NATURE OF DRIVING FORCE FOR PROTEIN FOLDING : A RESULT FROM ANALYZING THE STATISTICAL POTENTIAL

By Hao Li, Chao Tang, Ned S. Wingreen. 1995

Proteins fold into specific three dimensional structures to perform their diverse biological functions. It is now well established that for small proteins the information contained in the amino acid sequence is sufficient to determine the folded structure, which is the structure with minimum free energy [1]. Thus the native structure is dictated by the physical interactions between amino acids in the sequence, and understanding the nature of such interactions is crucial for protein structure prediction. As a protein contains thousands of atoms and interacts with huge number of water molecules, it is not feasible to calculate the free energy function from first principles. An often adapted practical approach is to derive a coarse grained potential (often on the level of amino acids) using the known structures in the existing protein data banks. In such an approach, the energy of a particular substructure in proteins is derived from the number of its appearances in the structure data bank via a Boltzmann factor [2 ‐ 4]. A classic example of such a statistical potential is the Miyazawa-Jernigan (MJ) matrix, a 20 3 20 inter-residue contact-energy matrix derived by Miyazawa and Jernigan [2,5]. This matrix tabulates the interaction strength between any two types of amino acids in proteins, and has been widely applied in protein design and folding simulations [6,7]. In this Letter, we apply a general method of matrix analysis, namely, eigenvalue decomposition, to the MJ matrix. The analysis reveals an intrinsic regularity of the MJ matrix, which yields basic information about the nature of the driving force for protein folding. We show that despite the complicated interactions in proteins, the major driving force is hydrophobic interaction and a force of demixing, the latter obeying Hildebrand’s solubility theory of simple liquids [8]. The result allows us to attribute the interactions responsible for folding to quantifiable properties of individual amino acids. These properties suggest further experimental tests, and can be used for analyzing sequence-structure relation. Eigenvalue decomposition is a general approach to analyzing matrices. A given N 3 N real symmetric ma

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Physics-driven coarse-grained model for biomolecular phase separation with near-quantitative accuracy (2021, background): “Although it has been suggested 24,43 that simple arithmetic combination rules are often sufficient, unlike in previous models, the pairwise interactions for those residue pairs that dominate the phase behavior are explicitly specified, giving the model greater flexibility.”
Pair potentials for protein folding: Choice of reference states and sensitivity of predicted native states to variations in the interaction schemes (2008, methodology): “When compared to the mixing energy obtained from Hildebrand’s solubility theory, theui can be related to the vaporization energies~Li et al., 1997!.”