Construction and comparison of the statistical coil states of unfolded and intrinsically disordered proteins from nearest-neighbor corrected conformational propensities of short peptides.

Abstract

Assessing the influence of nearest neighbors on the conformational ensemble of amino acid residues in unfolded and intrinsically disordered proteins and peptides is pivotal for a thorough understanding of the statistical coil state of unfolded proteins as well as of the energetics of the folding process. Research aimed at exploring nearest neighbor interactions has mostly focused on the analysis of restricted coil libraries that reflect conformational distributions in loops connecting more regular secondary structure segments. Recently, however, Toal et al. reported an experimentally based structural analysis of selected xy-pairs in GxyG tetrapeptides, which revealed quantitative information about conformational changes induced by nearest-neighbor interactions (Eur. J. Chem., 2015, 21, 5173-5192). Here, we perform analyses of Ramachandran plots of xy-pairs in GxyG and in coil libraries (Ting et al., PLOS CompBiol, 2010, 6, e1000763) using Hellinger distances as a quantitative measure of dissimilarities between Ramachandran distributions. Our analysis reveals that nearest-neighbor effects inferred from the above coil library are much less pronounced than corresponding structural changes observed for GxyG peptides. To determine whether nearest-neighbor induced conformational changes observed for GxyG can be utilized for the analysis of unfolded proteins, we analyzed sets of 3J(HHHα) coupling constants of three different unfolded proteins, namely the 130-residue fragment of the Staphylococcus aureus fibronectin-binding protein (FnBPc), denatured hen lysozyme, and the htau40 protein. For the first two proteins we found statistically meaningful correlations between predicted nearest-neighbor induced changes of 3J(HHHα) and experimentally observed deviations from corresponding coupling constants of GxG peptides in water, which we used as reference system with minimal nearest-neighbor interactions. This observation is in line with the NMR based understanding of these proteins being predominantly statistical coils. For htau40, however, which is known to exhibit residual structure and large deviations form statistical coil expectations, these correlations are weak or absent. Our results thus underscore the importance of nearest-neighbor interactions for a complete physical description of an ideal statistical coil state of a protein.