Ab initio protein folding simulations using atomic burials as informational intermediates between sequence and structure

Abstract

The three-dimensional structure of proteins is determined by their linear amino acid sequences but decipherment of the underlying protein folding code has remained elusive. Recent studies have suggested that burials, as expressed by atomic distances to the molecular center, are sufficiently informative for structural determination while potentially obtainable from sequences. Here we provide direct evidence for this distinctive role of burials in the folding code, demonstrating that burial propensities estimated from local sequence can indeed be used to fold globular proteins in ab initio simulations. We have used a statistical scheme based on a Hidden Markov Model (HMM) to classify all heavy atoms of a protein into a small number of burial atomic types depending on sequence context. Molecular dynamics simulations were then performed with a potential that forces all atoms of each type towards their predicted burial level, while simple geometric constraints were imposed on covalent structure and hydrogen bond formation. The correct folded conformation was obtained and distinguished in simulations that started from extended chains for a selection of structures comprising all three folding classes and high burial prediction quality. These results demonstrate that atomic burials can act as informational intermediates between sequence and structure, providing a new conceptual framework for improving structural prediction and understanding the fundamentals of protein folding. Proteins 2014; 82:1186-1199. (c) 2013 Wiley Periodicals, Inc.