Many types of biotherapeutics, either currently under development or in clinical use, are glycoproteins. This class of biopolymer is composed of a protein backbone decorated with one or more glycan (carbohydrate) chains. The heterogeneous nature of g
NATURE OF GLYCOSYLATION
Glycans are not produced in the same highly ordered manner as the protein chain; rather they are the product of the action of a series of sequentially localised biosynthetic enzymes involved in remodelling of the glycan chain as the glycoprotein passes through the endoplasmic reticulum, Golgi and trans-Golgi network. This form of structural processing during bulk flow results in a heterogenous population of glycans on the glycoprotein. The heterogeneity observed is also related to the nature of the protein chain. Glycoproteins with sites of glycan attachment on exposed surfaces have more processed (and therefore usually more heterogenous) glycans. Monoclonal antibodies, however, have the sites of glycan attachment in the Fc region of the protein and thus the action of glycosytransferases is hindered by stearic effects leading to the production of smaller, less heterogenous glycans.
This heterogeneity has led to the idea that the complexity of glycans puts them beyond the reach of characterisation techniques. This is not so and with the use of some carefully designed experiments and an understanding of glycan biosynthesis it is possible to characterise the glycans of a glycoprotein and fulfil the expectations for glycan analysis described in ICH Q6B .
Monoclonal antibodies carry so called N-glycans on their protein backbone. These are a class of glycan that contain a trimannosyl core attached to two N-Acetylglucosamine (GlcNAc) residues. One of these GlcNAc residues is the link through which the glycan is attached to the asparagine side chain of the protein. The mannose arms can be doubly, triply or quadruply substituted with GlcNAc residues, each of which can be extended with galactose (and possibly also GlcNAc) and capped with sialic acid and/or fucose. A fucose residue can also be found on the GlcNAc attached to the asparagine. The high number of possible glycan variants is limited by the nature of the location of the glycans on an antibody (as mentioned above).
ANALYSIS OF THE GLYCAN POPULATION
An initial screening of the glycans can be obtained using Matrix Assisted Laser Desorption Ionisation Mass Spectrometry) (MALDI-MS). Initially, the antibody is reduced and the subsequent free thiols chemically blocked. The glycoprotein is then digested with a protease to produce a series of peptides and glycopeptides. The enzyme peptide N-glycosidase F is then used to release the N-glycans. However, this enzyme does exhibit specificity in its substrate and will not release N-glycans with a fucose in α1-3 linkage to the core GlcNAc. This is a type of linkage often found in plant glycoproteins and would then require the enzyme peptide N-glycosidase A be used for N-glycan release instead. Once released, the glycans can be purified from peptides using solid phase extraction.
At this point, in our laboratory, it is common practise to permethylate the released N-glycans prior to mass spectrometric analysis. Firstly, all hydroxyl groups and other reactive species are converted to their methyl ethers and methyl esters respectively. This serves to “level the playing field” in terms of ionisation from the MALDI matrix and also increases sensitivity. This type of glycan analysis can be seen as semi-quantitative (but only semi-quantitative) and relative ratios, based on peak intensity, can be measured. Secondly, certain combinations of monosaccharides can produce N-glycans differing in mass by just one Dalton. This difference may give rise to some ambiguity in assignment depending on the quality or strength of the data. Therefore, permethylation of the glycans causes these structures to be more widely separated in mass and thus the potential for misassignments is avoided.
The data obtained from MALDI-MS analysis can be used to assign peaks based on the known residue masses of the monosaccharides and knowledge of the biosynthetic pathways of N-glycan production (Figure 1). An assignment of the signals in the data only allows compositions of N-glycans to be determined. It does not allow the assignment of particular monosaccharides within the structure. For example, is fucose (if present this is seen as a deoxyhexose mass in the assigned composition) on the core or the antenna? Is there any evidence of the Galα1-3Gal epitope which is known to be immunogenic?  Is there any evidence of a bisecting GlcNAc? Additionally, we cannot assign linkages to particular monosaccharides from the MALDI-MS data and this information could also be key for bioactivity, targeting or a necessary piece of information if a biosimilar is being produced. To that end we cannot say if, for example, sialic acid is linked to the three or six position of galactose. These questions can be addressed by further utilising mass spectrometric techniques.
IDENTIFICATION OF ANTENNAL STRUCTURES
Following MALDI-MS analysis, the permethylated N-glycans are analysed using electrospray ionisation mass spectrometry (ES-MS). This ionisation method produces a fine spray of ions that can undergo fragmentation in the source of the mass spectrometer. This is another advantage of the permethylated derivative since the fragmentation pathways are simple and well understood. The most abundant ions produced represent fragments of the N-glycan antennae, which is where the heterogeneity lies and is therefore the region of most interest. Careful control of source conditions allows fragmentation of antennae from the total glycan population. This data can be used to identify compositions of antennal structures that can be related back to compositions of the total population observed in the MALDI-MS data. The ions not observed in the ES-MS data should also be noted since this can help to rule out the presence of certain structural types. For example, lack of signals consistent with fucosylated antennae implies that, for a composition containing fucose, that residue must be on the core. In this technique the fragment ion composition consistent with the Galα1-3Gal epitope gives a signal at m/z 668. If this signal is seen then, whilst not being in itself definitive, should act as a warning sign for the presence of this epitope.
One other mass spectrometric technique that may provide useful structural information is the fragmentation of selected ions (the parent ion) using so-called MS/MS. This technique requires a mass spectrometer capable of selecting a parent ion, fragmenting this ion and scanning a mass range to identify the products of fragmentation (daughter ions). This technique is useful since it allows particular structures to be investigated. For example, if the in-source fragmentation data indicates the presence of antennal fragment ions of particular interest then these ions can be localised to certain structures by selecting specific precursor ions for MS/MS analysis and assessing for the presence of the ions of interest in the daughter ion spectra.
The use of MALDI-MS and ES-MS allows the assignments of glycan compositions and provides information on the antennal structures, helping to rationalise and confirm the postulated compositions. However, there is still one piece of structural information missing, namely the identification of the linkages of the monosaccharides within the glycan population.
IDENTIFICATION OF MONOSACCHARIDE LINKAGES
Monosaccharide linkage analysis is also performed on the permethylated glycans. These permethyl derivatives are structurally very stable and as such further chemical procedures can be performed on them. The process for linkage determination uses chemical conversion of the permethylated glycans to linear monosaccharides specifically tagged with acetyl groups wherever the monosaccharides were originally linked to one another. These monosaccharide derivatives are known as partially methylated alditol acetates (PMAAs). They can be separated, fragmented and analysed by gas chromatography mass spectrometry (GC-MS) which uses electron impact as the method of ionisation. The fragment ions are readily identifiable as originating from specific linked structures. For example, the fragmentation pattern of a hexose that was linked at its carbon 2 is very different from a hexose that is linked at its carbon 3. The use of gas chromatography allows separation of the different species and the mass spectrum allows identification of the linkage of each species (Figure 2).
The data generated from all of the above techniques, when combined, allows structures to be drawn for the glycan population of the glycoprotein under investigation. Monoclonal IgG antibodies most commonly have only one glycan chain per heavy chain (in the Fc region) and as such there is no need to extend the analytical process described above to separated glycopeptides, unless the IgG sequence indicates a possible second site of glycan attachment (eg in the variable region).
Glycan analysis is a detailed process and it is not always necessary to perform this level of investigation depending on what information is required at the time. For example, if cell line screening is taking place or bioreactor conditions are being investigated then it might be necessary only to perform a MALDI-MS analysis to assess the composition of the glycans and select the cell line or bioreactor process for further work.
During the course of the structural analysis of an antibody, a peptide map will be performed to ensure integrity of the protein sequence, most likely using reverse phase chromatography with on-line electrospray mass spectrometric analysis (LC/ES-MS). This data will also contain information on the glycans through the presence of a glycopeptide signal. Whilst this information is useful, it is important not to be tempted into using this data as a definitive measure of the glycan structures present on the sample. This is because glycopeptides readily fragment in the mass spectrometer source to produce fragment ions which are indistinguishable from naturally produced truncated glycans (Figure 3). Thus, glycan analysis should always be performed as a specific, standalone investigation.
TECHNIQUES FOR BATCH-TO-BATCH CHARACTERISATION
Characterisation of the N-glycans from a monoclonal antibody is necessary as part of a package of work providing full structural details of the molecule under investigation. However, once the stage of routine testing is reached, a simple chromatographic approach can be used to compare the glycan profiles across different batches. To this end, glycans from the glycoprotein under investigation should also be analysed chromatographically, producing a fingerprint profile against which the glycan profiles from other batches can be compared. For batch release this method must be validated. A useful chromatographic technique is the analysis of released glycans labelled with the fluorescent compound 2-aminobenzamide. These tagged glycans can be separated using HILIC in conjunction with fluorescence detection and mass spectrometric analysis to confirm the mass of the glycan. Since tagging is stoichiometric, the relative ratios of the different peaks can also be used when comparing batches.
Monoclonal antibody therapies form a significant sector of the global biopharmaceutical market and thus there is much investment in the investigation of molecular structure. The use of tailored mass spectrometric and chromatographic techniques allows for detailed investigation and on-going monitoring of the heterogeneous glycan structures present on this important class of biotherapeutics.
For further information, please contact:
Team Leader, Carbohydrate Analysis
SGS Life Science Services
t: +44 (0) 118 989 6940
1 Place des Alpes P.O. Box 2152, Geneva 1211, Switzerland