Session B43: Focus Session: Multiscale modeling--Coarse-graining in Space and Time II

Multiscale simulations of ion channel opening and closing provide insights into the molecular mechanisms of gating

We develop and implement a multiscale molecular simulation approach to study the opening and closing of a ligand-gated ion channel at atomic resolution. Ligand-gated channels are essential in biological signaling pathways that range from chemical sensing in bacteria to the firing of neurons in humans. On the basis of recently determined crystal structures and with the help of multiscale molecular simulations we study the conformational changes associated with GLIC ion channel gating transition. Starting from a coarse-grained transition pathway constructed on the basis of a multistate elastic network model, we perform string-method molecular dynamics simulations to refine the pathway at full atomic resolution. We find that the channel closes in an iris-like fashion as a result of a two-stage tilting of the pore lining helices. Water plays a central role in the gating transition. We find that the hydrophobic gate of the pore undergoes highly cooperative transitions between a densely filled and an empty state. The subtle tilting of the helices shifts the balance to the dry state, in which a 1.5 nm long hydrophobic stretch of the pore completely empties. By calculating the ionic conductance and the underlying free energy surface, we quantitatively demonstrate that this drying of the hydrophobic constriction, not sterics, is the major determinant of ion conductivity in the GLIC pentameric ion channel.   

Multiscale modeling of macromolecular dynamics

The understanding of emerging collective behaviors in biomolecular complexes represents a major challenge in modern biophysics. As a first step toward the study of such processes we have applied multi-resolution nonlinear dimensionality reduction and diffusion analysis to obtain reliable low-dimensional representations and models for the dynamics of apparently high-dimensional complex systems such as proteins in a biological environment. The results clearly show that the proposed methods can efficiently find low dimensional representations of complex processes such as protein folding, and suggest strategies to simplify significantly the study of such processes.   

Transferability of Coarse Grained Models: a Challenge for Simulation of Phase Transitions or Phase Separation Processes

Upon developing a coarse grained (CG) model, representability and transferability limitations are a problem that is inherent to the process of reducing the number of degrees of freedom. In this context, representability refers to the question which structural or thermodynamic properties of a higher resolution reference are reproduced by the CG model, and transferability refers to the question to which extent a CG model is applicable at a state-point that differs from the one where it was parametrized. This is naturally a highly relevant problem in simulations that involve phase transitions or structure formation processes driven by phase separation, for example in liquid crystalline systems or in biomolecular aggregation. I will show with a few examples how one can achieve and rationalize state-point transferability for CG models that have been parameterized in a bottom-up procedure from atomistic reference simulations, for example by choosing an appropriate reference state point.   

Mesoscopic Dynamics of Biopolymers and Protein Molecular Machines

The dynamics of biopolymers in solution and in crowded molecular environments, which mimic some features of the interior of a biochemical cell, will be discussed. In particular, the dynamics of protein machines that utilize chemical energy to effect cyclic conformational changes to carry out their catalytic functions will be described. The investigation of the dynamics of such complex systems requires knowledge of the time evolution on physically relevant long distance and time scales. This often necessitates a coarse grained or mesoscopic treatment of the dynamics. A hybrid particle-based mesoscopic dynamical method, which combines molecular dynamics for a coarse-grain model of the proteins with multiparticle collision dynamics for the solvent, will be described and utilized to study the dynamics of such systems. See, C. Echeverria, Y. Togashi, A. S. Mikhailov, and R. Kapral, Phys. Chem. Chem. Phys 13, 10527 (2011); C. Echeverria and R. Kapral, Phys. Chem. Chem. Phys., 14, 6755 (2012); J. M. Schofield, P. Inder and R. Kapral, J. Chem. Phys. 136, 205101 (2012).   

Multiscale Modeling of Deformation of Glassy Polymers

We examine the ability of chemically informed coarse-grained (CG) models to quantitatively describe correct mechanical properties of glassy polymer systems. The force-matching and the structure-matching procedures were used to obtain CG potentials at different levels of resolution. Equilibrium molecular dynamics simulations of amorphous polymers modeled at the all-atom level provided the necessary reference data. This work explores what characteristics are necessary for quantitative agreement of stress-strain curves between scales. For large coarse-graining (17 atoms per CG site of polystyrene) the force-matching procedure produces a potential that does not contain enough attraction to predict the correct elastic properties. Systematic methods were employed to match mechanical properties and their effects on polymer structure were examined. Higher resolution coarse-graining (5-11 atoms per CG site) is better able to reproduce atomistic mechanical data.   

Force fields for describing the solution-phase synthesis of shape-selective metal nanoparticles

Polyvinylpyrrolidone (PVP) and polyethylene oxide (PEO) are structure-directing agents that exhibit different performance in the polyol synthesis of Ag nanostructures. The success of these structure-directing agents in selective nanostructure synthesis is often attributed to their selective binding to Ag(100) facets. We use first-principles, density-functional theory (DFT) calculations in a vacuum environment to show that PVP has a stronger preference to bind to Ag(100) than to Ag(111), whereas PEO exhibits much weaker selectivity. To understand the role of solvent in the surface-sensitive binding, we develop classical force fields to describe the interactions of the structure-directing (PVP and PEO) and solvent (ethylene glycol) molecules with various Ag substrates. We parameterize the force fields through force-and-energy matching to DFT results using simulated annealing. We validate the force fields by comparisons to DFT and experimental binding energies. Our force fields reproduce the surface-sensitive binding predicted by DFT calculations. Molecular dynamics simulations based on these force fields can be used to reveal the role of solvent, polymer chain length, and polymer concentration in the selective synthesis of Ag nanostructures.   

Enhancing and reversing the electric field at liquid/liquid interfaces

The ion distribution at the interface between two immiscible electrolyte solutions determines the macroscopic properties of these liquid interfaces. The classical Poisson-Boltzmann theory has been widely used to describe it, even though it neglects the polarization and ion correlations typical of these ionic solutions. Here, we provide an enhanced description of a liquid/liquid interface in the presence of an electric field from first principles--that is, without needing any fitting parameter--including ion correlations, image charges and realistic ion-sizes in Monte Carlo simulations. Our data agree well with experimental excess surface tension measurements for a wide range of electrolyte concentrations, contrasting with the results of the classical Poisson-Boltzmann theory. More importantly, we observe that, in the vicinity of the point of zero charge, the electric field can increase significantly in strength near the liquid interface, or it can even reverse locally, at high salt concentration.