Benchmarking pK_a prediction methods for Lys115 in acetoacetate decarboxylase

Three different pK_a prediction methods were used to calculate the pK_a of Lys115 in acetoacetate decarboxylase (AADase): the empirical method PROPKA, the multiconformation continuum electrostatics (MCCE) method, and the molecular dynamics/thermodynamic integration (MD/TI) method with implicit solvent. As expected, accurate pK_a prediction of Lys115 depends on the protonation patterns of other ionizable groups, especially the nearby Glu76. However, since the prediction methods do not explicitly sample the protonation patterns of nearby residues, this must be done manually. When Glu76 is deprotonated, all three methods give an incorrect pK_a value for Lys115. If protonated, Glu76 is used in an MD/TI calculation, the pK_a of Lys115 is predicted to be 5.3, which agrees well with the experimental value of 5.9. This result agrees with previous site-directed mutagenesis studies, where the mutation of Glu76 (negative charge when deprotonated) to Gln (neutral) causes no change in K_m, suggesting that Glu76 has no effect on the pK_a shift of Lys115. Thus, we postulate that the pK_a of Glu76 is also shifted so that Glu76 is protonated (neutral) in AADase. Graphical abstract Simulated abundances of protonated species as pH is varied.