Markovian milestoning with Voronoi tessellations

Computation of transit times using the milestoning method with applications to polymer translocation

Milestoning is an efficient approximation for computing long-time kinetics and thermodynamics of large molecular systems, which are inaccessible to brute-force molecular dynamics simulations. A common use of milestoning is to compute the mean first passage time (MFPT) for a conformational transition of interest. However, the MFPT is not always the experimentally observed timescale. In particular, the duration of the transition path, or the mean transit time, can be measured in single-molecule experiments, such as studies of polymers translocating through pores and fluorescence resonance energy transfer studies of protein folding. Here we show how to use milestoning to compute transit times and illustrate our approach by applying it to the translocation of a polymer through a narrow pore.

A molecular simulation protocol to avoid sampling redundancy and discover new states

BACKGROUND

For biomacromolecules or their assemblies, experimental knowledge is often restricted to specific states. Ambiguity pervades simulations of these complex systems because there is no prior knowledge of relevant phase space domains, and sampling recurrence is difficult to achieve. In molecular dynamics methods, ruggedness of the free energy surface exacerbates this problem by slowing down the unbiased exploration of phase space. Sampling is inefficient if dwell times in metastable states are large.

METHODS

We suggest a heuristic algorithm to terminate and reseed trajectories run in multiple copies in parallel. It uses a recent method to order snapshots, which provides notions of "interesting" and "unique" for individual simulations. We define criteria to guide the reseeding of runs from more "interesting" points if they sample overlapping regions of phase space.

RESULTS

Using a pedagogical example and an α-helical peptide, the approach is demonstrated to amplify the rate of exploration of phase space and to discover metastable states not found by conventional sampling schemes. Evidence is provided that accurate kinetics and pathways can be extracted from the simulations.

CONCLUSIONS

The method, termed PIGS for Progress Index Guided Sampling, proceeds in unsupervised fashion, is scalable, and benefits synergistically from larger numbers of replicas. Results confirm that the underlying ideas are appropriate and sufficient to enhance sampling.

GENERAL SIGNIFICANCE

In molecular simulations, errors caused by not exploring relevant domains in phase space are always unquantifiable and can be arbitrarily large. Our protocol adds to the toolkit available to researchers in reducing these types of errors. This article is part of a Special Issue entitled "Recent developments of molecular dynamics".

Recent applications of boxed molecular dynamics: a simple multiscale technique for atomistic simulations

In this paper, we briefly review the boxed molecular dynamics (BXD) method which allows analysis of thermodynamics and kinetics in complicated molecular systems. BXD is a multiscale technique, in which thermodynamics and long-time dynamics are recovered from a set of short-time simulations. In this paper, we review previous applications of BXD to peptide cyclization, solution phase organic reaction dynamics and desorption of ions from self-assembled monolayers (SAMs). We also report preliminary results of simulations of diamond etching mechanisms and protein unfolding in atomic force microscopy experiments. The latter demonstrate a correlation between the protein's structural motifs and its potential of mean force. Simulations of these processes by standard molecular dynamics (MD) is typically not possible, because the experimental time scales are very long. However, BXD yields well-converged and physically meaningful results. Compared with other methods of accelerated MD, our BXD approach is very simple; it is easy to implement, and it provides an integrated approach for simultaneously obtaining both thermodynamics and kinetics. It also provides a strategy for obtaining statistically meaningful dynamical results in regions of configuration space that standard MD approaches would visit only very rarely.

Oxygen Pathways and Allostery in Monomeric Sarcosine Oxidase via Single-Sweep Free-Energy Reconstruction

Monomeric sarcosine oxidase (MSOX) is a flavoprotein D-amino acid oxidase with reported sarcosine and oxygen activation sites on the re and si faces of the flavin ring, respectively. O2 transport routes to the catalytic interior are not well understood and are difficult to ascertain solely from MSOX crystal structures. A composite free-energy method known as single-sweep is used to map and thermodynamically characterize oxygen sites and routes leading to the catalytically active Lys265 from the protein surface. The result is a network of pathways and free energies within MSOX illustrating that oxygen can access two free-energy minima on the re face of the reduced flavin from four separate solvent portals. No such minimum is observed on the si face. The pathways are geometrically similar for three major states of the enzyme: (1) apo with a closed flavin cleft, (2) apo with an open flavin cleft, and (3) inhibitor-bound with a closed flavin cleft. Interestingly, free energies along these transport pathways display significantly deeper minima when the substrate-mimicking inhibitor 2-furoic acid is bound at the sarcosine site, even at locations far from this site. This suggests a substrate-dependent allosteric modulation of the kinetics of O2 transport from the solvent to the active site.

Temperature-accelerated molecular dynamics gives insights into globular conformations sampled in the free state of the AC catalytic domain

The catalytic domain of the adenyl cyclase (AC) toxin from Bordetella pertussis is activated by interaction with calmodulin (CaM), resulting in cAMP overproduction in the infected cell. In the X-ray crystallographic structure of the complex between AC and the C terminal lobe of CaM, the toxin displays a markedly elongated shape. As for the structure of the isolated protein, experimental results support the hypothesis that more globular conformations are sampled, but information at atomic resolution is still lacking. Here, we use temperature-accelerated molecular dynamics (TAMD) simulations to generate putative all-atom models of globular conformations sampled by CaM-free AC. As collective variables, we use centers of mass coordinates of groups of residues selected from the analysis of standard molecular dynamics (MD) simulations. Results show that TAMD allows extended conformational sampling and generates AC conformations that are more globular than in the complexed state. These structures are then refined via energy minimization and further unrestrained MD simulations to optimize inter-domain packing interactions, thus resulting in the identification of a set of hydrogen bonds present in the globular conformations.

Investigations of α-helix↔β-sheet transition pathways in a miniprotein using the finite-temperature string method

A parallel implementation of the finite-temperature string method is described, which takes into account the invariance of coordinates with respect to rigid-body motions. The method is applied to the complex α-helix↔β-sheet transition in a β-hairpin miniprotein in implicit solvent, which exhibits much of the complexity of conformational changes in proteins. Two transition paths are considered, one derived from a linear interpolant between the endpoint structures and the other derived from a targeted dynamics simulation. Two methods for computing the conformational free energy (FE) along the string are compared, a restrained method, and a tessellation method introduced by E. Vanden-Eijnden and M. Venturoli [J. Chem. Phys. 130, 194103 (2009)]. It is found that obtaining meaningful free energy profiles using the present atom-based coordinates requires restricting sampling to a vicinity of the converged path, where the hyperplanar approximation to the isocommittor surface is sufficiently accurate. This sampling restriction can be easily achieved using restraints or constraints. The endpoint FE differences computed from the FE profiles are validated by comparison with previous calculations using a path-independent confinement method. The FE profiles are decomposed into the enthalpic and entropic contributions, and it is shown that the entropy difference contribution can be as large as 10 kcal/mol for intermediate regions along the path, compared to 15-20 kcal/mol for the enthalpy contribution. This result demonstrates that enthalpic barriers for transitions are offset by entropic contributions arising from the existence of different paths across a barrier. The possibility of using systematically coarse-grained representations of amino acids, in the spirit of multiple interaction site residue models, is proposed as a means to avoid ad hoc sampling restrictions to narrow transition tubes.

WExplore: hierarchical exploration of high-dimensional spaces using the weighted ensemble algorithm

As most relevant motions in biomolecular systems are inaccessible to conventional molecular dynamics simulations, algorithms that enhance sampling of rare events are indispensable. Increasing interest in intrinsically disordered systems and the desire to target ensembles of protein conformations (rather than single structures) in drug development motivate the need for enhanced sampling algorithms that are not limited to "two-basin" problems, and can efficiently determine structural ensembles. For systems that are not well-studied, this must often be done with little or no information about the dynamics of interest. Here we present a novel strategy to determine structural ensembles that uses dynamically defined sampling regions that are organized in a hierarchical framework. It is based on the weighted ensemble algorithm, where an ensemble of copies of the system ("replicas") is directed to new regions of configuration space through merging and cloning operations. The sampling hierarchy allows for a large number of regions to be defined, while using only a small number of replicas that can be balanced over multiple length scales. We demonstrate this algorithm on two model systems that are analytically solvable and examine the 10-residue peptide chignolin in explicit solvent. The latter system is analyzed using a configuration space network, and novel hydrogen bonds are found that facilitate folding.

Nano-pipette directed transport of nanotube transmembrane channels and hybrid vesicles

Using computational modeling, we simulate the interactions between a nanopipette and transmembrane, end-functionalized nanotubes that are localized within flat bilayers or nanoscopic vesicles. The functional groups (hairs) provide a "handle" for the moving pipette to controllably pick up and move the nanotubes to specific locations in the flat membrane, or the hybrid vesicle to specified regions on a surface. The ability to localize these hybrid vesicles on surfaces paves the way for creating nanoreactor arrays in fluidic devices.

Insight into the molecular mechanism of water evaporation via the finite temperature string method

The process of water's evaporation at its liquid/air interface has proven challenging to study experimentally and, because it constitutes a rare event on molecular time scales, presents a challenge for computer simulations as well. In this work, we simulated water's evaporation using the classical extended simple point charge model water model, and identified a minimum free energy path for this process in terms of 10 descriptive order parameters. The measured free energy change was 7.4 kcal/mol at 298 K, in reasonable agreement with the experimental value of 6.3 kcal/mol, and the mean first-passage time was 1375 ns for a single molecule, corresponding to an evaporation coefficient of 0.25. In the observed minimum free energy process, the water molecule diffuses to the surface, and tends to rotate so that its dipole and one O-H bond are oriented outward as it crosses the Gibbs dividing surface. As the water molecule moves further outward through the interfacial region, its local density is higher than the time-averaged density, indicating a local solvation shell that protrudes from the interface. The water molecule loses donor and acceptor hydrogen bonds, and then, with its dipole nearly normal to the interface, stops donating its remaining hydrogen bond. At that point, when the final, accepted hydrogen bond is broken, the water molecule is free. We also analyzed which order parameters are most important in the process and in reactive trajectories, and found that the relative orientation of water molecules near the evaporating molecule, and the number of accepted hydrogen bonds, were important variables in reactive trajectories and in kinetic descriptions of the process.

Simulation study of free-energy barriers in the wetting transition of an oily fluid on a rough surface with reentrant geometry

When in contact with a rough solid surface, fluids with low surface tension, such as oils and alkanes, have their lowest free energy in the fully wetted state. For applications where nonwetting by these phillic fluids is desired, some barrier must be introduced to maintain the nonwetted composite state. One way to create this free-energy barrier is to fabricate roughness with reentrant geometry, but the question remains as to whether the free-energy barrier is sufficiently high to prevent wetting. Our goal is to quantify the free-energy landscape for the wetting transition of an oily fluid on a surface of nails and identify significant surface features and conditions that maximize the wetting free-energy barrier (ΔGfwd*). This is a departure from most work on wetting, which focuses on the equilibrium composite and wetted states. We use boxed molecular dynamics (BXD) (Glowacki, D. R.; Paci, E.; Shalashilin, D. V. J. Phys. Chem. B2009, 113, 16603-16611) with a modified control scheme to evaluate both the thermodynamics and kinetics of the transition over a range of surface affinities (chemistry). We find that the reentrant geometry of the nails does create a free-energy barrier to transition for phillic chemistry whereas a corresponding system on straight posts wets spontaneously and, that doubling the nail height more than doubles ΔGfwd*. For neutral to phillic chemistry, the dewetting free-energy barrier is at least an order of magnitude higher than that for wetting, indicating an essentially irreversible wetting transition. Transition rates from BXD simulations and the associated trends agree well with those in our previous study that used forward flux sampling to compute transition rates for similar systems.

Minimum free energy path of ligand-induced transition in adenylate kinase

Large-scale conformational changes in proteins involve barrier-crossing transitions on the complex free energy surfaces of high-dimensional space. Such rare events cannot be efficiently captured by conventional molecular dynamics simulations. Here we show that, by combining the on-the-fly string method and the multi-state Bennett acceptance ratio (MBAR) method, the free energy profile of a conformational transition pathway in Escherichia coli adenylate kinase can be characterized in a high-dimensional space. The minimum free energy paths of the conformational transitions in adenylate kinase were explored by the on-the-fly string method in 20-dimensional space spanned by the 20 largest-amplitude principal modes, and the free energy and various kinds of average physical quantities along the pathways were successfully evaluated by the MBAR method. The influence of ligand binding on the pathways was characterized in terms of rigid-body motions of the lid-shaped ATP-binding domain (LID) and the AMP-binding (AMPbd) domains. It was found that the LID domain was able to partially close without the ligand, while the closure of the AMPbd domain required the ligand binding. The transition state ensemble of the ligand bound form was identified as those structures characterized by highly specific binding of the ligand to the AMPbd domain, and was validated by unrestrained MD simulations. It was also found that complete closure of the LID domain required the dehydration of solvents around the P-loop. These findings suggest that the interplay of the two different types of domain motion is an essential feature in the conformational transition of the enzyme.

Free Energy Guided Sampling

A free energy-guided sampling (FEGS) method is proposed for accelerating exploration of conformational space in unbiased molecular dynamics. Using the cut-based free energy profile and Markov state models, FEGS speeds up sampling of the canonical ensemble by iteratively restarting multiple short simulations in parallel from regions of the free energy surface visited rarely. This exploration stage is followed by a refinement stage in which multiple independent runs are initiated from Boltzmann distributed conformations. Notably, FEGS does not require either collective variables or reaction coordinates and can control the kinetic distance from the starting conformation. We applied FEGS to the alanine dipeptide, which has a human-comprehensible two-dimensional free energy landscape, and a three-stranded antiparallel β-sheet peptide of 20 residues whose folding/unfolding process is governed by a delicate interplay of enthalpy and entropy. For these two systems, FEGS speeds up the exploration of conformational space by 1 to 2 orders of magnitude with respect to conventional sampling and preserves the basins and barriers on the free energy profile.

Milestoning with transition memory

Milestoning is a method used to calculate the kinetics and thermodynamics of molecular processes occurring on time scales that are not accessible to brute force molecular dynamics (MD). In milestoning, the conformation space of the system is sectioned by hypersurfaces (milestones), an ensemble of trajectories is initialized on each milestone, and MD simulations are performed to calculate transitions between milestones. The transition probabilities and transition time distributions are then used to model the dynamics of the system with a Markov renewal process, wherein a long trajectory of the system is approximated as a succession of independent transitions between milestones. This approximation is justified if the transition probabilities and transition times are statistically independent. In practice, this amounts to a requirement that milestones are spaced such that trajectories lose position and velocity memory between subsequent transitions. Unfortunately, limiting the number of milestones limits both the resolution at which a system's properties can be analyzed, and the computational speedup achieved by the method. We propose a generalized milestoning procedure, milestoning with transition memory (MTM), which accounts for memory of previous transitions made by the system. When a reaction coordinate is used to define the milestones, the MTM procedure can be carried out at no significant additional expense as compared to conventional milestoning. To test MTM, we have applied its version that allows for the memory of the previous step to the toy model of a polymer chain undergoing Langevin dynamics in solution. We have computed the mean first passage time for the chain to attain a cyclic conformation and found that the number of milestones that can be used, without incurring significant errors in the first passage time is at least 8 times that permitted by conventional milestoning. We further demonstrate that, unlike conventional milestoning, MTM permits milestones to be spaced such that trajectories do not have enough time to lose their velocity memory between successively crossed milestones.

A conformational transition in the myosin VI converter contributes to the variable step size

Myosin VI (MVI) is a dimeric molecular motor that translocates backwards on actin filaments with a surprisingly large and variable step size, given its short lever arm. A recent x-ray structure of MVI indicates that the large step size can be explained in part by a novel conformation of the converter subdomain in the prepowerstroke state, in which a 53-residue insert, unique to MVI, reorients the lever arm nearly parallel to the actin filament. To determine whether the existence of the novel converter conformation could contribute to the step-size variability, we used a path-based free-energy simulation tool, the string method, to show that there is a small free-energy difference between the novel converter conformation and the conventional conformation found in other myosins. This result suggests that MVI can bind to actin with the converter in either conformation. Models of MVI/MV chimeric dimers show that the variability in the tilting angle of the lever arm that results from the two converter conformations can lead to step-size variations of ∼12 nm. These variations, in combination with other proposed mechanisms, could explain the experimentally determined step-size variability of ∼25 nm for wild-type MVI. Mutations to test the findings by experiment are suggested.

Free energy of conformational transition paths in biomolecules: the string method and its application to myosin VI

A set of techniques developed under the umbrella of the string method is used in combination with all-atom molecular dynamics simulations to analyze the conformation change between the prepowerstroke (PPS) and rigor (R) structures of the converter domain of myosin VI. The challenges specific to the application of these techniques to such a large and complex biomolecule are addressed in detail. These challenges include (i) identifying a proper set of collective variables to apply the string method, (ii) finding a suitable initial string, (iii) obtaining converged profiles of the free energy along the transition path, (iv) validating and interpreting the free energy profiles, and (v) computing the mean first passage time of the transition. A detailed description of the PPS↔R transition in the converter domain of myosin VI is obtained, including the transition path, the free energy along the path, and the rates of interconversion. The methodology developed here is expected to be useful more generally in studies of conformational transitions in complex biomolecules.

Simulations of allosteric transitions

Allosteric transitions are one of the subtler mechanisms used by nature to fine tune protein activity. Effector binding to a specific site on the protein surface induces significant activity change, and initiates a conformational transition that frequently includes domain motions and is very large. From a theoretical and biophysical perspective two problems are particularly intriguing. The first is the way in which a launching signal, which is spatially confined and includes only a few interacting atoms, is propagated to a large-scale conformational transition we frequently see in allosteric transitions. Hence, there is the question of how a small perturbation is magnified to yield motions of thousands of atoms. The second puzzle is of focus, coherence, and efficiency. The impact of the binding of the effector is sometimes extended over tens of angstroms. How the signal is transmitted and kept significant over such large distances in the 'noisy' molecular environment is another major direction of investigation. In the present review we examined different theoretical and computational attempts to solve the questions.

Forming transmembrane channels using end-functionalized nanotubes

Using dissipative particle dynamics (DPD) simulations, we examine the interaction between amphiphilic nanotubes and lipid bilayer membranes. The nanotubes are represented by a hydrophobic shaft that is end-functionalized with hydrophilic groups. Nanotubes that are capped by a monolayer of hydrophilic beads or also encompass hydrophilic "hairs" on just one end of the shaft are found to spontaneously penetrate and assume a transmembrane position; the process, however, depends critically on the membrane tension. On the other hand, nanotubes that include hydrophilic hairs at both ends of the hydrophobic shaft are not observed to spontaneously self-organize into the bilayer. When the membrane is stretched to form a pore, the nanotubes with two hairy ends adsorb on the edge of the pore and become localized in the membrane, thus forming a transmembrane channel. The findings from these studies provide guidelines for creating biomimetic nanotube channels that are capable of selectively transporting molecules through the membrane in response to changes in the local environment.

Milestoning without a Reaction Coordinate

Milestoning is a method for calculating kinetics and thermodynamics of long time processes typically not accessible for straightforward Molecular Dynamics (MD) simulation. In the Milestoning approach, the system of interest is partitioned into cells by dividing hypersurfaces (Milestones) and transitions are computed between nearby hypersurfaces. Kinetics and thermodynamics are derived from the statistics of these transitions. The original Milestoning work concentrated on systems in which a one-dimensional reaction coordinate or an order parameter could be identified. In many biomolecular processes the reaction proceeds via multiple channels or following more than a single order parameter. A description based on a one-dimensional reaction coordinate may be insufficient. In the present paper we introduce a variation that overcomes this limitation.Following the ideas of Vanden-Eijnden and Venturoli on Voronoi cells that avoid the use of an order parameter (J. Chem. Phys. 2009, 130, 194101), we describe another way to "Milestone" systems without a reaction coordinate. We examine the assumptions of the Milestoning calculations of mean first passage times (MFPT) and describe strategies to weaken these assumptions. The method described in this paper, Directional Milestoning, arranges hypersurfaces in higher dimensions that "tag" trajectories such that efficient calculations can be done and at the same time the assumptions required for exact calculations of MFPTs are satisfied approximately.In the original Milestoning papers trajectories are initiated from an equilibrium set of conformations. Here a more accurate distribution, that mimics the first hitting point distribution, is used. We demonstrate the usage of Directional Milestoning in conformational transitions of alanine dipeptide (in vacuum and in aqueous solution) and compare the correctness, efficiency, and statistical stability of the method with exact MD and with a related method.

Predicting the reaction coordinates of millisecond light-induced conformational changes in photoactive yellow protein

Understanding the dynamics of large-scale conformational changes in proteins still poses a challenge for molecular simulations. We employ transition path sampling of explicit solvent molecular dynamics trajectories to obtain atomistic insight in the reaction network of the millisecond timescale partial unfolding transition in the photocycle of the bacterial sensor photoactive yellow protein. Likelihood maximization analysis predicts the best model for the reaction coordinates of each substep as well as tentative transition states, without further simulation. We find that the unfolding of the alpha-helical region 43-51 is followed by sequential solvent exposure of both Glu46 and the chromophore. Which of these two residues is exposed first is correlated with the presence of a salt bridge that is part of the N-terminal domain. Additional molecular dynamics simulations indicate that the exposure of the chromophore does not result in a productive pathway. We discuss several possibilities for experimental validation of these predictions. Our results open the way for studying millisecond conformational changes in other medium-sized (signaling) proteins.

Forward flux sampling for rare event simulations

Rare events are ubiquitous in many different fields, yet they are notoriously difficult to simulate because few, if any, events are observed in a conventional simulation run. Over the past several decades, specialized simulation methods have been developed to overcome this problem. We review one recently developed class of such methods, known as forward flux sampling. Forward flux sampling uses a series of interfaces between the initial and final states to calculate rate constants and generate transition paths for rare events in equilibrium or nonequilibrium systems with stochastic dynamics. This review draws together a number of recent advances, summarizes several applications of the method and highlights challenges that remain to be overcome.

Some recent techniques for free energy calculations

A few recent techniques to calculate free energies in the context of molecular dynamics simulations are discussed: temperature-accelerated molecular dynamics, which is a method to explore fast the important regions in the free energy landscape associated with a set of continuous collective variables without having to know where these regions are beforehand; the single sweep method, which is a variational method to interpolate the free energy globally given a set of mean forces (i.e., a set of gradients of the free energy) calculated at specific points, or centers, on the free energy landscape; and a Voronoi-based free energy method for the calculation of the free energy of the Voronoi tessellation associated with a set of centers. We also discuss how this last technique can be used in conjunction with the string method, and how kinetic information such as reaction rates can be calculated by milestoning using the edges of a Voronoi tessellation as milestones.

Free energy and kinetics of conformational transitions from Voronoi tessellated milestoning with restraining potentials

Milestoning is a method aimed at reconstructing the statistical properties of the long-time dynamics of a system by exploiting the crossing statistics of a set of hypersurfaces, called the "milestones", placed along the reaction coordinate [Faradjian and Elber, J. Chem. Phys.2004, 120, 10880]. Recently, Vanden-Eijnden and Venturoli [J. Chem. Phys.2009, 130, 194101] showed that when a complete Voronoi tessellation of the configurational space is available, milestoning can be reformulated to utilise the statistics from a series of independent simulations, each confined within a single cell via strict reflections at the boundaries. As a byproduct, this "Voronoi tessellated milestoning" method also permits to compute the free energy of the tessellation. Here, the method is extended to support the usage of differentiable restraining potentials to confine the trajectories within each cell.

On the assumptions underlying milestoning

Milestoning is a procedure to compute the time evolution of complicated processes such as barrier crossing events or long diffusive transitions between predefined states. Milestoning reduces the dynamics to transition events between intermediates (the milestones) and computes the local kinetic information to describe these transitions via short molecular dynamics (MD) runs between the milestones. The procedure relies on the ability to reinitialize MD trajectories on the milestones to get the right kinetic information about the transitions. It also rests on the assumptions that the transition events between successive milestones and the time lags between these transitions are statistically independent. In this paper, we analyze the validity of these assumptions. We show that sets of optimal milestones exist, i.e., sets such that successive transitions are indeed statistically independent. The proof of this claim relies on the results of transition path theory and uses the isocommittor surfaces of the reaction as milestones. For systems in the overdamped limit, we also obtain the probability distribution to reinitialize the MD trajectories on the milestones, and we discuss why this distribution is not available in closed form for systems with inertia. We explain why the time lags between transitions are not statistically independent even for optimal milestones, but we show that working with such milestones allows one to compute mean first passage times between milestones exactly. Finally, we discuss some practical implications of our results and we compare milestoning with Markov state models in view of our findings.