Molecular Breakdown of Double Electron-Electron Resonance Data with Atomistic Simulations

Abstract

Double Electron-Electron Resonance (DEER) spectroscopy has become a landmark technique to investigate the structure and dynamics of membrane proteins, in different functional states and physiological conditions. From DEER experiments it is possible to obtain any given number of probability distributions for distances between spin labels attached to biomolecules. In contrast to diffraction methods or NMR spectroscopy, DEER is neither limited by the need of crystallization nor by the protein size. This notwithstanding, it is often not straightforward to interpret DEER data, as it reflects a plethora of protein conformations and rotameric states of the spin labels. Several strategies to disentangle this variability and to derive a clear structural interpretation have been put forward recently. Some of them entail the calculation of distance distributions from approximate protein models and rotamer libraries for the spin-labels. Other more rigorous approaches, such as Ensemble-Biased MetaDynamics (EBMetaD), are based on atomistic simulation methods, specifically designed to reproduce the DEER distributions exactly and with a minimal bias. Both kinds of approaches, however, rely on the probability distributions that are inferred from the actual measured data, and do not take into account the experimental noise. Here, we present a powerful and simple approach to minimally bias an atomistic simulation to sample a conformational ensemble that directly reproduces the DEER time-signal within the experimental uncertainty. The method is based on the maximum-entropy principle and extends the EBMetaD approach. We assess the performance of the method using spin-labeled T4 lysozyme in explicit water. The results demonstrate the accuracy and efficiency of the method. In summary, we propose a novel, rigorous technique to directly interpret experimental data from DEER measurements in terms of physically realistic conformational ensembles.