doi.org
Localizing the Glycan Binding Sites of SARS-CoV-2 Receptor-Binding Domain
Mapping the binding sites on viral proteins that interact with host glycans is essential for understanding host–pathogen interactions and guiding therapeutic design. While the SARS-CoV-2 spike (S) protein is known to bind host glycans through its N-terminal domain and receptor-binding domain (RBD), the specific glycan-interacting residues on the RBD remain undefined. Here, we introduce antibody masking-native mass spectrometry (AM-nMS), a spatially resolved assay for mapping glycan-binding sites by combining a panel of nanobodies targeting distinct regions within the RBD with center-of-mass monitoring and catch-and-release strategies. Applied to wild-type (Wuhan-Hu-1) SARS-CoV-2 RBD, AM-nMS reveals distinct yet partially overlapping glycan-binding surfaces. Heparan sulfate (HS) oligosaccharides bind within a shallow groove (encompassing residues 420–430, 441–452, and 490–512), which partially overlaps the receptor-binding motif (RBM). Sialylated glycans, including ganglioside oligosaccharides, engage a distal edge of the RBM (residues 470–486 and 490–512) that includes a portion of HS binding groove. Globoside glycans bind at a third site (residues 513–530 and 441–452) that also partially overlaps with the HS-binding region. These multiple binding sites near the RBM enable the RBD to interact with a wide range of host glycans. Site assignments are supported by library screening across the RBD from several SARS-CoV-2 variants of concern (VOCs), competitive binding nMS, mass photometry and molecular dynamics simulations. This work presents the first experimental residue-level map of glycan engagement by the SARS-CoV-2 RBD and establishes AM-nMS as a broadly applicable strategy for localizing glycan-binding sites on proteins.
![Figure 3. (a) Analysis of structures from experiments deposited in the PDB from GlycoNavi (https://glyconavi.org/TCarp/Statistics/) shows that only 6873 proteins contain N-glycans, for a total of 41,951 glycan structures. Of those 51% consist of only one monosaccharide, with less than 2054 structures containing 7 or more monosaccharide units. (b) Qualitative diagram illustrating different dynamic time-scales of biological systems in function of their size. Experiments relevant to structural biology are represented in the graph within their corresponding range of application in space/time. The design of the diagram was inspired by a similar example in Ref. [66] (c) 3D structure of the human growth hormone (hGH) from PDB 1HGU (cyan), with N-glycans at N93, N99, and N104 reconstructed with Re-Glyco Ensemble [67] using 50 frames selected from the MD structural ensemble, rendered with surface and sticks in white. The type of N-glycans are represented by SNFG symbols [41] in 2D and with the corresponding GlyTouCan [42] IDs. (d) An example of the N-glycans repertoire that can be found at highly exposed sites such as N93, N(98)99, N104. The relative populations of the corresponding glycoforms can be determined by glycoproteomics. (e) Relative populations of the N-glycans at different sites introduced by mutagenesis in Ref. [68]. Occupancy is shown in pink and the absence of glycans in grey. The length of the bars correspond to 100%. Glycosylation at positions N98 and N99 is mutually exclusive because of site proximity. (f) Structure of the complex of hGH (cyan) with the hGH receptor (white with transparent surface and cartoons) from PDB 1A22 indicating the surface area where protein–protein interactions occur and that needs to remain unoccupied for function. The orientation of the hGH in 1.c and 1.f is the same. (g) (..)](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:xorrvcukxonsyeoudfzmnmhz/bafkreidgymutov5movbrrild4wlrzy6ursngbrdicd6znazmel6mrwun5e)
![Figure 2. (a) The three main types of human N-glycans are oligomannose, hybrid and complex. Representative structures are shown with 2D SNFG symbols [41] and corresponding GlyTouCan [42] IDs. Arrows indicate multi-step processes (not direct reactions) occurring in the ER (light blue background) and Golgi (light yellow background), involving numerous enzymes. ERMan-1 (UniProt Q9UKM7) and GNT-1 (UniProt P26572) are shown as the enzymes initiating key steps in N-glycan processing. (b) Man9 N-glycan [3] (mannose in green and GlcNAc in blue sticks) at position N234, chain B, of the SARS-Cov-2 S (WHu-1) trapped inside a pocket upon opening of the RBD chain A (yellow surface). The protein structure is shown as a surface, with Chain A in yellow, B in white and C in pink. All other glycans are not shown for clarity. The 3D structure is from a representative frame from an MD simulation, based on PDB 3VYB, [1]. All other glycans are not shown for clarity. (c) Man5 in FcγRIII (PDB 3SGK) at N40 with the α(1–6) arm trapped in a pocket, exposing the α(1–3) arm to the solvent, leading to a local enrichment in hybrid N-glycans. The 3D structure is from a representative frame from an MD simulation, based on PDB 3SGK, [34]. All other glycans are not shown for clarity. (d) MS spectrum of the recombinant CD52-Fc active fraction (black) overlaid with the spectrum of the inactive fraction (red), which is the most abundant by weight. N-glycan compositions are annotated. (e) Top 10 N-glycan compositions with more than 2.5% relative abundance in active fractions, ordered by increasing m/z (M−H). The fraction in blue is the most active and shows a higher content of sialylated epitopes. (f) Proposed structure of the complex between the HMGB1-Box B (cyan cartoons) and the hyperglycosylated CD52 peptide (green cartoons). (..) Panels d–g are adapted from Ref. [43].](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:xorrvcukxonsyeoudfzmnmhz/bafkreie62w6gq3lx4yrgaabng5stqog6v7fwgctsfmek4szmzxr5wkojmq)
![Figure 1. Glycan macroheterogeneity is encoded in the protein sequence. (a) Bar graph of the amino acid abundance in eukaryotic proteins from the Swiss-Prot dataset (purple bars; number of proteins: 199,526; number of aa: 88,436,717) and the UniProt50 dataset (blue bars; number of proteins: 26, 204,260; number of aa: 8,800,588,153). (b) Average number of consensus sequon for N-glycosylation (N-x-(S/T)), C-mannosylation (W-xx-W/C), and O-fucosylation (C-xx-(S/T)C and C-xxxx-(S/T)C) found in each protein in the Swiss-Prot dataset (purple bars) and the UniProt50 dataset (blue bars). (c) Close-up view of the binding site of the eukaryotic oligosaccharyltransferase (OST; PDB 8AGC), shown with its surface in yellow, bound to the YJR1_N99 substrate peptide shown in cyan, with the backbone in cartoons and side chains in sticks. The peptide sequence is shown in the bottom-left corner. The structure of the complex is from a representative frame from an MD simulation in Ref. [20]. (d) Close-up view of the binding site of the complex between the C-mannosyltransferase (CMT; PDB 7ZLG), shown with its surface in purple, and a substrate peptide, shown in yellow with the backbone in cartoons and side chains in sticks. The peptide sequence is shown in yellow in the bottom-left corner. Molecular rendering was done with VMD [21] (https://www.ks.uiuc.edu/Research/vmd/), and statistical data visualisation done with seaborn MD, molecular dynamics; VMD, visual molecular dynamics. (https://seaborn.pydata.org/).](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:xorrvcukxonsyeoudfzmnmhz/bafkreifl7yuzwyccfp7jaw53tuqbfbp6colgtbbgbiwc3flcclh4rsul3i)
