Unraveling protein folding mechanism by analyzing the hierarchy of models with increasing level of detail

Taking protein G with 56 residues for a case study, we investigate the mechanism of protein folding. In addition to its native structure possessing α-helix and β-sheet contents of 27% and 39%, respectively, we construct a number of misfolded decoys with a wide variety of α-helix and β-sheet contents. We then consider a hierarchy of 8 different models with increasing level of detail in terms of the number of entropic and energetic physical factors incorporated. The polyatomic structure is always taken into account, but the side chains are removed in half of the models. The solvent is formed by either neutral hard spheres or water molecules. Protein intramolecular hydrogen bonds (H-bonds) and protein-solvent H-bonds (the latter is present only in water) are accounted for or not, depending on the model considered. We then apply a physics-based free-energy function (FEF) corresponding to each model and investigate which structures are most stabilized. This special approach taken on a step-by-step basis enables us to clarify the role of each physical factor in contributing to the structural stability and separately elucidate its effect. Depending on the model employed, significantly different structures such as very compact configurations with no secondary structures and configurations of associated α-helices are optimally stabilized. The native structure can be identified as that with lowest FEF only when the most detailed model is employed. This result is significant for at least the two reasons: The most detailed model considered here is able to capture the fundamental aspects of protein folding notwithstanding its simplicity; and it is shown that the native structure is stabilized by a complex interplay of minimal multiple factors that must be all included in the description. In the absence of even a single of these factors, the protein is likely to be driven towards a different, more stable state