Extending the osmophobic effect to protein side chains with a physical transfer model across osmolyte classes

Abstract

Understanding the role of the protein backbone and side chains on cosolvent-induced stabilization is essential for a molecular picture of osmolyte action. The prevailing view holds that protecting osmolytes stabilize proteins mainly through unfavorable backbone interactions - the osmophobic effect - with side chains playing a minor or opposing role. By revisiting the decomposition of amino acid transfer free energies to properly account for mutual shielding between backbone and side chains, we derive a transfer model consistent with experimental denaturation m-values for both urea and protecting osmolytes - a feat neither existing model achieves alone. Backbone accessibility is mechanism-dependent: geometric for excluded cosolvents, complete for binders with specific cosolvent-backbone interactions. For strong protecting osmolytes - TMAO, sarcosine, sucrose, trehalose, and sorbitol - both backbone and side chains contribute favorably to stabilization, with side-chain contributions comparable or even greater than the backbone's. For urea, accounting for the preferential binding and directional nature of urea-backbone hydrogen bonding - which makes the backbone accessible regardless of side-chain shielding - recovers the known balanced backbone and side-chain contributions to denaturation. Weaker protectants such as proline, betaine, and glycerol show competing backbone and side-chain effects that partially cancel. These results extend the osmophobic effect to side chains and yield a three-tier classification of osmolyte action: cooperative stabilization, cooperative destabilization, and competing backbone/side-chain contributions. The model's greater sensitivity to side-chain composition offers avenues for experimental validation and for protein engineering.