Stability and Conformational Specificity in Protein Design: Models for Binary Patterning and Electrostatics

Abstract

Binary patterning (the arrangement of hydrophobic and polar amino acids) and electrostatics are important determinants of the stability and conformational specificity of designed proteins. We have developed methods to to select the optimal binary pattern and model electrostatics in protein design studies. The Genclass method of binary patterning uses a solvent accessible surface generated from backbone coordinates of the target fold and side chains, constructs whose size and shape are similar to an average amino acid. Each position is classified according to the solvent exposure of its generic side chain. The method was tested by analyzing several proteins in the Protein Data Bank and by experimentally characterizing homeodomain variants whose binary patterns were systematically varied. Selection of the optimal binary pattern results in a designed protein that is monomeric, well-folded, and hyperthermophilic. Homeodomain variants with fewer hydrophobic residues are destabilized, additional hydrophobic residues induce aggregation. The optimal variant was further characterized by nuclear magnetic resonance spectroscopy. Binary patterning, in conjunction with a force field that models folded state energies, appears sufficient to satisfy two basic goals of protein design: stability and conformational specificity. Electrostatic interactions are critical determinants of protein structure and function. Computational protein design algorithms typically use fast methods based on Coulomb's law to model electrostatic interactions. These methods fail to accurately account for desolvation and solvent screening, which strongly attenuate electrostatic interactions in proteins. Using the current force field, we designed a 25-fold mutant with moderate stability similar to the wild type protein. Incorporating two classes of electrostatic interactions using simple rules yielded a nine-fold mutant of the initial design that is over 3 kcal mol^(-1) more stable. The simple electrostatic model used in the ORBIT force field is unable to predict the experimentally determined stabilities of the designed variants. Finite difference Poisson-Boltzmann (FDPB) methods have substantially better predictive power, but are far too slow for problems with high combinatorial complexity. We have developed new strategies for modeling electrostatics in protein design problems that utilize one- and two-body decomposable FDPB methods. Computational results indicate that this method has the accuracy and speed required for design calculations.