New Approaches to Glycan Analysis

Protein glycosylation is responsible for the chemical diversity of many proteins, including—especially—monoclonal antibodies. Although several types of glycosylation are possible, N- and O-glycosylation (occurring on nitrogen and oxygen-containing residues, respectively) are the most studied and are considered critical quality attributes of mAbs and other therapeutic proteins.

Glycan characterization is typically achieved through one of three LC-MS workflows. The most straightforward approach, intact mass analysis, requires a minimum of sample preparation since it involves the entire, unmodified protein. Intact mass is routinely performed on monoclonal antibodies, says Min Du, Ph.D., Sr. Manager for Pharma/Biopharma Applications at Thermo Fisher Scientific. “This rapid analysis reveals the global glycoform distribution for major glycans, but suffers from sensitivity issues with low-incidence glycans.”

At the other end of the spectrum is released glycan analysis, whose first step is the enzymatic release of glycans from their conjugation points on the protein, followed by LC-MS analysis of either the intact glycans or their constituent monosaccharides.

In between, peptide mapping uses specific proteases, including trypsin, to generate small peptides for subsequent analysis. Peptide mapping yields highly sensitive, site-specific glycan profiling.

Multi-attribute assays

Released glycan approaches employ high-performance anion-exchange chromatography with pulsed amperometric detection (HPAE-PAD), as opposed to standard normal- or reverse-phase HPLC. HPAE-PAD, which combines high-sensitivity anion exchange separation with electrochemical detection, avoids derivatization prior to detection, and quantifies both total glycosylation and individual monosaccharides.

HPAE provides complementary information to LC-MS for characterizing glycans, with the following advantages:

• Seamless integration with MS
• Derivatization-free sample preparation, which reduces data noise while avoiding sample degradation
• Suitability to a broad range of glycans, from neutral to heavily charged, which facilitates analysis of the acidic glycome (e.g., sialylated, sulfated, and phosphorylated)
• Through specialized column technology, the ability to resolve structural isomers

An emerging approach is the multi-attribute method (MAM), which extracts data on glycosylation and protein sequence, in addition to other critical quality attributes such as truncation, isomerization, oxidation, deamidation, and other post-translational modifications.

“One of the main drivers for MAM, compared with conventional released-glycan analysis, is that multiple attributes can be directly monitored with a single LC-MS method during product development, manufacturing, and release testing to improve operational productivity and ensure product quality and safety,” Du tells Biocompare. Some companies are incorporating MAM for at-line process analytics.

As Amgen’s Anurag S. Rathore noted in a January 2021 LCGC article, “MAM allows for monitoring of a biopharmaceutical throughout its life cycle, and with advances in automation (automated sampling) and data analytics, sampling from the bioreactor, followed by sample cleanup, MAM will be able to provide near real-time measurements of the product. In time, MAM will be more routinely considered a tool in our process analytical technology (PAT) toolbox.”

Ease of use

Glycan diversity is both a curse and a blessing. The combinatorial possibilities, even for glycans consisting of two sugars, is theoretically infinite (although limited in practice by the organism and available sugars). On the other hand, the aggregate of glycosylated protein species has been used as a quality signature to compare batches or production methods, for example in establishing biosimilarity.

According to Andrew Hanneman, Ph.D., Scientific Advisor, Biologics Solutions, Charles River Laboratories, hydrophobic interaction HPLC profiling by fluorescent labeling has come to dominate routine quality analysis and comparability studies of mAbs due to the method’s ease-of-use. "High-resolution MS in a QC lab provides intact and subunit LC-MS, including native mass spectrometry, for rapid lot-to-lot glycan analysis. Other techniques include MALDI-TOF MS and capillary electrophoresis (CE) of fluorescently labeled glycans.”

At the same time in-line glycan analysis remains challenging since sugars lack a suitable chromophore. “But at-line MS is emerging, especially using small and robust spectrometers. High-resolution instruments are getting closer to the production line when rapid knowledge about glycosylation as a critical quality attribute is vital.

Additionally, rapid fluorescence labeling now provides faster readouts, while de novo glycan analysis using sequential mass spectrometry can handle new and unusual structures, supplanting NMR, in some instances, at small scale.

O-glycans—a special case?

Glycan release methods work exceptionally well for N-glycans because PNGase efficiently removes a broad range of N-glycans. Enzymes that remove O-glycans exist as well, but they show high selectivity for certain glycoproteins (sequence bias), they remove only mono- or disaccharide core glycans, and do not release O-glycans bearing sialic acid.

Chemical dissociation, which is quite efficient, induces a downstream cascade of chemical eliminations that degrade the target analyte. This effect is minimized by introducing reducing agents, but that limits subsequent fluorescent labeling. Plus (and this is true of all release-and-analyze methods) once a glycan is released there is no way to tell which amino acid it “belonged” to.

A group at New England Biolabs (NEB) has commercialized a novel O-glycoprotease technique, based on an immunomodulating metalloprotease (IMPa) from the bacterium, Pseudomonas aeruginosa. A recent journal article highlights this work.

Researchers characterized the specificity of IMPa using a series of synthetic test glycopeptides. They found that IMPa tolerated any residue adjacent to the cleavage site, except for aspartic acid, and was fully active on glycoproteins of varying O-glycan structural complexity, including branching and sialic acids. The paper also outlined a one-step O-glycoproteomic workflow for two therapeutic glycoproteins.

The only drawback, says Christopher Taron, Scientific Director at NEB, is that the method works less well with aspartic acid at the cleavage location.

“This means that some O-glycans attached at Ser/Thr residues preceded by aspartic acid might be missed. This is not a major drawback, but rather something to know so proteomic data may be interpreted appropriately.”

What happens after cleavage by the O-glycoprotease?

“You get a glycopeptide mixture, with each species incorporating an O-glycan on its N-terminal amino acid,” Taron adds. This glycopeptide pool can be analyzed using MS-based proteomics to determine O-glycosites (e.g., mapping the peptides present) and the range of structures that occupy each site. So, despite the desire of analysts to have more robust O-glycan release-and-analyze methods, the approach utilizing O-glycoprotease is more data-rich in that it provides glycosite location and O-glycan structural information in the same experiment.”

This is “an important layer of information,” Taron says, that typical release-and-analyze approaches miss for O-glycans. “Compared with N-glycans, O-glycans are often highly clustered in many glycoproteins. The serine and threonine residues on which O-glycans reside are not easily predicted computationally—in contrast to N-glycans that have a conserved canonical attachment site. As such, two critical layers of information exist for profiling O-glycans: the position of each O-glycosite and the O-glycan structures that can occupy each site.”