Nuts and Bolts of Carbohydrate Analysis in Core Facilities

by Steven Carr (SmithKline/Beecham)

This workshop was presented by the Carbohydrate Research Committee. The featured speakers were Dr. Roberta Merkle (Complex Carbohydrate Research Center, University of Georgia), Professor Reid Townsend (University of California, San Francisco), and Dr. Michael Rohde (Amgen). Approximately 90 scientists attended the workshop. Below are summaries of each of the speaker's presentations.

Roberta Merkle's talk was entitled "Analysis of Glycoprotein Oligosaccharides: Methods of Release, Separation, and Structural Characterization". The complete elucidation of the oligosac-charide structure(s) found on a glycoprotein involves determining the identity and ring size of the monosaccharide components, the linkage type and anomeric configuration for each intersugar glycosidic bond, as well as the sequential arrangement of sugar residues. Such a complete characterization is usually a research project in itself and is beyond the scope of the non-carbohydrate-specialist. However, a variety of means to partially characterize glycoprotein oligosaccharides is available to the non-specialist.

An initial characterization could address the monosaccharide composition (identity and quantity of monosaccharide residues) of a glycoconjugate. Composition analysis requires glycosidic bond cleavage (e.g., acid hydrolysis or methanolysis) followed by derivatization to make volatile components that can then be analyzed by gas chromatography/mass spectrometry. The limitations of composition analysis are that little overall information is obtained in terms of oligosaccharide structure and that analysis by GC/MS requires specialized equipment and a high level of expertise. Alternative methods for determining composition that overcome such equipment and expertise requirements employ LC or gel electrophoresis of fluorescently-derivatized released monosaccharides.

Lectin blotting protocols combine electrophoretic separation of the glycoprotein with sensitive detection and allows elucidation of some general structural features of glycoprotein oligo-saccharides. This methodology takes advantage of the exquisite binding specificities of lectins and is advantageous in that most protein researchers already have the equipment and general skills needed to carry out such analyses. A limitation of this technique is that a glycoprotein often has a mixture of oligosaccharides present and only general features of structure are suggested by the results. Additionally, appropriate controls need to be included, and these are often not described in the commercially available kits.

More detailed structural analysis of glycoprotein oligosaccharides requires the release and separation of the oligosaccharides from the protein. Release may be effected by enzymatic treatments or by such chemical means as hydrazinolysis or mild alkaline/borohydride treatment. Hydrazinolysis results in cleavage of the N-linkage of GlcNAc to asparagine with concomitant cleavage of peptide bonds and de-N-acetylation of the N-acetylhexosamines GlcNAc and GalNAc. O-linked oligosaccharides can be released via a -elimination reaction using mild alkaline/NaBH4. However the released oligosaccharide is reduced, which no longer permits derivatization of the reducing end by, for example, a fluorescent tag.

After release of oligosaccharides it is necessary to fractionate and analyze them. One means of fractionation is lectin affinity chromatography. Combining several immobilized lectins in a sequential fashion can lead to a high degree of purification. Charged (sialylated, phosphorylated, and sulfated) oligosac-charides can be separated by anion-exchange chromatography. Released oligosaccharides can be tagged with a charged fluorophore such as 8-aminonaphthalene-1,3,6-trisulphonic acid (ANTS) and then separated and characterized by gel electrophoresis. Such a gel method could be combined with approaches employing exoglycosidase treatment or lectin affinity chromatographic separations to allow for further definition of structural features.

A technique for oligosaccharide sequencing, the reagent array analysis method uses set combinations of exoglycosidases to treat aliquots of an unknown oligosaccharide. The resultant digest mixture is fractionated by size-exclusion chromatography and analyzed by comparison to a database. The rationale for structural assignment is that a given enzyme combination will generate a certain limit oligosaccharide, and the liquid chromatographic elution pattern of oligosaccharides produced from the unknown is compared to those from authentic standards. The principle disadvantage of this approach is the requirement for ultrapure glycosidases of defined specificity. In our opinion, quality control of commercially available enzymes is often in doubt. Reid Townsend discussed "Carbohydrate Analysis of Glycoproteins using Gels and Blots". The essence of Reid's talk was 1) that changes in mobility and morphology of gel bands, after treatment with specific endo-glycosidases, offer important clues to oligosaccharide structures, and 2) that classification of oligosaccharide structures can be accomplished using monosaccharide composition and "oligosaccharide mapping" of stained bands on PVDF membranes. Glycosylated proteins bind reduced amounts of sodium dodecyl sulfate and display reduced mobility during SDS-PAGE (1). Presumably, the heterogeneity of the oligosaccharide chains affects the stoichiometry of SDS binding and produces widely dispersed bands. The presence of covalently-linked carbohydrate can be confirmed by a sharpening of the "fuzzy" band after treatment with peptide-N4-(N-acetyl--D-glucosaminly)asparagine amidase (PNGase). The class of oligosaccharides can, in some cases, be deduced by treatment with specific endo--N-acetyl-glucosaminidases. Endo H releases oligomannosidic and hybrid oligosaccharides from glycoproteins by cleaving between the two N-acetyl-glucosamine residues of the chitobiose core (2). Partial treatment of glycoproteins with multiple N-glycosylation sites gives a "ladder" of bands by SDS-PAGE. However, glycoproteins which contain both oligoman-nosidic- and lactosamine-type oligosaccharides may show minimal, if any change in band morphology after treatment with Endo H. Endo--galactosidase cleaves polylactosamine chains (3) and, in some cases, the presence of these elongated structures can be deduced by SDS-PAGE of the treated glycoprotein. The action of this enzyme can be affected by branching of the polylactosamine chain (3), and negative results must be appropriately interpreted. Recently, a family of Endo-F enzymes has been described that display specificity for the branching pattern of lactosamine-type oligosaccharides (4). For example, Endo F2 preferentially cleaves biantennary oligosaccharide chains.

Monosaccharide analysis can be determined on glycoproteins that have been purified by SDS-PAGE after electroblotting onto PVDF membranes (5). After appropriate acid hydrolysis, neutral, amino-, deoxy- and anionic sugars can be identified and quantified using two Coomassie stained bands (about 1 nmol of protein). Using high-pH anion-exchange chromatography with pulsed amperometric detection (HPAEC/PAD), these analyses can be successfully performed on glycoproteins with less than 10% carbohydrate by weight. Additional information can be obtained by treating stained bands with the releasing enzymes described above followed by "oligosaccharide mapping" (5). For example, a single Coomassie blue-stained band of recombinant tissue-type plasminogen activator was sequentially treated with Endo H and Endo F2, to release oligomannosidic-type/hybrid and then biantennary chains. Oligosaccharide profiles, obtained using HPAEC/PAD after Endo H treatment, revealed the expected type and amount of high-mannose oligosaccharides (6). Endo F2 treatment of the same blot and HPAEC/PAD analysis revealed the complement of biantennary oligosaccharides (6). Oligosaccharide analysis of endo--galactosidase-treated glycoproteins is more likely to reveal the presence of polylactosamine structures than a chan ge in band morphology. There is minimal change in the band migration of erythropoietin after digestion with endo--galactosidase. However, treatment of the PNGase-released oligosaccharides (from a blot of recombinant erythropoietin) with endo--Gal revealed a dramatic difference in the HPAEC/PAD chromatographic profiles. Reid concluded by emphasizing that only through detailed structure analysis can we begin to get useful insights into glycan function (7).

Mike Rohde presented a generalized discussion of how carbohydrate analysis may be approached for a laboratory not specifically skilled in carbohydrate research. This was from the perspective of a laboratory that was experienced in analyzing predominately proteins. Normal equipment included protein sequencers, HPLCs, amino acid analyzers, SDS/PAGE, capillary electrophoresis, and a mass spectrometer. Using only these tools, one may still gain a considerable amount of information about carbohydrates on proteins. Mike emphasized that core facilities should not be intimidated by the presence of carbohydrate on proteins, especially since we already have tools that will give us a great deal of information as to its nature.

Often the first clue that a protein is glycosylated will come from the behavior of the protein fractions on SDS/PAGE at some stage of purification. These clues can take the form of bands that are broad or fuzzy or in multiple locations on the gel. In the example given, SDS/PAGE of fractions across an HPLC peak gave a distribution of three to four MW bands that all appeared to have the same protein activity. With information such as this, the next step might be to treat aliquots of the protein with various combinations of glycosidases, such as neuraminidase, O-glycanase, and N-glycanase. The effect of each of these on the migration of the bands on the SDS gel will confirm the presence of carbohydrate and provide insight into the types of structures on the protein.

Once it is clear that you are dealing with a glycoprotein, the first reaction is to ask how to get rid of the carbohydrate. A more appropriate response might be to ask whether we want to remove it and whether it will interfere with other analyses. Examples were given to explore some of these questions and to demonstrate a lack of significant interferences.

Two examples of sequence analysis of glycosylated peptides from a proteolytic map of a recombinant protein were shown. In the case of an Asn linked glycopeptide, and a Ser linked glycopeptide, it is possible to infer the sites of glycosylation by blanks in the same sequence run on a gas or pulsed liquid sequencer, without interference with adjacent amino acid assignments in the sequence run. Note was made of reports from A. Gooley and K. L. Williams, Macquarie University, Sydney, Australia, in which they can observe the glycosylated PTH amino acids using a protein sequencer with TFA extraction of a peptide covalently coupled to the sequencing support. A paper describing their methodology will be in Techniques in Protein Chemistry, volume VI.

Once the suspected glycosylation site in a peptide is located by sequencing, MALDI/TOF mass spectrometry can provide additional information about the added mass arising from the carbohydrate. Information about the carbohydrate heterogeneity can be provided by examining mass spacing due to sialic acids, lactosamine (Gal-GlcNAc), and fucose additions. Branching information is not obtained directly from these molecular weight measurements, but the observed masses may be consistent with higher branched glycoforms. Therefore the MALDI/TOF data can suggest that further experiments are needed to gain this information.

Alternate mass spectra methods were shown, including the stepped collision energy, electrospray LC/MS method of S.A. Carr et al., which directly identifies the peptides that contain carbohydrate in a complex mixture. The method is sensitive enough to use only 25 pmol of material and still provide masses of all observed peaks and identification of glycosylated ones. A paper describing their methodology will also be in Techniques in Protein Chemistry, volume VI. Glycopeptides may also be identified by examining a "contour map" of mass distribution as a function of retention time. One usually sees a vertical ladder of ions for non glycosylated peptides, but a skewed ladder for glycosylated peptides. This skewing is due to the slight retention time difference for higher substitutions of carbohydrates on a single amino acid sequence peptide. In complex digests these off-axis ladders are sometimes difficult to recognize. Fragmentation of the glycopeptide ions in the second quadrupole region of the electrospray instrument can provide some limited linkage data for suspected glycopeptides.

A further level of sophistication in information on the released carbohydrate chains is possible with the addition of anion exchange at pH 12 and electrochemical detection. The method is sensitive (100 pmol) and provides a level of discrimination not available in the other methods discussed so far. This method becomes extremely powerful when combined with specific glycosides.

Finally, Mike presented some preliminary results using capillary electrophoresis. Running a peptide map in the presence of an ion pairing agent not only resolves all the non-glycosylated peptides, but also apparently provides some separation of individual glycopeptides based to some extent on the differing carbohydrates present. Detailed analysis of the peaks separated by this method will be a challenge but may be accessible by CZE/MS or again, the application of specific glycosidases.


  1. Segrest, J.P. and Jackson, R.L. (1972) Methods in Enzymol. 28, 54-62.
  2. Maley, F., Trible, R.B., Tarentino, A.L. and Plummer Jr., T.H. (1989) Anal. Biochem. 180, 195-204.
  3. Fukuda, M.N. and Matsumura, G. (1976) J. Biol. Chem. 251, 6218-6225.
  4. Tarentino, A.L., Quinones, G. Schrader, W.P., Changchien, L., and Plummer, T.H., Jr. (1992) J. Biol. Chem. 267, 3868-3872.
  5. Weitzhandler, M., Kadlecek, D., Avdalovic, N., Forte, J.G., Chow, D. and Townsend, R.R. (1993) J. Biol. Chem. 268, 5121-5130.
  6. Spellman, M.W., Basa, L.J., Leonard, C.K., Chakel, J.A., Wilson, S. and van Halbeek, H. (1989) J. Biol. Chem. 264, 14100-14111.
  7. van Hoek, A.N., Wiener, M.C., Verbavatz, J.M., Brown, D., Lipniunas, P.H., Townsend, R.R. and Verkman, A.S. "Purification and structure-function analysis of native, PNGase F- and endo--galactosidase-treated CHIP28 water channels" submitted for publication.

Return to the The ABRF Home Page

Created: 27th July 1995
Last modified: 27th July 1995