In silico method for identifying the key residues in a knotted protein: with MJ0366 as an example.

The knotted proteins are a class of rare but biologically important proteins, due to the special topology of their native structure. Here we present a simple in silico method to identify the key residues for knotting and unknotting in a knotted protein, using the trefoil protein MJ0366 as an example. We first simulate the folding process via the annealing molecular dynamics (AMD) simulations in the coarse-grained "Go"-like model. From the folding trajectories, we monitor the knotting process using the quantity "length of knot tails". In the meantime, we analyze the evolution of the local geometry of the C α trace with the help of the Discrete Frenet Frame (DFF). We identify the key residues by correlating the local geometry at each residue with the variable "length of knot tails" in the folding process, where a higher correlation coefficient indicates that the residue is more important for knotting. We validate our method by comparing with the experimental results in the literature. With the same method, we further predict the key residues for unknotting MJ0366 using the AMD simulations in both the coarse-grained "Go"-like model and all-atom (AA) force field model, respectively. We find that the key residues for unknotting are partially overlapped with those for knotting, indicating that the pathways for unknotting and knotting are generally similar except for the existence of some non-native contact interactions in the unknotting process. This in silico method can provide a new insight for understanding the knotting and unknotting processes of a knotted protein.