Search for: All records

Coronavirus spike glycoproteins presented on the virion surface mediate receptor binding, and membrane fusion during virus entry and constitute the primary target for vaccine and drug development. How the structure dynamics of the full-length spikes incorporated in viral lipid envelope correlates with the virus infectivity remains poorly understood. Here we present structures and distributions of native spike conformations on vitrified human coronavirus NL63 (HCoV-NL63) virions without chemical fixation by cryogenic electron tomography (cryoET) and subtomogram averaging, along with site-specific glycan composition and occupancy determined by mass spectrometry. The higher oligomannose glycan shield on HCoV-NL63 spikes than on SARS-CoV-2 spikes correlates with stronger immune evasion of HCoV-NL63. Incorporation of cryoET-derived native spike conformations into all-atom molecular dynamic simulations elucidate the conformational landscape of the glycosylated, full-length spike that reveals a role of hinge glycans in modulating spike bending. We show that glycosylation at N1242 at the upper portion of the stalk is responsible for the extensive orientational freedom of the spike crown. Subsequent infectivity assays implicated involvement of N1242-glyan in virus entry. Our results suggest a potential therapeutic target site for HCoV-NL63.

Accurate prediction of peptide binding affinity to the major histocompatibility complex (MHC) proteins has the potential to design better therapeutic vaccines. Previous work has shown that pan‐specific prediction algorithms can achieve better prediction performance than other approaches. However, most of the top algorithms are neural networks based black box models. Here, we propose DeepAttentionPan, an improved pan‐specific model, based on convolutional neural networks and attention mechanisms for more flexible, stable and interpretable MHC‐I binding prediction. With the attention mechanism, our ensemble model consisting of 20 trained networks achieves high and more stabilized prediction performance. Extensive tests on IEDB's weekly benchmark dataset show that our method achieves state‐of‐the‐art prediction performance on 21 test allele datasets. Analysis of the peptide positional attention weights learned by our model demonstrates its capability to capture critical binding positions of the peptides, which leads to mechanistic understanding of MHC‐peptide binding with high alignment with experimentally verified results. Furthermore, we show that with transfer learning, our pan model can be fine‐tuned for alleles with few samples to achieve additional performance improvement. DeepAttentionPan is freely available as an open‐source software athttps://github.com/jjin49/DeepAttentionPan.