Simulation of protein structure by a simplified model: |application to the rop dimer

Abstract

In this thesis, the estimation of the native three-dimensional structure of a dimeric protein, Rop, from its amino-acid sequence is performed. Rop is a small RNA-binding protein that is coded by certain plasmids and is involved in plasmid replication. This is a four-helix bundle dimer, the monomers being identical oppositely oriented helical hairpins of 63 residues, each. The native state is considered to be the most thermodynamically stable state corresponding to the lowest energy conformation. Here, a molecular model which significantly decreases the number of variables required for describing a given configuration is introduced. This is the so-called coarse-grained or low-resolution approach which is based on the idea that each residue is composed of two heterogeneous interaction sites, one on the backbone atom and the other on the side group selected atoms. A framework model of rigid building blocks is adopted in the computational approach to the protein folding problem. The blocks are the secondary structural units which possess sufficient internal stability so as to form at early stages of folding and maintain their integrity during the large-scale structural reorganization of the protein. Within the approximation of the model, the problem reduces to determination of the optimal organization of the already predicted secondary structural elements in space. The lowest energy conformation of the Rop dimer is generated by considering all the possible threedimensional rearrangements that can exist between the two monomers. The potential energies between the interaction sites of the protein are extracted from a database of known protein structures, using the Brookhaven Protein Data Bank (PDB). These potential energies are referred to as knowledge-based potentials. The predicted lowest energy conformation obtained by the knowledge-based potentials is in reasonable agreement with the native structure of Rop dimer. It is concluded that the simplified knowledge-based potentials can be satisfactorily used for a low-resolution estimation of the native state of globular proteins. Computations are repeated by using the effective interresidue contact energies which replace the knowledge-based potentials. The effective interresidue contact energies are defined for two distance ranges, 2 A... and 4.4 A... , only for a given pair of amino acid side chains i and j separated by a distance rij· Accordingly, they provide a relatively simplified basis for evaluating the conformations, as opposed to the knowledge-based potentials which are determined at 0.4 A throughout the ranges 2.0 A ...A. In fact, it is found that the lowest energy conformation obtained with the effective interresidue contact energies conform with the native structure of the protein and it is concluded that the contact energies are sufficient for estimating the three-dimensional structure of globular proteins. Hovewer, PDB extracted knowedge-based potentials are shown to yield a more satisfactory account of the native state behaviour.