In the context of SARS-CoV-2's evolution and immune evasion, the glycosylation of the spike protein's receptor binding domain (RBD) is a focal point. The study delves into the specific role of N343 glycosylation across various variants, utilizing over 45 seconds of molecular dynamics (MD) simulations. Results indicate the amphipathic N-glycan at N343 preserves the RBD's structural integrity, with its removal causing a uniform conformational change, affecting the receptor binding motif (RBM) in early strains (WHu-1, alpha, beta) but not in later ones (delta, omicron). Omicron variants, with key mutations near N343, maintain RBD architecture independent of the glycan, underscoring an evolutionary adaptation. Empirical data on RBD binding to monosialylated ganglioside co-receptors show dependency on N343 glycosylation for WHu-1 but not for delta, suggesting additional evolutionary changes in co-receptor interactions.

Panel a) Atomistic model of the SARS-CoV-2 (WHu-1) S glycoprotein trimer embedded in a lipid bilayer as reported in ref(Casalino et al., 2020). In the conformation shown, the S bears the RBD of chain A in an open conformation, highlighted with a solvent accessible surface rendering. The topological S1 and S2 subdomains are indicated on the left-hand side. Glycans are represented with sticks in white, the protein is represented with cartoon rendering with different shades of cyan to highlight the chains. Panel b) Close-up of the open RBD (WHu-1) in a ACE2-bound conformation (PDB 6M0J), with regions colour-coded as described in the legend. Key residues for anchoring the FA2G2 (GlyTouCan-ID G00998NI) N343 glycan, namely S371, S373 and S375, across the beta sheet core are highlighted also in the Symbol Nomenclature for Glycans (SNFG) diagram on the bottom-right with links to the monosaccharides corresponding to primary contacts. Key residues of the hydrophobic patch (orange) found to be inverted in the recently isolated FLip XBB1.5 variant are also indicated. Panel c) Heat map indicating the interactions frequency (%) classified in terms of hydrogen bonding and van der Waals contacts between the N343 glycan and the RBD residues 365 to 375 for each VoC, over the cumulative conventional MD (cMD) and enhanced GaMD sampling. Panel d) Side view of the RBD with the antigenic Region 1 (green), Region 2 (or RBM in yellow), and Region 3 (orange) highlighted. Key residues Y351 and L452 at the intersection between Region 1 and the Receptor Binding Motif (RBM) are indicated, together with the predicted site for the GM1 co-receptor binding. Rendering with VMD (https://www.ks.uiuc.edu/Research/vmd/).

Molecular insights were acquired using all-atom force fields for protein, glycan, and solvent molecules, with analysis extending to variant under monitoring (VUM) omicron BA.2.86 ('pirola'). This variant has gained a new N-glycosylation site at N354, indicating the glycan shield's adaptive evolution.

Panel a) Kernel Density Estimates (KDE) plot of the backbone RMSD values calculated relative to frame 1 (t = 0) of the trajectory for Region 1 (green) aa 337-353, Region 2 (yellow) aa 439-506, and Region 3 (orange) aa 411-426 of the glycosylated (left plot) and non-glycosylated (right plot) WHu-1 RBDs. Duration of the MD sampling is indicated on the top-right corner of each plot with the conformational equilibration time subtracted as the corresponding data were not included in the analysis. Representative structures from the MD trajectories of the WHu-1 RBD glycosylated (cyan) and non-glycosylated (blue) at N343 are shown on the right-hand side of the panel. The N343 glycan (GlyTouCan-ID G00998NI) is rendered with sticks in white, the hydrophobic residues underneath the N343 glycan are highlighted with VDW spheres, while the protein structure is represented with cartoons. Panel b) KDE plot of the backbone RMSD values (see details in panel a) above) calculated for the alpha (B.1.1.7) RBD glycosylated (left) and non-glycosylated (right) at N343. Representative structures from the MD simulation of the alpha RBDs are shown on the right-hand side of the panel, with the N343 glycosylated RBD shown with pink cartoons and the non-glycosylated alpha RBD in purple cartoons. Panel c) KDE plots of the backbone RMSD values calculated for the beta (B.1.351) RBD glycosylated (left) and non-glycosylated (right) at N343. Representative structures from the MD simulation of the beta RBDs are shown on the right-hand side of the panel, with the N343 glycosylated RBD shown with orange cartoons and the non-glycosylated alpha RBD in red cartoons. Panel d) Binding affinities (1/Kd, x103 M-1) for interactions between different RBDs (including intact and endoF3 treated WHu-1 RBD and alpha and beta RBD) and the GM1os (GlyTouCan-ID G46613JI) and GM2os (GlyTouCan-ID G61168WC) oligosaccharides. HEK293a samples(Nguyen et al., 2021) and shown here as reference. HEK293b samples all carry FLAG and His tags and are shown for WHu-1 (glycosylated and treated with endoF3 treated), alpha and beta sequences. Further details in Supplementary Material. Panel e) Predicted complex between the WHu-1 RBD and GM1os, with GM1os represented with sticks in SNFG colours, the protein represented with cartoons (cyan) and the N343 with sticks (white). Residues directly involved in the GM1os binding or proximal are labelled and highlighted with sticks. All N343 glycosylated RBDs carry also a FA2G2 N-glycan (GlyTouCan-ID G00998NI) at N331, which is not shown for clarity. Rendering done with VMD (https://www.ks.uiuc.edu/Research/vmd/), KDE analysis with seaborn (https://seaborn.pydata.org/) and bar plot with MS Excel.

Panel a) KDE plot of the backbone RMSD values calculated relative to frame 1 (t = 0) of the MD1 trajectory for Region 1 (green) aa 337-353, Region 2 (yellow) aa 439-506, and Region 3 (orange) aa 411-426 of the N343 glycosylated delta (B.1.617.2) RBD. The MD1 simulation was started from the open RBD conformation from the cryo-EM structure PDB 7V7Q. Based on the conformation of the N-glycan reconstructed at N353, the first 100 ns of the MD1 production trajectory were considered part of the conformational equilibration and not included in the data analysis. Panel b) KDE plot of the backbone RMSD values calculated relative to frame 1 (t = 0) of the MD2 trajectory for Regions 1-3 (see details above) of the N343 glycosylated delta (B.1.617.2) RBD. The MD2 simulation was started from the open RBD conformation from the cryo-EM structure PDB 7V7Q with different velocities relative to MD1. The first 350 ns of the MD2 production trajectory were considered part of the conformational equilibration and not included in the data analysis. Panel c) KDE plot of the backbone RMSD values calculated relative to frame 1 (t = 0) of the GaMD trajectory for Regions 1-3 of the N343 glycosylated delta (B.1.617.2) RBD. The first 400 ns of the GaMD production trajectory were considered part of the conformational equilibration and not included in the data analysis. Panel e) Graphical representation of the delta RBD with the protein structure (lime cartoon) from a representative snapshot from MD1. The N343 FA2G2 glycan (GlyTouCan-ID G00998NI) is represented in different colours, corresponding to the different MD trajectories, as described in the legend, with snapshots taken at intervals of 100 ns. Residues in the hydrophobic core of the delta RBD are represented with VDW spheres partially visible under the N-glycans overlay. Panel f) Insert showing the junction between Regions 1 and 2 from the left-hand side of the RBD in panel e). The residues involved in the network solidifying the junction are highlighted with sticks and labelled. Panel f) Affinities (1/Kd, x103 M-1) for interactions between GM1os (GlyTouCan-ID G46613JI) and GM2os (GlyTouCan-ID G61168WC) oligosaccharides and the intact and endoF3-treated delta RBD and omicron RBD. Rendering done with VMD (https://www.ks.uiuc.edu/Research/vmd/), KDE analysis with seaborn (https://seaborn.pydata.org/) and bar plot with MS Excel.

Panel a) KDE plot of the backbone RMSD values calculated relative to frame 1 (t = 0) of the GaMD trajectory for Region 1 (green) aa 337-353, Region 2 (yellow) aa 439-506, and Region 3 (orange) aa 411-426 of the glycosylated omicron (BA.1) RBD. Panel b) KDE plot of the backbone RMSD values calculated relative to frame 1 (t = 0) of the GaMD trajectory (see details above) of the non-glycosylated omicron (BA.1) RBD. Panel c) Graphical representation of the glycosylated (protein in yellow cartoons and N343-FA2G2 in white sticks, N331 omitted for clarity) and non-glycosylated (protein in cyan cartoons) of the omicron (BA.1) RBD. Structures correspond to the last frame of the GaMD trajectories, see details in the legend. Panel d) KDE plot of the backbone RMSD values calculated relative to frame 1 (t = 0) of the MD trajectory of the omicron BA.2.86 RBD glycosylated with FA2G2 N-glycans at N343, N354 and N331(not shown). Panel e) Graphical representation of the omicron BA.2.86 RBD (protein in violet cartoons and N-glycans in violet sticks) structurally aligned to the glycosylated omicron (BA.1) RBD (protein in yellow cartoons) for reference. The N343 and N354 glycans are intertwined throughout the trajectory. Panel f) Same graphical representation of the omicron BA.2.86 and BA.1 RBDs with the N-glycans not shown. The purple arrow points to the displacement of the loop in response to the presence of the N354 glycan in BA.2.86. Rendering with VMD (https://www.ks.uiuc.edu/Research/vmd/) and KDE analysis with seaborn (https://seaborn.pydata.org/).

This paper's findings expand our understanding of glycosylation's critical roles in structure-function relationships of viral proteins. The potential for N343 to become a mutational hotspot and the structural dispensability of its glycosylation site may inform immune surveillance and vaccine design, considering the balance between protein fold integrity and immune recognition.