Comparison of orthogonal chromatographic and lectin-affinity microarray methods for glycan profiling of a therapeutic...

The N-linked glycosylation of four lots of a marketed human therapeutic monoclonal antibody (mAb) was assessed by three orthogonal chromatographic methods and a commercial lectin microarray. For chromatography, the N-glycans were removed enzymatically from the mAbs using PNGase F. Native glycans were determined by HPAEC-PAD using a panel of 21 N-glycan standards and a multi-stage linear gradient eluent profile for sequential analyses of typical neutral and sialylated glycans in one chromatographic run. The monosaccharide contents of these glycans following acid hydrolysis were confirmed by HPAEC-PAD with monosaccharide standards. Glycosylation analysis by HILIC-FD after stoichiometric labelling with two different fluorescent tags (2-AA and 2-AB) enabled direct quantitation. The 2-AA- and 2-AB-labelled versions of the same glycan standard panel yielded distinctive separation profiles suitable for orthogonal identification of mAb glycans. Glycan profiling with the lectin microarray required partial denaturation of the intact mAbs to expose the sequestered Fc N-glycans. Glycosylation fingerprints were obtained using a fluorescently labelled antibody directed against human IgG Fc. Fluorescence intensities from the fingerprints were deconvoluted with a proprietary algorithm to obtain semi-quantitative "glycan structural class" information. Glycosylation analyses of the four mAb lots by these four methods, which separate and detect oligosaccharides according to different principles, provided complementary and corroboratory qualitative and quantitative information. The predominant N-linked structures were core-fucosylated asialo diantennary structures with varying galactosylation. There were also trace amounts of afucosyl and bisected glycans, but no detectable sialylation by any of the four methods. The therapeutic mAb demonstrated a high degree of consistency in the types and amounts of N-linked glycans in the four lots (<6% CV), and between all four analysis methods (<6% CV). The described methods are co-supported by the excellent quantitative agreement of their results, which is particularly notable considering the orthogonality of their separation and detection mechanisms.