Analysis of mucin-domain glycoproteins using mass spectrometry

Abstract

Mucin-domain glycoproteins are characterized by their high density of glycosylated serine and threonine residues, which complicates their analysis by mass spectrometry. The dense glycosylation renders the protein backbone inaccessible to workhorse proteases like trypsin, the vast heterogeneity of glycosylation often results in ion suppression from unmodified peptides, and search algorithms struggle to confidently analyze and site-localize O-glycosites. We have made a number of advances to address these challenges, rendering mucinomics possible for the first time. Here, we summarize these contributions and provide a detailed protocol for mass spectrometric analysis of mucin-domain glycoproteins.

Protocol 1: Enrichment of mucin-domain glycoproteins

Protocol 2: Enzymatic digestion of proteins

Protocol 3: Mass spectrometry data acquisition

Protocol 4: Data analysis of glycopeptides

INTRODUCTION:

Protein glycosylation is a ubiquitous post-translational modification found on over 50% of the proteome and on over 80% of extracellular proteins.(Varki & Gagneux, 2015) Diseased cells have significantly altered glycosylation patterns when compared to normal cells, often expressing glycans with new skeletal arrangements and/or enhancement/depletion of specific carbohydrates. These changes in glycosylation have been linked to the progression of numerous diseases including cancer, inflammatory bowel disease, and cystic fibrosis.(Kudelka et al., 2015; Kufe, 2009; Varki et al., 2015) For instance, evidence suggests that the changes in cell surface glycosylation confer increased mobility and immunological evasion to tumor cells, in part due to loss of cell polarization and enhanced metastasis.(Hollingsworth & Swanson, 2004; Kufe, 2009) While aberrant glycosylation has been recognized for decades, we still do not understand how diseased cells modulate the structures and site-specificity of glycosylation. This gap in knowledge is largely attributable to the difficulties associated with studying cell-surface glycans and their associated structures.

A major focus of the Malaker laboratory is to develop mass spectrometry (MS)-based methods that allow for the glycoproteomic sequencing of mucins (“mucinomics”), a class of densely O-glycosylated proteins.(Bagdonaite et al., 2022; Calle et al., 2021; Chongsaritsinsuk, Steigmeyer, Mahoney, Rosenfeld, Lucas, Ince, et al., 2023; Hollenhorst et al., 2023; Ince et al., 2022; Malaker et al., 2019, 2021, 2022; Rangel-Angarita & Malaker, 2021; Shon et al., 2021) Aberrant mucin glycosylation and expression has been linked to nearly all epithelial cancers including ovarian, breast, and colorectal carcinomas.(Bhatia et al., 2019; Hollingsworth & Swanson, 2004) Mucins can have thousands of dense O-glycosites, each of which can bear hundreds of different glycan structures, which creates a combinatorial explosion of possible proteoforms.(Malaker et al., 2022) This complexity makes studying mucins particularly challenging because (a) existing glycoproteomic software is unreliable, thus requiring MS data to be manually curated, (b) glycopeptides are less abundant when compared to unmodified counterparts, necessitating enrichment; and (c) the dense glycosylation in mucins renders them inaccessible to workhorse proteases like trypsin.(Malaker et al., 2019; Rangel-Angarita et al., 2022; Rangel-Angarita & Malaker, 2021; Riley et al., 2021; Shon et al., 2021) We have made substantial progress toward addressing these challenges by establishing novel technologies for performing O-glycoprotein analysis by MS.

Several recent developments in the field of glycoproteomics have allowed for the MS analysis of methods. First, several mucinases have been characterized, which comprise a class of proteases that specifically degrade mucin domains and enable MS analysis of the densely glycosylated peptides.(Chongsaritsinsuk, Steigmeyer, Mahoney, Rosenfeld, Lucas, Smith, et al., 2023; Malaker et al., 2019; Shon et al., 2020, 2021) Additionally, we have developed and optimized an enrichment technique to selectively isolate mucin-domain glycoproteins from non-mucin proteins.(Mahoney et al., 2023; Malaker et al., 2022) Here, we first describe our protein-level mucin enrichment technique, wherein we conjugate inactive point mutants of mucinases to solid support and use this to selectively isolate mucins from complex samples such as cell lysate and human serum. In the second protocol a mucinase and/or trypsin digest of proteins is detailed, followed by the optimal MS settings for O-glycopeptide analysis. Finally, we detail how to best use search algorithms as a starting point for data analysis, including our detailed procedures for manual annotation. Overall, we hope this serves as a valuable resource to those performing mucin glycoproteomics.

BASIC PROTOCOL 1

Basic protocol title: Enrichment of mucin-domain glycoproteins

In complex samples such as cell lysates, serum, plasma, or ascites fluid, it is often beneficial to perform a protein-level enrichment for mucin glycoproteins. This will assure a depletion of unmodified proteins, such as albumin, which will often overtake signal in the mass spectrometer and impede analysis of relevant glycopeptides. To do this, we employ an inactive point mutant of mucinase StcE, which we have demonstrated can selectively enrich mucins from various samples.(Mahoney et al., 2023; Malaker et al., 2022; Rangel-Angarita & Malaker, 2021)

Materials:

• Human serum (Innovative Research, ISER10ML)

• PBS (Gibco, 10010–023)

• 10 x TBS (BioRad 1706435)

• Tris pH 7.4 (AmericanBio, AB14044–01000)

• Sodium deoxycholate (Research Products International, D91500)

• EDTA (Invitrogen, 15575020)

• StcE mutant (StcE^E447D, Millipore Sigma SAE0212)

• Micropipettes (1 mL, 200 μL, and 20 μL)

• Water purification system (Thermo Scientific, 50132370)

• 0.22 μm syringe filter (Millipore, SLGVR33RS) or 0.45 μm spin filter (Thermo Scientific, F2517–6)

• 2 mL microcentrifuge tube (Eppendorf, 22363352)

• NHS-ester bead slurry (GE Healthcare, 17090601)

• Centrifuge (Eppendorf, 5427R)

• 1.5 mL microcentrifuge tube (Eppendorf, 22364111)

• Tube rotator (Thermo Scientific, 88881001)

• Heated mixing block (Eppendorf ThermoMixer, 5382000023)

• 4 °C refrigerator or cold room (many suitable options available; example: Fisher Scientific FBV72RPGA)

Preparation of biomaterials

The protocol can be used for a variety of biofluids, but here we focus on preparation of samples from human serum. Prior to proceeding with the protocol, the biofluid should be submicron filtered (0.45 μm or 0.22 μm). For smaller samples (<5 mL), a lower volume submicron filter should be used to maximize recovery. Sample concentrations should be between 0.5–50 mg/mL.

Derivatization of beads with inactive point mutant of StcE

For all steps removing supernatant, displace as few beads during supernatant removal as possible. In general, this means removing the supernatant slowly, making sure not to push down on the pipette plunger, and leaving a small amount of supernatant over the beads. If the bead pellet is disturbed, centrifuge again at the same speed and time before trying to remove more supernatant. Experimenting with different tip sizes can help; some people have better results with small tip ends (e.g., gel loading tips), while others have better results using large diameter tips (e.g., p1000).

• 1StcE ^E447D can be made recombinantly as described in Malaker et al., 2019 or purchased directly from Sigma-Aldrich (part number SAE0212).
• 2Move 1 mL NHS-ester bead slurry to a 2 mL microcentrifuge tube

  Make sure the beads are fully suspended before pipetting. First shake the bottle, then pipette mix before drawing solution. Depending on the number and size of enrichments to be performed, the bead volume can be scaled alongside protein amount. Make sure that each enrichment will have at least 30–50 μL packed bead volume for easier supernatant removal.

  Different amine-reactive functional groups and brands can be used for this conjugation with minimal effect on results.

• 3Centrifuge the bead slurry at 2500 rcf for 2–3 minutes to pellet beads and remove the supernatant.

  We recommend allowing the beads to sit on a benchtop for an additional 30 seconds before removing supernatant.

• 4Rinse 3 x with 1 mL of PBS, discarding the supernatant after each rinse.
• 5Add 1–1.5 mg of StcE^E447D in solution to bead slurry

  Make sure the volume is sufficient to allow the liquid to move in the tube– if the volume is too small (~200 μL), the sample will not leave the bottom of the tube and will not be adequately mixed. Dilute with PBS as necessary.

  The procedure is amenable to a wide array of buffers (e.g., PBS, water, etc.) within a pH range of 3–9. Tris should not be used as this will dissociate the enzyme from the solid support. Many dilutions of the inactive enzyme can be used, as long as it is concentrated enough to fit in the Eppendorf tube; typically, the enzyme is stored at a concentration of ~1 mg/mL so that 1–1.5 mL is added to the tube.

• 6Rotate at 4 °C for 6 hours-overnight.
• 7Centrifuge the bead slurry at 2500 rcf for 2–3 minutes to pellet beads and remove the supernatant.

  If the beads are not firmly pelleted, repeat the spin or allow the beads to sit on the bench for 1 minute to settle. Different manufacturers or bead types may require higher speeds and/or times to pellet properly. Make sure not to exceed the manufacturer recommendations for centrifugation speed or the beads may be damaged.

• 8Add 1 mL of 100 mM Tris, pH 7.4, to bead slurry.

  This step assures that all NHS-esters are capped and unreactive to prevent non-specific binding to the solid support.

• 9Rotate at 4 °C for 30 minutes.
• 10Remove supernatant and repeat blocking step with 100 mM Tris pH 7.4 overnight.

  Steps 8–9 can be performed in two one-hour blocking steps with similar efficiency.

• 11Centrifuge the bead slurry at 2500 rcf for 2–3 minutes to pellet beads and remove the supernatant.
• 12Rinse 3 × 1 mL of TBS (20mM Tris, 500mM NaCl).

  It is common for solid supports to require a longer centrifugation time in high salt solutions.

• 13Add 800 μL of 20 mM Tris pH 7.4, 100 mM NaCl to beads.
• 14Beads can be stored at 4 °C for at least one month.
• 15Before using, we recommended rinsing beads with 1 mL of 20 mM Tris pH 7.4, 100 mM NaCl.

  This should remove any protein that has leached from the beads during storage.

Enrichment of mucins from complex samples

• 16Reconstitute 500 μg of serum to a final volume of 400 μL with 20mM Tris, 100mM NaCl in a 1.5 mL microcentrifuge tube.
• 17Add 10 μL of 500 mM EDTA for a final concentration of ~10 mM EDTA in the final sample.

  The inactive point mutant retains a low level of activity; the EDTA sequesters any metal so that the metalloprotease cannot erroneously digest the mucin proteins. If the biofluid contains higher levels of divalent cations than human serum, the EDTA concentration should be adjusted accordingly.

• 18Add 100 μL of the StcE-conjugated beads to the sample.
• 19Rotate 6 hrs to overnight.
• 20Spin beads at 2,500 rcf for 4 mins at 4 °C and save the supernatant.

  The supernatant can be used to investigate binding efficiency if something has gone wrong in earlier steps.

• 21Rinse 3x with 1 mL of TBS to remove nonspecific binding.

  Here, add solution directly onto beads to disrupt the settling, and only vortex if there is a pellet remaining.

• 22Transfer to a new microcentrifuge tube after the last rinse with TBS.
• 23Rinse 2x with 1 mL of 20mM Tris.
• 24Add 100 uL 0.5% (5 mg/mL) sodium deoxycholate in 20mM Tris.

  Concentrations from 0.5–1.5% SDC show similar results, but the lower concentrations are more easily removed in the later steps.

• 25Boil the beads at 95 °C, 1100 rpm for 5 minutes in the heated mixing block (ThermoMixer).

  Here, make sure that the shaking is strong enough to keep the beads suspended in the solution. If the beads have settled, vortex the tube for 30 seconds in 5 second bursts before pelleting.

• 26Spin beads at 2,500 rcf., 4 mins, 4 °C, and save the supernatant/eluant.
• 27Repeat steps 9–11.

  The sample is now ready for proteolytic digestion, but SDC will need to be precipitated from the sample prior to MS analysis, detailed in Protocol 2.

BASIC PROTOCOL 2

Basic protocol title: Enzymatic digestion of mucin-domain glycoprotein(s)

Bottom-up proteomics involves the digestion of proteins into peptides prior to mass spectrometry analysis. The most common enzyme for this purpose is trypsin, however, most mucin-domain glycoproteins are incompletely digested with trypsin due to the dense glycosylation and lack of Arg/Lys residues.(Malaker et al., 2019) Thus, we and others have introduced a series of mucinase and O-glycoprotease enzymes that selectively cleave within the mucin domains with varying specificity.(Haurat et al., 2020; Malaker et al., 2019; Medley et al., 2022; Shon et al., 2020, 2021; Vainauskas et al., 2022; Yang et al., 2018) The selection of enzyme is critical to attain the highest sequence coverage possible; in some cases, when sample amount allows, it is often beneficial to use more than one mucinase (or protease) in separate or combined digests. For our purposes, we prefer the use of enzyme SmE, but have found that ImpA provides complementary glycopeptide coverage and can be a useful addition.(Chongsaritsinsuk, Steigmeyer, Mahoney, Rosenfeld, Lucas, Smith, et al., 2023) Further, trypsinizing after mucinase digestion can result in higher coverage of more sparsely glycosylated regions.(Mahoney et al., 2023) Finally, N-glycosylation can convolute data analysis and potentially suppress O-glycopeptides desired but can be removed by a universal N-glycosidase called PNGaseF.

Materials:

• Sample prepared in Basic Protocol 1

• Dithiothreitol (DTT, Sigma, D0632)

• Iodoacetamide (IAA, Sigma, I1149)

• Ammonium bicarbonate (Honeywell Fluka, 40867)

• PNGase F (New England Biolabs, P0709S)

• Calcium chloride (MP Biomedicals, 153502)

• Trypsin (Promega, V511A)

• Formic acid (FA, Thermo Scientific, 85178)

• LCMS grade Acetonitrile (ACN, Honeywell, LC015)

• Ultrahigh purity water (Fisher, W6–1)

• ImpA/O-glycoprotease (New England Biolabs, P0761S)

• Analytical balance (Mettler Toledo, XPE26)

• Heated mixing block (Eppendorf, 5382000023)

• Digital incubator (Thermo Scientific, 50125590)

• 50 mL centrifuge tube (Falcon, 352070)

• 18G hypodermic needle (BD, 305195)

• SPE cartridges (Phenomenex, 8B-S100-AAK)

• Vacuum concentrator (Labconco, 7810012)

• Cold trap (Labconco, 7460020)

Reduction and alkylation

• 1To make fresh 50 mM DTT, first weigh a small amount of DTT (~1–5 mg) and then add 130 μL of water for each mg of DTT. Vortex and spin.
• 2Add DTT to the sample at a final concentration of 1 mM. Vortex and spin.

  When working with larger sample volumes at low concentrations (e.g., ≥50 μL, ≤0.5 mg/mL), add DTT to a final concentration of 0.5 mM and IAA (vide infra) to a final concentration of 0.75 mM.

  The solutions should be pipetted into the solution, not onto the walls of the sample tube.

• 3Incubate at 65 °C for 20 minutes while shaking at 1100 rpm in the heated mixing block (ThermoMixer).
• 4Remove sample(s) from the heat block and cool at RT for ~5–10 minutes.

  This step allows the sample to cool, which helps avoid overalkylation.

• 5While performing step 3 or 4, make fresh 75 mM IAA. To do so, weigh a small amount of IAA (~1–5 mg) and add 75 μL of water for each mg IAA. Vortex and spin.
• 6Add IAA to a concentration of 1.5 mM (same volume as DTT). Vortex and spin. React in the dark at RT for 15 mins.

Mucinase Digestion

If sample amount is unknown, these quantities must be estimated, which is often the case after performing an enrichment with StcE^E447D. For an enrichment of 500 μg of cell lysate or biofluids, 0.2 μg of SmE is used, followed by 0.1 μg of trypsin. Both trypsin and SmE are fairly robust and work similarly over a range of concentrations. In general, a lower enzyme to protein ratio (e.g., 1:20) is used for smaller volumes and sample amounts, while a higher ratio (e.g., 1:200) can be used for larger volumes and higher concentrations. It is important to keep enzyme stock solution concentrations above 0.025 mg/mL to avoid pipetting errors and/or sample losses. Additionally, dilute the protease so that the total volume added to sample is at least 1 μL, which assists in accurate pipetting and thus enzyme concentrations. For people experienced with pipetting low volumes, volumes down to 0.5 μL can still provide reproducible results. Low volumes should always be pipetted directly into the solution.

• 7Express SmE as described in reference Chongsaritsinsuk, Steigmeyer, Mahoney, Rosenfeld, Lucas, Ince, et al., 2023; and/or purchase O-glycoprotease (ImpA) from NEB.

  Typically, mucinases are stored at approximately 1 mg/mL in PBS.

• 8Make fresh 50 mM ammonium bicarbonate (AmBic) by dissolving 0.395 g in 100 mL ultrapure water.
• 9Dilute concentrated mucinase 1:10 in 50 mM AmBic.
• 10Vortex and spin mucinase, then add enzyme to sample at a 1:20 (wt:wt) ratio.
• 11If PNGaseF treatment is desired, add 1 μL of concentrated stock (500 U) and 1 μL of 20 mM CaCl₂ to 8 μL of 50mM AmBic. Add 1 μL of dilute stock per 2 μg of protein. For enriched samples with unknown content, add 0.5 μL of concentrated stock.

  Often it is useful to remove N-glycans for analysis as it makes data interpretation easier.

  Adding divalent cations is necessary here since the EDTA from the PNGaseF storage buffer will sequester the zinc ions from the mucinase or O-glycoprotease, rendering it inactive.

• 12Vortex sample and spin down.
• 13Incubate at 37 °C for 3 hours-overnight.

Trypsin Digestion:

Often, mucin-domain glycoproteins contain N-glycans and/or large unmodified regions of the protein. If the goals of your experiment are to (a) identify full sequence coverage of a single protein or (b) identify all proteins present in a sample, then a tryptic digest can be added. Further, in cases where an isolated mucin domain is <50 amino acids in size, a trypsin digest is necessary in order to generate enough carrier peptides to get through HILIC enrichment and/or SPE clean-up. The addition of trypsin, therefore, is very much dependent on the questions being asked by the investigators.

• 14Add 2 μL of 0.05 μg/μL trypsin. Vortex and spin.
• 15Incubate at 37 °C for 6 hours to overnight.

Sample clean up

• 16Poke holes in the lid of a 50 mL Falcon tube using a hypodermic needle and put a new needle through one of the holes.

  Caution: sharps should be handled carefully and disposed of properly. Never recap an old needle.

• 17Affix a 10 mg SPE cartridge into the end of the needle.
• 18Add 1000 μL of ACN, allow liquid to drain through the column and needle.

  For each step, all of the liquid should be allowed to drip through until none remains on top. However, the next solution should be added quickly thereafter to prevent the material from fully drying.

• 19Add 1000 μL of 0.1% formic acid in H₂O.
• 20Add 1000 μL of 40% ACN, 0.1% formic acid in H₂O.

  Failure to do steps 4–5 can cause hydrophobic contaminants to leach from the column and be seen late in the gradient during LC-MS analysis, which can lead to shorter column life.

• 21Equilibrate by adding 1000 μL of 0.1% formic acid in H₂O.
• 22Add 100 μL of 0.5% formic acid to sample. Vortex and spin down.

  When removing sodium deoxycholate from the sample, add neat formic acid to 1% final concentration. A fluffy while precipitate should be visible immediately. Vortex vigorously, then centrifuge at 5000 rcf for one minute. Collect the supernatant, being sure not to disturb the white precipitate. Put this supernatant in a new tube, then spin again at 5000 rcf for 1 minute. If any white precipitate is still visible, do not include it when transferring the sample to the cleanup column.

• 23Transfer sample onto SPE column.
• 24Add 200 μL of 0.1% formic acid to sample tube, then to column.

  If performing a cleanup after removing sodium deoxycholate, do not rinse the sample tube, add the solution directly to the SPE column.

• 25Add 200 μL of 0.1% formic to the column.
• 26Move the column from the 50 mL falcon tube to a fresh microcentrifuge tube and elute 2x with 150 μL of 40% ACN in 0.1% formic acid.
• 27Dry down in vacuum concentrator until dry. (~1 hour) Samples can be kept at RT for < 1 day, −20 °C for a month, or −80 °C for long term storage.

BASIC PROTOCOL 3

Basic protocol title: Mass spectrometry data collection for O-glycopeptides

The field of O-glycoproteomics has seen many notable advances in the last decade that have allowed the site-specific localization of O-glycans on a myriad of proteins.(Bagdonaite et al., 2022; Rangel-Angarita & Malaker, 2021) Here, we describe best practices for operation of the high performance liquid chromatography (HPLC) and mass spectrometer for O-glycoproteomics in hopes to provide the reader a guide to generating high quality data. This includes optimal parameters for HPLC separation prior to electrospray ionization, as well as glycopeptide analysis within the MS, including dissociation via beam-type collision induced dissociation (beamCID), electron transfer dissociation (ETD), and/or ETD with supplemental collisional activation (EThcD).(Riley et al., 2020) We note that the parameters described here are specific to the instruments in our laboratory (i.e., Dionex Ultimate 3000 coupled to a Thermo Orbitrap Tribrid Eclipse), however, most of the parameters can be adapted for different HPLC systems and instruments with ETD enabled.

Materials:

• Sample prepared in Basic Protocol 2

• LCMS grade water (Fisher, T51140-K2)

• Formic acid (FA, Thermo Scientific, 85178)

• Acetonitrile (ACN, Honeywell, LC015)

• Acclaim PepMap 100 column packed with 2 cm of 5 μm C18 material (Thermo Fisher, 154750)

• 15 cm PepMap RSLC EASY-Spray C18 column (Thermo Fisher, ES904)

• Dionex Ultimate 3000

• Thermo Orbitrap Tribrid Eclipse with ETD enabled

HPLC parameters:

• 1
  Set up an HPLC gradient from 0–35% solvent B (0.1% formic acid with 80% ACN) in 60 min.

  1. For the UltiMate 3000, this is achieved by running solvents at 2-2-35-100-100-2%B in 0-5-65-70-75-80 minutes, with the column oven valve switching to 10_1 at 4 minutes and returning to 1_2 at 95 minutes. Loading solvent A (0.1% formic acid in water) is flowed at 5 μL/min throughout the gradient. Spectra are collected for 100 minutes.
     A 100%B rinse is included at the end of the 0–35%B gradient to wash any additional non-peptidic contaminants from the column, which extends column life.
• 2Reconstitute samples in solvent A (0.1% formic acid in water).
• 3Load samples via autosampler and inject using a flow rate of 0.3 μL/min onto a 75 μm × 150 mm EASY-Spray column containing 2 μm C18 beads. Hold columns at 35 °C using a column heater in the EASY-Spray ionization source.

  The sample volume will depend on the experiment in question. For recombinant mucin-domain glycoproteins, we typically digest 1 μg, reconstitute in 10 μL, and inject 40% (i.e., 400 ng). For more complex samples wherein we are enriching mucins from serum or lysate, we will start with 50–500 μg of material, reconstitute in 8 μL, and inject 6 μL; the maximum inject volume on a 20 μL sample loop when using the ‘μL pickup’ injection setting is 6.8 μL. Sample volumes should be adjusted accordingly if using a different sized sample loop and/or injection settings.

Mass spectrometer parameters:

• 4
  Full scan MS1 spectra:

  • a
    Resolution – 60,000 at FWHM
  • b
    Automatic gain control (AGC) target – 3e5
  • c
    Mass range – 300 to 1500 m/z
  • d
    Dynamic exclusion – enabled with repeat count of 2, duration of 7 s, and exclusion duration of 7 s
  • e
    Charge state – 2 to 6, unidentified charge states will not be selected
• 5
  MS2 fragmentation spectra:

  • f
    Higher-energy collision induced dissociation (HCD) is performed on all selected precursor ions:

    1. Isolation window – 2 m/z
    2. Normalized collision – 25–30-40% stepped collision energy
    3. Orbitrap detection, resolution – 7,500 at FWHM
    4. Maximum inject time – 75 ms
    5. Automatic gain control (AGC) target – default
  • g
    Electron-based dissociation (ExD) is triggered on the same precursor if 1) the precursor m/z was between 300 and 1,500 m/z and 2) 3 of 8 HexNAc or NeuAc fingerprint ions (126.055, 138.055, 144.07, 168.065, 186.076, 204.086, 274.092, and 292.103) were present at ± 0.1 m/z and greater than 5% relative intensity

    1. For samples between 300 and 850 m/z, ETD is selected with the following parameters:

       • Charge-dependent reaction times
       • Maximum inject time – 100 ms
       • Injection targets – default
       • Ion trap detection
       • Scan rate – normal
    2. For samples over 850 m/z, ETD with supplemental HCD activation was enabled with the following parameters:

       • Charge-dependent reaction times
       • Maximum inject time – 150 ms
       • Injection targets – default
       • Normalized collision – 15% nce
       • Orbitrap detection, resolution - 7,500 at FWHM

  Note that siloxanes in antiperspirants often generate significant background. We recommend MS-compatible products that do not contain dimethicone or cyclopentasiloxane when operating or nearby instrumentation.

BASIC PROTOCOL 4

Basic protocol title: Mass spectrometry data analysis of O-glycopeptides

One of the biggest hurdles associated with glycoproteomics is the annotation of glycopeptide fragmentation spectra, especially in large datasets containing thousands of potential glycopeptides. This process is highly error prone due to the challenging task of correctly assigning the glycan composition, modification site(s), as well as the peptide backbone sequence. As a result, glycoproteomics publications frequently report incorrect glycopeptide identifications and/or suffer from ambiguous annotation even when attempting to control the false discovery rate (FDR) of assignments.(Kawahara et al., 2021) O-glycopeptides present additional issues when compared to the more well-studied N-glycosylation, especially with regard to the use of alternative proteases (e.g., O-glycoproteases), site-localization of O-glycopeptides, and necessity for multiple modes of fragmentation (e.g., HCD and ET(hc)D).(Bagdonaite et al., 2022) This is a well-recognized issue within the field and drove us to perform a systematic comparison of software programs with the ability to analyze O-glycopeptides. In this study, we found that O-Pair performed best when analyzing single recombinant mucin-domain glycoproteins but suffered as sample complexity increased. For more complex samples, Byonic was the most useful in correctly identifying glycoproteins.(Rangel-Angarita et al., 2022) Thus, we often use multiple search algorithms to help analyze our data, along with a large amount of manual validation and/or de novo sequencing.

Materials:

• Byonic (Protein Metrics)

• Metamorpheus with O-Pair

• Thermo Xcalibur QualBrowser

Search parameters for O-Pair in Metamorpheus:

1. Create directed .fasta files containing the recombinant mucin glycoprotein(s) of interest.

   We do not recommend searching an entire proteome in O-Pair as a primary analysis tool, but it can be used to supplement Byonic or other search algorithms.

2. Mass tolerance – 10 ppm (MS1), 20 ppm (MS2)

3. Variable modifications – Met oxidation

   For samples treated with PNGaseF, Asn deamidation should be included as a variable modification.

4. Fixed modifications – Cys carbamidomethylation

5. Glycan database – Default O-glycan (contains 12 structures)

   If glycomic analysis has been performed, generate a custom database with the detected structures.

6. Cleavage specificity – “Semi-specific” cleavage N-terminally to Ser/Thr, 6 allowed missed cleavages

   For samples treated with trypsin, C-terminal Arg/Lys cleavage should also be included.

7. Output should be filtered to a q value <0.01 and manually validated (detailed below). Further, if samples have not been PNGaseF treated, we advise removing glycopeptides with the N_X_S/T sequon.

Search parameters for Byonic:

1. Create directed .fasta files containing the proteins of interest or download the proteome of interest from a database such as Uniprot.

   Alternatively, users can use the “Create focused database function” in Byonic; here, unmodified peptides are used to generate a protein database for a specific sample. This reduces search time, however, in mucinase-only digestion samples, it is possible that only modified peptides are present in the sample and thus will not be included in the focused database.

2. Mass tolerance – 10 ppm (MS1), 20 ppm (MS2)

3. Variable modifications – Met oxidation

   For samples treated with PNGaseF, Asn deamidation should be included as a variable modification.

4. Fixed modifications – Cys carbamidomethylation

5. Glycan database – O-glycan “most common”, contains 9 structures.

   For samples without PNGaseF added, include an N-glycan database dependent on the sample, i.e., for human serum/plasma, “57 most common human plasma” is most appropriate.

   If glycomic analysis has been performed, generate a custom database with the detected structures.

6. Cleavage specificity – N-terminal to Ser and Thr, “specific” (fastest), 6 allowed missed cleavages

   For samples treated with trypsin, C-terminal Arg/Lys cleavage should also be included.

7. Filter results for those with a score >200 and a logprob of >2.

Manual validation of glycopeptides in Thermo Xcalibur:

1. First, use the MS1 to confirm the precursor mass and chosen isotope is correct. Identify whether co-isolated species are present.

2. If the prior condition is met, extract the associated HCD spectra, and average them if multiple spectra were taken.

   1. Confirm initial glycopeptide identification by confirming that the precursor mass without a glycan is present (i.e., Y0) and identify that most b/y ions are present (sans glycan).
   2. For longer peptides, confirm Y0 is present and fragments that are expected to be abundant (e.g., N-terminally to Pro, C-terminally to Asp).
   3. If the peptide contains a Pro at the C-terminus, the bn-1 should be considered sufficient
3. Next, use the ETD spectra to site-localize glycosylation. Again, extract all spectra and average if possible.

   1. Consider all plausible localizations, regardless of search output.
   2. Sites can be confirmed if c/z ions are present between potential glycosites.

Other important considerations during manual validation of search results:

1. After the initial identification of a particular peptide sequence in a strong spectrum, less stringent conditions can be used if the same peptide occurred with a different glycan structure. Use the stronger fragmentation spectrum to determine characteristic fragmentation masses, and then weaker spectra were assigned based on fragment abundance similarity (akin to manual spectral matching).

2. In EThcD spectra, ions that have glycosylation present on the fragment (i.e., c/z ions) were considered more important for localization than the ones that show the fragment without glycosylation (i.e., b/y ions).

3. For peptides with a 138/144 ratio under 1.2, assume that all glycans in the spectrum are core 1 structures. That is, if two sites do not have coverage in ET(hc)D but the glycan composition was N2, H2N2, or H2N2A4, it generally can be assigned as two N1, H1N1, or H1N1A2 structures to both sites. However, this will not be the case if there was an oxonium ion present at 407 m/z, which would indicate the presence of 2 HexNAcs in a single glycan structure.

Further manual identification of glycopeptides:

1. Extract the HexNAc fingerprint ion (204.0867) from MS2 spectra, which will give the overall intensity of the oxonium ion throughout the run. Alternatively, one can extract any HCD scans that triggered an ET(hc)D scan. If any of these highly abundant glycosylated species were not identified by the search algorithm, we recommend manually sequencing the HCD and ETD spectra.

2. To assure the highest number of glycoforms (i.e., peptides with different glycan compositions and/or glycosites), extract expected fragments and any spectra with a new glycan composition and validate/localize as described above.

COMMENTARY:

A complete commentary section is required of all protocol-style articles and must include each of the sub-sections listed below.

Background Information:

The study of mucins previously suffered from a dearth of tools that could allow for their visualization, enrichment, and degradation. Thus, Bertozzi and colleagues sought an enzyme that would selectively cleave mucin proteins. In 2019, we introduced StcE as a mucin-selective protease and characterized its cleavage motif as reliant on both peptide sequence and the presence of glycosylation. We then leveraged StcE to improve mass spectrometric analysis of mucin-domain glycoproteins.(Malaker et al., 2019) Shortly thereafter, we expanded our mucinase toolkit to include enzymes from several commensal and pathogenic bacteria, each of which has a slightly different cleavage motif. We also used inactive point mutants of these enzymes to stain mucins via flow cytometry, Western blot, and immunohistochemistry.(Shon et al., 2020) To extend the use of these point mutants, we reasoned that they could also be used to enrich mucins from complex samples, which allowed us to define the human “mucinome”.(Malaker et al., 2022) More recently, the Malaker laboratory introduced the mucinase SmE, which is thus far the most useful enzyme for digestion of mucin proteins.(Chongsaritsinsuk, Steigmeyer, Mahoney, Rosenfeld, Lucas, Smith, et al., 2023) We also optimized the mucin enrichment procedure to increase throughput and allow for in-solution digestion with mucinases.(Mahoney et al., 2023) Overall, the last five years has seen considerable improvements in the analysis and enrichment of mucins.

Critical Parameters:

A critical parameter to consider is the starting amount of sample for enrichment and/or digestion. In some cases, e.g., human serum, the overall protein content might be high due to albumin or other contaminating proteins, but the concentration of mucins might be relatively low. Optimization might be necessary to achieve best results. As a starting point, we would recommend 50 μg of a concentrated biofluid (e.g., serum) or 500 μg of a cell lysate.

As mentioned above, one of the most important considerations is the selection of enzyme(s) for protein digestion. The researcher must consider (a) the sample being digested and (b) the goals of the experiment. For instance, if one seeks to identify all O-glycosites on a mucin-domain glycoprotein, it is likely beneficial to perform multiple digests with several mucinases/O-glycoproteases. In our work on the T-cell immunoglobulin and mucin-domain containing (TIM) protein family, we found that SmE was the best choice to analyze the mucin domain, but that ImpA was complementary and helpful to identify O-glycosites outside of this region.(Chongsaritsinsuk, Steigmeyer, Mahoney, Rosenfeld, Lucas, Smith, et al., 2023) The sequence of the protein should also be taken into account; if few tryptic sites exist, then alternative proteases such as GluC, chymotrypsin, thermolysin, or AspN could be considered.

Finally, data analysis is by far the most sensitive aspect of this experiment to user manipulation. We must emphasize that changing search parameters can dramatically affect output from all search algorithms, and that results generated by these programs must be taken with a grain of salt. Manual validation and/or de novo sequencing is still a mandatory aspect of our analyses, and we urge other researchers in the field to similarly examine their data.

Troubleshooting:

See Tables 1, 2, and 3 for troubleshooting tips.

Table 1.

Troubleshooting Guide for enrichment of mucin-domain glycoproteins

Problem	Possible Cause	Solution
StcE cleavage observed in results	StcE mutant not completely inactivated	Add more EDTA to sample during purification to chelate zinc ions Make sure high salt rinses have 1–2 mM EDTA Generate E447A mutant to abrogate activity further
Lipids/ lysis detergents observed in chromatogram	Insufficient/ incomplete rinsing prior to elution	Perform additional high salt rinses and/or additional tube transfers after enrichment.
Low yield	Loss of beads during supernatant removal	Make sure beads are fully pelleted before removing supernatant. Leave sufficient volume to avoid disturbing beads.

Table 2:

Troubleshooting protein digest/cleanup

Problem	Possible Cause	Solution
Incomplete alkylation	Insufficient reduction	Allow reduction to proceed for 1 hour before adding alkylating agent
Overalkylation	Overreaction with IAA	Allow sample to cool before adding IAA Quench alkylation reaction by adding DTT
Incomplete digestion	Issues during digestion protocol	Check pH after adding protease, adjust to physiological pH
Salts observed in chromatogram	Poor rinsing during desalting	During desalting, make sure to add sample directly to the packed bed, and add rinses circularly to the walls of the cartridge. Perform an additional rinse.
Sodium deoxycholate in final sample at a high level	Inefficient separation before cleanup	Chill sample to 4 °C prior to centrifuging

Table 3:

Troubleshooting problems during data interpretation

Problem	Possible Cause	Solution
Few identifications with search results despite good chromatogram	Incorrect digestion parameters or protein database	Search with wider parameters (nonspecific, mammalian/bacterial databases) or de novo abundant peaks to identify the issue

Understanding Results:

To demonstrate exactly what a user should achieve if these protocols are replicated, two types of data should be evaluated. The first is the raw data from the MS-analysis, as shown in Figure 1. Using Thermo Xcalibur, one can evaluate the total ion chromatogram (TIC, Figure 1A), base peak chromatogram (BP, Figure 1B), and HexNAc fingerprint (204 trace, Figure 1C). The TIC and BP allows a user to evaluate the overall level of protein/peptide in their sample, where a good level is usually around an NL of e9-e10. One can also identify whether the chromatographic separation is satisfactory. As seen in Figure 1, the peak shapes are narrow and symmetrical – if this is not the case, the separation is likely not ideal and the chromatography column might need to be changed. Since each spectrum with a 204 ion present is indicative of a glycopeptide, the HexNAc fingerprint serves as a quick gauge for how successful the enrichment was. If the HexNac signal is <e5 levels, it is unlikely that any glycopeptides can be sequenced; NLs of e6-e7 are desirable (as in Figure 1). Additionally, O-glycopeptides tend to elute early in the gradient; an abundance of signal between 15–40 min is often indicative of a successful experiment.

Figure 1. MS analysis of mucin-enriched human serum.

Human serum was subjected to enrichment followed by digestion with mucinase SmE and trypsin. Resultant peptides were analyzed by LC-MS/MS on a Thermo Orbitrap Eclipse Tribrid using an HCD-pd-ET(hc)D method; the raw data from Thermo Xcalibur is shown. (A) Total ion chromatogram, (B) base peak chromatogram, and (C) HexNAc fingerprint trace are depicted.

Another metric to consider is the output from search algorithms like Byonic and O-Pair. In an effective enrichment experiment, the most abundant and/or highest scoring proteins should be mucin-domain glycoproteins as defined in reference 11. As demonstrated in Figure 2A, the search results from unenriched human serum have very few mucins and the most abundant protein (by an order of magnitude) is albumin. After enrichment (Figure 2B), the abundance of albumin drops by an order of magnitude, and several of the top scoring proteins are now mucin-domain glycoproteins including apolipoprotein(a), plasma protease C1 inhibitor, immunoglobulin A1, and proteoglycan-4. Presence of abundant mucin proteins is indicative of a successful enrichment; note that non-specific binding is expected, so non-mucin proteins will still appear in the search output. Overall, this type of results demonstrate that mucins were effectively enriched.

Figure 2. Byonic search results from unenriched and enriched human serum.

(A) Human serum digested with trypsin and resultant peptides were analyzed by LC-MS/MS on a Thermo Orbitrap Eclipse Tribrid using an HCD-pd-ET(hc)D method, followed by searching with Byonic. Output from Byonic are depicted here, the top 30 proteins by abundance and scores are shown. (B) Human serum was subjected to enrichment followed by digestion with mucinase SmE and trypsin, then analyzed as described in (A). The top 30 proteins encompass many mucin-domain glycoproteins as defined in Malaker et al., 2022.

Time Considerations:

The protocols listed here are quite flexible with regard to time and can be manipulated to fit the researcher’s schedule. The StcE-conjugated beads are generally made overnight and capped the following day; at this point they are ready for enrichment but can be stored for up to one month at 4 °C if desired. Binding of mucins to the beads can be performed for 6 h to overnight without a decrease in efficacy, and the washes/elution takes approximately 1 h. For proteolytic digestion, reduction/alkylation takes approximately 1 h, and both trypsin/mucinase digestion can be performed anywhere from 3 h to overnight. Desalting with SPE cartridges takes about 30 min and HPLC separation/MS analysis is usually 60–120 min. By far the most time-consuming aspect of these protocols is the data analysis, which can take weeks depending on the complexity of the sample and the number of potential O-glycopeptides identified. Search algorithms vary dramatically in the length of time they need to finish; Byonic can search for days when searching large glycan databases against the human proteome, whereas O-Pair will finish within minutes. Manual validation, as mentioned above, is an arduous process and can take several weeks for a complex sample.

DATA AVAILABILITY STATEMENT:

Data discussed here was reported previously in reference 21 and is available on PRIDE with identifier PXD046534.