PHYS 360 |
| Biological function of a protein is often driven by its response to a specific signal. Changes in the protein structure may occur in response to such signals and may be important for protein function. In larger proteins, local conformational rearrangements of atoms initiated by, e.g., interaction with a ligand may result in a movement of entire sub-domains making it difficult to predict conformational plasticity of a protein based on limited structural information, e.g. when a structure of only one or a few conformations is known. Coarse-grained single structure models such as Anisotropic Network Model (ANM) and rigid cluster decomposition (FIRST) (Jacobs and Thorpe. 1995 Phys.Rev.Lett. 75:4051-4054) have been developed to predict protein flexibility, large scale conformational rearrangements and dynamics, but avoid costly MD simulations. In this work using extensive MD simulations we explore a mechanistic flexibility of several globular proteins of various sizes and structures: barnase, GluR2 S1S2, DHFR and HIV-1 RT. We compare and contrast different coarse graining computational techniques for their ability to reliably predict functionally relevant protein dynamics. ANM indeed compares well with the top modes of Essential Dynamics from MD simulation. However, conformational rearrangements required by function of the protein may not be among the top modes of ED. We show that the rigid cluster decomposition method combined with MD simulations in a hierarchical manner (termed MD/FIRST (Mamonova et all. Phys.Biol. 2005 2:S137-47) can reliably predict structural and dynamical functionally important sub-domain movements. |
|
PHYS Poster Session - Multiscale Modeling in Biophysics
7:30 PM-10:00 PM, Wednesday, April 9, 2008 Morial Convention Center -- Hall A, Poster
Division of Physical Chemistry |