimwipes tissues (Kimberly-Clark)

• Freezer−80 °C (Forma Scientific)

• Mass spectrometer: The protocol below is optimized for either (1) Orbitrap Elite Hybrid Ion Trap-Orbitrap mass spectrometer equipped with EASY-nLCII system (Thermo Fisher Scientific) or (2) Orbitrap Fusion Tribrid mass spectrometer equipped with EASY 1000 system (Thermo Fisher Scientific)

Software

• RawConverter, version 1.1.0.19 http://fields.scripps.edu/rawconv/

• Notepad++, version 7.5.4 https://notepad-plus-plus.org/download/v7.5.3.html

• glyco_motif_filter, version 1.0 http://fields.scripps.edu/yates/wp/?page_id=687

• IP2 – Integrated Proteomics Pipeline, version 5.0.1 http://goldfish.scripps.edu/ip2/login.jsp

Reagent setup

• Ammonium acetate (100 mM, pH 6) Dissolve 0.077 g of ammonium acetate in 10 ml of water. Adjust the buffer pH to 6 by adding acetic acid. Dispense into aliquots and store the buffer at −20 °C for up to 2 months.

• Urea (8 M) Dissolve 0.048 g of urea in 100 μl of 100 mM ammonium acetate (pH 6). Prepare the solution fresh.

• Dithiothreitol (500 mM) Dissolve 0.0077 g of dithiothreitol in 100 μl of water. Prepare the solution fresh.

• Iodoacetamide (500 mM) Dissolve 0.00925 g of iodoacetamide in 100 μl of water. Prepare the solution fresh.

• Ammonium bicarbonate (100 mM, pH 8) Dissolve 0.079 g of ammonium bicarbonate in 10 ml of water. Measure the buffer pH. Dispense into aliquots and store the buffer at −20 °C for up to 2 months.

• 100 mM Tris-HCl (pH 7.8) Dissolve 12.114 g of Tris in 800 ml of water. Adjust pH to 7.8 with concentrated HCl. Bring final volume to 1 L with water. Store the buffer at 4 °C for up to 2 months.

• CaCl₂ (1 M) Dissolve 0.147 g of calcium chloride dehydrate in 1 ml of water. Dispense into aliquots and store the buffer at −20 °C for up to 2 months.

• EDTA (1 M, pH 8) Add 186.12 g of EDTA•2H₂O to 300 mL of H₂O. Stir vigorously on a magnetic stirrer. Adjust the pH to 8.0 with NaOH pellets. The disodium salt of EDTA will not go into solution until the pH of the solution is adjusted to ~8.0 by the addition of NaOH. Bring final volume to 500 ml with water. Store the buffer at 4 °C for up to 2 months.

• Arg-C (0.5 μg/ul) Dissolve 10 μg of Arg-C in 20 μl of the buffer that contains 50 mM Tris-HCl (pH 7.8), 5 mM CaCl₂, 2 mM EDTA. Dispense into aliquots and store the solution at −80 °C for up to 2 months.

• Trypsin (0.5 μg/ul) Dissolve 20 μg of trypsin in the resuspension buffer provided by manufacturer. Dispense into aliquots and store the solution at −80 °C for up to 2 months.

• Elastase (0.5 μg/ul) Dissolve 0.5 mg of elastase in 1 ml of water. Dispense into aliquots and store the solution at −80 °C for up to 2 months.

• Subtilisin (0.5 μg/ul) Dissolve 0.5 mg of subtilisin in 1 ml of water. Dispense into aliquots and store the solution at −80 °C for up to 2 months.

• Chymotrypsin (0.5 μg/ul) Dissolve 25 μg of chymotrypsin in 50 μl of 1 mM HCl. Dispense into aliquots and store the solution at −80 °C for up to 2 months.

• Ammonium acetate (100 mM, pH 5.5) Dissolve 0.077 g of ammonium acetate in 10 ml of water. Adjust the buffer pH to 5.5 by adding acetic acid. Dispense into aliquots and store the buffer at −20 °C for up to 2 months.

• ¹⁸O-water (97 atom % of ¹⁸O) Seal the aliquots of ¹⁸O-water and place them into a desiccator.

• Ammonium bicarbonate (100 mM, prepared with ¹⁸O-water, pH 8) Place ammonium bicarbonate into a desiccator for at least 48 h to dry it before use. Dissolve 0.0079 g of ammonium bicarbonate in 1 ml of ¹⁸O-water. Place it into a desiccator and store the buffer at room temperature (20-25 °C) for up to 2 months.

• PNGase F (glycerol-free) Lyophilize the solution that contains PNGase F and re-dissolved it into equal volume of ¹⁸O-water. Prepare the solution fresh.

• Endo H activity assays A set of control experiments should be done before using this protocol, in which a defined amount of the desired glycoprotein is treated with a series of Endo H concentrations according to the manufacturer’s instruction. Gel shifts of high mannose glycans on SDS-PAGE can be used to determine completion of deglycosylation.

• 0.1% (vol/vol) formic acid in 80% (vol/vol) ACN Mix 800 ml of 100% (vol/vol) ACN and 200 ml of water. Add 1 ml of formic acid. Store the solution at room temperature for up to 1 month.

• 0.1% (vol/vol) Formic acid in 5% (vol/vol) acetonitrile Mix 50 ml of 100% (vol/vol) ACN and 950 ml of water. Add 1 ml of formic acid. Store the solution at room temperature for up to 1 month.

• 0.1% (vol/vol) Formic acid Add 1 ml of formic acid to 1 L of water. Store the solution at room temperature for up to 1 month.

• 0.1% (vol/vol) Formic Acid in Acetonitrile Add 1 ml of formic acid to 1 L of Acetonitrile. Store the solution at room temperature for up to 1 month.

Equipment setup

LC setup

The given information is for use with an Easy 1000 or Easy nLCII pump (Thermo), parameters may have to be adjusted with other LC systems. Easy 1000 and Easy nLCII are very similar in nature and only differ in pressure limits and thus type of column that can be used.

	Easy 1000	Easy nLCII
Column	BEH 1.7 μm C18 resin (Waters) pressure loaded into a 100 μm inner diameter × 25 cm length capillary column	4 μm Jupiter C18 (Phenomenex) pressure loaded into a 100um inner diameter × 15cm length capillary column
Mobile phases	A: 0.1% (vol/vol) Formic acid in water B: 0.1% (vol/vol) Formic acid in acetonitrile	A: 0.1% (vol/vol) Formic acid in 5% (vol/vol) acetonitrile B: 0.1% (vol/vol) Formic acid in 80% (vol/vol) acetonitrile
Flow rate	200 nl/min	400 nl/min
Gradient program	A gradient of 5-25% B over 150 min, an increase to 40% B over 50 min, an increase to 100% B over 10 min and held at 100% B for a final 30 min of washing was	Column was initially held at 0%B for 10 min. A gradient of 0–10% B over 10 min, an increase to 60% B over 85 min, an increase to 100% B
	used for 240 min total run time. Column was re-equilibrated with 20 μl buffer A prior to the injection of sample.	over 20 min, held at 100% B for 5 min of washing before returning to 0%B for 5 min and held at 0% B for 5min was used for 140 min total run time.
Sample injection volume	3-5 μl with 1-2 μg sample	10 μl with 1-2 μg sample

Gradient Program for Easy 1000:

Step	Time (min)	% Solvent B
Gradient	0	5
	150	25
	200	40
	210	100
	240	100

Gradient program for Easy nLCII

Step	Time (min)	% Solvent B
Gradient	0	0
	10	0
	20	10
	105	60
	125	100
	130	100
	135	0
	140	0

Mass spectrometer setup

Instrument should be mass calibrated prior to use. Detailed information below was used on Orbitrap Fusion and Orbitrap Elite (Thermo).

	Orbitrap Fusion	Orbitrap Elite
Ionization mode	Positive	Positive
Electrospray voltage	2.5 kV	2.5 kV
MS1 Resolution	120 k	120 k
MS2 Resolution	15 k	15 k
MS1 scan range	400-1500	400-2000
MS1 maximum injection time	50 ms	200 ms
MS2 maximum injection time	250 ms	1000 ms
MS1 AGC target	4e5	1e6
MS2 AGC target	5e4	5e4
Isolation	2 m/z with quadrupole	2 m/z with ion trap
Fragmentation	CID at 35%	CID at 35%
Dynamic Exclusion	5 seconds	60 seconds
Charge states	2-7 selected	+1 and unassigned are rejected
Intensity Threshold	5e4	500
DDA method	3 second cycle time	Top ten

Procedure

Buffer exchange for glycoproteins (Timing ~ 11 hours)

• ◆Critical To minimize sample loss, low protein-binding microcentrifuge tubes and pipet tips should be used during sample preparation. Only volatile salts are used in the protocol, which can be removed by lyophilization, resulting in higher sensitivity due to minimal protein loss. High-purity reagents should be used for sample preparation and MS analysis to minimize signals derived from contaminants.
• ◆Critical If the protein is a lyophilized powder, then directly go to the ‘Denaturation and alkylation of glycoprotein’ section (Step 13).

• 1Insert a Centrifugal Filter-10kDa into a tube.
• 2Rinse the device by adding 100 μl of 100 mM ammonium acetate immediately before use. Seal the tube with the attached cap.
• 3Centrifuge the device at 8000 g for 50 min at 4 °C.
• 4
  Pipette the solution containing approximately 30 μg of target protein into the device. Seal the tube with the attached cap. If the volume of the solution is less than 100 μl, bring final volume to 100 μl with water.

  • ◆
    CRITICAL STEP Protein concentration is estimated with NanoDrop A280.
• 5Centrifuge the device at 8000 g for 50 min at 4 °C.
• 6Pipette 100 μl of 100 mM ammonium acetate into the device.
• 7Centrifuge the device at 8000 g for 50 min at 4 °C.
• 8
  Repeat steps 6 through 7 at least two times.

  • ◆
    CRITICAL STEP In order to complete buffer exchange, it is important to extensively wash the device that contains glycoproteins (at least threetimes) with 100 mM ammonium acetate (pH 6).
  • ◆
    CRITICAL STEP To minimize non-enzymatic deamidation, the acidic buffer, 100 mM ammonium acetate (pH 6), instead of mildly alkaline buffers should be used during buffer exchange.
• 9Wash the filter membrane with 100 μl of 100 mM ammonium acetate (pH 6).
• 10Collect the solution and store it in a low protein-binding microcentrifuge tube.
• 11
  Repeat steps 9 through 10 at least four times and combine the fractions.

  • ◆
    CRITICAL STEP Wash the filter membrane with the buffer at least five times after buffer exchange in order to achieve maximum recovery of proteins.
  • ◆
    PAUSE POINT The solution can be stored up to one week at −80 °C or in liquid nitrogen.
• 12
  Lyophilize 500 μl of the solution at room temperature for at least 5 hours.

  • ◆
    CRITICAL STEP If volatile salts such as ammonium acetate are added to the buffer, it is important to completely remove them with the lyophilizer.

Denaturation and alkylation of glycoprotein (Timing ~ 7 hours)

• 13
  Dissolve the glycoprotein in 100 μl of 8 M urea in 100 mM ammonium acetate (pH 6) and place the solution at room temperature for 1 hour.

  • ◆
    CRITICAL STEP The acidic buffer, 100 mM ammonium acetate, instead of mildly alkaline buffers should be used for sample preparation if possible in order to keep non-enzymatic deamidation to a minimum.
• 14
  Add 2 μl of 500 mM DTT (to a final concentration of ~10 mM) and incubate the solution at 56 °C for 1 hour.

  • ◆
    CRITICAL STEP Spin down all the solution from the wall of microcentrifuge tube before adding DTT.
• 15
  Add 11 μl of freshly prepared 500 mM Iodoacetamide (to a final concentration of ~50 mM) and incubate the solution in the dark at room temperature for 45 min.

  • ◆
    CRITICAL STEP To preserve activity of iodoacetamide, prepare iodoacetamide solution immediately before use as it is unstable. Perform the alkylation step in the dark as iodoacetamide is light-sensitive.
• 16
  Buffer exchange to 100 mM ammonium bicarbonate (pH 8) by using centrifugal filter with a membrane NMWL of 10 kDa.

  • ◆
    CRITICAL STEP In order to complete buffer exchange, it is important to wash the filter at least three times with 100 μl of 100 mM ammonium bicarbonate at 8000 g for 50 min at 4 °C.
• 17
  Dispense the sample into five equal aliquots for the following proteolytic digestion.

  • ◆
    CRITICAL STEP Do not dry glycoproteins after denaturation as they may not be fully soluble in the buffer after lyophilization.
  • ◆
    PAUSE POINT The samples can be stored at −80 °C or in liquid nitrogen for at least 2 weeks.

Protease treatments (Timing ~24 hours)

• 18Five aliquots (containing 6 μg of denatured glycoproteins each) are subjected to treatments with multiple proteases and combinations of proteases involving: Arg-C followed by trypsin (Option A), elastase (Option B), and subtilisin (Option C). Aliquots A-C will later be combined into a “triple digestion” sample (Step 21). The remaining two aliquots are digested with chymotrypsin (Option D) or a combination of trypsin and chymotrypsin (Option E).

(A) Arg-C followed by trypsin

1. Add Arg-C to one aliquot that contains approximately 6 μg of denatured glycoproteins at enzyme/protein ratio of 1:20 (w/w). Bring the final volume to 100 μl with 100 mM ammonium bicarbonate (pH 8). Add DTT and EDTA to final concentrations of 5 mM and 0.2 mM, respectively.

   • ◆
     CRITICAL STEP Arg-C is able to cleave at the C-terminus of arginine residues, including sites next to proline, resulting in increased sequence coverage when combined with trypsin digestion.

2. Incubate the solution at 37 °C for 4 hours.

3. Lyophilize the resulting peptide mixture for at least 3 hours to remove water and volatile salt.

4. Redissolve the peptide mixture in 500 μl of 100 mM ammonium acetate (pH 6).

5. Add sequencing-grade modified trypsin to the solution at the trypsin/protein ratio of 1:10 (w/w).

   • ◆
     CRITICAL STEP Sequencing-grade modified trypsin is able to digest glycoproteins at pH 6 while preserving adequate digestion efficiency.

6. Incubate the reaction at 37 °C for 16 hours.

(B) Elastase

1. Add elastase to the second aliquot of the denatured glycoprotein at an elastase/protein ratio of 1:20 (w/w).

   • ◆
     CRITICAL STEP the utility of triple digestion can generate higher sequence coverage than with any of single enzymes alone.

2. Bring the final volume to 500 μl with 100 mM ammonium bicarbonate (pH 8).

3. Incubate the reaction at 37 °C for 16 hours.

(C) Subtilisin

1. Add subtilisin to the third aliquot of the denatured glycoprotein at a subtilisin/protein ratio of 1:20 (w/w).

2. Bring the final volume to 500 μl with 100 mM ammonium bicarbonate (pH 8).

3. Incubate the reaction at 37 °C for 4 hours.

   • CRITICAL STEP Do not incubate the reaction longer than 4 hours in order to obtain appropriate lengths of peptides for MS detection.

(D) Chymotrypsin

1. Add chymotrypsin to the fourth aliquot of the denatured glycoprotein at a chymotrypsin/protein ratio of 1:13 (w/w).

   • ◆
     CRITICAL STEP Do not add too much chymotrypsin to the solution as it can be self-digested, which will suppress ESI signals of analytes.

2. Bring the final volume to 500 μl with 100 mM ammonium bicarbonate (pH 8).

3. Incubate the reaction in a 30 °C water bath for 10 hours.

(E) A combination of trypsin and chymotrypsin

1. Add trypsin and chymotrypsin to the fifth aliquot of the denatured glycoprotein at enzyme/protein ratios of 1:20 (w/w) and 1:13 (w/w) respectively.

2. Bring the final volume to 500 μl with 100 mM ammonium bicarbonate (pH 8).

3. Incubate the reaction at 37 °C for 16 hours.

   • ◆
     PAUSE POINT The peptide mixtures derived from combination proteolytic digestion can be stored at −80 °C or in liquid nitrogen for at least 2 weeks.

Denaturation of proteases (Timing ~10 hours)

• 19Lyophilize the peptide mixtures derived from each of the five protease digestions (Step 18 Options A-E) for at least 5 hours.
• 20Redissolve each sample in 100 μl of water.
• 21Combine the peptide mixtures derived from Step 18 Option A-C into a “triple digestion” sample.
• 22Heat the combined triple digestion sample (Step 21) and the samples generated from digestion with chymotrypsin (Step 18 Option D), and a combination of trypsin and chymotrypsin (Step 18 Option E) at 100 °C for 30 seconds.
• 23Cool the samples at room temperature for 30 seconds.
• 24
  Repeat steps 22 and 23 at least five times.

  • ◆
    CRITICAL STEP In order to completely deactivate the proteases used, steps 22 and 23 should be repeated at least five times. Any remaining active proteases will accelerate incorporation of ¹⁸O-water into the C-termini of peptides during the following PNGase F treatment conducted in ¹⁸O-water.
• 25
  Lyophilize the samples for at least 3 hours to remove any remaining volatile salts in the samples.

  • ◆
    CRITICAL STEP Complete removal of volatile salts in the samples is important for the following Endo H treatment, which has optimal activity at pH 5.5.
  • ◆
    PAUSE POINT The peptide mixtures can be stored at −80 °C or in liquid nitrogen for at least 2 weeks.

Sequential endoglycosidase treatment (Timing ~7 hours)

• 26Redissolve the three samples in 20 μl of 100 mM ammonium acetate (pH 5.5).
• 27
  Add Endo H to the peptide mixtures at a minimum enzyme/glycoprotein ratio of 250 NEB units/10 μg.

  • ◆
    CRITICAL STEP The quantity of Endo H needed for deglycosylation may vary from protein to protein. A set of control experiments should be done before using this protocol (see Reagent Setup).
• 28
  Incubate the reaction at 37 °C for 1 hour.

  • ◆
    PAUSE POINT The Endo H-treated peptides can be stored up to one week at −80 °C or in liquid nitrogen.
• 29
  Lyophilize the Endo H-treated samples for at least 3 hours immediately before use. In the meantime, proceed to Step 30.

  • ◆
    CRITICAL STEP Complete removal of ammonium acetate in the samples ensures the following PNGase F treatment to completion.
• 30Lyophilize the PNGase F solution for at least 1 hour immediately before use and redissolve the PNGase F enzyme in the same volume of ¹⁸O-water (see Reagent Setup).
• 31
  Add 20 μl of 100 mM ammonium bicarbonate (pH 8) prepared with ¹⁸O-water to the PNGase F solution.

  • ◆
    CRITICAL STEP Steps 31-32 should be done as quickly as possible to reduce contact of the reaction mixture with air.
• 32
  Add PNGase F to the Endo H-treated peptide mixtures at a minimum enzyme/glycoprotein ratio of 500 NEB units/10 μg.

  • ◆
    CRITICAL STEP The quantity of PNGase F that is needed for deglycosylation can be determined by treating a defined amount of the desired glycoprotein with a series of PNGase F concentrations. Gel shifts of N-glycans on SDS-PAGE can be used to determine completion of deglycosylation.
• 33Seal the microcentrifuge tube that contains the mixture.
• 34Incubate the reaction at 37 °C for 1 hour.
• 35
  Dispense into aliquots than contain ~2 μg of peptides and store them in liquid nitrogen immediately.

  • ◆
    PAUSE POINT The samples can be stored in liquid nitrogen for at least one month.

LC-MS/MS analysis (Timing 6-48 hours per glycoprotein)

• 36
  Set up LC-MS/MS system to characterize deglycosylated peptides as described under EQUIPMENT SETUP.

  • ◆
    CRITICAL STEP Each glycoprotein is digested in two or three technical replicates and analyzed by the same MS.
  • ◆
    TROUBLESHOOTING

Data analysis (Timing variable; hours to one day per glycoprotein)

• 37Extract the MS1 and MS2 spectra from the MS raw files using the spectrum-converting software RawConverter (Supplementary Fig. 2a).
• 38Add the MS raw files to the “files to covert” of RawConverter.
• 39Set “Experiment Type” as data dependent. Enable “Select monoisotopic m/z in DDA”. “Output formats” should be set as “MS1, MS2, and MS3”, in which both MS1 and MS2 data are extracted from the MS raw files.
• 40Use a text editor such as Notepad++ to prepare a file containing the sequences of target glycoproteins in Fasta format (Supplementary Fig. 2b).
• 41Add the resulting file to a predefined database, such as the European Bioinformatic Institute (IPI) Bos Taurus protein database, with the “database” of IP2 (Supplementary Fig. 2c).
• 42
  Set the parameters as: Source, Uniprot; Organism Name, Bos Taurus; Generate reverse (decoy) sequences, yes; Add contaminant proteins, No.

  • ◆
    CRITICAL STEP Reverse (decoy) sequences should be generated and included in the final database in order to estimate peptide probabilities and FDRs.
• 43Upload the resulting file that contains the sequences of target glycoproteins to the “database” of IP2.
• 44Upload the database to IP2.
• 45Upload the MS1 and MS2 files to IP2 (Supplementary Fig. 2d and 3e).
• 46Start a ProLuCID search in IP2-Intergreted Proteomics Pipeline (version 5.0.1) software package.
• 47Set the mass tolerance at 50 p.p.m. for precursor ions and 20 p.p.m. for fragment ions (MS2 spectra is detected in the Orbitrap). No enzyme specificity is considered for searching. Set carboxyamidomethylation (+ 57.02146 C) as a fixed modification. Set oxidation (+ 15.9994 M), deamidation (+ 2.988261 N), GlcNAc (+ 203.079373 N), and pyroglutamate formation from N-terminal glutamine residue (- 17.026549 Q), as variable modifications (Supplementary Fig. 2f).
• 48
  Filter the results generated from the ProLuCID search by using DTASelect (version 2.0). The parameters are set as: minimum number of peptide per protein ≥ 2, spectrum false positive rate ≤ 0.05, and precursor delta mass cutoff ≤ 10 p.p.m. (Supplementary Fig. 2g).

  • ◆
    CRITICIAL STEP Sequence coverage of target proteins should be >95%.
  • ◆
    Troubleshooting
• 49Filter the results generated from DTASelect with the software “Glyco_motif_filter” to remove those peptides with N+3 and/or N+203 modifications that are not located at the motif (N-X-S/T, X can be any amino acid residue except proline).

Figure 3.

Validation of sequential endoglycosidase treatment. (a) Scatter plot of the site-specific N-glycan processing of invertase produced by the yeast S. cerevisiae. (b) Color-coded bar graph of the site-specific N-glycan processing of invertase. (c) Scatter plot of the site-specific N-glycan processing of alpha-1-acid glycoprotein. (d) Color-coded bar graph of the site-specific N-glycan processing of alpha-1-acid glycoprotein. A set of peptides with N+0, N+3, and N+203 modifications were displayed only when at least one of the three had a peak area of at least > 5E8. Data was obtained from 6 independent experiments. Mean ± SEM were plotted. Fig. 3a and 3b adapted with permission from ref. 10, Nature Publishing Group.

CRITICAL STEP

All peptides with N+203 modification that are not located at the consensus motif should be manually checked before they are removed. Asparagine residues that are not located at the motif should be considered as potential glycosylation sites when multiple spectra hits with N+203 modification that do not contain the motif are consistently detected. Further verification is needed for these potential glycosylation sites.

• 50
  Start a label-free analysis using Census (another component of the IP2 software package). The parameters are set as: enabled “find missing peptide”, mass tolerance ≤ 10 p.p.m., retention time tolerance ≤ 0.1 min (Supplementary Fig. 2h).

  • ◆
    CRITICAL STEP Ion injection time is used to further normalize the resulting peak area.
• 51
  Determine the abundance of each peptide from each raw file by the sum of the ion intensity peak area over all identified charge states.

  • ◆
    CRITICAL STEP In order to improve the accuracy of the method, a set of peptides with N+0, N+3, and N+203 modifications are considered only when at least one of the three has a peak area of at least 5E8. Of note, approximately two microgram of purified glycoproteins are loaded onto the column in the present study. This value was empirically determined as optimal to distinguish information from spectral noise and will vary from instrument to instrument. A control experiment should be done by using well-characterized model glycoproteins, such as invertase (occupied by high mannose glycans) and alpha-1-acid glycoprotein (occupied by complex type glycans), before setting the peak area threshold. Peak area values will be dependent on the type of LC and mass spectrometer used and the appropriate threshold will need to be determined for other instrument types.
  • ◆
    TROUBLESHOOTING
• 52Combine the data derived from two or three technical replicates after analysis of each MS run separately.

Timing

• Steps 1-12, buffer exchange for glycoproteins: 11 hours

• Steps 13-17, denaturation and alkylation of glycoprotein: 7 hours

• Step 18, combination proteolytic digestion: 24 hours

• Steps 19-25, denaturation of proteases: 10 hours

• Steps 26-35, sequential endoglycosidase treatment: 7 hours

• Step 36, LC-MS/MS analysis: 6-48 hours per glycoprotein

• Steps 37-52, data analysis: variable; hours to one day per glycoprotein (depending on complexity)

Troubleshooting

Troubleshooting advice can be found in Table 1.

Table 1.

Troubleshooting table.

Step	Problem	Possible reason	Solution
36	No peptides detectable	The LC-MS instrument is not performing properly	Make sure that the LC-MS instrument is well-calibrated and working properly according to specifications
48	Low sequence coverage of target glycoproteins	Glycoproteins may not have been washed out completely after buffer exchange (step 16)	Wash the filter membrane at least five times after buffer exchange to maximize recovery of target glycoproteins
48	No peptides detected after combination proteolytic digestion	Proteases stored in the buffer at −80 °C for too long may not be active (step 18)	Make fresh protease solution
48	Too many peaks that belong to non-enzymatic deamidation are detected	The microcentrifuge tube that contains 18O-water is not completely sealed or exposed to air for too long (step 30-33)	Seal the microcentrifuge tube completely and place it into a desiccator. Perform steps 30-33 as quickly as possible to reduce contact of the reaction mixture with air
51	The output file generated by Census is empty or cannot be downloaded	The MS1 file is not extracted correctly by RawConverter or the names of target glycoproteins are not recognized by Census	Extract the MS1 spectra from raw files again using RawConverter. Take out spaces that is not recognized by Census from the names of target glycoproteins

Anticipated results

This protocol is used to determine site-specific N-glycan processing of glycoproteins. In the most widely used strategy, glycoproteins are digested with specific proteases such as trypsin, resulting in (glyco)peptides that are suitable for LC-MS/MS analysis. Glycosylated peptides, however, have much lower ionization efficiency during MS analysis relative to peptides, and thus milligram quantities of materials are typically used for typical glycoproteomics methods^29,61. Characterization of glycopeptides with multiple glycosites by MS/MS is still challenging even with combination of different types of fragmentation techniques²⁸. Quantitative measurement of glycopeptides is complicated by the fact that ionization efficiencies of glycopeptides differ with variable glycoforms³⁸. This protocol describes an alternative way to overcome these problems by the use of combination proteolytic digestion followed by sequential endoglycosidase treatment.

Validation of sequential endoglycosidase treatment

One of the distinguishing features of this protocol is the use of sequential treatment of glycopeptides with endoglycosidases, Endo H followed by PNGase F, to create unique mass signatures for glycosites that have no N-glycan, high mannose type glycan, or complex type glycan. This strategy converts the glycoproteomics analysis to a proteomics analysis, resulting in higher sensitivity of the protocol (needs only 30 μg of sample for a complete analysis). The key for success is that the sequential endoglycosidase treatments go to completion, avoiding mis-assignment of N-glycan processing status. To this end, we applied the protocol to assess site-specific N-glycan processing of two well-characterized model glycoproteins, invertase produced by the yeast S. cerevisiae and alpha-1-acid glycoprotein from bovine serum⁷¹. Glycosites on invertase are occupied by under-processed oligomannose and those on alpha-1-acid glycoprotein are fully processed complex type glycosylation (Fig. 3). All N-glycosites on both glycoproteins were identified with multiple MS/MS spectra ranging from 54 to more than 10,000 per site (Supplementary Table 2 and 3). Low percentages of spectra hits that contain non-enzymatic deamidation or ¹⁸O-incorporation into the C-termini of peptides were found in total of all spectra hits identified, which is attributed to the complete denaturation of proteases used (Supplementary Table 4 and 5). As described in the procedure section as well as the previous study¹⁰, a set of peptides with N+0, N+3, and N+203 modifications were considered only when at least one of the three had a peak area of at least 5E8. As expected, the 14 N-glycosites of invertase were identified to be entirely high-mannose type glycosylation, and site occupancy was > 90% for all glycosites except the sites N64 and N275 (Fig. 3a and 3b). In contrast, the five N-glycosites of alpha-1-acid glycoprotein were completely complex-type glycosylation, and site occupancy was > 98% for all five sites (Fig. 3c and 3d). These results indicated that sequential endoglycosidase treatment reached completion.

Validation of MS detection of glycotypes

Another major assumption is that the endoglycosidase-treated peptide glycosites that are unoccupied or occupied by high mannose glycans or complex type glycans are detected equally during MS analysis. To test this assumption, the HIV-1 Env trimer, BG505 SOSIP.664, expressed in the presence of kifunensine (high mannose only) was selected as a model protein (hereafter referred to as Kif_BG505). N-glycans on Kif_BG505 were first removed by using sequential endoglycosidase treatment, and as expected, glycosites comprised > 95% high mannose type glycosylation (green bars), indicating that the kifunensine treatment was effective and Kif_BG505 had high mannose glycans (Fig. 4a). On the other hand, PNGase F treatment only was also applied to release N-glycans on Kif_BG505, resulting in homogenous N+3 (> 98% of purple bars, Fig. 4b). The resulting two samples were then mixed at a molar ratio of 1:1 in order to assess MS detection of glycotypes (Fig. 4c). Both peptides with N+3 and N+203 modifications are detectable for each glycosite, with a ratio of 1.0 to 1.2, suggesting slightly increased sensitivity for peptides with the N+3 modification. Synthetic peptides that carry asparagine (unoccupied), aspartic acid (PNGase F treated), and N-acetylglucosamine-linked asparagine residues, respectively, at the glycosylation site, display similar ionization efficiency during ESI-MS analysis, further indicating that the protocol is able to semi-quantitatively assess site-specific N-glycan processing for glycoproteins⁶¹.

Figure 4.

Validation of MS detection of glycotypes. (a) Site-specific N-glycan processing of Kif_BG505 that was treated with Endo H followed by PNGase F. (b) Site-specific glycosylation of Kif_BG505 that was treated with PNGase F only. (c) MS detection for peptides that contain N+3 and N+203 modifications at a molar ratio of 1:1. Peptides that had potential glycosites, but were not glycosylated were not included. The proportions of high mannose and complex type glycans at those glycosites highlighted in yellow were assigned based on the proportion of spectra hits since peak area did not reach the threshold of 5E8. Data was obtained from 9 independent experiments. Mean ± SEM were plotted. Image adapted with permission from ref. 10, Nature Publishing Group.

Examples of the protocol

Although the protocol was initially developed for analysis of site-specific N-glycan processing of the HIV Env trimer¹⁰, it is applicable to analysis of site-specific N-glycan processing of recombinant glycoprotein therapeutics (Fig. 5a and 5b), serum glycoproteins (Fig. 5c and 5d), and soluble or membrane bound envelope glycoproteins from viruses (Fig. 6). It is also likely to be useful in characterization of glycoprotein processing in more complex systems like whole cells due to its high sensitivity.

Figure 5.

Application of the protocol for characterization of site-specific N-glycan processing of recombinant glycoprotein therapeutics and serum glycoproteins. Site-specific N-glycan processing of recombinant glycoprotein therapeutics, including IgG (a) and IgM (b), as well as serum glycoproteins, including transferrin (c) and fetuin (d). A set of peptides with N+0, N+3, and N+203 modifications were displayed only when at least one of the three had a peak area of at least > 5E8. Data was obtained from at least 6 independent experiments. Mean ± SEM were plotted. Fig. 5d adapted with permission from ref. 10, Nature Publishing Group.

Figure 6.

Application of the protocol for characterization of site-specific N-glycan processing of virus envelope glycoproteins. Site-specific N-glycan processing of recombinant envelope glycoproteins derived from viruses, including (a) the prefusion-stabilized spike glycoprotein ectodomain of Middle East respiratory syndrome coronavirus (MERS-CoV S-2P protein), (b) influenza virus hemagglutinin from H3N2 strain A/Victoria/361/2011, and (c) HIV-1 envelope glycoprotein. The proportions of high mannose and complex type glycans at those glycosites highlighted in yellow were assigned based on the proportion of spectra hits since peak area did not reach the threshold of 5E8. Data was obtained from at least 6 independent experiments. Mean ± SEM were plotted. Fig. 6b and 6c adapted with permission from ref. 10, Nature Publishing Group.

Site-specific N-glycan processing of recombinantly produced therapeutic glycoproteins, including IgG and IgM, was determined (Fig. 5a and 5b). Human serum IgG, which contains IgG1-4 with IgG1 and IgG2 as the major isotypes, was found to be entirely complex type glycosylation, consistent with previous studies (Fig. 5a)^32,49,52. Of the five N-glycosites on the IgM, which is the major antibody produced in the primary immune response, three sites were shown to be completely occupied by complex type structures (N171, N332, and N395), whereas other two sites, N402 and N563 were primarily high mannose type glycosylation (Fig. 5b). The glycosite N563 that is proximal to the CH4 domain and thus is a poor substrate for oligosaccharyltransferase was found to be partially glycosylated, in agreement with previous studies^24,72.

Abnormal glycosylation of serum glycoproteins is a common feature in various human diseases, such as cancers and congenital disorders of glycosylation (CDGs). In particular, serum transferrin was first used to diagnose abnormal glycosylation in CDG patients back in the 1990s⁷³. Normal transferrin has two N-glycosylation sites, each of which is fully occupied^74,75, whereas in type I CDGs, an increase of mono-glycosylated transferrin was found due to defects of oligosaccharide assembly and transfer to glycoproteins in those patients⁷⁶. Analysis of site-specific N-glycan processing of commercially available human serum transferrin revealed that the two glycosites of this protein were entirely complex type glycosylation (Fig. 5c), in line with the previous studies^74,75. Another serum glycoprotein, fetuin, was also found to be entirely complex type glycosylation, with full occupancy of two sites (N99 and N156) and partial glycosylation of the third glycosite (N176, 89% of site occupancy, Fig. 5d).

Membrane-bound envelope glycoproteins of various viruses, such as MERS-CoV S-2P protein and HIV Env trimer, are the target of neutralizing antibodies, and thus are the focus of vaccine development^55,70. N-linked glycans on those envelope glycoproteins serve as a shield to protect the underlying protein from immune surveillance, and thus confound development of effective vaccines to those viruses⁵⁴. Importantly, some neutralizing antibodies to those viruses have glycan-dependent epitopes^15,57,70, suggesting that vaccine design efforts would benefit greatly from understanding the N-glycan processing status at each glycosylation site. MERS-CoV S-2P protein is a large trimer (~600 kDa) with ~25 N-linked glycans per monomer, each of which comprises two noncovalently associated subunits, S1 and S2⁷⁰. Characterization of site-specific N-glycan processing of a prefusion-stabilized MERS-CoV S protein ectodomain (MERS S-2P) revealed that all 23 glycosites on the protein were fully occupied except the N104 site (Fig. 6a). High-mannose glycans were predominantly found in the S1-NTD (residues 18-353), whereas other regions of the S protein, including the RBD (367-588), the two subdomains (589-751), and S2 (752-1291), contained glycosites occupied largely by complex-type glycans. The glycan N1176, which is in the epitope for antibody G4, and was reported to mask antibody recognition⁷⁰, was found to have a complex-type structure. Of note, we did not observe that the proteases used had biases on specific glycotypes (Supplementary Fig. 2). Of the 12 glycosites on the recombinant influenza haemagglutinin (HA) of A/Victoria/361/2011, three were >85% high mannose (N45, N165, and N285), four were fully complex type glycosylation (N22, N38, N63, and N483), and the rest were occupied by a mixture of high mannose and complex type glycans (Fig. 6b). It is striking to observe that the site N122 was not occupied while another site N144 was only 32% occupied on this HA glycoprotein. We also applied the protocol to the benchmark HIV Env trimer, BG505 SOSIP.664, resulting in identification of all 28 glycosites with up to 2000 spectra hits per glycosite (Fig. 6c). All 28 glycosites were > 90% occupied except the sites N185e, N197, N618, and N625. Of those that were largely occupied, fourteen were >75% high mannose, four were >75% complex type glycosylation, and six other sites had a mixture of high mannose and complex type glycans. In particular, the glycosites, N295, N332, N339, N386, and N392, in the high mannose patch region, were found to be occupied predominantly by under-processed oligomannose, consistent with previous studies^14,77. The N160 glycan, which is critical for binding of bnAbs PG9 and PG16 to HIV Env, was composed predominantly of high mannose structures, confirming the glycan composition at this site described in previous structural studies^58,78,79. Interestingly, the high mannose and complex type glycans identified at each glycosylation site of BG505 SOSIP.664 matched the pathway of N-glycan processing, in which high mannose structures are first trimmed from Man₉ to Man₅ before addition of the terminal monosaccharides that define complex type/hybrid glycans, when compared to the results of the same protein obtained on intact glycopeptide level²⁹. Thus, the sites were predominantly occupied by Man₉ if they are 100% high mannose glycosylation, while other sites were occupied by mixtures of processed high mannose structures (Man₈ to Man₅) and simple complex type structures if they were occupied by a mixture of high mannose and complex type glycans¹⁰.

We believe this protocol will be of wide interest to the proteomics and glycomics fields, and will be used by many outside those fields who want to gain high level information about the glycoproteins they investigate.

Supplementary Material

Supporting Info

Supplementary Figure 1 Identification of peptides with multiple glycosylation sites with the protocol. MS/MS spectra and fragment assignment of the (a) diglycosylated, (b) tri-glycosylated, and (c) tetra-glycosylated peptides derived from BG505 SOSIP.664.

Supplementary Figure 2 LC-MS/MS data processing with RawConverter and IP2. (a) Screenshot of RawConverter for extracting the MS1 and MS2 spectra from MS raw files, (b) screenshot of Notepad++ for preparing a file that contains the sequences of target glycoproteins in Fasta format, (c) screenshot of IP2 for generating and unloading the database, (d) and (e) screenshot of IP2 for uploading the MS1 and MS2 files, (f) screenshot of IP2 for a ProLuCID search, (g) screenshot of IP2 for DTASelect, and (h) screenshot of IP2 for a label-free analysis using Census.

Supplementary Figure 3 Scatter plot of site-specific N-glycan processing of MERS-CoV S-2P protein with a breakdown on the used proteases: (a) triple digestion, (b) the combination of trypsin and chymotrypsin, (c) chymotrypsin. The proportions of high mannose and complex type glycans at those glycosites highlighted in yellow were assigned based on the proportion of spectra hits since peak area did not reach the threshold of 5E8. Data was obtained from 6 independent experiments. Mean ± SEM were plotted.

Supplementary Table 1 Identification of the peptide glycosite N155 on MERS-CoV S-2P protein by using the protocol.

Supplementary Table 2 The number of MS/MS spectra that could be detected from each glycosylation site of invertase produced by the yeast S. cerevisiae.

Supplementary Table 3 The number of MS/MS spectra that could be detected from each glycosylation site of alpha-1-acid glycoprotein from bovine serum.

Supplementary Table 4 Percentages of spectra hits with non-enzymatic deamidation in total of all spectra hits identified.

Supplementary Table 5 Percentages of spectra hits with ¹⁸O-incorporation into the C-termini of peptides in total of all spectra hits identified.

EDITORIAL SUMMARY.

This protocol describes a semi-quantitative glycoproteomics method using sequential treatment with endoglycosidases to create unique mass signatures to determine glycan occupancy and proportion of high mannose and complex glycans at each glycosite.