Universal Metrics of Interstructure Distance for Flexible and Intrinsically Disordered Proteins

Abstract

Proteins exist along a continuum from order to disorder, and many are now known to contain both folded and unfolded regions. Molecular dynamics (MD) simulations of intrinsically disordered proteins reveal deficiencies in most current analysis tools. Proteins with disordered regions lack stable references structures and analysis results are influenced by the underlying dependence of most structural measurements on molecule size and polymer length. Our solution (1) identifies an optimal measurement for comparing highly dissimilar conformations, (2) performs a pairwise structural comparison which does not require a reference structure [Fisher and Stultz. 2011], (3) uses an entropic chain model to estimate maximal protein motion, and (4) normalizes to the minimal thermal protein motion of folded proteins based on MD simulation data from the Dynameomics database [van der Kamp, et al. 2010]. This approach provides a universal measurement of interstructure distance (ISD). Per residue ISD provides an improved means of identifying regions of local disorder and flexibility. The output is analogous to root mean square fluctuation but resilient to the errors associated with fitting to a reference structure. Whole molecule ISD provides a basis for multidimensional scaling and conformational clustering. The tool g_isd implements a library of ISD analysis methods designed to be familiar to users of the Gromacs MD package [Hess, et al. 2008].