The Transition State of the ras Binding Domain of Raf Is Structurally Polarized Based on Φ-Values but Is Energetically Diffuse

Abstract

The ras binding domain (RBD) of the Ser/Thr kinase c-Raf/Raf-1 spans 78 residues and adopts a structure characteristic of the β-grasp ubiquitin-like topology. Recently, the primary sequence of Raf RBD has been nearly exhaustively mutated experimentally by insertion of stretches of degenerate codons, which revealed sequence conservation and hydrophobic core organization similar to that found in an alignment of β-grasp ubiquitin-like proteins. These results now allow us to examine the relationship between sequence conservation and the folding process, particularly viewed through the analysis of transition state (TS) structure. Specifically, we present herein a protein engineering study combining classic truncation (Ala/Gly) and atypical mutants to predict folding TS ensemble properties. Based on classical Φ-value analysis, Raf RBD TS structure is particularly polarized around the N-terminal β-hairpin. However, all residues constituting the inner layer of the hydrophobic core are involved in TS stabilization, although they are clearly found in a less native-like environment. The TS structure can also be probed by a direct measure of its destabilization upon mutation, ΔΔ G U-‡. Viewed through this analysis, Raf RBD TS is a more diffuse structure, in which all residues of the hydrophobic core including β-strands 1, 2, 3 and 5 and the major α-helix play similar roles in TS stabilization. In addition, Φ-values and ΔΔ G U-‡ reveal striking similarities in the TS of Raf RBD and ubiquitin, a structural analogue displaying insignificant sequence identity (<12%). However, ubiquitin TS appears more denatured-like and polarized around the N-terminal β-hairpin. We suggest that analysis of Φ-values should also consider the direct impact of mutations on differences in free energy between the unfolded and TS (ΔΔG U-‡) to ensure that the description of TS properties is accurate. Finally, the impact of these findings on the modeling of protein folding is discussed.