Firstly, for the reaction of ethylcarbamate in simplified heme active site, we computed all the minima and transition state structures along the potential energy surface with Gaussian09. The potential energy surface was elucidated for both the doublet and quartet spin states of the iron compound. For both spin states, the energy barriers were around 20 kcal/mol, which was in line with previous literature. After the first hydrogen abstraction transition state, the resulting intermediate can undergo either isomerization to attain the productive conformation for either the hydroxylation or desaturation step without having to bypass another transition state – a common feature of a bifurcation. Thus, traditional DFT calculations could not help explaining product distribution by simply comparing which transition state was more energetically feasible, or which product was more thermodynamically stable. Herein, we employed quasi-classical dynamics simulations to understand product distribution. 100 dynamics trajectories were simulated from the first transition state structure. The OH-rebound pathway toward hydroxylation product was found to dominate. This result was in line with previous literature reporting that hydroxylation product was the main product. Our results also suggested that the pathway toward desaturation product was less efficient as it required more time for intermediate reorganization.
To understand the difference between heme and non-heme enzymes, we needed to further examine the reactions catalyzed by non-heme active site. Potential energy surfaces were elucidated for the stepwise N-demethylation reactions for KMe3 and KMe3 methylated lysine with (AcO)2(imidazole)2Fe=O. Transition state structures and corresponding minima were computed for quintet and triplet surfaces, using the same method as above. Quasi-classical dynamics were then simulated for both substrates, starting from the hydrogen abstraction transition state. 100 trajectories were simulated for each system. Comparing the KMe3 and KMe2 system, an additional side product was observed for the KMe2 substrate due to the presence of the NH proton. This prompted reexamination of the potential energy surface with DFT for the formation of this side product. The results are discussed in more details in the upcoming publication.
Following the idea of site selectivity in enzyme active site and how that would affect product distribution, we collaborated with an experimental group in the same department to investigate the role of oxonium ions as reactive intermediates in several natural product biosynthesis. I performed quantum mechanical calculations to help confirm the identity of the oxonium ion intermediates by sampling various conformations of oxonium ions, computing chemical shifts for the ensemble of energetically relevant conformers, and then comparing the computational NMR data to their experimental data. This workflow was repeated for all 5 oxonium ions. Transition state structures (TSSs) were calculated for 1,2-hydride shifts to help understand the decomposition of the reactive oxonium ion. TSSs were also calculated for the reactions of forming oxonium ions and for the reactions between oxonium ions and nucleophiles. Though there are three possible sites on the oxonium ions for nucleophilic attack, the observed product distribution in earlier experimental study suggested that one site was more favorable. Distortion-interaction analyses revealed that reaction at this site involved the least structural reorganization of the substrate in enzyme active site. Manuscript with detailed discussion of these results are under reviews for publication.
