Elastic network model based approaches for conformer generation and docking applications

Abstract

The dynamic nature of proteins poses challenging problems in computational biology, especially in terms of conformational sampling and transitions. In this thesis, an Elastic Network Model (ENM)-based computational method, namely ClustENM, was developed for sampling large conformational changes of biomolecules with various sizes and oligomerization states. ClustENM is an iterative method that combines ENM with energy minimization and clustering steps. It is an unbiased technique, which necessitates only an initial structure of the biomolecule as input but no information on target. To test the performance of ClustENM in conformational sampling, it was applied to six systems, namely adenylate kinase (AK), calmodulin, p38 MAP kinase, HIV-1 reverse transcriptase, triosephosphate isomerase (TIM), and supramolecule 70S ribosome. The generated atomistic conformers were found to be in agreement with experimental data (971 structures) and molecular dynamics (MD) simulations. ClustENM was used to model the trigger factor-50S subunit of ribosome complex, leading to structures consistent with the data from cryo-EM. Additionally, ligand e ects on TIM conformational dynamics were investigated based on MD simulations of its apo form and complexes with an inhibitor or its substrate. Generated conformers from ClustENM were further used in docking applications for AK, LAO-binding protein, dipeptide binding protein and biotin carboxylase. Close-to-native ligand binding poses were obtained especially in the rst three cases. Thus, ClustENM emerges as a computationally e cient method applicable to extremely large systems or transitions. Its utility relies on the generation of a manageable number of atomistic conformers that are entropically accessible to a folded starting structure, which can also assist ligand docking applications.