Computational Calculations of pKa Values of Metal Ligands in Proteins

Metalloenzymes are enzymes that use a metal cation as a cofactor in their active site. Theycatalyze various reactions, such as hydrolysis, oxidation, and reduction. Some examples ofmetalloenzymes are hydrogenases, which facilitate the uptake of molecular hydrogen;superoxide dismutase, which inhibits the oxidative damage to cells; nitrogenase, which enablesthe fixation of atmospheric nitrogen; proteases, which break down peptide bonds; andphosphodiesterases, which cleave phosphate ester bonds [۱-۴]. Many metalloenzymes areinvolved in acid-base reactions and have experimentally determined pKa values. In this study,we calculate the pKa values of ۱۸ acid-base reactions in nine different metalloenzymes withvarying metal ligands, metal centers (Zn, Mn, or Fe), protein cofactors, enzyme variants (wildtype or mutated), or redox states of the metal centers. We compare our calculated values with theexperimental ones. The metalloenzymes we study are alcohol dehydrogenase, superoxidedismutase, Rieske protein, thioredoxin-like ferredoxins, carbonic anhydrase, heme nitricoxide/oxygen binding protein, cytochrome P۴۵۰, and myoglobin. We use quantummechanics/molecular mechanics (QM/MM) calculations and the QM-cluster method in acontinuum solvent to perform the calculations. We explore different variations of the approach,such as the QM method, the basis set, the size of the QM system, and the inclusion of theenvironment either implicitly as a continuum dielectric or explicitly as point charges. We alsouse the QM/MM thermodynamic cycle perturbation (QTCP) method to examine the effect ofenvironmental dynamics on the calculations. Additionally, we use the Big-QM method, whichemploys a large QM system to do single-point QM/MM calculations on the structure that hasbeen optimized using the original QM/MM technique. This study aims to determine which of themethods we test are the most accurate and whether they can be used for predictive analyses. Wehave obtained a set of data for pKa using various methods, but we have not yet reached aconclusion on which method is the best.