Deciphering the Folding Mechanism of Proteins G and L and Their Mutants
journal contributionposted on 05.08.2022, 20:40 authored by Liwei Chang, Alberto Perez
Much of our understanding of folding mechanisms comes from interpretations of experimental ϕ and ψ value analysis, relating the differences in stability of the transition state ensemble (TSE) and folded state. We introduce a unified approach combining simulations and Bayesian inference to provide atomistic detail for the folding mechanism of proteins G and L and their mutants. Proteins G and L fold to similar topologies despite low sequence similarity, but differ in their folding pathways. A fast folding redesign of protein G, NuG2, switches folding pathways and folds through a similar pathway with protein L. A redesign of protein L also leads to faster folding, respecting the original folding pathway. Our Bayesian inference approach starts from the same prior on all systems and correctly identifies the folding mechanism for each of the four proteins, a success of the force field and sampling strategy. The approach is computationally efficient and correctly identifies the TSE and intermediate structures along the folding pathway in good agreement with experiments. We complement our findings by using two orthogonal approaches that differ in computational cost and interpretability. Adaptive sampling MD combined with the Markov state model provides a kinetic model that confirms the more complex folding mechanism of protein G and its mutant. Finally, a novel fragment decomposition approach using AlphaFold identifies preferences for secondary structure element combinations that follow the order of events observed in the folding pathways.
Read the peer-reviewed publication
ψ value analysisprovide atomistic detailintermediate structures alongfolding mechanisms comestransition state ensembleswitches folding pathwayscomplex folding mechanismoriginal folding pathwayfast folding redesignfolding pathwaysfolding mechanismfolding pathwayfaster foldingfolded statesimilar pathwaysampling strategyproteins gprotein gprior </kinetic modelgood agreementfour proteinsforce fieldexperimental ϕevents observedcorrectly identifiescomputationally efficientcomputational costbayesian inference