Most membrane proteins are proteins with N-glycosylation or O-glycosylation modifications. Glycans have no specific secondary structure and have very strong mobility, which can shield most areas of the protein surface. The high degree of freedom of glycan conformation hinders the elucidation of the overall glycoprotein structure. Glycoprotein structure can be analyzed by computer simulation, but it needs to be performed on a supercomputer and takes hundreds of thousands of hours, so this method cannot be used routinely.
On February 29, 2024, a collaborative team of Mateusz Sikora from Max Planck Institute for Biophysics, Cyril Hanus from Paris Cité University, and Shang-Te Danny Hsu from Academia Sinica in Taiwan, published an article in Cell entitled "Rapid simulation of glycoprotein structures by grafting and steric exclusion of glycan conformer libraries". This article introduces a method - GlycoSHIELD, which can realize the grafting of real glycan conformation clusters onto static protein structures in a few minutes on a personal computer.
The authors first generated a library of 68 common and physiologically relevant N/O-glycans by molecular dynamics simulation (MDS) and developed an open-source toolkit and a web application GlycoSHIELD to graft glycan conformation clusters onto any static protein structure. Next, the authors studied N-cadherin to determine whether GlycoSHIELD could reconstruct the glycan shielding conformations predicted by MDS. N-cadherin is an adhesion glycoprotein composed of five dense and stable immunoglobulin-like domains (EC1-EC5). The authors performed MDS and GlycoSHIELD on the EC4-EC5 domains, respectively, and found that GlycoSHIELD almost reproduced the structure obtained by MDS, with only minor differences in the shielding of protein epitopes by Glycans between the two.
Fig. 1 GlycoSHIELD generates realistic glycan shield models. (Tsai, et al., 2024)
Subsequently, the authors first used MDS to compare the interdomain motion of non-glycosylated and glycosylated N-cadherin EC4-EC5, and found that non-glycosylated EC4-EC5 rapidly folded into a relatively compact and stable conformation, while glycosylated proteins remained extended and flexible. Similarly, the authors used GlycoSHIELD to graft glycans onto non-sugar EC4-EC5 structures, and quantified the glycan acceptability during the simulation, and found that the density of EC4-EC5 measured by the radius of rotation of the protein backbone was negatively correlated with the glycan acceptability. These results show that GlycoSHIELD can reliably predict the effect of glycans on protein conformation.
The SARS-CoV-2 Spike (S) protein is a trimer with 66 N-glycosylation sites. N-glycosylation modification affects the interaction of the S protein with its cellular receptor angiotensin converting enzyme 2 (ACE2), thereby affecting the immune response. The authors compared the glycan shield conformations of the SARS-CoV-2 S protein generated by GlycoSHIELD and MDS and found that the morphologies of the glycan shielded proteins obtained by the two methods were very consistent. Glycan shielding may affect antibody recognition of the S protein, thereby affecting the host immune response. Therefore, the authors performed radiographic accessibility analysis and found that the epitope shielding predicted by MDS was almost completely reproduced by GlycoSHIELD.
In situ cryo-electron tomography and MDS showed that the S protein handle is very flexible, which may promote its interaction with ACE2, thereby increasing the affinity of SARS-CoV-2 for target cells, while the N-glycans located in the transmembrane region of the S protein may conflict with the viral envelope when over-tilted, thereby restricting the orientation of the protein. The authors used GlycoSHIELD to verify this hypothesis. They first generated a truncated model consisting of the HR2 region and the transmembrane domain (TMD) of the S protein, and used GlycoSHIELD to graft glycan conformations on each protein conformation in order to simulate the flexibility of the HR2-TMD hinge and change the angle between HR2 and TMD. It was finally found that the glycosylated form remained more upright than the non-glycosylated form, which is consistent with the typical orientation of native proteins observed in situ. The above results also show that GlycoSHIELD provides a reliable prediction of the effects of glycans on protein dynamics.
Fig. 2 N-glycans may affect the conformation of SARS-CoV-2 S protein and its recognition by antibodies. (Tsai, et al., 2024)
The development of cryo-electron microscopy (cryo-EM) has accelerated the characterization of intact glycoprotein structures, but because glycans are very dynamic, they are usually only partially resolved in protein structures elucidated by cryo-EM. By comparing glycoprotein models and experimental density maps, the authors found that GlycoSHIELD can recover important information about sugar dynamics that is lost during cryo-EM data processing.
The analysis of Protein Structure and glycan composition by combining cryo-EM and mass spectrometry (MS) techniques provides important information on the effects of glycans on protein accessibility. To more systematically characterize the structural effects of glycans on hCoV S protein, the authors combined GlycoSHIELD with MS and cryo-EM to compare the S proteins and their glycan shielding conformations of five coronaviruses (SARSCoV, MERS-CoV, hCoV-HKU1, hCoV-NL, and hCoV-229E), providing new insights into the binding of S proteins to receptors and antibodies in cells.
Finally, the authors also modeled a fully glycosylated type A γ-aminobutyric acid (GABAA) receptor (the main inhibitory neurotransmitter receptor in mammals) and found that glycans may modulate the binding of specific ligands and may occupy a larger channel lumen volume than previously expected, proposing an important hypothesis for the mechanism of neurotransmitter receptors.
GlycoSHIELD provides a direct and feasible method for obtaining quantitative information on glycoprotein morphology and structural dynamics. The advantage of this method is that it greatly reduces the required computational resources, time, and specific technical knowledge, however, the disadvantage is that it ignores the non-steric interactions of glycans with proteins and the interactions between glycans. Although GlycoSHIELD cannot replace the extended MDS of the complete system, it enables more non-expert users to generate reliable glycan shield models for the vast majority of glycoproteins.
Reference
