Kinetic models for intrinsically disordered peptides from molecular dynamics simulations

Intrinsically disordered proteins (IDPs) are proteins which lack a well-defined three-dimensional structure. Despite being disordered in solution, their conformational ensemble is usually not fully random, but they transiently sample various secondary structure elements. These proteins are often involved in signal transduction or molecular recognition and it has been proposed that their function relies is determined by their molecular kinetics. Since the experimental characterization of molecular kinetics in a diverse conformational ensemble is extremely difficult, molecular dynamics simulations have become an indispensable tool for the investigation of IDPs.

From Markov state model analyses of molecular-dynamics simulations, we know that the molecular kinetics can be described as a superposition of dynamical modes (the eigenvectors of the Markov model transition matrix). This description does not rank the long-lived conformations according their free energies but yields a hierarchy of kinetic exchange processes between groups of long-lived conformations. In this sense it is complementary to the usual free energy picture. Using a recently published variational approach to conformational dynamics, the slow dynamic modes can be expressed as a linear combination of functions defined on the conformational space. By defining functions which have a structural interpretation, one can directly interpret the linear combination coefficients as contributions of a specific structural transition to the overall dynamic mode. We have developed such a set of functions for the conformational dynamics of peptides. Each functions is a constructed from the possible local dynamic modes within a residue: a stationary mode and two dynamically excited modes. The results of the method show that the overall slow dynamic processes are a combination of only a few locally excited modes, implying that the slow conformational dynamics of peptides is governed by a few critical residues. I will demonstrate the method on a 15-residue fragment from the IDP human islet amyloid polypeptide and compare the results to the analogous fragment from cat islet amyloid polypeptide, which has a slightly different sequence.