Insights into protein folding mechanisms from large scale analysis of mutational effects

Protein folding mechanisms are probed experimentally using single-point mutant perturbations. The relative effects on the folding (ϕ-values) and unfolding (1 - ϕ) rates are used to infer the detailed structure of the transition-state ensemble (TSE). Here we analyze kinetic data on > 800 mutations carried out for 24 proteins with simple kinetic behavior. We find two surprising results: (i) all mutant effects are described by the equation: . Therefore all data are consistent with a single ϕ-value (0.24) with accuracy comparable to experimental precision, suggesting that the structural information in conventional ϕ-values is low. (ii) ϕ-values change with stability, increasing in mean value and spread from native to unfolding conditions, and thus cannot be interpreted without proper normalization. We eliminate stability effects calculating the ϕ-values at the mutant denaturation midpoints; i.e., conditions of zero stability (ϕ0). We then show that the intrinsic variability is ϕ0 = 0.36 ± 0.11, being somewhat larger for β-sheet-rich proteins than for α-helical proteins. Importantly, we discover that ϕ0-values are proportional to how many of the residues surrounding the mutated site are local in sequence. High ϕ0-values correspond to protein surface sites, which have few nonlocal neighbors, whereas core residues with many tertiary interactions produce the lowest ϕ0-values. These results suggest a general mechanism in which the TSE at zero stability is a broad conformational ensemble stabilized by local interactions and without specific tertiary interactions, reconciling ϕ-values with many other empirical observations.