Extended Sampling of Macromolecular Conformations from Uniformly Distributed Points on Multidimensional Normal Mode Hyperspheres.

Proteins are dynamic entities that adopt diverse conformations, which play a pivotal role in their function. Understanding these conformations is essential, and protein collective motions, particularly those captured by normal mode (NM) and their linear combinations, provide a robust means for conformational sampling. This work introduces a novel approach to obtaining a uniformly oriented set of a given number of lowest frequency NM combined vectors and generating harmonically equidistant restrained structures along them. They are all thus uniformly located on concentric hyperspheres, systematically covering the defined NM space fully. The generated structures are further relaxed with standard molecular dynamics (MD) simulations to explore the conformational space. The efficiency of the approach we termed "distributed points Molecular Dynamics using Normal Modes" (dpMDNM) was assessed by applying it to hen egg-white lysozyme and human cytochrome P450 3A4 (CYP3A4). To this purpose, we compared our new approach with other methods and analyzed the sampling of existing experimental structures. Our results demonstrate the efficacy of dpMDNM in extensive conformational sampling, particularly as more NMs are considered. Ensembles generated by dpMDNM exhibited a broad coverage of the experimental structures, providing valuable insights into the functional aspects of lysozyme and CYP3A4. Furthermore, dpMDNM also covered lysozyme structures with relatively elevated energies corresponding to transient states not easily obtained by standard MD simulations, in conformity with nuclear magnetic resonance structural indications. This method offers an efficient and rational framework for comprehensive protein conformational sampling, contributing significantly to our understanding of protein dynamics and function.