Transition Path Sampling Guided by Structural Motifs

Meaningful insight into the function of a dynamic protein requires knowledge of the conformations that it can adopt and how it transitions between these conformations. Despite tremendous strides in structure determination in the past decade, it is still uncommon to resolve a protein in all of its key conformational states. However, it is not uncommon for family members to be solved in many different states throughout a reaction cycle, but if their sequence identity is low, it is difficult to leverage this information to determine how any single member moves through this conformational landscape. Here we develop a simulation technique that uses distance matrices to a target state to define a structural similarity metric (SSM) to guide the dynamics of a protein from a known conformation to a target, based on a related protein. Applying the method to the well-studied β-β-α (BBA) protein, we fold the β-hairpin portion of BBA to the correct native state with the correct sequence alignment without specifying which residue pairs correspond to which distance values in the matrix. We then generalize the SSM to multiple dimensions based on structural motifs within the protein and simulate the full folding. We compare different string-based, enhanced sampling methods coupled to the SSM and discuss their advantages and disadvantages. We show that the method can simulate the transition to a new conformation based on a homologous protein structure even when the sequence similarity is too low for alignment-based methods to guide the transition. We end by suggesting that the deconstruction of state space into distinct structural motifs is a natural and potentially efficient framework for searching through conformational space.