Abstract
Engineered proteins are increasingly being required to function or pass through environmental stresses for which the underlying protein has not evolved. A major example in health are antibody therapeutics, where a low pH step is used for purification and viral inactivation. In order to develop a computational model for analysis of pH-stability, predictions are compared with experimental data for the relative pH-sensitivities of antibody domains. The model is then applied to proteases that have evolved to be functional in an acid environment, showing a clear signature for low pH-dependence of stability in the neutral to acidic pH region, largely through reduction of salt-bridges. Interestingly, an extensively acidic protein surface can maintain contribution to structural stabilisation at acidic pH through replacement of basic sidechains with polar, hydrogen-bonding groups. These observations form a design principle for engineering acid-stable proteins.