Modeling studies of membrane proteins employing quantum mechanical-molecular mechanical simulations and elastic network models
Abstract
This study is a theoretical investigation into the conductance of a newly discovered class of trans-membrane proton transporters, as well the development of a number of novel methods for improving the normal mode analysis of trans-membrane proteins. Accurate modeling of trans-membrane proteins presents unique challenges, as these proteins are embedded in a highly inhomogeneous environment which plays an important role in their function. A proposed model for the conduction of the Hv1 family of proton transporters is derived from the known primary structure of a Kv family voltage sensing domain, as there is sequence similarity to suggest they share a similar structure. Various theoretical methods are applied such as QM/MM, SCC-DFTB, GSBP, umbrella sampling and WHAM in order to compute the potential of mean force and diffusion profile, which are then used in a Nernst-Planck formalism to estimate the overall conductance. An evaluation of the accuracy of SCC-DFTB was also performed by comparing the computed energies with those using the B3LYP density functional, in order to better understand the computed potential of mean force. Three novel methods are shown to improve the accuracy of the coarse grained elastic network model, as it is applied to compute the root mean squared fluctuations of a protein's alpha-carbons. The first two methods are designed to reduce the "tip effect", by employing various methods of tethering residues which show excessive amplitude in the computed RMSFs. The third method models the inhomogeneous environment by including viscosity, which is dependent on the position and solvent accessible surface area of the residue.
- Publication:
-
Ph.D. Thesis
- Pub Date:
- October 2008
- Bibcode:
- 2008PhDT........86R