Protein motions are complex, including occurring at different time scales, and their roles in enzyme-catalyzed reactions have always been of great interest among enzymologists. In order to characterize the potential factors that play a role on the chemical step of enzymatic reactions, variants of dihydrofolate reductase have been used as a benchmark system to study the motions of proteins correlated with the chemical step. A “global dynamic network” of coupled residues in Escherichia coli dihydrofolate reductase (ecDHFR), which assists in catalyzing the chemical step, has been demonstrated through quantum mechanical/molecular mechanical and molecular dynamic (QM/MM/MD) simulations, as well as bioinformatic analyses. A few specific residues — M42, G121, and I14 — were shown to function synergistically with measurements of single turnover rates and the temperature dependence of intrinsic kinetic isotope effects (KIEsint) of site-directed mutants. Although similar networks have been found in other enzymes, the general features of these networks are still unclear. This project focuses on exploring homologous residues of the proposed global network in human DHFR through computer simulations and measurements of the temperature dependence of KIEsint. The mutants M53W and S145V, both remote residues, showed significant decreases in catalytic efficiency. Non-additive isotope effects on activation energy were observed between M53 and S145, indicating their synergistic effect on hydride transfer in human DHFR.
Apart from the effects of the conserved residues, we also extend our studies to exploring three potential phylogenetic events that account for the discrepancies between E. coli and human DHFR. They are L28, PP insertion and PEKN insertions by phylogenetic sequence analysis. Two of them (N23PP and G51PEKN E. coli DHFR) have been proved to be important both by MD simulation and experimental probe of KIEs measurement. The experiments have found that PP insertion itself rigidified the M20 loop and motions coupled to hydride transfer were impaired, however, loop rigidification was improved after incorporating PEKN. Furthermore, deletion of PP and PEKN of the engineered human enzyme also show a similar outcome. However, the effect of the key residue of L28 is not clear. In this project, we have step-wise engineered the human DHFR to be like hagfish (F31M) and E. coli (F32L). And it is found out that there is an increase in the temperature dependence of KIEs when the enzyme was bacterilized into a more primitive variant. This indicates that not only is residue F32 important and correlated with the chemical step as indicated by bioinformatic studies, but it is possible to trace the evolutionary trajectory. A triple mutation F32L-PP26N-PEKN62G on the human DHFR was also conducted, and it is not surprising to find out that the temperature dependence of KIEs has retained its behavior like wild-type human DHFR. These results suggest that the three predicted phylogenetically coherent events coevolved together to maintain the evolutionary preservation of the protein dynamics to enable H-tunneling from well-reorganized active sites.
As has been indicated by the previous project, as the enzyme evolves, the active site of the enzyme will “reorganize” to form the optimal transition state for chemical step (from F32L-F32M-wild type DHFR). Here in this project, we aimed to systematically address this point of view through a series of cyclic permutation DHFR from directed evolutions. As this primitive enzyme is 7 orders of magnitude less efficient than the well-evolved human DHFR, together with four generations of evolved variants (cp, cp’ and cp”), this provides a good model system for explorations of the molecular basis of enzyme evolution. It is found that the organizations of transition state are improved before the catalytic efficiency is enhanced as the enzyme evolves.