A computational characterization of hydrogen bonding in serine proteases

Several hydrogen bonding properties of serine proteases are studied by QM/MM calculations that make use of the Effective Fragment Potential (EFP) methodology.; First, two divide-and-conquer (DAQ) approaches for building multipole-based molecular electrostatic potentials of proteins are presented and evaluated. Both methods create systems with integer charges without using charge renormalization and are used to assemble pieces in the EFP methodology. The new DAQ approaches are tested in calculations of the proton affinity of N ζ of Lys55 in the inhibitor turkey ovomucoid third domain, used to build a variety of MM regions, applied to calculations of the pK_a of Lys55, and compared to other computational methodologies in which force field charges are employed.; Second, the relationship between hydrogen bonding and the experimentally observed NMR chemical shifts in the catalytic triad of low-pH α-chymotrypsin is investigated by combined use of the EFP and ONIOM-NMR methods. NMR chemical shifts are reproduced and their physical meaning examined. Bond lengthening and polarization due to the molecular environment are postulated as the causes of the extreme down-field chemical shift of Hδ1 (18.2 ppm). The unusual down-field shift of Hϵ1 (9.2 ppm) is shown to be induced by interactions with the C=O group of Ser214. The free energy cost of moving Hδ1 from His57 to Asp102 is predicted to be 5.5 kcal/mol.; Third, a theoretical model for the systematic study of various structural and spectroscopic properties of strong hydrogen bonds in enzymes is presented. The model is applied to the Nδ1-H-Oδ1 hydrogen bond between His57 and Asp102 in the active sites of low-pH α-chymotrypsin and α-lytic protease. The minimum energy structures of both enzymes reproduce the experimental Nδ1-Oδ1 distance and are used to obtain computational values for the H-D fractionation factor (&phis;), the proton chemical shift (δ_H) of the H δ1, and changes in δ_H upon isotope substitution (Δδ_H-D and Δδ_H-T). For both enzymes, calculated parameters are in good agreement with available experimental data. The theoretical model is used to make predictions for other properties for which experimental values are not available.