Flash
Using large-scale electronic structure analysis to understand unexpectedly short noncovalent interactions in the structurally characterized proteome

Protein crystal structures have provided new insight into protein function, either inferred from interatomic distances or as a starting point for atomistic simulation of dynamics and reactivity. In the decades since the first X-ray crystal structures of proteins were solved, over 120,000 structures have been made publically accessible, of which over 50,000 are high-resolution (i.e., sub-2.0 Å). Even in this high-resolution dataset of protein structures, unusually short, non-bonded distances, i.e. close contacts, are pervasive. These close contacts are expected to correspond to unusually energetically strong interactions, such as strong hydrogen bonds exhibiting charge transfer, or other interactions that are otherwise not fully understood and are not well characterized by commonly employed molecular mechanics force fields. With recent advances that allow first-principles modeling of thousands of atoms, large-scale electronic structure models can provide valuable insights in the nature of these interactions. We screen the structurally characterized proteome to identify and examine all unusually short intra-protein residue-residue interactions in high-resolution crystal structures. We i) resolve the residue and secondary-structure dependent nature of these unusually short interactions; ii) quantify the extent to which components from energy decomposition (i.e., electrostatic versus van der Waals) in a classical and first-principles picture differ; and iii) validate and explain representative newly discovered interaction motifs through large-scale simulation of the effect of the greater protein environment. These newly discovered interactions will provide key new insights into established protein structure-function relationships.