Predicting protein folding pathway using a novel folding force field model derived from known protein universe

Abstract

AbstractThe protein folding problem has emerged as a new challenge with the significant advances in deep learning driven protein structure prediction methods. While the structures of almost all known proteins have been successfully predicted, the process by which they fold remains an enigma. Understanding the intricate folding mechanism is of paramount importance, as it directly impacts the stable expression and biological function of proteins. Here, we propose FoldPAthreader, a protein folding pathway prediction method that designs a novel folding force field model by exploring the intrinsic relationship between protein evolutionary history and folding mechanisms from the known protein universe. Further, the folding force field is used to guide Monte Carlo conformational sampling, driving the protein chain fold into its native state by exploring a series of transition states and potential intermediates. On the 30 targets we collected, FoldPAthreader can successfully predict 70% of the proteins whose folding pathway is consistent with wet-lab experimental data. The results show that the folding force field can capture key dynamic features of hydrogen bonding and hydrophobic interactions. Importantly, for the widely studied BPTI and TIM proteins, the folding pathway predicted by FoldPAthreader have the same microscopic dynamic properties as those simulated by molecular dynamics.Significance StatementProtein folding is the process by which a protein acquires its functional conformations by gradually transforming from random coils into a specific three-dimensional structure. In the post-Alphafold2 era, functional analysis of protein macromolecules should not only rely on the final state structure, but should pay more attention to the structural folding process, that is, the various intermediate states formed during the folding process. At present, there is no folding force field specifically used for protein folding pathway prediction in computational biology. Here we extracted folding information from 100-million-level structure database and designed a new folding force field for folding pathway prediction, proving a hypothesis that the protein evolutionary history implicitly contains folding information of individual protein. This study may provide new insights into the understanding of protein folding mechanisms, which is expected to advance drug discovery.