Josh P. Roberts has an M.A. in the history and philosophy of science, and he also went through the Ph.D. program in molecular, cellular, developmental biology, and genetics at the University of Minnesota, with dissertation research in ocular immunology.
As tools to recognize and analyze the carbohydrates (called glycans) that decorate proteins are improving, researchers are becoming ever more appreciative of the roles these moieties play in biology and medicine. Glycans are involved in cell signaling and adhesion, and they mark malignant tissue. The glycan complement found on antibodies, for example, “affect[s] the biological activity, the pharmacokinetics and the stability of the molecule,” says T. Shantha Raju, scientific director at Janssen Pharmaceuticals.
The vast majority of carbohydrates attach to the protein at an asparagine residue (N-linked glycans), or serines or threonines (O-linked glycans), through a core sugar structure. From these cores stem one or more antennae, themselves comprising varying numbers of linked sugar residues. Beyond this, the rules that govern composition and placement are barely known, and the task of determining which carbohydrate structure, if any, is found at which potential glycosylation site on any given protein is not a trivial one. In fact, it’s likely to be a heterogeneous mix of related structures that occupy a given site at different times.
Scientists from R&D to QC use techniques ranging from gel electrophoresis to tandem mass spectrometry (MS) to determine the presence and makeup of these sugar trees, with the specific approach depending in part on how much is already known and how much detail is needed. Here we look at the tried-and-true as well as the cutting edge among these methods.
Glycomics
There are two principal approaches to analyzing glycans, says Robert Gates, market segment manager for proteins and enzymes at Sigma-Aldrich: Either release them from the protein backbone and study the free glycans, or digest the protein and study the resulting glycopeptides.
The more streamlined “glycomics” (as opposed to “glycoproteomics”; see below) workflow is often used, in which carbohydrates are cleaved by a peptide:N-glycosidase such as PNGase F to study the overall heterogeneity of N-linked glycans decorating a protein. No corresponding enzyme will cleave all the O-linked glycans from the protein, so these are instead generally removed by chemical means. The resulting carbohydrates are then collected and studied.
N-glycan preparation is typically a low-throughput process. ProZyme’s GlykoPrep™ product line uses Agilent Technologies’ AssayMAP cartridges to digest and clean up as many as 192 samples in less than three hours. “The workflow is essentially the same theory and chemistry that people are doing now across three to four days,” says Justin Hyche, ProZyme’s director of sales and marketing.
Released glycans can be fluorescently labeled (most often with 2-aminobenzamide (2-AB) or 2-aminobenzoic acid (2-AA)) to make them easier to visualize during both separation and MS. These tags traditionally have required the toxic reducing agent sodium cyanoborohydride, says Mike Gibson, president and founder of QA-Bio, a distributor of glycan-analysis tools. His company sells the new LudgerTag Glycan Labeling Kits that instead use the nonhazardous reductant picoline borane in its 2-AA and 2-AB kits.
Running the labeled glycans through HPLC may yield a fingerprint of, say, five to 20 peaks, each of which is a unique glycoform, Gibson says. These can be matched against libraries of previous results—to the extent that they exist—as well as standards. Comparing the run time to a glucose ladder, for example, will indicate the number of monosaccharides the glycan contains.
Lectins—naturally occurring proteins that specifically bind to particular carbohydrate moieties—can be used to determine whether a glycan contains particular structures. Like antibodies, these can serve both as capture and detection reagents, and a wide variety is available from different vendors.
Another common analysis technique is “sequencing,” accomplished by applying exoglycosidases—either sequentially or in a matrix—that will specifically cleave a given terminal sugar moiety to reveal the identity and order of the monosaccharides that make up the glycan. Individual exoglycosidases and standards can be purchased from different vendors, and kits such as QA-Bio’s CarboSeq™ N Kit and ProZyme’s FACE® N-Linked Oligosaccharide Sequencing Kit are available, as well.
A lot of information can be obtained from the free glycans alone. “You can get the amount of sialylation, fucosylation, the number of antennae. You can characterize whether it’s high-mannose, complex or hybrid,” says Carlito Lebrilla, distinguished professor of analytical chemistry at the University of California, Davis. Lebrilla typically follows up his own glycan separations with MS.
Glycoproteomics
To pin down the site at which a glycan is attached, researchers often use proteases to cleave the glycoprotein, producing a glycopeptide that can be queried by MS. The asparagine to which an N-glycan is attached tends to be part of a consensus sequence, aiding in the analysis; no such sequence has been discovered for O-glycans.
But though a specific protease (such as trypsin) can yield “pretty good heterogeneity at specific sites,” says Lebrilla, the resulting data are not comprehensive. “We might see a lot of glycopeptides from a specific site, but you won’t see all the sites.” A nonspecific protease like pronase E or cathepsin, on the other hand, breaks the protein into smaller pieces to give better coverage; but in that case, Lebrilla says, “the problem is that the MS is very difficult to interpret.”
Lebrilla frequently uses porous graphitized carbon (PGC) to separate the glycans and smaller glycopeptides. With PGC, he notes, “you can actually separate out the different glycan isomers,” something that isn’t accomplished by looking at mass alone. He sometimes adds a third dimension, such as ion mobility separation, to the analysis, as well.
But ultimately, says Vernon Reinhold, a structural glycobiologist at the University of New Hampshire, “a comprehensive sequence will only be achieved by MSn, that is, multiple stages of disassembly.” The alternative, using HPLC-MS or MS/MS, results only in "a series of cartoons constructed on inference.”
It’s not practical to do de novo structural elucidation each time, and many like Reinhold believe the solution is to construct searchable libraries and use those for rapid identification. Fortunately, the numbers are not infinite: Researchers estimate there are fewer than 5,000 N-glycan structures in humans, with about 100 of these representing 99% of the abundance, notes Lebrilla. In other words, the goal is achievable—formidable, but achievable.