Very fast empirical prediction and rationalization of protein pKa values

A very fast empirical method is presented for structure-based protein pK(a) prediction and rationalization. The desolvation effects and intra-protein interactions, which cause variations in pKa values of protein ionizable groups, are empirically related to the positions and chemical nature of the groups proximate to the pK(a) sites. A computer program is written to automatically predict pK(a) values based on these empirical relationships within a couple of seconds. Unusual pKa values at buried active sites, which are among the most interesting protein pKa values, are predicted very well with the empirical method. A test on 233 carboxyl, 12 cysteine, 45 histidine, and 24 lysine pK(a) values in various proteins shows a root-mean-square deviation (RMSD) of 0.89 from experimental values. Removal of the 29 pK(a) values that are upper or lower limits results in an RMSD = 0.79 for the remaining 285 pK(a) values.