Louise J. Persson: Atomistic Modeling of Multimeric Proteins in the Gas Phase: From Simulation Strategies to Molecular Insights

Abstract

Gas-phase experiments have become increasingly important for studying proteins and protein complexes. Established techniques such as native mass spectrometry and ion mobility spectrometry are now complemented by emerging imaging methods, including electrospray-ion beam deposition coupled with cryo-electron microscopy and femtosecond X-ray diffraction. Molecular dynamics simulations provide a powerful counterpart to these methods, offering detailed insight into protein behavior under vacuum conditions with unparalleled spatial and temporal resolution.

This work optimizes strategies for simulating protein complexes in the gas phase. We validate the fast multipole method for long-range force computations and demonstrate significant performance enhancements for large systems, enabling all-atom simulations of multimeric proteins that were previously computationally prohibitive. In addition, we introduce a tool for the automated optimization of parameters for the fast multipole method, which balances computational performance with accuracy. We further examine the impact of thermodynamic ensemble on gas-phase proteins, highlighting the importance of matching simulation settings to the relevant physical conditions and the benefits of thermostatted dynamics for achieving extensive sampling.

We leverage these methodological enhancements in simulations of soluble and membrane protein complexes in the gas phase over extended timescales, and compare structural properties against experimental data from the aforementioned techniques. Particular focus is placed on temperature-dependent conformational changes and their effect on collision cross sections, as well as the interactions between membrane proteins and detergent in the gas phase.

Overall, this work strengthens the role of molecular dynamics simulations in gas-phase protein studies by enabling simulations of large multimeric proteins, and bridging a gap in capabilities between computation and experiment. The applied models deepen understanding of protein behavior in vacuum, inform the continued development and application of these experiments, and support their use in probing native properties of protein complexes that remain inaccessible to condensed-phase methods.