Mass Spectrometry of Intact Glycoproteins
Most high-performance mass spectrometers are now capable of resolving gly- coforms of glycoproteins of considerable molecular weight, provided that the number of glycosylation sites is limited. Thus, for example, Thompson et al. have resolved glycoforms of half antibodies using an orbitrap-based instrument (m/z of the 18+ charge state in the region of 4.2 kDa) , and the same group have resolved 49 glycoforms of chicken ovalbumin using similar detection . Differences in measured mass and that of the protein can give an indication of the glycans present. More complex glycoproteins usually give unresolved broad mass peaks, particularly by matrix-assisted laser desorption/ ionization (MALDI), and further analysis requires either proteolysis or removal of the intact carbohydrates.
Site analysis involves the determination of the glycan occupancy at each consensus sequence site; several methods are available. Usually, the protein is digested with a protease such as trypsin, which cleaves peptide bonds
C-terminal to lysine and arginine leaving an amino group (basic) at each terminus, thus promoting the formation of doubly charged ions by mass spectrometry (MS). The idea is to isolate each glycosylation site to a different glycopeptide with analysis usually by liquid chromatography/mass spectrometry (LC/MS). Missed cleavages are common, particularly when the cleavage site is near a glycan, and some proteins such as a1-acid glycoprotein  are difficult to digest. However, the use of a new surfactant “RapiGest SFTM” greatly aids digestion as does the use of a microwave oven . Problems arise when it is not possible to produce peptides containing only one glycosylation site and when the peptides are too large for efficient mass spectral analysis. Other proteases are available but are not used as frequently. The most common are Glu-C, which cleaves peptide bonds C-terminal to aspartic or glutamic acids; Lys-C, which cleaves C-terminal to lysine; and Arg-C, which cleaves C-terminal to arginine.
Mass spectrometric detection of glycopeptides in the presence of peptides is often difficult because of suppression effects . However, glycopeptides can be identified by the presence of oxonium ions as described in Section 18.104.22.168 . Site analysis has been greatly aided in recent years by the advent of electron capture dissociation (ECD) and electron transfer dissociation (ETD) fragmentation, which preferentially fragments the peptide chain rather than the glycan as is common with collision-induced dissociation (CID) [58-60].
Removal of glycans from either the intact glycoprotein or a glycopeptide with the endoglycosidase PNGase F leaves aspartic acid in place of the asparagine at the N-linked site of the protein. The concomitant increase in mass by 1 Da can be detected by MS to identify the site. The percent occupancy at the site can also be deduced. In cases where the peptide sequence is not known, the aspartic acid can be identified by partial 18O incorporation if the digestion is performed in 40% 18O-enriched water . In other methods, glycans have been removed with the endoglycosidase endoH to leave a GlcNAc residue at the linkage site, and Lee et al.  have achieved the same effect by heating glycopeptides with TFA in a microwave oven to degrade the glycans (from horseradish peroxidase) to the same residue. A method for locating O-glycosylation sites described by Muller et al.  also involves partial deglycosylation, this time with trifluo- romethanesulfonic acid to the level of core-GlcNAc residues. The glycoprotein was then cleaved with an Arg-C-specific endopeptidase, clostripain, to yield tandem repeat icosapeptides, which were analyzed by MALDI/postsource decay (PSD) from a-cyano-4-hydroxycinnamic acid (CHCA) matrix.
An alternative technique uses nonspecific protease such as the cocktail of enzymes derived from Streptomyces griseus, known as pronase to cleave the protein, leaving glycans containing asparagine or a very short peptide chain [64-66]. Problem with this method is that it can produce several peptide fragments from each site, thus complicating the analysis or degrading the protein to just asparagine, thus removing any data on the site of attachment. By using extended hydrolysis periods, Schiel et al.  have produced both N- and O- glycans containing only asparagine at the reducing terminus. Hua et al.  have evaluated the use of seven nonspecific proteases with ribonuclease B as a test compound as a method for site analysis. The proteinases elastase, papain, pepsin, pronase, proteinase K, subtilisin, or thermolysin gave small peptides from which those containing the glycosylation site, with its attached glycans, could be obtained. The authors claim that rather than being nonspecific, these proteinases are multispecific and able to hydrolyze proteins at a large but finite number of sites, thus having the ability to leave asparagine or small peptides (normally two to six amino acids) attached to the glycan. Dodds et al.  have described immobilized pronase that retains its activity after repeated use for at least 6 weeks.
Attempts to locate O-glycans on peptide chains by fragmentation sometimes fail  because of the preferential loss of the glycans catalyzed by proton migration from [M+«H]”+ ions. Elimination of the proton has been proposed as a way of overcoming this reaction and has been achieved by ECD in an ion cyclotron resonance (ICR) instrument . It has been argued that ionization by charge localization could achieve the same result by eliminating the ionizing protons. Consequently, Czeszak et al.  derivatized glycopeptides at their amino terminus with a phosphonium group and showed that the resulting ions, when studied by MALDI/PSD, undergo predictable a-type fragmentation of the peptide chain without loss of the attached glycans. In contrast, CID on the doubly charged protonated phosphonium cation gave predominant loss of the sugar moiety. Experiments were conducted with only GalNAc attached to the peptide, but the method may be applicable to peptides carrying larger glycans. Alternatively, the O-glycans could be degraded to GalNAc as in the method described by Muller et al. .
A method named the GlycoFilter has been described for rapid analysis of glycosylation sites and glycan structure . Glycoproteins were trapped in a spin filter, reduced, and alkylated and glycans were then released with PNGase F. Recovery was by a second centrifugation and analysis was performed by MALDI after permethylation. Finally, the residual protein was hydrolyzed with trypsin in H218O, and the peptides were recovered by a third centrifugation step. Each enzymatic step could be accelerated by heating in a domestic microwave oven. A total of 865 and 295 N-glycosites were identified from urine and plasma samples, respectively.