Glycosylation and Serological Reactivity of an Expression-enhanced SARS-CoV-2 Viral Spike Mimetic

Abstract

Extensive glycosylation of viral glycoproteins is a key feature of the antigenic surface of viruses and yet glycan processing can also be influenced by the manner of their recombinant production. The low yields of the soluble form of the trimeric spike (S) glycoprotein from SARS-CoV-2 has prompted advances in protein engineering that have greatly enhanced the stability and yields of the glycoprotein. The latest expression-enhanced version of the spike incorporates six proline substitutions to stabilize the prefusion conformation (termed SARS-CoV-2 S HexaPro). Although the substitutions greatly enhanced expression whilst not compromising protein structure, the influence of these substitutions on glycan processing has not been explored. Here, we show that the site-specific N-linked glycosylation of the expression-enhanced HexaPro resembles that of an earlier version containing two proline substitutions (2P), and that both capture features of native viral glycosylation. However, there are site-specific differences in glycosylation of HexaPro when compared to 2P. Despite these discrepancies, analysis of the serological reactivity of clinical samples from infected individuals confirmed that both HexaPro and 2P protein are equally able to detect IgG, IgA, and IgM responses in all sera analysed. Moreover, we extend this observation to include an analysis of glycan engineered S protein, whereby all N-linked glycans were converted to oligomannose-type and conclude that serological activity is not impacted by large scale changes in glycosylation. These observations suggest that variations in glycan processing will not impact the serological assessments currently being performed across the globe.

Keywords: SARS-CoV-2; antibody; glycoprotein; glycosylation; serology.

Graphical abstract

Figure 1

Representation and characterization of recombinant SARS-CoV-2 spike protein. (A) The protein domains are represented as N-terminal domain (NTD), receptor-binding domain (RBD), fusion peptide (FP), heptad repeat 1 (HR1), central helix (CH), connector domain (CD), and transmembrane domain (TM). The fusion cleavage site is illustrated as dashed lines (blue). N-linked glycosylation sequons (N-X-S/T, where X ≠ P) are shown as branches. SARS-CoV-2 WT presents S1 and S2 domain with furin cleavage site (RRAR) and transmembrane domain at C-terminal end. SARS-CoV-2 2P prefusion stabilized protein with proline substitutions at residues 986 and 987 and, a “GSAS” mutation at furin cleavage site (residues 682–685). HexaPro prefusion stabilized protein of SARS-CoV-2 with a “GSAS” mutation at furin cleavage site and six proline substitutions, highlighted in red. (B) Structural representation of HexaPro S protein illustrating six proline substitutions (red spheres) in SARS-CoV-2 ectodomain (PDB ID: 6XKL). The S1 subunit along with N-glycans are shown as transparent molecular surface. The S2 subunit is shown in dark grey. Different domains present in S1/S2 subunits are highlighted in respective colors in ribbon diagram of only one protomer.

Figure 2

Characterization of recombinant SARS-COV-2 S protein, HexaPro. (A) SEC of affinity-purified recombinantly expressed S protein. Dashed lines indicate fractions collected for subsequent use. (B) SPR of HexaPro (ligand) with soluble ACE2 receptor (analyte). Dark blue to light blue lines represent the serial dilutions of ACE2 protein from 200 nM to 3.125 nM, respectively. Black lines are fitted values of the respective concentration to illustrate the best fit to a 1:1 binding model. Three repeats were performed and averaged to determine the k_a, k_d and K_D values. (C) Residual plot illustrating the deviation of the fitted data to the raw values of the experimental data at different concentrations. (D) Representation of K_D values determined using various repeats, grey dots illustrate the binding of 2P with ACE2 (values reproduced from 56) and the binding of HexaPro with ACE2 are shown as blue triangles. The mean of K_D values of 2P is plotted as a black line and the error bars represent ±standard deviation calculated using GraphPad Prism.

Figure 3

Site-specific glycosylation of expression-enhanced recombinant trimer of SARS-CoV-2 S protein (HexaPro). (A) Relative quantification of the N-linked glycosylation sites of trimeric S protein, produced in HEK293F cells. The bar graph represents the mean of three independently expressed replicates with error bars representing the standard error of the mean. The color codes in the schematic illustrates the processing state of glycans from least processed to most processed, oligomannose (green), hybrid (dashed pink), and complex glycans (pink). The proportion of unoccupied N-linked glycan sites are represented in grey. The pie charts summarize the quantification of these categories. The N-linked glycan site labels are colored based on the oligomannose-type glycan content, green (80–100%), orange (30–79%) and magenta (0–29%). (B) The model was constructed using the prefusion structure of trimeric SARS-CoV-2 S glycoprotein as described in Materials and Methods. The S1 and S2 subunits are shown as light and dark grey, respectively. The glycans sites are categorized as high-mannose type glycans (green), hybrid glycans (orange), and complex-type glycans (pink). The ACE2 receptor binding site is shown in blue.

Figure 4

Comparison of glycan composition across prefusion-stabilized SARS-CoV-2 S protein. (A) The percentage point change in oligomannose-type glycan content between SARS-CoV-2 S protein, HexaPro and 2P. The percentage point (p.p.) difference on the y-axis is the arithmetic difference between the percentiles of oligomannose-type glycans between the two populations (here defined as: p.p. = % HexaPro – % 2P). Positive values (blue) indicate a higher abundance of oligomannose-type glycans in HexaPro relative to 2P. Negative values (red) indicate a lower abundance of oligomannose-type glycans in HexaPro relative to 2P. (B) A full length model of SARS-CoV-2 S protein with N-glycans colored based on the percentage point change values. The scale represents the differences in oligomannose-type glycans observed in HexaPro when compared to 2P protein. Colors correspond to p.p. values in Panel A. The model was constructed using prefusion structure of trimeric SARS-CoV-2 S glycoprotein, as detailed in Materials and Methods. (C) Correlation of ASA values between HexaPro and 2P S protein. The average ASA values of all glycans from three replica simulations of two-RBD-up HexaPro (left side) and one-RBD-up (right side) structures plotted against the average ASA values from simulations of the respective 2P structures. (D) The average ASA values for 2P (black) and HexaPro two-RBD-up (red) with error bars showing standard deviations along the trajectories and across three repeat simulations. The displayed sites are those with changes in the oligomannose content across both versions. N74 (high ASA values) and N234 (lowest ASA values) were used as a reference. The chain (A, B, or C) of the trimeric S protein is indicated along the x-axis.

Figure 5

Antibody binding to spike glycoproteins. Individual serological responses from pre-2019 donors (Pre19, n = 8), non-hospitalized convalescent donors (NHC, n = 16) or PCR + hospitalized subjects (HS, n = 16) as determined by ELISA using HRP-labelled combined anti-IgG, IgA and IgM. A) Absorbance values of sera serially diluted from a starting dilution of 1:40 against 0.1 µg 2P (cyan bars) or HexaPro (blue bars). B) Absorbance values of sera serially diluted from a starting dilution of 1:40 from Pre-19 (black circle, dashed lines), NHC of HexaPro protein (blue squares) and Kifunensine-treated HexaPro (green squares) as determined by ELISA using HRP-labelled combined anti-IgG, IgA, IgM, and GAM. C) Area Under the Curve (AUC) of responses shown in figure B. The blue bars representing the AUC of HexaPro with IgG, IgA, IgM and GAM. The green bars representing the AUC of kif-treated HexaPro with different immunoglobulins. The mean ± standard deviation from the mean (SD) is plotted. D) HILIC-UPLC profile of N-linked glycans from WT (wildtype) HexaPro and Kifunensine-treated HexaPro produced in HEK 293F cells and purified by Ni⁺² column followed by SEC. The blue peaks representing glycan spectra of WT-HexaPro. The green peaks representing glycan spectra of kifunensine-treated HexaPro showing only Man₉GlcNAc₂ (M9) glycans.