Abstract
Since pH sensitivity has a fundamental role in biology, much effort has been committed to establishing physical models to rationalize and predict pH dependence from molecular structures. Two of the key challenges are to accurately calculate ionizable group solvation and hydration and then to apply this modeling to all conformations relevant to the process in question. Explicit solvent methods coupled to molecular dynamics simulation are increasingly complementing lower resolution implicit solvent techniques, but equally, the scale of biological data acquisition leaves a role for high-throughput modeling. Additionally, determination of ranges of structures for a system allows sampling of key stages in solvation. In a review of the area, it is emphasized that pH sensors in biology beyond the most obvious candidate (histidine side chain, with an unshifted pK (a) near neutral pH) should be considered; that modeling can benefit from other concepts in bioinformatics, in particular modulation of interactions and function in families of homologs; and that it can also be beneficial to incorporate as many experimental structures as possible, to mitigate against small variations in conformation and to analyze larger, functional, conformational changes. These aspects are then demonstrated with new work on the spike protein of SARS-CoV-2, looking at the pH dependence of variants, including prediction of a change in the balance of locked, closed, and open forms at neutral pH for the Omicron variant spike protein.