A coarse-grained approach to protein design: learning from design to understand folding.

Ivan Coluzza

doi:10.1371/journal.pone.0020853

Abstract

Computational studies have given a great contribution in building our current understanding of the complex behavior of protein molecules; nevertheless, a complete characterization of their free energy landscape still represents a major challenge. Here, we introduce a new coarse-grained approach that allows for an extensive sampling of the conformational space of a large number of sequences. We explicitly discuss its application in protein design, and by studying four representative proteins, we show that the method generates sequences with a relatively smooth free energy surface directed towards the target structures.

Highlights

Protein molecules play a central role in the large majority of biochemical reactions in living organisms [1]
We compared the artificial hydrophobic/ philic profile (HP) profiles to the average profile obtained from the Pfam alignment data (PF00542) for protein 1CTF; the curve for W.S. sequences is qualitatively comparable to one of the real proteins, as the discrepancies occur in regions where the wild type proteins express hydrophobic residues even if highly exposed to the solvent, which could be the results of functionalities that we did not include in the design procedure. At this point it is natural to ask if the caterpillar model, with the solvation term, is able to reproduce the folded structures of real proteins, since we have shown that designed sequences refold to the target structure, and the design produces protein like sequences
Each of the tested sequences reached the target structure with a very high precision considering the simplicity of the model, demonstrating that the procedure is universal for proteins with different proportions of alpha helices and beta sheets

Summary

Introduction

Protein molecules play a central role in the large majority of biochemical reactions in living organisms [1]. Performance of these functions generally requires folding of the proteins into a specific three-dimensional structure, the so-called native state [2,3], (a number of exceptions involving the so-called ‘‘disordered’’ proteins has been discovered [4]). Folding becomes more complex and requires an extensive search in the space of possible sequences to obtain a folding chain. By using lattice models it was possible to design heteropolymers with a large variety of target configurations, and to generate lattice proteins with more complex self-assembly properties [14,15]. Some successful designs of novel artificial enzymes have been obtained by introducing residues expected to play a catalytic role in a specific reaction [16] in sequences with known folds

Methods

Results

Conclusion