Profiling of Permethylated Mucin O-glycans Using Matrix-assisted Laser Desorption/Ionization Time-of-flight Mass Spectrometry

Mucins are large glycoproteins that make up most of the structural components in mucus. They are produced by specialized epithelial cells called goblet cells and undergo extensive O-glycosylation before being secreted at the mucosal surface. The glycosylation of mucins takes place in the Golgi apparatus, where a family of specialized glycosyltransferases, polypeptide-N-acetylgalactosamine transferases (ppGalNAc-T) initiates the process by attaching a GalNAc residue to a serine (Ser) or a threonine (Thr)¹. Then, an arsenal of glycosyltransferases extends the glycan structures by adding galactose (Gal), N-acetylglucosamine (GlcNAc), GalNAc, N-acetylneuraminic acid (Neu5Ac), and fucose (Fuc) residues, while other carbohydrate-modifying enzymes can carry out reactions such as sulfation of the glycans¹. The glycans that decorate the mucin protein backbone are primarily found in clusters in protein regions rich in Ser and Thr residues, as the hydroxyl groups on the side chain of these residues serve as O-glycosylation sites. This dense glycosylation gives mucins the appearance of a "bottle-brush" and is responsible for their physical properties, such as the ability to retain large amounts of water, their gel-forming ability, and high viscoelasticity^2,3. Due to these properties, mucins provide lubrication, protect the epithelium from dehydration and physical injury, and form a physical barrier between the epithelium and its microenvironment^4,5. Mucin glycans can also serve as binding sites and a carbon source for commensal or pathogenic bacteria^6,7,8. The organization and composition of mucus are instrumental in protecting the epithelium by keeping microbes nearby but away from the epithelium surface. Mucin glycans also participate in cell-cell communication and adhesion⁹.

Mucin glycans play a pivotal role in maintaining mucosal homeostasis, and aberrant mucin glycosylation profiles are associated with impaired barrier function, altered microbiota composition, increased tumor metastasis, inflammatory bowel disease (IBD), and other diseases^10,11,12. Therefore, the study of mucin glycosylation is important in understanding the role of mucus and mucins in human health and can lead to novel approaches to the management or treatment of diseases.

Mass spectrometry-based techniques are among the most popular options for studying mucin glycans. MALDI-TOF MS is a soft ionization technique that enables the analysis of large, nonvolatile biomolecules such as glycans and glycoproteins¹³. This approach offers high sensitivity and a wide mass range, allowing the detection and identification of low-abundant and diverse glycan mixtures. In addition, the fragmentation patterns obtained from tandem mass spectrometry (MS/MS) experiments can provide valuable structural information about the glycan composition and sequence¹³.

Typically, the analysis of O-linked glycans from mucins by MALDI-TOF MS involves a multi-step workflow, where the O-glycans are first chemically released from the protein backbone using methods like reductive β-elimination, then desalted to remove the salts from the chemical release and derivatized to enhance ionization and detection. The purified, derivatized glycans are then co-crystallized with a suitable MALDI matrix on a MALDI target plate and mass spectra are acquired that provide information on the masses of the intact glycan ions. Following this, the glycans are fragmented in MS/MS experiments to yield information on the structure of the glycans¹⁴. Here we provide a detailed protocol to release, desalt, and permethylate glycans and analyze them by MALDI-TOF MS.

NOTE: See Figure 1 for a schematic representation of the workflow.

Figure 1: Schematic representation of the experimental workflow. Please click here to view a larger version of this figure.

1. Glycan release

1. Preparation of the β-elimination buffer
   NOTE: This must be prepared fresh before each use.
   1. Weigh a pellet of sodium hydroxide (NaOH) and dissolve it in a suitable amount of H₂O in a 50 mL tube to get a 0.1 M solution.
   2. In a 5 mL tube, dissolve 189 mg of sodium borohydride (NaBH₄) in 5 mL of the alkaline solution prepared above in step 1.1.1.
      CAUTION: NaOH is corrosive and NaBH₄ is corrosive, toxic, flammable, and poses a health hazard. Both chemicals should be handled under a fume cupboard using suitable personal protective equipment. Do not dissolve NaBH₄ directly in H₂O, as it decomposes and releases hydrogen that can be self-ignited.
2. Setup of the β-elimination reaction
   1. In a 4 mL glass vial, dissolve up to 100 µg of purified mucins in 250 µL of H₂O.
      NOTE: Some mucins may not dissolve fully and instead form a suspension of very small particles.
   2. Add 250 µL of the β-elimination buffer and close the vial tightly with a PTFE-lined cap. Incubate the reaction at 45 ^oC for 16 h.
3. Stopping the reaction
   1. In a 100 mL glass bottle, add 5 mL of glacial acetic acid to 95 mL of H₂O (5% (v/v) concentration).
      NOTE: This can be stored at room temperature and reused.  CAUTION: Acetic acid is flammable and corrosive and should be used in a fume cupboard with the necessary personal protective equipment.
   2. On ice, add 1 mL of the 5% acetic acid solution to each reaction (step 1.2.3) in a dropwise manner. At this point, look for effervescence from the release of hydrogen from the excess of NaBH₄. If effervescence has not stopped after the addition of 1 mL of 5% acetic acid, repeat this step.
      NOTE: If there is no effervescence upon the addition of 5% acetic acid, then the β-elimination reaction might be incomplete or unsuccessful.
4. Desalting of the samples
   1. In a 100 mL glass bottle, prepare a dense suspension of cation exchange resin, H⁺ form, 200-400 mesh in 5% (v/v) acetic acid.
   2. Plug a glass pipette with a small amount of glass wool. Add 1.5 mL of the resin suspension onto the glass pipette from the top and let it settle and drain. This will make the desalting column.
   3. Wash the resin with 2 mL of 5% acetic acid. Discard the flowthrough.
   4. Add the mucin sample to the desalting column. Collect the flowthrough in a 5 mL tube.
   5. Rinse the sample vials with 1 mL of 5% acetic acid and load this to the desalting column. Collect the flowthrough in the same tube.
   6. Wash the column 2x with 1 mL of 5% acetic acid. Collect the flowthrough in the same tube.
   7. Use a centrifugal evaporator to remove the acetic acid and concentrate the samples (5 mbar, 30 ^oC, 2 h).
   8. Freeze the samples at -80 ^oC and dry them overnight on a freeze dryer.
      NOTE: Alternatively, the centrifugal evaporation can continue to dryness, but it will make redissolving the samples more difficult.
   9. In a 100 mL glass bottle, add 5 mL of acetic acid to 95 mL of methanol (5% (v/v) concentration).
      CAUTION: Methanol is flammable and toxic. It should be used in a fume cabinet with the necessary personal protective equipment.
   10. Dissolve the samples in 0.5 mL of methanolic acetic acid and dry the samples under a stream of nitrogen. Repeat this step 3-5x. Store the dried samples at 4 ^oC.
5. Permethylation of the glycans
   1. In an 8 mL glass vial, mix 50% NaOH solution (0.2 mL) with 0.4 mL of methanol and vortex thoroughly. A white precipitate will form that will quickly dissolve again.
   2. Using a glass syringe, add 4 mL of anhydrous dimethyl sulfoxide (DMSO) and vortex thoroughly.
   3. Centrifuge the solution for 2 min at 2,000 × g. This will form a gel at the bottom of the tube, a liquid phase in the middle, and salts accumulated at the top.
   4. Carefully decant the supernatant and the salts, without disturbing the gel.
   5. Repeat steps 1.5.2-1.5.4 5x until no more salts are formed. If the salts stick to the walls of the tube, gently swirl the tube to bring the salts in suspension, without disturbing the gel at the bottom, or remove them using a glass Pasteur pipette.
   6. After no more salts are formed, resuspend the gel in 4 mL of anhydrous DMSO. This suspension is the permethylation base.
      NOTE: This amount of permethylation base is enough for ~25 samples. Quantities can be adjusted accordingly.
   7. To the dried samples, add 100 µL of anhydrous DMSO and 150 µL of the permethylation base. Using a glass pipette, add 80 µL of iodomethane as quickly as possible, to avoid moisture absorption from the environment.
      CAUTION: Iodomethane is toxic and poses a health and environmental hazard. It should be used in a fume cabinet with the necessary personal protective equipment.
   8. Incubate the permethylation reactions for 30 min at room temperature with vigorous shaking.
   9. Add 0.5 mL of H₂O to terminate the reaction. The samples will turn cloudy.
      NOTE: If the samples do not become cloudy, it is possible that the permethylation reaction was incomplete.
   10. Flush the samples with nitrogen until they become clear.
   11. Load the samples on a Hydrophilic-Lipophilic-Balanced (HLB) 96-well extraction plate (60 mg) or HLB cartridges. Apply positive pressure to push the samples through the solid phase at a flow rate of ~1 mL/ min. If using HLB cartridges, use a rubber teat or vacuum manifold to apply pressure to the samples. Discard the flowthrough.
   12. Wash wells with 2 x 1 mL of H₂O. Apply positive pressure and discard the flowthrough.
   13. Elute the glycans from the HLB plate using 1 mL of methanol. Apply pressure is only until methanol starts coming out of the plate. Then, the gravity flow is enough to give the required flow rate. Collect the flowthrough.
   14. Repeat the elution step once more.
   15. Dry the samples using a centrifugal evaporator
   16. In a 4 mL glass vial, prepare TA50 buffer by mixing 1 mL of acetonitrile, 998 µL of H₂O, and 2 µL of trifluoroacetic acid.
       CAUTION: Acetonitrile is flammable and trifluoroacetic acid is corrosive. Both should be used in a fume cupboard, using suitable personal protective equipment.
   17. In a microcentrifuge tube, prepare the matrix solution by dissolving 10 mg of super-DHB [9:1 (w/w) mixture of 2,5-dihydroxy benzoic acid and 2-hydroxy-5-methoxybenzoic acid] in 1 mL of TA50.
   18. Dissolve the dried samples in 10 µL of TA50.
   19. Mix 1 µL of the sample with 1 µL of the matrix solution, spot it onto a steel MALDI target plate, and let it dry.

2. MALDI TOF analysis of the samples

1. Setting up to acquire parent-ion data
   1. Fire 1,000 shots from the laser at 50% intensity.
      NOTE: These can vary depending on the global instrument settings and can be adjusted as required.
   2. Under the Laser carrier tab, set the random walk setting to complete sample.
   3. Under the Detection tab, set the mass range to 500 - 3,000; increase this range if strong peaks are observed at 2,500-3,000 m/z. Set the detector gain to 20x.
      NOTE: The gain can vary based on the global settings of the instrument and can be adjusted as required.
   4. Under the Spectrometer tab, set the polarity to positive and the Matrix Suppression to Deflection, 300 Da.
   5. Click on Start | once data acquisition finishes, Save as to save the data in an appropriate directory.
2. Opening the parent spectra to acquire fragmentation spectra
   1. After right-clicking on the spectra, select Find Mass list.
   2. From the generated list, select the relevant ions for fragmentation and click on add to MS/MS list after right-clicking on the selected peaks.
   3. Right-click again on the mass list and select MS/MS list.
   4. Click on Send to flexControl.
   5. On flexControl, load the LIFT method and navigate to the LIFT tab.
   6. Select each loaded peak one by one and acquire both parent and fragment ions by switching between the Parent and Fragments modes. Adjust the laser intensity and number of shots as required. Click on Save to save the data after the fragmentation of each parent ion.
   7. Export the spectra in '.ascII' format.
      NOTE: This can be later imported into Glycoworkbench for further analysis.
3. Data analysis using glycoworkbench
   1. In Glycoworkbench, load the exported .ascII file of the mass spectrum and carry out a baseline correction and noise filtering using the relevant buttons at the bottom of the window, before computing the peak centroids. In the pop-up window, choose the appropriate mass range, Min signal/noise ratio, and Min MS peak intensity.
   2. Once the peak centroids are computed, use the Glyco-Peakfinder tool to find structure compositions matching the peak list. Select perMe as the Derivatization, redEnd as the Reducing End, and the min and max number of expected residues.
      1. For mucins, select Hexose, N-Acetyl-Hexosamine, Deoxy-Hexose, and N-Acetyl-Neuraminic acid. For sulfated glycans, use the other residue option with a mass of 87.9. For MALDI-TOF data, set Max # Na ions and Max # charges are set to 1 and the accuracy to 0.5 Da.
         ​NOTE: The accuracy may need to be further increased, depending on the calibration of the mass spectrometer.
   3. Select all the correct annotations by holding down the control key and clicking on each annotation; then right-click on the annotation and click on Add to the annotated peak list.
   4. To generate a report comparing multiple annotated spectra, in the tools tab, under reporting, select Create a report comparing different profiles. Print the report as PDF to save it.
   5. After identifying the glycan peaks, calculate the relative abundance of each individual structure as a percentage of the total area under the curve of all glycan peaks.
   6. For the analysis of fragmentation spectra, load the spectra onto Glycoworkbench and generate the peak list as in step 2.3.1.
   7. Draw putative glycan structures that correspond to the annotated glycan compositions. Generate fragments for each drawn structure by selecting the relevant compute fragments option under the Fragments tool. Then, select all fragments, right-click on them, and click on Copy fragments into canvas.
   8. To further support the analysis of fragmentation, select a peak of interest by clicking on it in the spectrum viewer. Next, by right-clicking and selecting Find all structures matching the peaks, interrogate the m/z of the selected peak against online databases for putative glycan structures.
   9. Select putative structures of interest and copy them to the canvas window of the software following a right-click and selecting the appropriate option.
   10. To compute the possible fragments that can be generated from each glycan structure, select the structure of interest in the canvas, and then, under the Tools tab, choose Fragments and Compute fragments for the current structure.
   11. If fragments for more than one glycan structure have been computed, view the comparison table in the summary tab, under the Fragments table view.
   12. Select the fragments that match the peak list for each potential glycan structure and select Copy fragments to canvas from the right-click dropdown menu.
   13. In the Tools tab, under profiler, select annotate peaks with all structures and then click on Tools | Reporting | Create a report of the annotations. Print the report as PDF to save it.

In the following results, porcine gastric mucin (PGM) was used to optimize the desalting and permethylation steps. To compare two commonly used desalting approaches, after release of the glycans, the samples were desalted either by cation exchange (as described in the protocol above) and co-evaporation of borate with methanol or by solid phase extraction with porous graphitized carbon (PGC) cartridges. This requires preconditioning of the cartridge with 80% acetonitrile and 100% H2O, a wash step with 0.1% (v/v) trifluoroacetic acid (TFA), and elution of the glycans with 2 mL of 25% acetonitrile in 0.1% TFA, 2 mL of 50% acetonitrile in 0.1% TFA, and 2 mL of 80% acetonitrile in 0.1% TFA.

The resulting spectra after permethylation were analyzed in the m/z 500 to m/z 2,500 range. A total of 33 glycan structures were identified in the sample desalted by ion exchange and borate removal whereas 32 structures were identified in the sample desalted by PGC extraction (Figure 2A). One glycan structure was not identified in the PGC desalted sample (marked with an asterisk in Figure 2A). This comparative analysis showed that recovery of larger glycans was not as efficient after PGC extraction, compared to ion exchange. This is also evidenced by the relative abundance of larger glycans, which accounts for 21% of total glycans after PGC extraction, as opposed to 31% for ion exchange. Furthermore, desalting by ion exchange and borate removal led to spectra with increased signal-to-noise ratios compared to those produced after desalting by solid phase extraction with PGC (Figure 2B).

To optimize the permethylation reaction time, aliquots of the same permethylation reaction were quenched by adding H2O at different time points, (i.e., 5 min, 30 min, or 90 min). The results showed that 33 glycan peaks could be identified in the m/z 500 to m/z 2,500 range, in the samples permethylated for 30 or 90 min, with the resulting spectra being similar. In contrast, only 31 glycan peaks could be identified when permethylation was carried out for 5 min (Figure 3). Interestingly, the two structures that were not identified in this shorter reaction time (marked with an asterisk in Figure 3) are the two sialylated structures, suggesting that the modification of sialic acids is not efficient in 5 min, under the conditions tested.

Fragmentation was used for each PGM glycan peak to confirm the glycan composition and determine the glycan structure. Two examples are presented here (Figure 4). In Figure 4A, the fragmentation spectrum of a peak at 983 m/z is presented. The glycan composition is assigned as HexNAc2Hex2. There are two potential structures for this glycan composition, either Gal-(Gal-GlcNAc-Gal-)-GalNAc (glycan 1) or Gal-GlcNAc-Gal-GalNAc (glycan 2). Fragmentation of this compound produced a mass spectrum with peaks that can be assigned to different fragments of the glycan. The peak at m/z 529 and the peak at m/z 708 are characteristic of structures 1 and 2, respectively, and we can conclude that both glycans 1 and 2 are present in the PGM glycan mix.

In Figure 4B, the fragmentation spectrum of a peak at m/z 2,026 is presented. Glycoworkbench computed two potential glycan compositions for this peak, Hex3HexNAc1Neu5Ac3 and Hex3HexNAc4dHex2. The sialylated composition has a lower accuracy (~160 ppm and 178 ppm, respectively), suggesting that the first one must be the correct choice. However, the fragmentation profile indicates the presence of Fuc, not Neu5Ac; therefore, the second composition is the correct one. Here, four putative glycan structures are presented, and the spectrum has been annotated with the computed fragments from each putative glycan structure. In the annotated spectrum, we can see that peaks characteristic only of glycans 4 and 6 are present, while no peaks characteristic of glycans 3 and 5 were identified. Based on these data, we can conclude that glycans 4 and 6 are present.

Figure 2: Comparison of the glycan peaks according to the desalting and solid phase extraction. (A) Comparison of the PGM glycan profile after desalting by either cation exchange chromatography and borate removal or by PGC solid phase extraction. (B) Comparison of the two desalting approaches in the resulting signal-to-noise ratio for each identified glycan peak. The asterisk specifies the glycan that was only identified in the samples desalted by cation exchange and removal of borate. Please click here to view a larger version of this figure.

Figure 3: Comparison of the PGM glycan profile after permethylation for 5, 30, or 90 min. The asterisks indicate the sialylated glycans that were only identified after 30 or 90 min of permethylation. Please click here to view a larger version of this figure.

Figure 4: Examples of fragmentation spectra from PGM glycans. (A) Fragmentation of the glycan peak at m/z 983. (B) Fragmentation of the glycan peak at m/z 2026. The boxes indicate potential glycan structures. The numbers over the fragments indicate from which glycan structure they could be derived. Please click here to view a larger version of this figure.

To study glycans, researchers employ a variety of methodologies, ranging from the use of lectins to identify specific epitopes to more sophisticated liquid chromatography-tandem mass spectrometry (LC-MS) and nuclear magnetic resonance (NMR) to obtain detailed characterization of glycan structures. The main advantages of MALDI-TOF MS over alternative approaches are the sensitivity and the high-throughput nature of the method¹³. However, as there is no capability to separate isobaric compounds, different glycan structures with the same m/z value produce a mix of fragments under MS/MS that is then difficult to deconvolute¹³. If determining the exact structure of mucin glycans is required, LC-MS should be the method of choice.

Here, we present a protocol for the analysis of mucin glycans by MALDI-TOF MS that includes all the critical steps in the process, from glycan release, desalting, and permethylation, to data acquisition and analysis. It is important to acknowledge that each of these steps can be carried out using different approaches, and they all come with associated benefits and limitations.

In this protocol, we used reductive β-elimination to release the glycans. The reaction takes place under basic conditions. If the mucin samples are already in suspension and have not been previously purified, the pH must be checked using a pH indicator paper; if acidic, it needs to be increased to a pH of ~8-9. A limitation of this approach is that the basic conditions required for the release of the glycans also lead to the hydrolysis and, therefore, loss of O-acetyl groups often found on sialic acids, in a reaction known as saponification.

While other methods have been described in the literature for the release of glycans^15,16,17,18, some of which can preserve the labile O-acetyl groups, reductive β-elimination remains the gold standard in the field. This approach requires a high concentration of salts. Salts and other ion-suppression contaminants then need to be removed prior to MALDI-TOF analysis¹⁹, as they can affect the ionization efficiency and quantification of glycans. Here, cation exchange chromatography, followed by removal of borate by methanol evaporation was shown to be the most efficient approach for desalting the released glycans from mucin. In this case, the glycans do not interact with the resin and instead flow through it, while sodium ions are trapped onto the resin. However, the preparation of the ion exchange columns is laborious and makes this approach inefficient for large-scale experiments with numerous samples. In such cases, PGC desalting may be preferred, as this resin is also available in a 96-well plate format. However, as presented above, desalting by PGC extraction can lead to inefficient elution, particularly of larger glycans, and this should be considered when interpreting the results.

Although native glycans can also be analyzed by MALDI-TOF, permethylation is usually employed, as this reaction significantly increases the sensitivity of detection, since native glycans have a low ionization efficiency²⁰. Furthermore, permethylation makes negatively charged glycans, such as sulfated or sialylated glycans, easier to detect in positive mode and preserves labile glycan epitopes, such as sialic acids or fucose²⁰. Most commonly, permethylation is allowed to run for 5-10 min^14,21, with some reports allowing the reaction to continue up to 40 min²². Here, we showed that, under the conditions tested, sialylated glycans were not efficiently detected after 5 min of permethylation, and a longer incubation was required.

In addition to the reaction time, another factor that can affect the permethylation efficiency is the content of water in the reaction. The presence of water can lead to undermethylation or partial methylation of the glycans, where not all hydroxyl groups have been converted to methyl groups²³. This can easily be determined as if undermethylation has occurred, groups of peaks that differ by 14 Da (the mass difference between hydroxyl and methyl groups) can be observed in the mass spectra. This issue can be addressed by repeating the permethylation reaction and re-analyzing the data.

A limiting factor in glycan analysis is the interpretation of the spectra. Despite recent advances, this process is still largely manual and requires good knowledge and understanding of the glycan biosynthetic pathways to accurately annotate the spectra. Tools like Glycoworkbench²⁴ can assist by providing possible annotations, glycan structures, and compositions, but due to the inherent heterogeneity of glycan mixtures, researchers still need to make the decision of which choice is the correct one. In the example presented above using PGM, Glycoworkbench offered two potential glycan compositions for the peak at mz 2,025.7620, and based on the fragmentation profile, we concluded that the fucosylated structure was present, rather than the sialylated one. However, if a fragmentation spectrum was not available (which is often the case for low abundant glycans), researchers can base the annotation on the knowledge that glycans with a composition of Hex3HexNAc1Neu5Ac3 are found on gangliosides, rather than mucins, and this option should therefore not be considered in the context of PGM. Thus, identified glycan structures or compositions should be cross-referenced to previous publications and should agree with the principles of O-glycan biosynthesis. Overall, this protocol serves as a valuable resource, especially for researchers who are new to the field of glycobiology and with an interest in the study of interactions occurring at the mucosal interface.