Login / Signup

Underestimated Noncovalent Interactions in Protein Data Bank.

Zhijian XuQian ZhangJiye ShiJianming Zhu
Published in: Journal of chemical information and modeling (2019)
Noncovalent interactions (NCIs) play essential roles in the structure and function of biomacromolecules. There are various NCIs, e.g., hydrogen bonds (HBs), cation-π and π-π interactions, and ionic bonds, among which HBs are the most widespread and well-studied. By utilizing the ratio of the observed HBs over pseudo HBs (1.0 Å longer than the HB distance criteria without angle constraints), we demonstrated that HBs in both protein-ligand and protein-protein interfaces are overlooked in structures deposited in PDB. After the QM/MM optimization of 12 protein-ligand complexes, we showed that the overlooked HBs could be recovered. With a systematic search in the PDB, we found that the HB number per residue (NHB/R) in proteins decreases as structural resolution becomes lower, implying that HBs are overlooked even today, regardless of the type of refinement approach used. Similarly, cation-π, π-π, and ionic interactions were found to be significantly lost, manifesting the universal underestimation of various NCIs. Considering the vital role of NCIs, it is important to recover the NCIs to facilitate drug design, to explore protein-protein interaction, and to study protein structure and function.
Keyphrases
  • protein protein
  • small molecule
  • ionic liquid
  • amino acid
  • high resolution
  • binding protein
  • emergency department
  • machine learning
  • electronic health record
  • deep learning
  • drug induced
  • transition metal