Gaetano T. Montelione, Ph.D.
Abstract
Protein structures are not static objects, but rather sample a conformational energy landscape that underpins their functions. Recent advances in molecular modeling using deep learning can revolutionize our understanding of dynamic protein structures. NMR is particularly well-suited for studying dynamic features of biomolecular structures. The conventional process for determining biomolecular structures from NMR data involves its representation as conformation- dependent restraints and generation of structural models directed by these restraints. We have developed an alternative approach: generating a distribution of realistic protein conformational models using AI based methods and then selecting the set(s) of conformers that best explain the experimental data. We have applied this conformational selection approach to redetermine the solution NMR structures of several proteins. First, we generate a diverse set of conformer models using AlphaFold2 with an enhanced sampling protocol. The models that best-fit chemical shift data are scored and selected with a Bayesian scoring metric, and then cross-validated with conformer-specific NOESY data. For some proteins, the “AlphaFold-NMR” protocol has identified multiple conformational states which, considered as a multistate ensemble, fit the experimental data better than the conventional “restraint-based” NMR structures. These previously unrecognized alternative conformational states provide novel insights into protein structure-dynamic-function relationships. Our studies of the conformational landscapes accessible by AI-based modeling together with experimental data demonstrate provide an alternative to conventional restraint-satisfaction protocols for protein NMR structure determination, and identify conformations that are otherwise averaged out in the data analysis, providing novel structural dynamics and function information.
YJ Huang, GT Montelione. Hidden structural states of proteins revealed by conformer selection with AlphaFold-NMR
(2024) bioRxiv 2024.06.26.600902; doi: https://doi.org/10.1101/2024.06.26.600902
When: October 10, 2025
Where: North Classroom
Time: 11:00 am - 12:00pm