Uncovering patterns of atomic interactions in static and dynamic structures of proteins

Aj Venkatakrishnan, Rasmus Fonseca, Karen Anthony, Scott A. Hollingsworth, Augustine Chemparathy, Daniel Hilger, Albert J. Kooistra, Ramin Ahmari, M. Madan Babu, Brian K. Kobilka, Ron O. DrorbioRxiv 2019 · DOI

TL;DR

This paper introduces "GetContacts" and the "Contact Comparison Atlas," a powerful software suite and web resource for rapidly computing, comparing, and visualizing atomic interactions within and across protein structures and simulations. This enables researchers to uncover functional patterns in protein families like GPCRs, G proteins, and kinases, aiding in drug discovery and protein engineering.

What Problem Does This Paper Address?

The explosion of protein structural data from techniques like cryo-EM and the increasing use of molecular dynamics simulations present a massive analytical challenge. While comparing interatomic interactions across different protein structures or simulations is a powerful way to understand protein function, it's currently cumbersome and difficult for all but a few well-studied systems. This bottleneck hinders discoveries in areas like drug design, protein engineering, and understanding disease mechanisms. The authors aim to bridge this gap by providing tools to automate and simplify these comparative analyses.

Methods

The core of this work lies in two main components:

  • GetContacts: A software package designed for rapid computation of atomic interactions (hydrogen bonds, salt bridges, aromatic interactions, etc.) in both static protein structures and molecular dynamics simulations. It leverages geometric criteria to identify these interactions. For simulations, it efficiently computes contacts frame-by-frame.
  • Contact Comparison Atlas: A web-based resource that utilizes GetContacts to compare interaction patterns across multiple structures and simulations. It offers interactive visualization tools, including:
  • Flareplots: A 2D visualization that displays residues in a circular layout and interactions as chords connecting them.
  • 3D Structural Rendering: Visualizing interaction patterns directly on the protein's 3D structure.
  • Interaction Fingerprint Panel: A way to systematically identify and track shared and unique interactions across the compared datasets.

The authors also address the challenge of comparing related but distinct proteins by using protein-independent numbering schemes (like GPCRdb numbers) derived from structural alignments.

Datasets

The paper utilizes publicly available protein structures from the Protein Data Bank (PDB). Specifically, they pre-populate the Contact Comparison Atlas with structures and simulations from three key protein families:

  • G protein-coupled receptors (GPCRs)
  • G proteins
  • Kinases

They also mention using molecular dynamics simulation trajectories for muscarinic acetylcholine receptors (mAChRs) and other GPCRs.

Key Insights and Ideas

  • The power of comparative interaction analysis: The paper strongly emphasizes that the real value in structural data comes from comparing it across multiple states or related proteins, not just analyzing individual structures.
  • Water as a key mediator: They highlight conserved water-mediated interactions in the binding pockets of opioid receptors, demonstrating water's crucial role in ligand binding.
  • Unique interactions drive function: The study reveals that unique interaction networks, even in highly similar proteins (like viral vs. human chemokine receptors), can explain distinct functional properties, such as constitutive activity.
  • Subtype-specific interactions are key for selectivity: For muscarinic receptors, they found that while ligand binding is largely conserved, distinct interaction patterns at the G-protein binding site correlate with different G-protein coupling preferences, and unique interactions can be exploited for subtype-selective drug design.
  • Conserved salt bridge in tyrosine kinases: A specific salt bridge is identified as a unique feature of tyrosine kinases, suggesting a functional role in their distinct properties and potentially being a target for inhibitors.

Key Results

  • GetContacts efficiency: The software is capable of rapidly computing interactions, which is crucial for handling the large number of frames in molecular dynamics simulations.
  • Contact Comparison Atlas utility: The web resource successfully facilitated the identification of conserved and unique interaction patterns in GPCRs, G proteins, and kinases.
  • Functional insights: The analysis led to specific findings, such as:
  • A conserved water-mediated interaction in opioid receptor binding pockets.
  • Unique interaction networks contributing to the constitutive activity of the viral chemokine receptor US28.
  • Identification of shared ligand-binding interactions and distinct G-protein coupling interactions in muscarinic receptors.
  • Discovery of a unique salt bridge in tyrosine kinases.

Limitations and Future Work

  • Quality of input data: The authors acknowledge that the interpretation of interactions is dependent on the quality of the input structural data (resolution) and simulation data (force field accuracy, convergence).
  • Binary interaction definition: Interactions are marked as present or absent based on distance and angle cutoffs, which is a simplification.
  • Future Applications: The authors suggest that GetContacts and the Contact Comparison Atlas have broad applications in:
  • Structure modeling (refining homology models, identifying conformational substates).
  • Drug design (improving docking studies, designing compounds with enhanced specificity).
  • Protein engineering (stabilizing specific conformational states).
  • Interpreting pathological mutations and genetic variation.

Summary updated

Abstract

Abstract The number of structures and molecular dynamics simulations of proteins is exploding owing to dramatic advances in cryo-electron microscopy, crystallography, and computing. One of the most powerful ways to analyze structural information involves comparisons of interatomic interactions across different structures or simulations of the same protein or related proteins from the same family ( e.g. different GPCRs). Such comparative analyses are of interest to a wide range of researchers but currently prove challenging for all but a few. To facilitate comparative structural analyses, we have developed tools for (i) rapidly computing and comparing interatomic interactions and (ii) interactively visualizing interactions to enable structure-based interpretations. Using these tools, we have developed the Contact Comparison Atlas, a web-based resource for the comparative analysis of interactions in structures and simulations of proteins. Using the Contact Comparison Atlas and our tools, we have identified patterns of interactions with functional implications in structures of G-protein-coupled receptors, G proteins and kinases and in the dynamics of muscarinic receptors. The Contact Comparison Atlas can be used to enable structure modeling, drug discovery, protein engineering, and the prediction of disease-associated mutations. Contact Comparison Atlas website: https://getcontacts.github.io/atlas/
BibTeX
@misc{venkatakrishnan2019uncovering,
  author = {AJ Venkatakrishnan and Rasmus Fonseca and Karen Anthony and Scott A. Hollingsworth and Augustine Chemparathy and Daniel Hilger and Albert J. Kooistra and Ramin Ahmari and M. Madan Babu and Brian K. Kobilka and Ron O. Dror},
  title = {{Uncovering patterns of atomic interactions in static and dynamic structures of proteins}},
  year = {2019},
  doi = {10.1101/840694},
  howpublished = {bioRxiv},
}
Figure