Simplifying glycan monitoring of complex antigens such as the SARS-CoV-2 spike to accelerate vaccine development

Abstract

Glycosylation is a key quality attribute that must be closely monitored for protein therapeutics. Established assays such as HILIC-Fld of released glycans and LC-MS of glycopeptides work well for glycoproteins with a few glycosylation sites but are less amenable for those with multiple glycosylation sites, resulting in complex datasets that are time consuming to generate and difficult to analyze. As part of efforts to improve preparedness for future pandemics, researchers are currently assessing where time can be saved in the vaccine development and production process. In this context, we evaluated if neutral and acidic monosaccharides analysis via HPAEC-PAD could be used as a rapid and robust alternative to LC-MS and HILIC-Fld for monitoring glycosylation between protein production batches. Using glycoengineered spike proteins we show that the HPAEC-PAD monosaccharide assays could quickly and reproducibly detect both major and minor glycosylation differences between batches. Moreover, the monosaccharide results aligned well with those obtained by HILIC-Fld and LC-MS.

Fig. 1. Comparison of the sample preparation required for LC-MS, HILIC-Fld, and HPAEC-PAD.

a Illustrates the seven required steps to obtain the results via LC-MS, encompassing denaturation, reduction, PNGase F and protease digestion, centrifugation, injection, and analyses. b The nine necessary steps to acquire the results are outlined, involving denaturation, PNGase F digestion, purification, evaporation, labeling, purification, sialidase treatment, injection, and analyses. Lastly, c highlights the four essential steps prior to obtaining the results via HPAEC-PAD, encompassing TFA or sialidase cleavage, evaporation, injection, and analyses.

Fig. 2. Monosaccharide analysis via HPAEC-PAD of glycovariants of the SARS-CoV-2 spike protein.

Results reported in mol monosaccharide mol⁻¹ monomer (protomer MW = 143,230.6 Da). Results are means of three triplicate injections and three triplicate reactions. Error bars are the standard deviation observed between the triplicate hydrolyses values. a Fuc, GlcN, Gal, and Man content obtained after TFA hydrolysis. The data were analyzed using two-way ANOVA, Tukey’s multiple comparisons test, comparing the content of each monosaccharide type between variants. Not all comparisons shown. For full comparisons see Supplementary Table 2. b Neu5Ac content obtained after enzymatic hydrolysis. For a full comparison see Supplementary Table 3. P values: <0.03 (*), <0.002 (**), and <0.0001(****).

Fig. 3. HILIC-UPLC-Fld chromatograms of PRO1-468 to PRO1-472 samples.

The chromatograms of PRO1-468 (purple), PRO1-469 (green), PRO1-470 (blue), PRO1-471 (red), and PRO1-472 (black) are represented with no offset on the x axis, but offsets on the y-axis to facilitate comparison. An expected representative structure is depicted on the right. To note, this representative structure indicates only a single version of many different glycoforms present. Glycans are depicted by the symbolic nomenclature for glycans using DrawGlycan-SNFG software. Green circle: Man, blue square: GlcNAc, yellow circle: Gal, red triangle: Fuc, purple diamond: Neu5Ac.

Fig. 4. Comparison of HILIC-UPLC-Fld chromatograms with no offset on the x axis, but offsets on the y axis to facilitate comparison of PRO1-471 (red) after sialidase treatment and PRO1-469 (non-treated, black) with expected typical structure.

Glycans depicted as by the symbolic nomenclature for glycans using DrawGlycan-SNFG software. Green circle: Man, blue square: GlcNAc, yellow circle: Gal, red triangle: Fuc, purple diamond: Neu5Ac.

Fig. 5. Example of glycan distribution per site.

Pie charts showing how glycans are distributed for six different sites. AL: α-lytic protease, GC: Glu-C, TR: trypsin. biantennary (black, wide diagonal stripes downward), triantennary (red, 30% dotted), tetra-antennary (blue, horizontal brick), hybrid (orange, narrow horizontal stripes), other glycans (includes glycans such as A1 and A1G, purple, solid diamond grid) and high-Man (green, solid fill).

Fig. 6. Monosaccharide analysis via HPAEC-PAD of three independent spike batches.

Results reported in mol monosaccharide mol⁻¹ monomer (MW = 143,230.6 Da). Results are the means of three triplicate injections and three triplicate reactions. Error bars are the standard deviation observed between the triplicate hydrolyses values. The data were analyzed using two-way ANOVA, Tukey’s multiple comparisons test, comparing the content of each monosaccharide type between variants. Not all comparisons are shown. For full comparison see Supplementary Table 7. a Fuc, GlcN, Gal, and Man content obtained after TFA hydrolysis. b Neu5Ac content obtained after enzymatic hydrolysis. The data were analyzed using a one-way ANOVA, Tukey’s multiple comparisons test, showing all comparisons. For exact p values see Supplementary Table 8. P values: <0.002 (**), <0.0002 (***) and <0.0001(****).

Fig. 7. Comparison of HILIC-UPLC-Fld chromatograms of 3 batches of the spike protein produced in CHO pools.

The three HILIC traces for samples PRO1-392 (blue), PRO1-394 (red), and PRO1-412 (black) with offsets on the y axis to facilitate comparison and no offsets on the x axis are depicted.