Starting from December 2019, coronavirus disease 2019 (COVID-19) has emerged as a once-in-a-century pandemic with deadly consequences, which urgently calls for new treatments, cures and supporting apparatuses. Remdesivir was reported by World Health Organization (WHO) as the most promising drug currently available for the treatment of COVID-19. Here, we use molecular dynamics simulations and free energy perturbation methods to study the inhibition mechanism of remdesivir to its target SARS-CoV-2 virus RNA-dependent RNA polymerase (RdRp). In the absence of a crystal structure of the SARS-CoV-2 RdRp, we first construct the homology model of this polymerase based on a previously available structure of SARS-CoV NSP12 RdRp (with a sequence identify of 95.8%). We then build the putative binding mode by aligning the remdesivir + RdRp complex to the ATP bound poliovirus RdRp. The putative binding structure is further optimized with molecular dynamics simulations and demonstrated to be stable, indicating a reasonable binding mode for remdesivir. The relative binding free energy of remdesivir is calculated to be -8.28 ± 0.65 kcal/mol, much stronger than the natural substrate ATP (-4.14 ± 0.89 kcal/mol) which is needed for the polymerization. The ~800-fold improvement in the Kd from remdesivir over ATP indicates an effective replacement of APT in blocking of the RdRp binding pocket. Key residues D618, S549 and R555 are found to be the contributors to the binding affinity of remdesivir. These findings demonstrate that remdesivir can potentially act as a SARS-CoV-2 RNA-chain terminator, effectively stopping its RNA reproduction, with key residues also identified for future lead optimization and/or drug resistance studies.

Immunologic recognition of peptide antigens bound to class I major histocompatibility complex (MHC) molecules is essential to both novel immunotherapeutic development and human health at large. Current methods for predicting antigen peptide immunogenicity rely primarily on simple sequence representations, which allow for some understanding of immunogenic features but provide inadequate consideration of the full scale of molecular mechanisms tied to peptide recognition. We here characterize contributions that unsupervised and supervised artificial intelligence (AI) methods can make toward understanding and predicting MHC(HLA-A2)-peptide complex immunogenicity when applied to large ensembles of molecular dynamics simulations. We first show that an unsupervised AI method allows us to identify subtle features that drive immunogenicity differences between a cancer neoantigen and its wild-type peptide counterpart. Next, we demonstrate that a supervised AI method for class I MHC(HLA-A2)-peptide complex classification significantly outperforms a sequence model on small datasets corrected for trivial sequence correlations. Furthermore, we show that both unsupervised and supervised approaches reveal determinants of immunogenicity based on time-dependent molecular fluctuations and anchor position dynamics outside the MHC binding groove. We discuss implications of these structural and dynamic immunogenicity correlates for the induction of T cell responses and therapeutic T cell receptor design.

An unsolved challenge in the development of antigen-specific immunotherapies is determining the optimal antigens to target. Comprehension of antigen–major histocompatibility complex (MHC) binding is paramount toward achieving this goal. Here, we apply CASTELO, a combined machine learning-molecular dynamics (ML-MD) approach, to identify per-residue antigen binding contributions and then design novel antigens of increased MHC-II binding affinity for a type 1 diabetes-implicated system. We build upon a small-molecule lead optimization algorithm by training a convolutional variational autoencoder (CVAE) on MD trajectories of 48 different systems across four antigens and four HLA serotypes. We develop several new machine learning metrics including a structure-based anchor residue classification model as well as cluster comparison scores. ML-MD predictions agree well with experimental binding results and free energy perturbation-predicted binding affinities. Moreover, ML-MD metrics are independent of traditional MD stability metrics such as contact area and root-mean-square fluctuations (RMSF), which do not reflect binding affinity data. Our work supports the role of structure-based deep learning techniques in antigen-specific immunotherapy design.