Computational Approach Choice in Modeling Flexible Enzyme Active Sites

The last step in the reductase step of the catalytic mechanism of MsrB was re-investigated using several computational approaches. Our previous QM-cluster paper showed that two possible mechanisms could occur, however the direct formation of disulfide from sulfonium cation was favored over sulfenic acid formation. In contrary, experimental studies suggest sulfenic acid formation. Therefore, first, we investigated the effect of level of theory, which confirmed previous conclusion. In addition, the effect of model choice was also investigated using ONIOM including a large QM layer around Cys440. Interestingly, deprotonating Cys440 leads to direct nucleophilic attack on Cys495 forming disulfide. Second, to eliminate the possibility that all previous results are an artifact of the used crystal structure in which the S...S distance is 3.29 Å, we ran a 5 ns MD simulation on the sulfonium cation intermediate. Surprisingly, our results show that the distance between the two sulfur is significantly increased to 4.88 Å. More importantly a water molecule is located in a proper position for nucleophilic attack. QM/MM calculations shows that sulfenic acid is formed via low barrier of 16.7 kJ mol-1. Finally, the effect of substrate binding on the two Cys's distance were investigated via running several MD simulations of possible intermediates, showing that substrate binding induces conformational changes increasing the sulfur's distance which is decreased upon substrate removal upon sulfenic acid formation. These results question the applicability of QM cluster approach in systems including flexible turns. It also emphasizes the importance of proper preparation of the starting structure.