All-Atom Simulations Reveal Protein Charge Decoration in the Folded and Unfolded Ensemble Is Key in Thermophilic Adaptation

Thermophilic proteins denature at much higher temperature compared to their mesophilic homologues, in spite of high structural and sequential similarity. Computational approaches to understand this puzzle face three major challenges: (i) unfolded ensembles are usually neglected, (ii) simulation studies of the folded states are often too short, and (iii) the majority of investigations focus on a few protein pairs, obscuring the prevalence of different strategies across multiple protein systems. We address these concerns by carrying out all-atom simulations to characterize physicochemical properties of both the folded and the disordered ensemble in multiple (12) thermophilic-mesophilic homologous protein pairs. We notice two clear trends in most pairs (10 out of 12). First, specific distribution of charges in the native basin - sampled from multimicrosecond long Molecular Dynamics (MD) simulation trajectories - leads to more favorable electrostatic interaction energy in thermophiles compared to mesophiles. Next, thermophilic proteins have lowered electrostatic interaction in their unfolded state - generated using Monte Carlo (MC) simulation - compared to their mesophilic counterparts. The net contribution of interaction energy to folding stability, however, remains more favorable in thermophiles compared to mesophiles. The overall contribution of electrostatics quantified by combining the net interaction energy and the solvation penalty of folding - due to differential charge burial in the folded and the unfolded ensemble - is also mostly favorable in thermophilic proteins compared to mesophiles. The systems that deviate from this trend provide interesting test cases to learn more about alternate design strategies when modification of charges is not viable due to functional reasons. The unequal contribution of the unfolded state to the stability in thermophiles and mesophiles highlights the importance of modeling the disordered ensemble to understand thermophilic adaptation as well as protein stability, in general. Our integrated approach - combining finite element analysis with MC and MD - can be useful in designing charge mutations to alter protein stability.