Coarse-grained Modeling of Protein Dynamics using Elastic Network Models

Dynamics is crucial for the functioning of biological macromolecules. Because of severe limitations in studying protein dynamics experimentally or with all-atom simulations, coarse-grained methods, especially elastic network models (ENMs), are frequently employed. In the last years, studies on various proteins showed that ENMs reliably reproduce experimental data, despite the simplified protein representation and the purely harmonic potential function. This work on two proteins with very different dynamical properties highlights the remarkable success of ENMs and shows which kind of questions can be answered using coarse-grained methods. The allosteric enzyme aminoglycoside phosphotransferase(3')-IIIa (APH), which confers resistance against a broad range of aminoglycoside antibiotics to pathogenic bacteria, drastically changes its flexibility upon binding of substrates, but without changing its average conformation. In contrast, the homotrimeric vesicular stomatitis virus glycoprotein G (VSV-G), which triggers the pH-dependent fusion of viral and host membrane, undergoes a large structural rearrangement. A striking difference between the two proteins is their shape. VSV-G contains weakly constrained protein segments, the fusion loops, which can undergo large-scale motions at low energetic cost, whereas APH is not obviously arranged into different protein segments. Nevertheless, ENM calculations show that also APH consists of independently moving segments with correlated internal motion, so-called dynamic domains. The concept of dynamic domains can explain the differential effects of ligand binding on the dynamics of APH. The first chapter of this thesis describes how experimental evidence for the importance of dynamics successively replaced the former idea of static proteins, and explains the dynamic basis of ligand binding, allostery and conformational changes. In the second chapter, theoretical methods for the analysis of protein dynamics are introduced, with emphasis on the ENM-based methods used in my studies. The studies are summarized in the third chapter. In the study on APH, I employ the Gaussian network model to analyze the ligand-dependent dynamics, the broad substrate specificity and the perturbation-sensitivity of the ligand binding sites. In a second study, ENM-based as well as all-atom molecular dynamics simulations are used to analyze the conformational change of VSV-G. Both approaches detect the fusion loops of VSV-G as most flexible parts of the protein, and thus as most likely starting point for the structural rearrangement, but only the all-atom model can generate deviations from the average structure at low pH. The last study describes the implementation and application of a dynamic domain assignment method, called CovarDom, which is based on covariances of residue fluctuations. Calculation of dynamic domains for a large protein set demonstrates the general applicability of CovarDom.

Level thesis:degree_level

thesis.doctoral

Grantor dc:publisher

Universität Bayreuth

Year

2013

Author dc:creator

• Wieninger, Silke

Contributors dc:contributor

• Ullmann, G. Matthias

Repository record source_url

https://epub.uni-bayreuth.de/id/eprint/124/

OAI identifier oai:identifier

oai:epub.uni-bayreuth.de:124