Removing N-glycosylation of the ZF2001 vaccine affects neutralizing antibody levels
Neutralizing antibody titers in the serum of immunized mice are typically used to evaluate humoral immunity. In this section, the immunization protocol used for mice was identical to that used for the RBD protein vaccine (Fig. 2a). We immunized mice with 1 μg or 5 μg of ZF2001 antigen (RBD-wild type (WT)) in the untreated groups (control group: control-1 μg, control-5 μg) and the peptide: N-glycanase (PNGase F)-treated groups (delete-group: delete-1 μg, delete-5 μg). Neutralizing antibody levels were detected in the serum of mice 14 days after the second immunization using a pseudovirus neutralization assay. Compared with those in the control group, the serum neutralizing antibody levels in the deletion group significantly decreased (p 1b). Specifically, the antibody level in the delete-1 μg group was 4.4-fold lower than that in the control-1 μg group, whereas the antibody level in the delete-5 μg group was 4.8-fold lower than that in the control-5 μg group (Fig. 1b). These findings indicated that removing the N-glycosylation of the antigen had a significant influence on neutralizing antibody levels. However, the glycosylation site with the most significant effect remains unknown. Thus, we conducted an in-depth exploration of the effects of specific glycosylation sites on the immunogenicity of the vaccine and its underlying mechanisms.
Preliminary study and effects of site-directed mutations on immune responses induced by the SARS-CoV-2 recombinant RBD nucleic acid vaccine. a Recombinant RBD nucleic acid immunization schedule and time points for blood collection and spleen extraction; b Left: Removing N-glycosylation from the ZF2001 vaccine affected neutralizing antibody levels: ZF2001 vaccine antigen untreated group (1 μg ; 5 μg), the N-glycosylation of ZF2001 vaccine antigen was removed by PNGase F group (delete-1 μg; delete-5 μg) (n=10); Right: The absence of residual PNGase F was confirmed by 12% SDS‒PAGE under nonreducing conditions; c Titers of binding antibodies (n = 10) (left: 5 μg; right: 20 μg); d Neutralizing antibodies (n = 10) (left: 14 days after two immunizations; right: 14 days after three immunizations) RBD-NA, green; O-T323-NA, red; N-N331-NA, purple; N-N343-NA, blue; the data were presented as the geometric mean ± geometric SD, and each symbol represented a mouse. e Cellular immune response assessment (n = 6) (RBD-NA, green; O-T323-NA, red; N-N331-NA, blue; N-N343-NA, pink) results of the 5 μg dose group (left); and the 20 μg dose group (right); the data were presented as the geometric mean ± SD, and each symbol represented a mouse. Statistical analysis was conducted using one-way ANOVA and Tukey’s multiple comparison tests for bar graphs; *p p p p
Impact of site-directed mutations on the immunogenicity of the recombinant COVID-19 RBD nucleic acid vaccine
Serum antibody titers in immunized mice serve as a common indicator of humoral immunity. In this study, mice were immunized with plasmids containing single glycosylation site mutations (N-N331-NA, N-N343-NA and O-T323-NA, as well as RBD-NA as a control) (Supplementary Fig. 2). A magnetic particle chemiluminescence assay for mouse RBD-IgG titers was subsequently used to determine antibody levels. Fourteen days after the two immunizations (Fig. 1c), the RBD-IgG titers of the RBD group at both doses were significantly greater than those of the other three groups (p 1c) in the 5 μg dose groups, the antibody levels in the RBD-NA group (2880 mIU/ml) were 2.2-fold, 2.0-fold, and 2.4-fold higher than those in the O-T323-NA group, N-N331-NA group, and N-N343-NA group, respectively. In the 20 μg dose groups, the antibody levels in the RBD-treated group (2950 mIU/ml) were 1.5-fold, 2.8-fold, and 3.2-fold greater than those in the O-T323-NA, N-N331-NA, and N-N343-NA groups, respectively. Antibody levels in the O-T323-NA group were also significantly greater than those in the N-N331-NA and N-N343-NA groups (p 1c).
Neutralizing antibody titers in the serum samples were evaluated using a pseudovirus neutralization assay, and geometric mean titers (GMTs) of the EC50 values were calculated. Fourteen days after the two immunizations (Fig. 1d), the GMTs of the RBD-NA group were 780 (20 μg dose) and 298 (5 μg dose), which were significantly greater than those of the N-N331-NA and N-N343-NA groups (p 1d), the GMTs of the RBD-NA group were 1420 (20 μg dose) and 700 (5 μg dose), which were significantly greater than those of the N-N331-NA and N-N343-NA groups (p p 1d).
In this study, three cytokines (IFN-γ, IL-2, and IL-4) secreted by the splenic lymphocytes of mice were detected using an enzyme-linked immunospot (ELISPOT) assay. The results indicated that mutations in glycosylation sites also have a significant negative effect on cellular immunity. Spot-forming cells (SFCs) of the three cytokines (Fig. 1e), IFN-γ, IL-2 and IL-4 in the RBD-NA group at both doses were significantly greater than those in the other three groups (p 1e). Together, these findings suggest that the N343 glycosylation site plays a more important role in the immunogenicity of the vaccine (Fig. 1).
Given that glycosylation deficiency can induce unfolded protein responses (UPR) and impair protein expression,2 which in turn may influence the immunogenicity of nucleic acid vaccines through altered in vivo antigen production, glycosylation-deficient RBD variants—N-N331-NA, N-N343-NA, O-T323-NA, and an RBD-NA control—were expressed in Chinese hamster ovary (CHO) cells to obtain purified recombinant proteins. These proteins were subsequently used to immunize mice to evaluate the immunogenic consequences and molecular mechanisms of site-specific glycosylation loss.
Impact of site-directed mutations on the immunogenicity of the recombinant COVID-19 RBD protein vaccine
RBD-IgG levels were detected using the same methodology as described above. Fourteen days after the two immunizations (Fig. 2b), the antibody concentration in the RBD-Pr group (2935 mIU/ml) was significantly greater than that in the other three groups, being 2.3-fold, 4.5-fold, and 10.9-fold greater than that in the O-T323-Pr group, N-N331-Pr group, and N-N343-Pr group, respectively. Antibody levels in the O-T323-Pr group were significantly greater than those in the N-N331-Pr and N-N343-Pr groups (p 2b).
Effects of site-directed mutations on immune responses induced by the recombinant RBD protein vaccine. a Recombinant RBD protein immunization schedule and time points for blood collection and spleen extraction; b Titers of binding antibodies (n = 10); c Neutralizing antibodies at different time points (RBD-Pr, green; O-T323-Pr, red; N-N331-Pr, purple; N-N343-Pr, blue). Line chart of the fold decrease in neutralizing antibody levels compared with those in the RBD group in each group; the data were presented as the geometric mean ± geometric SD, and each symbol represented a mouse. d Cellular immune response assessment (n = 6) (RBD-Pr, green; O-T323-Pr, red; N-N331-Pr, blue; N-N343-Pr, pink). The data were presented as the mean ± SD, and each symbol represented a mouse. Statistical analysis was conducted using one-way ANOVA and Tukey’s multiple comparison tests for bar graphs; *p p p p
Neutralizing antibody levels in the serum of the mice were measured using the pseudovirus neutralization assay as described above (Fig. 2c). At both time points, the serum GMTs of the mice in the RBD-Pr group were significantly greater than those in the other three groups (p 2c), the GMT of the RBD-Pr group (1792) was 1.8-fold, 4.3-fold, and 35.0-fold greater than those of the O-T323-Pr, N-N331-Pr, and N-N343-Pr groups, respectively, with the GMT of the O-T323-Pr group being significantly greater than those of the N-N331-Pr and N-N343-Pr groups (p 2c).
The ELISPOT results revealed that the SFCs of four cytokines in the RBD-Pr group (59, 45, 46, and 37) were significantly greater than those in the other groups (p 2d).
In conclusion, the negative impact of mutation of N-glycosylation sites was generally greater than that of mutation of O-glycosylation sites. However, in addition to the impact of N-glycosylation, there may be other reasons for the increased negative impact of mutation of the N343 site on the immunogenicity of the vaccine, such as antigen structure and protein stability.
Cross-neutralization of pseudoviruses with multiple variants and single glycosylation site mutations
To investigate whether the long-term protective effect of glycosylation on vaccine immunogenicity aligns with previous findings, mouse serum-neutralizing antibodies were evaluated using a pseudovirus neutralization assay 150 days after the two immunizations (Fig. 3a). Compared with those in the prototype strain group, the GMTs in each group decreased compared with those observed 14 days after the second immunization. Compared with the control group, the RBD-Pr, O-T323-Pr, N-N331-Pr, and N-N343-Pr groups presented decreases of 1.4-fold, 1.3-fold, 1.6-fold, and 1.2-fold, respectively, with the RBD-Pr group showing significantly higher levels (p 3a).
Neutralizing antibody levels 150 days after the two immunizations. a Neutralization of pseudoviruses with the WT-1 strain; Cross-neutralization of pseudoviruses with multiple variant strains, B.1.1.7 (b), B.1.351 (c), B.1.617.2 (d), BA.2 (e), and BA.4/5 (f); Cross-neutralization of pseudoviruses with single glycosylation site mutations, N331 (g), N343 (h), T323 (i), and WT-1(j). The data were presented as the geometric mean ± geometric SD, and each symbol represented a mouse (n = 8). Statistical analysis was conducted using one-way ANOVA and Tukey’s multiple comparison tests for bar graphs; *p p p p
Given the conserved nature of the N331, N343, and T323 glycosylation sites on the RBD, which remain unchanged across subsequent variant strains, we aimed to investigate whether the perspectives derived from previous results could be applied to different variants. Therefore, multiple variant pseudoviruses (B.1.1.7 [α], B.1.351 [β], B.1.617.2 [δ], BA.2 [O], and BA.4/5 [O]) were used to assess antibody levels 150 days after the two immunizations (Fig. 3b, f). For these five variants, the neutralizing antibody levels in the RBD-Pr group were significantly greater than those in the N-N331-Pr, N-N343-Pr, and O-T323-Pr groups (p 3b, f). The O-T323-Pr group levels were higher than those in the N-N331-Pr and N-N343-Pr groups (Fig. 3b, f). Compared with those against the WT-1 pseudovirus, overall neutralizing antibody levels against variants decreased, especially for Omicron.
To investigate the protective efficacy of current vaccines against future variants with mutations at the N331, N343, and T323 glycosylation sites, neutralizing antibody levels were detected 150 days after two immunizations using single-glycosylation-site-mutant pseudoviruses (Fig. 3g, i). With respect to T323, the GMT of the O-T323-Pr (1539.7) group was significantly greater than that of the RBD-Pr, N-N331-Pr, and N-N343-Pr groups (p 3i). With respect to N331, the GMT of the RBD-Pr (1553.9) group was greater than that of the N-N331-Pr (43.3) and N-N343-Pr (42.8) groups (p 3g). With respect to N343, the GMT of the RBD-Pr (1312.3) group was greater than that of the two N-glycosylation-mutated groups (p 3h). The results obtained 150 days after immunization revealed that mouse serum still strongly neutralized monoglycosylation site-mutated pseudoviruses.
In the T323 pseudovirus experiment (Fig. 3i), the O-T323-Pr group performed the best, likely because of antigen similarity. The absence of these findings in the N331 and N343 pseudoviruses may be due to the low immunogenicity of the N-N331-Pr and N-N343-Pr groups or because the T323 mutation affects mainly antigenicity, whereas the N331 and N343 mutations affect both immunogenicity and antigenicity. The RBD-Pr group still performed well, suggesting that existing wild-type vaccines may protect against future strains with RBD region glycosylation site mutations.
Effects of antigen deglycosylation on the activation of MHC-I- and MHC-II-expressing immune cells
The spleen plays a crucial role in immune regulation processes. When exogenous antigens invade the body, a series of immune responses are triggered. CD80 and CD86, surface markers of immune cells, play important roles in activating T lymphocytes. MHC-I binds to the CD8⁺ T-cell receptor, helping it recognize virus-infected cells. MHC-II binds to the CD4⁺ T-cell receptor, assisting in the initiation of an immune response and the regulation of cellular and humoral immune responses.17
Mouse splenic lymphocytes were stimulated in vitro with 10 μg/mL antigen (Fig. 4a). Among all the groups, the proportion of MHC-I-expressing cells with both CD80⁺ and CD86⁺ expression was significantly greater in the RBD-Pr group than in the O-T323-Pr group, N-N331-Pr group, and N-N343-Pr group. For MHC-II-expressing cells with CD80⁺ and CD86⁺ expression (Fig. 4a), there was no significant difference among the RBD-Pr group, O-T323-Pr group, and N-N331-Pr group. In this cell population, the results for only the RBD-Pr group were significantly greater than those for the N-N343-Pr group (Fig. 4a). Moreover, in the 50 μg dose group, the RBD-Pr group also performed the best (supplementary Fig. 1).
Influence of glycosylation site mutations on antigenicity. a Activation of MHC-I- and MHC-II-expressing immune cells; b Binding ability of the antigen to antibodies and the ACE2 receptor: when the RBD was used as the standard (white), if the affinity was greater than that of the RBD, the color was blue; if the affinity was less than that of the RBD, the color was red. The deeper the color was, the stronger (blue) or weaker (red) the degree of antigen binding. The data were presented as the mean ± SD, and each symbol represented a mouse (n = 3). Statistical analysis was conducted using one-way ANOVA and Tukey’s multiple comparison tests for bar graphs; *p p p p
Compared with the MHC-II-expressing cell population, the differences in MHC-I-expressing cells between the RBD-Pr group and the other three groups were more significant (Fig. 4a). It was speculated that mutation of the glycosylation sites of the antigen has a greater effect on antigen presentation to CD8⁺ T cells, reducing the expression of MHC-I on cells. This finding demonstrated that mutation of the N343 glycosylation site had the most significant effect. It is also hypothesized that the potential mechanism underlying the reduced T-cell responses observed with the N343 mutation may be associated with “altered antigen processing efficiency”.4
Effects of mutations at specific glycosylation sites on the antigenicity of the RBD
ELISAs were performed to evaluate the effects of mutation of the T323, N331, and N343 glycosylation sites on antigen receptor binding and antigenicity. Therefore, binding to ACE2 and six classes of RBD-targeting neutralizing antibodies (class Ⅰ–VI)18 was assessed (Fig. 4b).
Binding motifs and sites were determined using virus structure clustering. The antigenicity of deglycosylation decreased significantly. Binding to XGv191, XGv038, and XGv157 decreased 2.61-fold, 2.34-fold, and 1.57-fold, respectively. Binding to S309 and ACE2 decreased 5.94-fold and 3.70-fold, respectively. Mutation of N331 and T323 did not affect binding to ACE2 or the six antibody classes (ratio≈1.0, 0.95–1.04), likely because these are distant from the receptor binding motif (RBM) and are nonhot antibody-binding regions.19 The N343 mutation had a greater effect. Compared with RBD-Pr, N-N343-Pr exhibited 4.12-fold and 3.25-fold decreases in binding to S309 and ACE2, respectively. Binding to some Class Ⅳ and Ⅴ-Ⅵ antibodies also decreased significantly (XGv191: 2.11-fold; XGv038: 1.68-fold; XGv157: 1.36-fold), while binding to Class I-III and XGv286 was unaffected (Fig. 4b).
N343 mutation affects antigen-ACE2 binding, as the N343-linked glycan is involved in the RBD-ACE2 interaction. Mutation-induced amino acid changes at N343 led to the loss of RBD-S309 hydrogen bonds, reducing the binding ability (4.12-fold) (Fig. 4b). N343 mutation also affected the binding of class Ⅳ (XGv191), Ⅴ and Ⅵ antibodies, likely due to changes in antigen structure. These changes, near N343, affected the binding of class Ⅴ-Ⅵ and some class Ⅳ (XGv191) antibodies but not class I-III and some class Ⅳ (XGv286) antibodies. Further research on N343 mutation-induced protein structure changes is therefore warranted.
Glycan analysis of specific glycosylation sites
Glycan profiling of the T323, N331, and N343 sites of RBD-Pr was performed. Peptide mapping was performed using Biologics Explorer 4.0 software, followed by database retrieval (Supplementary Fig. 3, Supplementary Table 2). The glycan type numbers and glycosylation rates at each site were analyzed (Supplementary Table 1). At T323, this included Core1 (42.60%), Core1-S1 (11.87%), and Core1-S2 (0.02%) (Fig. 5a). N331 had 9 glycan types, with A2F1 (15.00%), A2G1F1 (10.70%), and A3G1F1 (6.20%) being the most common (Fig. 5c). N343 had 29 glycan types, with A2F1 (30.44%), A1F1 (19.12%), and A3F1 (12.90%) being the most common (Fig. 5b). The glycosylation coverage rates were as follows: T323 (54.5%), N331 (47.8%), and N343 (98.5%) (Fig. 5d). The results revealed that N343 had the highest degree of glycosylation coverage and the most complex types, including several mannose- and galactose-containing types. These findings also suggest that increasing the glycosylation coverage rate of antigens may increase vaccine immunogenicity. The glycan database of the software is presented in Supplementary Table 3.
Glycan forms of glycosylation sites in RBD samples, locations of glycosylation sites, and α1/α2 helices on the recombinant RBD. Glycan types and ratios at the T323 site (a), the N343 site (b), and the N331 site (c); d Glycan occupancy at each site; e Distances between the α1 and α2 helices for unmodified, T323A, N331Q, and N343Q recombinant RBDs across nine SARS-CoV-2 variants, the data were presented as the mean ± SD
Since phagocyte-based galactose and mannose receptors can be activated by these glycans for antigen presentation,20 mutation of N343 likely affects immunogenicity, antigenicity, and in vitro stimulation because of reduced galactose and mannose levels. Therefore, studying the effects of different antigen–glycan types on antigenicity and immunogenicity is crucial for the quality control of vaccines.17
Secondary structure analysis using microfluidic modulation spectroscopy
MMS, a protein secondary-structure study technique, was used to analyze the spectra of four protein samples (RBD-Pr, N-N331-Pr, N-N343-Pr, O-T323-Pr) for secondary-structure composition (α-helix, β-sheet, random coil, etc.). The results revealed that they were mainly β-sheets and β-turns, with detectable structural differences among the samples (Supplementary Fig. 5). Relative to those of RBD-Pr, N-N343-Pr exhibited the most distinct differences among the three mutant samples, with decreased intermolecular (β) and intramolecular (β-) β-sheet structures, increased β-turn structures, and main change-region peak values at 1635 cm⁻¹ and 1672 cm⁻¹. The differences in the other samples were less obvious (Supplementary Fig. 5). These results indicate that mutation of the N343 glycosylation site affects the secondary structure of the recombinant protein the most. This may be important because of its significant effect on the immunogenicity and antigenicity of the COVID-19 vaccine.
Molecular dynamics simulations and charge heterogeneity analysis
To elucidate the structural mechanisms by which site-specific glycosylation mutations influence RBD immunogenicity and antigenicity, 5-ns MD simulations were conducted on fully glycosylated and three site-specific deglycosylated RBDs (T323A, N331Q, and N343Q) derived from nine representative SARS-CoV-2 variants (WT, Delta, BA.2, BA.4/5, BQ.1.1, XBB.1, BA.2.86, JN.1, and KP.3). Compared with the fully glycosylated RBD, the deglycosylated forms exhibited overall increases in the distance between the α1 and α2 helices by 0.3 Å (T323A), 0.5 Å (N331Q), and 0.8 Å (N343Q). Notably, this displacement negatively correlated with the spatial proximity of the glycosylation site to the α1/α2 helical interface.
To investigate the molecular basis of these conformational changes, representative snapshots from MD trajectories of the fully glycosylated and N343Q-mutated RBDs were analyzed to assess alterations in local interactions. In the WT and Delta variants, the glycan at N343 formed a stabilizing hydrogen bond with residue S371. In contrast, in variants from BA.2 onward harboring the S371F mutation, the N343 glycan interacted instead with the N370 residue. This hydrogen bond appeared to function as a molecular “latch”, stabilizing the α1/α2 helical interface and maintaining the structural integrity of the RBD. As expected, disruption of this glycan interaction in the N343Q mutant led to a consistent outward shift of the α2 helix across all nine variants, suggesting progressive destabilization of the RBD structure over longer timescales, potentially compromising its functional conformation (Fig. 5e, Supplementary Fig. 6, Supplementary Table 4). Although MD simulations are inherently mimetic and thus have certain limitations, they nonetheless provide valuable atomistic insights that complement experimental observations and help elucidate molecular mechanisms at a level often inaccessible to experiments alone.
Isoelectric-focusing electrophoresis separates charged variants from different posttranslational modifications (charge heterogeneity).21 Experimentally, N-N343-Pr differed notably from the other three samples. N-N331-Pr and O-T323-Pr were similar to the RBD-Pr control. Mutation of N343 caused the greatest change in protein charge heterogeneity, reflecting both differences in glycosylation and antigen structure changes due to amino acid alterations postmutation (Supplementary Fig. 4).
