Conformational Landscapes of Ligand-Gated Ion Channel GLIC from Markov State Modeling

Abstract

Ligand-gated ion channels (LGICs) are responsible for propagating electrochemical signals in neurons and other excitable cells, by rapidly cycling between functional states upon appropriate chemical stimuli. However, few LGICs have been characterized in both open and closed conformations, and the timescales of the gating transitions, initiated by ligand-binding, are typically too slow for classical molecular dynamics simulations. Consequently, the precise landscape of LGIC transitions is poorly understood. We have utilized enhanced sampling to run unbiased molecular dynamics simulations in a massively parallel fashion, and construct Markov state models of a structurally rich, pH-gated bacterial LGIC, GLIC from Gloeobacter violaceus. We then tested the ability of these models to recapitulate functional effects of protonation and mutagenesis on channel opening. Consistent with functional recordings, our models reveal deepening of the open state free energy well upon protonation in the extracellular domain or mutation (I233T) of the transmembrane gate, with relatively low maximal open probability. Analysis of states along the main reaction coordinate confirms multiple previously suggested mechanisms, but we note that the blooming and twisting motions of the extracellular domain mainly depend on pH and not the conformational state of the channel. Additionally, analysis of the conformational symmetries reveals a highly asymmetric intermediate and symmetric open state, while the closed states display mixed asymmetric extracellular and symmetric transmembrane domains. This work demonstrates the construction of plausible Markov state models from limited sampling of LGIC variants, capable of predicting allosteric effects from chemical perturbation, and identifying geometric variables in functional transitions. Such statistical tools constitute a promising approach to elucidate LGIC gating mechanisms, with practical implications for modeling neuronal function and the development of state-selective drugs.