Thermodynamics of the Purine Nucleoside Phosphorylase Reaction Revealed by Computer Simulations

Enzymes are able to catalyze chemical reactions by reducing the activation free energy, yielding significant increases in the reaction rates. This can thermodynamically be accomplished by either reducing the activation enthalpy or increasing the activation entropy. The effect of remote mutations on the thermodynamic activation parameters of human purine nucleoside phosphorylase is examined using extensive molecular dynamics and free energy simulations. More than 2700 independent reaction free energy profiles for six different temperatures have been calculated to obtain high-precision computational Arrhenius plots. On the basis of these, the activation enthalpies and entropies were computed from linear regression of the plots with Δ<i>G</i>^⧧ as a function of 1/<i>T</i>, and the obtained thermodynamic activation parameters are in very good agreement with those from experiments. The Arrhenius plots immediately show that the 6-oxopurines (INO and GUO) have identical slopes, whereas the 6-aminopurine (ADO) has a significantly different slope, indicating that the substrate specificity is related to the difference in thermodynamic activation parameters. Furthermore, the calculations show that the human PNP specificity for 6-oxopurines over 6-aminopurines originates from significant differences in electrostatic preorganization. The effect of the remote double mutation, K22E and H104R (E:R), has also been examined, as it alters human PNP toward the bovine PNP. These residues are situated on the protein surface, 28–35 Å from the active site, and the mutation alters the enthalpy–entropy balance with little effect on the catalytic rates. It is thus quite remarkable that the empirical valence bond method can reproduce the enthalpies and entropies induced by these long-range mutations