Molecular Dynamics Simulations as a Tool for Probing Molecular Interactions in Hydrogen-Deuterium Exchange and Liquid Chromatography Workflows
Abstract
Proteins are biological macromolecules that are essential for countless physiological processes. Many health conditions are rooted in protein misfunction, resulting in the need to develop analytical techniques that can thoroughly study protein structures and functions. Liquid chromatography (LC)-mass spectrometry (MS) plays a key role in this context. By leveraging the separation power of LC, techniques such as hydrogen-deuterium exchange (HDX) MS can be used to decipher protein dynamics. However, the interpretation of experimental data obtained from these techniques is often hindered by an incomplete molecular understanding of the underlying physical processes. This dissertation explores how in silico tools, specifically molecular dynamics (MD) simulations, can provide novel insights into how these analytical techniques inform on the behavior of proteins and peptides.
The work in Chapter 2 examines HDX patterns of the model protein cytochrome c in its two canonical heme oxidation states, Fe(II) and Fe(III). Previous HDX work already pointed to increased stability in the Fe(II) state, however, the molecular foundation for this stability enhancement remained elusive. We performed HDX experiments in conjunction with MD simulations of cytochrome c with different heme coordination environments, and we elucidated how these changes account for the observed HDX patterns. Additionally, our MD work uncovered large-scale protein motions that are not detectable by HDX, thereby providing evidence for the existence of “HDX silent” protein motions.
Chapter 3 showcases the first high-fidelity MD modeling of a reversed-phase LC (RPLC) stationary phase and mobile phase for studying peptide interactions that are responsible for RPLC retention. Two tryptic peptides with different physicochemical properties were modeled in water and in a water/acetonitrile (ACN) mixture. It was found that peptide/stationary phase interactions primarily involved side chains with strongly hydrophobic character. While the data were insufficient for making retention predictions, the MD setup developed in this Chapter established the foundation for conducting detailed computational analyses of peptide interactions with nonpolar stationary phases under various solvent conditions and in the presence of formate as ion pairing agent.
Chapter 4 expands on the foundation established in Chapter 3 by using additional MD tools for making peptide RPLC retention predictions. In the past, peptide retention predictions relied on algorithms that involved empirical rules or large sets of RPLC training data; however, such traditional approaches are unable to uncover the physicochemical principles that drive retention. Using four tryptic peptides and five water/ACN solvent conditions, umbrella sampling MD was used to generate retention predictions from first principles, by determining the free energy of peptide/stationary phase binding equilibria under various mobile phase conditions. This study marks the first time that MD simulations alone have been effective at predicting peptide retention, highlighting how this technique can be used as a standalone tool or to enhance existing prediction algorithms. Overall, this dissertation highlights that MD can complement experimental techniques by providing a molecular level understanding of the underlying physicochemical principles.