(639d) The Full-Length Protein a Challenges Both the Structure-Function Paradigm and the Protein Structure Prediction Technique Alphafold: An Advanced Molecular Simulation Study

The bacterium Staphylococcus aureus, commonly known as “staph”, causes serious infections in humans ranging from pneumonia to bloodstream infections that can lead to deadly sepsis. The bacterial surface protein Staphylococcal protein A (SpA), enhances virulence by enabling the pathogen to suppress the immune response, mainly by binding to antibodies. Beyond its natural role, engineered SpA is now used in affinity chromatography to purify hundreds of monoclonal antibody medicines, including most of the top 10-selling drugs in the world. SpA has five homologous three-helix-bundle antibody-binding domains and a large disordered C-terminal anchor domain. A number of studies aimed at understanding SpA and the interactions that make SpA a potent virulence factor have focused on individual three-helix bundle domains. The individual domains’ binding affinities, folding stabilities, and crystal structures both in isolation and complexed with immunoglobulin Fc domains have been well studied. However, both in the engineered context and in the naturally-occurring form, structural and thermodynamic analysis of the intact full-length SpA has remained elusive. In this work, we present the first study of the folding process of full–length SpA using advanced all-atom molecular dynamics (MD) simulations. We first obtained the AlphaFold structure prediction for the full-length protein, which showed all five individual domains folded with flexible linkers between them along with a disordered anchor domain. Since there is no other reference structure for this full-length protein in the literature, we took this configuration as a reference folded structure and performed advanced simulations defining a reaction coordinate with this reference folded state. We then projected the free energy surface of full-length SpA on this coordinate and found that this structure has very high energy (i.e., is highly unstable) compared to a fully disordered compact state. None of the individual domains remained stable in the context of full-length SpA. In order to assess whether our methodology is capable of capturing the stability of individual domains in isolation, we also performed advanced sampling simulations of individual domains separately. We then calculated the thermal stability of the individual domains and found that our methodology reproduces experimentally known thermal stability nearly exactly, increasing confidence that the disordered nature of full-length SpA is not an artifact of our methodology. We believe that our rigorously-validated approach will enable the discovery of novel mechanisms that involve structural disorder (such as this one) as well as advancing understanding of the role of protein structural disorder in biomolecular interactions in immunology.