Developing and applying new theoretical and computational methods to study complex condensed phase systems

Gregory A. Voth

Gregory A. Voth
Haig P. Papazian Distinguished Service Professor
Department of Chemistry
Google Scholar Page

The research in the Voth Group involves theoretical and computer simulation studies of biomolecular, condensed phase, quantum mechanical, and materials systems. One of our goals is to develop new theory to describe such problems across multiple, connected length and time scales. Another related goal is to develop and apply new computational methods, tied to our multiscale theory, that can explain and predict complex phenomena occurring in these systems. Our methods are developed, for example, to probe protein-protein self-assembly, membrane-protein interactions, biomolecular and liquid state charge transport, complex liquids, self-assembly, and energy conversion materials. Our research is also often carried out in close collaboration with leading experimentalists from around the world. 

Material for Download

A Modern Perspective on the Hydrated Excess Proton (aka "Hydronium") 

Multi-scale Coarse-graining (MS-CG) Force Matching (FM) code is now publicly available for download 

Research News

Systematic Coarse-Grained Lipid Force Fields with Semiexplicit Solvation via Virtual Sites

Lipids, which are amphipathic biomolecules that assemble into cellular membranes, are central to many biophysical processes. To understand the fundamental molecular mechanisms that dictate the macroscopic behavior of lipid assemblies, we introduce a novel procedure for the systematic development of low-resolution coarse-grained (CG) lipid models that will enable simulations of biologically-relevant spatiotemporal scales with molecular fidelity. The central idea is to represent the structural features of the solvent-lipid interface through the introduction of virtual CG sites. We then leverage two systematic coarse-graining approaches, multiscale coarse-graining (MS-CG) and relative entropy minimization (REM), in a hybrid fashion to derive the effective interactions of our virtual-site CG (VCG) models from reference all-atom simulations. We demonstrate that VCG models recapitulate the rich biophysics of lipids, which enable self-assembly, morphological diversity, and multiple phases. Our findings further suggest that the VCG framework is a powerful approach for an investigation into macromolecular biophysics.

Gating mechanisms during actin filament elongation by formins

The complex interactions between actin and actin binding proteins regulate actin filament formation within the cytoskeletal network. A central part of this regulatory system is composed of formins, which are highly conserved multidomain proteins that nucleate and elongate unbranched actin filaments. In this paper, we utilize multiscale simulation techniques to understand the underlying factors that cause the difference in the formin-mediated polymerization rate of actin by studying the interactions between actin and the FH2 domains of the formins Cdc12, Bni1, and mDia1. These formins slow elongation of actin filament by 5 to 95% depending on the particular species. We found two mechanisms that can influence the rate of actin polymerization mediated by formins from our all-atom molecular dynamics and bottom-up coarse-grained simulations. According to these mechanisms, formins can sterically block the addition of new actin subunits, or flatten the helical twist of the terminal actin subunits. Making the filament ends unfavorable for the addition of a new actin subunit. Our findings help to explain why Cdc12 FH2 domains slow elongation of the actin filament much more than Bni1 and mDia1.

Multiscale simulation of actin filaments and actin-associated proteins

In this review, we gave an overview of the work mostly done in the Voth group that was highly motivated by the important contributions of Tom Pollard to understand the essential structural and chemical aspects of the actin cytoskeleton. We have discussed the use of various simulation techniques to understand the key structural and chemical features of actin, which are associated with biological processes occurring at different scales such as actin filament polymerization, the hydrolysis of bound nucleotide ATP and phosphate release. These simulation techniques include detailed atomistic simulations and advanced free energy sampling techniques to understand structural transformation of actin subunits, advanced coarse-graining methods such as ED-CG, hENM, and the UCG to model actin at various resolutions of coarse-graining, and hybrid quantum mechanics/molecular mechanics approach to study ATP hydrolysis. Furthermore, we summarized the modeling of actin binding proteins such as the Arp2/3 complex and formin to have a molecular level understanding of the fundamental mechanisms affecting the architecture of actin network at different length and time scales by using these methods. Many of these studies were a direct result of a collaboration with Tom Pollard and his group or inspired by informal interactions with him.

Modulating the Chemical Transport Properties of a Transmembrane Antiporter via Alternative Anion Flux

CIC is a family of proteins crucial to the homeostasis of ion concentrations and pH gradients in bacteria as well as in the human body. ClC-ec1, as a prototypical ClC protein, exchanges Cl– for H+ across the membrane. Other polyatomic anions, such as NO3– and SCN–, are also permeable through the protein, but with partially or entirely uncoupled H+ flux. In this work, we characterized the potential of mean forces of the proton transport (PT) process in ClC-ec1 and the essential mechanism of H+ transport affected by the alternative anion flux. We revealed that the energy barrier is related to the water connectivity along the PT pathway in the presence of the excess proton, which is significantly affected by the nature of the bound anion. This piece of work shows how chemical engineering can regulate PT in proteins by changing the hydration behavior.

Ultra-Coarse-Grained Liquid State Models with Implicit Hydrogen Bonding

Reproducibility in the coarse-grained model is essential, especially when vital information below the CG resolution is lost during the CG process. In the case of a two-site methanol molecule, this problem becomes apparent since the hydrogen-bonding topology (either donor or acceptor) among -OH beads is lost, resulting in inaccurate local structures. To surmount this challenge, we apply the Ultra-Coarse-Grained (UCG) theory to construct a high-fidelity CG model with implicit hydrogen-bonding. Since hydrogen bonding is a local phenomenon, the internal states of the hydrogen-bonding participating CG sites are determined by the local density of the CG sites. This strategy is then applied to two different hydrogen bonding motifs: chain-like and ring-like. For both motifs including amino acid building blocks, the UCG models show better structural correlations (pair, triplet, and local number density) compared to the conventional MS-CG models. These results strongly suggest that the UCG theory can significantly improve the reproducibility of the CG model in the complex condensed system. 

Mesopic coarse-grained representation of fluids rigorously derived from atomistic models

In this work, we built up the first rigorous bridge between atomistic and supramolecular mesoscopic models of fluids. A delicate dynamic coarse-graining mapping scheme, whose idea originates from the centroidal Voronoi tesellation algorithm in computational geometry, was designed to account for the microscopic momentum transport that governs the fluid motion at mesoscale. Besides, a systematic parameterization method based on Mori-Zwanzig formalism was developed and faithfully reproduces the statistical and dynamical characteristics of the coarse-grained trajectory. The new dynamical coarse-grained mapping scheme and the parameterization protocol open up an avenue for direct bottom-up construction of mesoscopic models of complex fluids in a Lagrangian description. 

Molecular transport through membranes: Accurate permeability coefficients from multidimensional potentials of mean force and local diffusion constants

Estimating the permeability coefficient of small molecules through lipid bilayer membranes plays an essential role in the development of effective drug candidates. However, the absolute permeability coefficients obtained from pre-existing computer simulation methods are usually off by orders of magnitude, mostly due to the poor convergence of permeation free energy profiles and over-simplified diffusion models. To overcome these obstacles, we describe the permeation process using multiple reaction coordinates and estimate the permeability along the minimum free energy path of the multidimensional potential of mean force. A combination of cutting-edge metadynamics enhanced sampling techniques, and improved representation of the permeation process leads to a considerably more accurate estimation of permeability coefficients compared to pre-existing methods.

Entropic forces drive clustering and spatial localization of influenza A M2 during viral budding

For influenza virus to release from the infected host cell, controlled viral budding must finalize with membrane scission of the viral envelope. Curiously, influenza carries its own protein, M2, which can sever the membrane of the constricted budding neck. Here we elucidate the physical mechanism of clustering and spatial localization of the M2 scission proteins through a combined computational and experimental approach.

Insights into the Cooperative Nature of ATP Hydrolysis in Actin Filaments

Hydrolysis of the nucleotide bound to each subunit of actin serves as an important clock that governs the remodeling of the cytoskeletal network. Whether the hydrolysis and the subsequent phosphate release act independently or are somehow coupled across different subunits in the filament network has been debated for over three decades. Here, we developed a systematic multi-scale modeling framework by combining atomistic simulations, the Ultra-Coarse-Graining approach, and Markov State Modeling, that addresses this issue.

Ultra-Coarse-Grained Models Allow for an Accurate and Transferable Treatment of Interfacial Systems

In this work, we have demonstrated the application of the recently developed Ultra-CG (UCG) theory to heterogeneous systems: liquid/vapor and liquid/liquid interfaces. Due to the inhomogeneous nature of an interfacial system, the conventional MS-CG framework fails to capture the structure and directionality of molecules, resulting in the breakdown of the interfacial density profile. In order to resolve the limitation of the MS-CG method, we have designed the UCG framework to systematically distinguish different local environments in interfacial systems and faithfully recapitulates structural correlations in the interfacial systems. More fascinatingly, the CG interactions obtained by the UCG methodology are transferable to corresponding bulk states. This transferability was observed in both liquid/vapor and liquid/liquid systems: the UCG methodology can impart transferable CG models while retaining high accuracy.

Past Research Highlights 

Center for Multiscale Theory and SimulationThe James Franck InstituteInstitute for Biophysical DynamicsComputation Institute