Analysis of Carbohydrate Content of Glycoproteins

Dr. Roberta Merkle of the Complex Carbohydrate Research Center, University of Georgia, Athens, Georgia, presented a comprehensive guide to the analysis of monosaccharides derived from the hydrolysis of glycoproteins, glycopeptides and oligosaccharides derived from those sources. These notes are my attempt to provide background information on the bullet points in her slide set. Dr. Merkle presented most of what is here in these notes but I have added some original material. These notes may be considered ad hoc listener's notes trying to emulate a set of unseen speaker's notes. These notes are my opinion and view and do not necessarily reflect the views or opinions of my employer, PE-Applied Biosystems. Joseph D. Olechno, 7/1/96.

SLIDE 1: Analysis of carbohydrate content of glycoproteins

Much of the information in this lecture can easily be extended to the carbohydrate composition analysis of glycolipids, polysaccharides, secondary plant metabolites, etc.

SLIDE 2: Monosaccharide composition

Composition data can be useful in the biotechnology industry as an indicator of changes in the oligosaccharides during production or processing. For the researcher, differences in composition can provide important insight into possible oligosaccharide structure.

Carbohydrate analysis, while a little confusing with the multitude of potential methodologies, is relatively easy and can provide extremely important information.

Monosaccharide composition analysis can be very useful as a diagnostic to determine what types of oligosaccharides (e.g., O-linked vs. N-linked; high mannose vs. complex) are bound to a protein. In some cases, it may be useful in picking up small but important changes such as the fucosylation of an epidermal growth factor (EGF) region or the N-acetylglucosaminylation of a nuclear protein. Changes in carbohydrate structure and content have been associated with many changes in the biological activity of proteins as well as the chemical and physical properties of the proteins. For a review of the biological effects of protein glycosylation see A. Varki, Glycobiology, 3, 97-130, 1993.

Readers of this tutorial should seriously consider trying some analyses in the lab.

SLIDE 3:N- and O-linked oligosaccharides

O-Linked oligosaccharides are linked primarily through the hydroxyl oxygen of the amino acids serine and threonine. O-Linkages can occur anywhere in the protein but it is common to have many O-linked sites in close proximity. Some other amino acids can also be O-glycosylated. Tyrosine is O-glucosylated as the first step in the production of glycogen. Hydroxyproline, hydroxylysine and a few other hydroxylated amino acids have also been reported. Many different sugars can be directly attached to the hydroxylated amino acid. N-Acetylgalactosamine (GalNAc) is one of the most common but N-acetylglucosamine (GlcNAc), fucose (Fuc), xylose (Xyl) and other sugars are occasionally linked to serine/threonine.

Many O-linked structures are usually relatively small with 2-6 monosaccharides making up a "normal" oligosaccharide. There are many exceptions. N-Linked oligosaccharides are always attached to the amide nitrogen of asparagine. The linkage is always through two N-acetylglucosamines (chitobiose core). Only asparagines which are part of a three amino acid consensus sequence can be glycosylated. N-Linked oligosaccharides come in three major types, high mannose (which contain the two GlcNAcs of the chitobiose core and from five to nine mannoses), complex (in which the mannose side chains have been clipped down and then extended with GlcNAcs, galactose (Gal), Fuc and sialic acids) and hybrid (in which some high mannose chains remain but the rest of the oligosaccharide resembles a complex structure). N-Linked oligosaccharides generally have 7-20 monosaccharide components. Small structure changes in N-linked oligosaccharides can have dramatic effects upon clearance of the glycoprotein to which they are attached. In general, it is relatively easy to clip N-linked oligosaccharides enzymatically from the asparagines to which they are attached. O-linked oligosaccharides can be enzymatically removed from their proteins only with a cocktail of various enzymes, each specific for a different amino acid and/or carbohydrate.

SLIDE 4: "Why is it so difficult..." Problems in monosaccharide composition analysis

While it might appear that the difficulties in determining carbohydrates should be small, a recollection of amino acid analysis shows that composition analysis is often more difficult that a cursory view might indicate. Dr. Merkle pointed out various places where the analysis can go awry.

Hydrolysis is the addition of water across bonds to result in the formation of (usually reducing ) sugars. Both anomers (alpha and beta) of the sugars exist in solution but the amount of each anomer is determined by the conditions of the final solution (pH, temperature, various salts) rather than by the original linkages. Most sugars will exist in the pyranose form. Methanolysis breaks sugar-sugar linkages by adding methanol across the bond and results in the formation of methyl glycosides. Again, it should be emphasized that the amount of alpha and beta methyl glycosides has nothing to do with the alpha and beta linkages originally found in the oligosaccharide but are due to the conditions of methanolysis.

"Why is it so difficult to perform a simple monosaccharide composition analysis? Accuracy depends on effective hydrolysis." Amen. And note that effective does not just mean removal from the protein. It is also important that the hydrolysis not destroy any carbohydrates (or, at least, destroy them in a readily reproducible manner).

SLIDE 5: Difficulties of hydrolysis

"No single hydrolysis condition is effective for all oligosaccharides and glycoproteins." For instance, the chitobiose core of N-linked oligosaccharides tends to be relatively resistant to many hydrolyses. Some people have suggested that this is due to the hydrolysis of the N-acetyl groups followed by the protonation of the amines. The protonated amines are suggested to keep away similarly charged protons from the sugar and reducing hydrolysis. Some carbohydrates are particularly prone to acid decomposition (e.g., sialic acids and fucose). Strong acid conditions required for the release of neutral and amino sugars will completely destroy sialic acids. Sialic acids (N-acylated neuraminic acids, deoxysugars with [usually] both amino groups and carboxylic acids) must be cleaved from oligosaccharides by mild conditions (enzymatic or dilute acid) and analyzed prior to treating the analyte with strong acid to release other sugars. Amino sugars seem to be very stable to acid hydrolysis conditions. Many deoxysugars (fucose, 2-deoxyglucose, etc.) seem to be more acid labile.

SLIDE 6: Classes of monosaccharides

A few other monosaccharides are associated with glycoproteins: inositol (in GPI anchors), glucose (very rarely), rhamnose, glucuronic and iduronic acid (as components of glycosaminoglycans [GAGs]) and double-bond containing octosulonic and nonasulonic acids (often grouped with sialic acids).

SLIDE 7: Selection of hydrolysis conditions:

It is wise to note Dr. Merkle's statement that "Optimum conditions are determined experimentally." While examples and general protocols can be suggested, it is up to the individual researcher or QC department to optimize the analysis for the protein(s) they are studying.

SLIDE 8: HPLC analysis

Note: Sulfuric acid is an oxidizing acid and generally gives poor results except when used at low temperatures and in low concentrations (e.g., the release of sialic acids). Many "non-oxidizing" acids contain strongly oxidizing impurities (e.g., ferric ion in HCl) which can have deleterious effects on the released monosaccharides. For this reason, extremely pure acids and other reagents are suggested. One attendee noted that triple-distilled trifluoroacetic acid gave significantly better results than reagent grade TFA.

SLIDE 9: Hydrolysis conditions

As recently as 1994, Y. C. Lee stated that it was absolutely necessary to use two different hydrolyses, one to release amino sugars (strong conditions that also released the neutral sugars but degraded them significantly) and a second, milder condition to release neutral monosaccharides which might not be strong enough to get complete release of all amino sugars. Fu and O'Neill reported that the use of triple-distilled TFA allowed them to use a single hydrolytic condition to release all amino sugars without significant decomposition of neutral sugars.

SLIDE 10: Hydrolysis advice

I strongly agree with Dr. Merkle that the researcher try several conditions with several samples and monitor the results. Taking aliquots of the hydrolysis sample and evaluating each aliquot for carbohydrate composition can give useful insights. This method, while time consuming, can also be used as a general technique because it allows the researcher to extrapolate the concentrations of each sugar back to "zero" time to eliminate quantitation errors caused by the decomposition of sugars.

SLIDE 11: Sialic acid hydrolysis

Sialic acids are easily destroyed under the acidic conditions needed to release neutral and amino monosaccharides. In some cases it may be possible to first release the sialic acids from a glycoprotein, analyze an aliquot of the solution and then to treat the rest of the sample under harsher conditions to release the other monosaccharides.

SLIDE 12: Analysis techniques

Speaks for itself.

SLIDES 13 & 14: Colorimetric assays

I may disagree somewhat about the ease of optimization. I find that sample differences can lead to changes in results and that it is not always easy to account for these differences. In general, colorimetric assays do not provide either the researcher or the QC department with enough quantifiable information for most applications.

SLIDES 15 & 16: GC-FID, GC-MS, GC analysis advantages

GLC (gas-liquid chromatography, sometimes referred to as just GC [gas chromatography]) has been a long time standard in the analysis of monosaccharides. Dr. Merkle may wish to adjust my notes but almost all (approaching 100%) glycoprotein oligosaccharides have some amino sugars. Oligosaccharides from other sources (e.g., starches, cell walls, etc.) often have no appreciable amino sugars and are often hydrolyzed, reduced and acetylated as described.

FID (flame ionization detection) provides no extra information to GC analysis other than the fact that the material coming off the column at that time is something that can be destroyed in a flame. MS is useful for distinguishing carbohydrate peaks from impurity peaks, however it should be noted that many monosaccharides have identical mass fragmentation patterns (Gal=Man=Glc, lcNAc=GalNAc), so the mass spectral data must be used in combination with the GC retention data for identifcation. The mass spectral data (electron-impact MS) can allow the determination that some peaks are carbohydrate even though they don't correspond to the retention data of the standards. Their identity would have to be determined by deduction (from the fragmentation pattern data based on knowledge of usual cleavage sites and eliminations giving secondary fragments) and by other means. This is especially useful when analyzing plant and microbial glycoconjugates, where unusual sugars

turn up.

Other advantages not listed include:

*mass analysis to eliminate non-carbohydrate analytes

*mass analysis to distinguish between amino and neutral carbohydrates

*advanced experiments to determine the position of sugar linkage in oligosaccharides (not discussed here)

SLIDE 17: Methanolysis illustration

While an alpha linkage is shown in both the original oligosaccharide and in the derivative methyl glycoside, it is important to remember that methanolysis results in the formation of mixed anomers which are independent of the original linkage type. The free sugars (like sucrose, a disaccharide, are not volatile); by converting them to trimethylsilyl ethers we can increase their volatility and make them amenable to GC analysis. Various trimethylsilylation reagents other than Tri-Sil are commercially available.

SLIDE 18: Gas chromatogram of TMS-derivatized monosaccharides released from the glycoprotein fetuin

Note that there are 2 mannose (Man) peaks, 5 galactose (Gal) peaks, 2 glucose (Glc) peaks, 3 N-acetylglucosamine (GlcNAc) and 2 N-acetylgalactosamine (GalNAc) peaks, and 1 N-acetyl neuraminic acid (NeuAc) peak. The inositol peak is an internal standard. Multiple peaks can arise for a number of reasons:

1. the alpha and beta anomers are usually easily resolved by GC, 2. while some sugars yield only pyranose derivatives, some sugars can yield both pyranose and furanose derivatives (and both in alpha and beta forms), 3. incomplete derivatization can yield multiple products, 4. incomplete acetylation can result in the formation of some amine derivatization by the silylating reagent, 5. uronic acids may trans-esterify to internal esters, 6. other ways also exist.

The area under each identified monosaccharide must be combined with the areas for other peaks derived from that same monosaccharide to accurately determine the actual amount of that monosaccharide.

One item of particular interest: glucose is not part of the oligosaccharides bound to fetuin. The glucose seen in this chromatogram is, most likely, due to sample contamination. Paper and cardboard dust are significant causes of carbohydrate contamination. Glucose is rarely a major constituent of glycoproteins and when large amounts are encountered it is good to look for contamination. Contamination can come from paper dust as mentioned but can also be found in water (which is one reason to use freshly distilled water). Deionized water may have substantial amounts of carbohydrate present in it. Proteins often carry along with their mass a substantial amount of glucose from isolations of the original material.

SLIDE 19: GC Disadvantages

This says it all.

SLIDE 20: HPAEC analysis

Actually, at all pH values, monosaccharides are acids. They are just so weak that we don't see any evidence of their anionic state until they are at very high pH. For example, here are the pKa values of some monosaccharides (Rohrer and Olechno, Anal. Chem., 64, 914-916, 1992):

galactose 12.39

glucose 12.28

xylose 12.15

mannose 12.08

fructose 12.03

glucose-1-d 12.31 (d= deuterium)

glucose-2-d 12.29

SLIDES 21 & 22: Pulsed amperometric detection

During the time potential is held at the analytical potential, signal is not immediately collected. There is a short delay time of a few milliseconds to allow the charging current to decay. Non-pulsed amperometric detection usually is not effective with carbohydrates due to a buildup of partially oxidized materials on the electrode. Recent work by Baldwin has indicated that copper electrodes may be useful for the non-pulsed detection of carbohydrates. Zare has used this for amperometric detection of carbohydrates by CE.

SLIDE 23: HPAEC advantages and disadvantages

Some problems with HPAEC (also called HPAE) include the following. Xylose and mannose co-elute under most conditions. Alditols tend to elute very early and are difficult to resolve from one another but new columns may eliminate that problem. O-Acylations (like those found on some sialic acids) are lost or transmigrate in high pH solutions and can lead to variable results. High pH can (slowly) decompose most glycoprotein derived monosaccharides at room temperature. If analytes are collected after separation for subsequent analysis, they should be neutralized. Reducing sugars with good leaving groups at the three position (e.g., glucose-3-O-phosphate) are quickly destroyed at high pH. I think that the estimate of $25K for a Dionex system might be lower than current list price.

Advantages include the following. There is no need to derivatize neutral sugars and no need to derivatize amino sugars back to N-acetyl derivatives. Since anomerization (interconversion of alpha and beta forms) occurs very rapidly, this leads to only a single peak for each sugar. A major advantage is that non-reducing sugars and ketoses are both easily detected. These sugars are derivatized only with difficulty and are, therefore, not generally amenable to analysis by other techniques.

SLIDE 24: Principles of HPLC analysis,

Many different reagents have been developed for the derivatization of monosaccharides prior to their analysis by either HPLC or electrophoresis (both capillary and slab). For more background see reviews in the literature. (For example, Chiesa et al. in "Capillary Electrophoresis in Analytical Biotechnology" ed. P. G. Righetti (covers both LC and CE). In most cases, for HPLC analysis, the extremely hydrophilic carbohydrates (which would elute in the void volume of a reversed phase column under most conditions) must be derivatized with hydrophobic groups to increase retention. These groups can also lead to increased UV (or fluorescence) sensitivity. Normal phase and metal-loaded cation exchange HPLC are both rarely used in modern carbohydrate analyses.

SLIDE 25:

Derivatization of monosaccharides with PMP

Depending on the article you are reading PMP is also called MPP. PMP reacts only with reducing aldoses. Therefore, methanolysis is not a useful preparative procedure. PMP reacts with aldehydes in a way analogous to the reaction of aldehydes with malonic acid esters. PMP could be considered a cryptic malonic acid ester. The oxygen in the product should be a hydroxyl and, yes, the ring is opened. Other organic solvents may be used for extraction but differences will be seen in the efficiency and amino sugars might be partially lost with chlorinated solvents (e.g., methylene chloride). Since two molecules of PMP are attached to each reducing position, a single analyte peak is seen for each sugars, i.e., no anomers. PMP will not react with ketoses (e.g., fructose, sialic acids). Sialic acids are first converted to N-acylated mannosamines (and pyruvic acid) and the mannosamines (aldoses) are derivatized. This technique makes it possible to analyze for O-acylated sialic acids (not achievable by HPAEC) but not results have been published.

Derivatization with Aryl Amines

Many other different reagents have been used for the derivatization of reducing sugars. Most are based upon reacting the reducing sugar with an (aryl) amine to form a Schiff base (a reversible reaction) followed by reduction to form the amine (an irreversible reaction). Depending upon the aryl group, the product may be strongly UV absorbing or fluorescent. This reaction is very good for aldoses. Some amines will react with ketoses but not all. For example, fructose (a ketose) can be derivatized with 4-aminobenzoic acid but not with 2-aminopyridine. The aryl (or "R" group) of the amine can be chosen to provide good separation. For example, hydrophobic R groups are useful for RP-HPLC but charged R groups are useful for electrophoresis (see below.)

SLIDE 26: HPLC advantages and disadvantages

Some success (admittedly poor in my hands) has been claimed for the HPLC separation of underivatized sugars on metal loaded cation exchange columns (e.g., BioRad) using refractive index detection. This eliminates the need for any derivatization (like HPAEC) but resolution tends to be poor and sensitivity is usually not very good but new generation RI detectors might change that.

Dr. Merkle notes that excess PMP must be resolved but that removal of excess reagent is not limited to PMP but includes many other derivatizing reagents as well. (Fluorogenic reagents, i.e., those that are not fluorescent until after reaction with the analyte, are one type of reagent that does not have to be removed before analysis. For example, o-phenylenediamine, a non-fluorescent reagent, can react with sialic acids to produce a fluorescent product.)

One family of derivatizing reagents not mentioned react with primary amines of sugars. These amines can come from either naturally occurring amino sugars, e.g., galactosamine or from the reductive amination of reducing sugars with ammonia. The most well-known of these reagents is o-phthaldialdehyde, a reagent also used for amino acids and CBQCA, a reagent used for carbohydrate analysis by CE (see below). Both of these reagents are fluorogenic.

SLIDE 27: Principle of carbohydrate electrophoresis - slab gel

The structure in the tutorial slide is incorrect. After derivatization, the sugar is no longer in a ring (pyranose) format, as shown, but is a straight chain. The 5-position oxygen (shown in illustration as "O") becomes a hydroxyl and there is no bond between it and the C1 carbon (CH2 here and pentavalent). With reductive amination (as with PMP described above), aldoses do not form anomers (there are two equivalent hydrogens at position 1). However, anomers will form with ketoses possibly leading to two resolvable analytes. I do not know whether AMAC is capable of derivatizing ketoses like fructose or the sialic acids (see section about different abilities of aryl amines to derivatize aldoses and ketoses). As with PMP and other reductive amination methods, non-reducing sugars (alditols formed by reductive alkaline hydrolysis of O-linked sugars, inositol from GPI-anchor proteins, etc.) cannot be analyzed.

Others have shown that monosaccharides can be resolved by both paper and slab gel electrophoresis with other derivatizing reagents. Tungstate and other oxyanions have been used in place of borate and complexation with metal cations has also been employed as a means of resolving monosaccharides.

Derivatization for electrophoresis is primarily done for ease of detection but there has been some work done with charged derivatizing reagents to improve resolution.

SLIDE 28: Electrophoresis advantages and disadvantages

Sensitivity in the picomole range is difficult without a gel reader. Running samples in parallel has the same problems often experienced with protein or DNA gels; "smiling" (where the edges of the gel run slower than the center), or slanting with standards running much faster on one side of the gel than the other. While this technique might be good for a researcher, it might not be robust enough for a QC lab. Other disadvantages include the inability to analyze non-reducing sugars and the difficulty in getting good quantitation from an gel imager. The separation tends to be relatively slow although the time required for derivatization is probably the most important.

SLIDE 29: Capillary electrophoresis (CE)

A lot has been written on the analysis of carbohydrates by CE. I did a review in 1994 (Olechno and Ulfelder in Handbook of Capillary Electrophoresis, ed. J. P. Landers, CRC Press) and used 124 references. I recently completed a second edition (due in September) that added 150 references over two years. Also see the article by Chiesa et al. referenced above. While Dr. Merkle is correct that the most common methods of analysis involve derivatization, a number of papers resolved and analyzed non-derivatized carbohydrates either at high pH (where, not only do they gain charge, but their absorbance increases) or with borate using only the natural absorbance. Composition data can be useful in the biotechnology industry as an indicator of changes in the oligosaccharides during production or processing. For the researcher, differences in composition can provide important insight into possible oligosaccharide structure.

SLIDE 30: CE advantages and disadvantages

While very high sensitivity has been achieved on a mass basis (sub-attomole as indicated if you derivatize with the appropriate reagent and have a laser and can do laser-induced fluorescence), this is in a very small volume (nanoliters to picoliters) which means relatively high concentrations (mM to mM) for most analyses. Also, while very little is injected, approximately 1000 times as much sample as is injected must be placed into the instrument so that an injection can be done. (Most instruments require at least 5 microliters for reproducible successful injections and 10 to 50 microliters is more realistic.) Secondly, when limits of detection are described they are usually done by serially diluting a known sample until the signal is 2X noise. But, as described by researchers at Beckman a few years ago, this limit is almost meaningless for real samples. If a sample containing sugar near the limit of detection as described above is derivatized, one usually finds that impurities in the derivatizing reagent are significantly more prominent than the analyte itself. These impurities were diluted away during serially analysis. They don't go away when the analyte is in low concentration to begin with. Note that this problem would occur with HPLC derivatizations as well as with CE or slab gels.

Dr. Merkle reports that reacetylation before analysis is needed for amino sugars. In many cases, this is not necessary. The amount of derivatizing is thousands of times higher than the concentration of sugar and, therefore, there is little lost to inter- and intra-reactions of amino sugars with reducing sugars. On the other hand, acetylation will remove any chance of reaction of amino sugars with reducing sugars.

Reproducibility of analysis has been a recurring problem with CE regardless of the analyte. Improved instrumentation, the use of internal standards and a better understanding of how best to design buffers have all lead to improved reproducibility.

SLIDE 31: Conclusions

These conclusions are mine (JDO) and do not necessarily reflect the views of PE-Applied Biosystems (my employer), the ABRF or Dr. Roberta Merkle, the presenter.

Monosaccharide composition analysis of glycoproteins can provide useful information about the types of oligosaccharides decorating a glycoprotein. QC groups could use monosaccharide composition as a method of analysis to monitor changes in glycosylation of a protein which may lead to biological changes in product.

Monosaccharide analysis at high levels (nanomoles-micromoles) is relatively easy and can be approached numerous ways. High sensitivity analyses are difficult with almost any technique available. Each lab should decide which technique is most appropriate for them based upon a few considerations: 1. lab expertise, 2. lab funding, 3. number of samples per day or week, 4. sensitivity needed, 5. support (technical) needed, types of analytes being analyzed (e.g., limited to glycoproteins or including cell wall analyses or starches or naturally occurring alditols). Those labs that will use carbohydrate analysis infrequently may find their needs best addressed by either outside "core labs" or by various derivatization techniques followed by HPLC (or if available, CE). Those labs that will do frequent carbohydrate analyses on many samples will probably find it easier to dedicate money and personnel to carbohydrate specific instrumentation including HPAEC or GC/MS. Colorimetric and slab gel techniques can provide useful qualitative results but are provide poor quantitation.

Significant work has been done in the areas of carbohydrate analysis. Before you do experiments, spend some time on the phone or in the library. Remember, "A few months of lab work can easily save you a day in the library. Ask your vendors for suggestions but be careful of the replies.