Erratum: Corrigendum: Cloud-based simulations on Google Exacycle reveal ligand modulation of GPCR activation pathways

Kai Kohlhoff, Diwakar Shukla, Morgan Lawrenz, Gregory R. Bowman, David E. Konerding, Dan Belov, Russ B. Altman, Vijay S. PandeNature Chemistry 2015 · DOI · OpenAlex

Abstract

This corrigendum addresses visual errors in the chemical structures presented in the original 2014 study on G protein-coupled receptor (GPCR) activation pathways. While the underlying computational simulations performed on Google Exacycle remain valid, this update provides corrected ligand diagrams, PubChem identifiers, and methodological clarifications regarding stereoisomer selection and ligand classification.

Introduction

The original research utilized massive-scale molecular dynamics (MD) simulations via Google’s Exacycle platform to explore how various ligands modulate GPCR activation. Given that GPCRs are targets for approximately one-third of FDA-approved drugs, accurate representation of these molecular interactions is critical. This corrigendum corrects errors in Figure 4 of the original publication and clarifies the handling of ligand stereochemistry and nomenclature.

Methodological Clarifications

The authors provide additional detail on the ligand preparation and docking protocols used in the study:

  • Protonation: All functional groups were protonated to reflect a physiological pH of 7.0.
  • Stereoisomer Enumeration: For compounds where specific stereoisomer activity was unknown, isomers were enumerated for molecules with up to four chiral centers.
  • Docking Strategy: Each isomer was docked into various protein conformations. The isomer with the highest docking score for a specific conformation was selected for analysis. This approach acknowledges that different receptor states may preferentially bind different stereoisomers, highlighting the dynamic nature of ligand-protein interactions.

Structural Corrections and Data Integrity

The corrigendum resolves several drawing errors in the original ligand diagrams. Importantly, the authors confirm that the original MD simulations were conducted using the correct chemical structures; the errors were confined to the visual representations in the paper. This reinforces the necessity of verifying that visual figures accurately reflect the computational inputs.

Specific corrections include:

  • Compounds 2–4: Corrected from hemiaminals to $\beta$-amino alcohols.
  • Compound 7: Removal of an extra benzylic carbon.
  • Compound 8: Removal of an extra ring.

To ensure reproducibility, the correct PubChem CIDs are provided:

  • Agonists: 19044758 (1), 44216210 (2), 44209282 (3), 44213610 (4).
  • Antagonists: 15020513 (5), 19823514 (6), 44209768 (7), 44209764 (8).

Nomenclature and Classification

The label for Figure 4b has been updated from "Inverse agonist" to "Antagonist" to maintain consistency with the source database (Gatica and Cavasotto, 2012). The authors note that while some molecules, such as carazolol, may exhibit inverse agonist activity, the broader "Antagonist" label is used for classification within the study's framework of 3D chemotypes. These chemotypes represent groups of ligands sharing similar binding poses across receptor conformations that have similar "progress scores"—a metric for the receptor's position along the activation pathway.

Conclusion

This update reinforces the validity of the original large-scale simulation results while providing the necessary corrections for structural accuracy. It also highlights a persistent challenge in computational pharmacology: the need for experimental data on the specific activities of individual stereoisomers—similar to existing studies on drugs like albuterol and fenoterol—to fully validate high-throughput docking results. Future work should focus on high-resolution experimental validation to complement these large-scale computational screens.

Summary updated

BibTeX
@inproceedings{kohlhoff2015erratum,
  author = {Kai Kohlhoff and Diwakar Shukla and Morgan Lawrenz and Gregory R. Bowman and David E. Konerding and Dan Belov and Russ B. Altman and Vijay S. Pande and Kai J Kohlhoff},
  title = {{Erratum: Corrigendum: Cloud-based simulations on Google Exacycle reveal ligand modulation of GPCR activation pathways}},
  booktitle = {Nature Chemistry},
  year = {2015},
  doi = {10.1038/nchem.2272},
}
Figure