Conformational transitions of proteins using multi-scale modeling approaches

Abstract

A detailed analysis on a dataset consisting of 30 proteins was performed using a hybrid simulation technique. This iterative technique, named as ANM-MC, generates a targeted pathway between two conformations, where the collective modes from the anisotropic network model (ANM) are used for deformation and the energy of the deformed structure is minimized via local moves using a knowledge-based Monte Carlo (MC) algorithm. The dataset consisted of hinge-bending proteins, DNA-binding proteins, enzymes showing loop closure, and shear-type proteins. The initial RMSDs (root mean square distance) between initial and target crystal structures span a broad range of 1.0-12.0 Å. When a relatively low cutoff radius of 10 Å was used in ANM, the target structures were approached successfully, for almost the whole dataset, both in forward and reverse runs. Forward transitions followed plausible pathways, namely minimum free energy path observed for adenylate kinase and many close crystal structures detected along paths of different proteins. For the hinge-bending type proteins, the open-to-closed transitions were directed by one predominant mode (first or second). However, significant changes in the dominant mode character were observed along the transition for most proteins. When a criterion of decreasing radius of gyration (Rg) was imposed for mode selection, closed structures could be predicted for 10 out of 12 hinge-bending proteins. Here deforming along a single mode lead to most successful predictions. For the rest of the proteins in the dataset, it was not possible to make predictions due to the higher number of modes involved in forward transitions and/or insignificant change in Rg. For similar reasons, prediction of an open structure from a closed one, i.e. reverse transitions, was only possible for a limited number of hinge-bending proteins. Application of ANM-MC technique to different ICL3 loop models of β2-adrenergic receptor indicated successful loop closure with implications on the allosteric mechanism of the receptor. Finally, the extreme inter-domain flexibility and consistent collective dynamics of nuclear factor-kappa B were elucidated via molecular dynamics simulations, ANM and ANM-MC simulations.