Basic ingredients of free energy calculations: A review

Efficient Estimation of Absolute Binding Free Energy for a Homeodomain-DNA Complex from Nonequilibrium Pulling Simulations

Estimation of binding free energies is one of the central aims of simulations of biomolecular complexes. We explore the accuracy and efficiency of setups based on nonequilibrium pulling simulations applied to the estimation of binding affinities of DNA-binding proteins. Absolute binding free energies are calculated over a range of temperatures and compared to results obtained previously using an equilibrium method. We show that realistic binding affinities can be obtained with the presented nonequilibrium approach, which also entails lower computational requirements. Errors of the binding free energy estimates are investigated and are shown to be comparable to those observed previously. Bounds are provided on the convergence of the errors with respect to the number of pulling simulations performed and with respect to the applied pull rate.

Umbrella Visualization: A method of analysis dedicated to glycan flexibility with UnityMol

N-glycosylation is a post-translational modification heavily impacting protein functions. Some alterations of glycosylation, such as sialic acid hydrolysis, are related to protein dysfunction. Because of their high flexibility and the many reactive groups of the glycan chains, studying glycans with in vitro methods is a challenging task. Molecular dynamics is a useful tool and probably the only one in biology able to overcome this problem and gives access to conformational information through exhaustive sampling. To better decipher the impact of N-glycans, the analysis and visualization of their influence over time on protein structure is a prerequisite. We developed the Umbrella Visualization, a graphical method that assigns the glycan intrinsic flexibility during a molecular dynamics trajectory. The density plot generated by this method brought relevant informations regarding glycans dynamics and flexibility, but needs further development in order to integrate an accurate description of the protein topology and its interactions. We propose here to transform this analysis method into a visualization mode in UnityMol. UnityMol is a molecular editor, viewer and prototyping platform, coded in C#. The new representation of glycan chains presented in this study takes into account both the main positions adopted by each antenna of a glycan and their statistical relevance. By displaying the collected data on the protein surface, one is then able to investigate the protein/glycan interactions.

Molecular Dynamics Simulations Applied to Structural and Dynamical Transitions of the Huntingtin Protein: A Review

Over the recent years, Huntington's disease (HD) has become widely discussed in the scientific literature especially because at the mutant level there are several contradictions regarding the aggregation mechanism. The specific role of the physiological huntingtin protein remains unknown, due to the lack of characterization of its entire crystallographic structure, making the experimental and theoretical research even harder when taking into consideration its involvement in multiple biological functions and its high affinity for different interacting partners. Different types of models, containing fewer (not more than 35 Qs) polyglutamine residues for the WT structure and above 35 Qs for the mutants, were subjected to classical or advanced MD simulations to establish the proteins' structural stability by evaluating their conformational changes. Outside the polyQ tract, there are two other regions of interest (the N17 domain and the polyP rich domain) considered to be essential for the aggregation kinetics at the mutant level. The polymerization process is considered to be dependent on the polyQ length. As the polyQ tract's dimension increases, the structures present more β-sheet conformations. Contrarily, it is also considered that the aggregation stability is not necessarily dependent on the number of Qs, while the initial stage of the aggregation seed might play the decisive role. A general assumption regarding the polyP domain is that it might preserve the polyQ structures soluble by acting as an antagonist for β-sheet formation.

Ligand Binding Thermodynamic Cycles: Hysteresis, the Locally Weighted Histogram Analysis Method, and the Overlapping States Matrix

Free energy perturbation (FEP) simulations have been widely applied to obtain predictions of the relative binding free energy for a series of congeneric ligands binding to the same receptor, which is an essential component for the lead optimization process in computer-aided drug discovery. In the case of several congeneric ligands forming a perturbation map involving a closed thermodynamic cycle, the summation of the estimated free energy change along each edge in the cycle using Bennett acceptance ratio (BAR) usually will deviate from zero due to systematic and random errors, which is the hysteresis of cycle closure. In this work, the advanced reweighting techniques binless weighted histogram analysis method (UWHAM) and locally weighted histogram analysis method (LWHAM) are applied to provide statistical estimators of the free energy change along each edge in order to eliminate the hysteresis effect. As an example, we analyze a closed thermodynamic cycle involving four congeneric ligands which bind to HIV-1 integrase, a promising target which has emerged for antiviral therapy. We demonstrate that, compared with FEP and BAR, more accurate and hysteresis-free estimates of free energy differences can be achieved by using UWHAM to find a single estimate of the density of states based on all of the data in the cycle. Furthermore, by comparison of LWHAM results obtained from the inclusion of different numbers of neighboring states with UWHAM estimation involving all the states, we show how to determine the optimal neighborhood size in the LWHAM analysis to balance the trade-offs between computational cost and accuracy of the free energy prediction. Even with the smallest neighborhood, LWHAM can improve the BAR free energy estimates using the same input data as BAR. We introduce an overlapping states matrix that is constructed by using the global jump formula of LWHAM and plot its heat map. The heat map provides a quantitative measure of the overlap between pairs of alchemical/thermodynamic states. We explain how to identify and improve the FEP calculations along the edges that most likely cause large systematic errors by using the heat map of the overlapping states matrix and by comparing the BAR and UWHAM estimates of the free energy change.

Predicting and Experimentally Validating Hot-Spot Residues at Protein-Protein Interfaces

Protein-protein interactions (PPIs) are vital to all biological processes. These interactions are often dynamic, sometimes transient, typically occur over large topographically shallow protein surfaces, and can exhibit a broad range of affinities. Considerable progress has been made in determining PPI structures. However, given the above properties, understanding the key determinants of their thermodynamic stability remains a challenge in chemical biology. An improved ability to identify and engineer PPIs would advance understanding of biological mechanisms and mutant phenotypes and also provide a firmer foundation for inhibitor design. In silico prediction of PPI hot-spot amino acids using computational alanine scanning (CAS) offers a rapid approach for predicting key residues that drive protein-protein association. This can be applied to all known PPI structures; however there is a trade-off between throughput and accuracy. Here we describe a comparative analysis of multiple CAS methods, which highlights effective approaches to improve the accuracy of predicting hot-spot residues. Alongside this, we introduce a new method, BUDE Alanine Scanning, which can be applied to single structures from crystallography and to structural ensembles from NMR or molecular dynamics data. The comparative analyses facilitate accurate prediction of hot-spots that we validate experimentally with three diverse targets: NOXA-B/MCL-1 (an α-helix-mediated PPI), SIMS/SUMO, and GKAP/SHANK-PDZ (both β-strand-mediated interactions). Finally, the approach is applied to the accurate prediction of hot-spot residues at a topographically novel Affimer/BCL-xL protein-protein interface.

Application of the Movable Type Free Energy Method to the Caspase-Inhibitor BindingAffinity Study

Many studies have provided evidence suggesting that caspases not only contribute to the neurodegeneration associated with Alzheimer’s disease (AD) but also play essential roles in promoting the underlying pathology of this disease. Studies regarding the caspase inhibition draw researchers’ attention through time due to its therapeutic value in the treatment of AD. In this work, we apply the “Movable Type” (MT) free energy method, a Monte Carlo sampling method extrapolating the binding free energy by simulating the partition functions for both free-state and bound-state protein and ligand configurations, to the caspase-inhibitor binding affinity study. Two test benchmarks are introduced to examine the robustness and sensitivity of the MT method concerning the caspase inhibition complexing. The first benchmark employs a large-scale test set including more than a hundred active inhibitors binding to caspase-3. The second benchmark includes several smaller test sets studying the relative binding free energy differences for minor structural changes at the caspase-inhibitor interaction interfaces. Calculation results show that the RMS errors for all test sets are below 1.5 kcal/mol compared to the experimental binding affinity values, demonstrating good performance in simulating the caspase-inhibitor complexing. For better understanding the protein-ligand interaction mechanism, we then take a closer look at the global minimum binding modes and free-state ligand conformations to study two pairs of caspase-inhibitor complexes with (1) different caspase targets binding to the same inhibitor, and (2) different polypeptide inhibitors targeting the same caspase target. By comparing the contact maps at the binding site of different complexes, we revealed how small structural changes affect the caspase-inhibitor interaction energies. Overall, this work provides a new free energy approach for studying the caspase inhibition, with structural insight revealed for both free-state and bound-state molecular configurations.

Principal component analysis of nonequilibrium molecular dynamics simulations

Principal component analysis (PCA) represents a standard approach to identify collective variables {x_i} = x, which can be used to construct the free energy landscape ΔG(x) of a molecular system. While PCA is routinely applied to equilibrium molecular dynamics (MD) simulations, it is less obvious as to how to extend the approach to nonequilibrium simulation techniques. This includes, e.g., the definition of the statistical averages employed in PCA as well as the relation between the equilibrium free energy landscape ΔG(x) and the energy landscapes ΔG(x) obtained from nonequilibrium MD. As an example for a nonequilibrium method, "targeted MD" is considered which employs a moving distance constraint to enforce rare transitions along some biasing coordinate s. The introduced bias can be described by a weighting function P(s), which provides a direct relation between equilibrium and nonequilibrium data, and thus establishes a well-defined way to perform PCA on nonequilibrium data. While the resulting distribution P(x) and energy ΔG∝lnP will not reflect the equilibrium state of the system, the nonequilibrium energy landscape ΔG(x) may directly reveal the molecular reaction mechanism. Applied to targeted MD simulations of the unfolding of decaalanine, for example, a PCA performed on backbone dihedral angles is shown to discriminate several unfolding pathways. Although the formulation is in principle exact, its practical use depends critically on the choice of the biasing coordinate s, which should account for a naturally occurring motion between two well-defined end-states of the system.

Molecular simulation as a computational pharmaceutics tool to predict drug solubility, solubilization processes and partitioning

In this review we will discuss how computational methods, and in particular classical molecular dynamics simulations, can be used to calculate solubility of pharmaceutically relevant molecules and systems. To the extent possible, we focus on the non-technical details of these calculations, and try to show also the added value of a more thorough and detailed understanding of the solubilization process obtained by using computational simulations. Although the main focus is on classical molecular dynamics simulations, we also provide the reader with some insights into other computational techniques, such as the COSMO-method, and also discuss Flory-Huggins theory and solubility parameters. We hope that this review will serve as a valuable starting point for any pharmaceutical researcher, who has not yet fully explored the possibilities offered by computational approaches to solubility calculations.

Alchemical Free-Energy Calculations by Multiple-Replica λ-Dynamics: The Conveyor Belt Thermodynamic Integration Scheme

A new method is proposed to calculate alchemical free-energy differences based on molecular dynamics (MD) simulations, called the conveyor belt thermodynamic integration (CBTI) scheme. As in thermodynamic integration (TI), K replicas of the system are simulated at different values of the alchemical coupling parameter λ. The number K is taken to be even, and the replicas are equally spaced on a forward-turn-backward-turn path, akin to a conveyor belt (CB) between the two physical end-states; and as in λ-dynamics (λD), the λ-values associated with the individual systems evolve in time along the simulation. However, they do so in a concerted fashion, determined by the evolution of a single dynamical variable Λ of period 2π controlling the advance of the entire CB. Thus, a change of Λ is always associated with K/2 equispaced replicas moving forward and K/2 equispaced replicas moving backward along λ. As a result, the effective free-energy profile of the replica system along Λ is periodic of period 2 πK^-1, and the magnitude of its variations decreases rapidly upon increasing K, at least as K^-1 in the limit of large K. When a sufficient number of replicas is used, these variations become small, which enables a complete and quasi-homogeneous coverage of the λ-range by the replica system, without application of any biasing potential. If desired, a memory-based biasing potential can still be added to further homogenize the sampling, the preoptimization of which is computationally inexpensive. The final free-energy profile along λ is calculated similarly to TI, by binning of the Hamiltonian λ-derivative as a function of λ considering all replicas simultaneously, followed by quadrature integration. The associated quadrature error can be kept very low owing to the continuous and quasi-homogeneous λ-sampling. The CBTI scheme can be viewed as a continuous/deterministic/dynamical analog of the Hamiltonian replica-exchange/permutation (HRE/HRP) schemes or as a correlated multiple-replica analog of the λD or λ-local elevation umbrella sampling (λ-LEUS) schemes. Compared to TI, it shares the advantage of the latter schemes in terms of enhanced orthogonal sampling, i.e. the availability of variable-λ paths to circumvent conformational barriers present at specific λ-values. Compared to HRE/HRP, it permits a deterministic and continuous sampling of the λ-range, is expected to be less sensitive to possible artifacts of the thermo- and barostating schemes, and bypasses the need to carefully preselect a λ-ladder and a swapping-attempt frequency. Compared to λ-LEUS, it eliminates (or drastically reduces) the dead time associated with the preoptimization of a biasing potential. The goal of this article is to provide the mathematical/physical formulation of the proposed CBTI scheme, along with an initial application of the method to the calculation of the hydration free energy of methanol.

Orientational and Folding Thermodynamics via Electric Dipole Moment Restraining

The projection of molecular processes onto a small set of relevant descriptors, the so-called reaction coordinates or collective variables (CVs), is a technique nowadays routinely employed by the biomolecular simulation community. In this work, we implemented two CVs to manipulate the orientation (i.e., angle) (μ⃗_a) and magnitude (|μ⃗|) of the electric dipole moment. In doing so, we studied the thermodynamics of water orientation under the application of external voltages and the folding of two polypeptides at zero-field conditions. The projection of the free-energy [potential of mean force (PMF)] along water orientation defined an upper limit of around 0.3 V for irrelevant thermodynamic effects. On the other hand, sufficiently strong μ⃗_a restraints applied on 12-alanine (Ala₁₂) triggered structural effects because of the alignment of local dipoles; for lower restraints, a full-body rotation is achieved. The manipulation of |μ⃗| produced strong perturbations on the secondary structure of Ala₁₂, promoting an enhanced sampling to its configurational space. Rigorous free-energy calculations in the form of 2-D PMFs for deca-alanine showed the utility of |μ⃗| as a reaction coordinate to study folding in small α helices. As a whole, we propose that the manipulation of both components of the dipole moment, μ⃗_a and |μ⃗|, provides thermodynamics insights into the structural conformation and stability of biomolecules. These new CVs are implemented in the Colvars module, available for NAMD and LAMMPS.

Can All-Atom Molecular Dynamics Simulations Quantitatively Describe Homeodomain-DNA Binding Equilibria?

We systematically investigate the applicability of a molecular dynamics-based setup for the calculations of standard binding free energies of biologically relevant protein-DNA complexes. The free energies are extracted from a potential of mean force calculated using umbrella sampling simulations. Two protein-DNA systems derived from a homeodomain transcription factor complex are studied in order to investigate the binding of both disordered and globular proteins. Free energies and trajectories obtained using two modern molecular mechanical force fields are compared to each other and to experimental data. The temperature dependence of the calculated standard binding free energies is investigated by performing all simulations over a range of temperatures. We show that the values of standard binding free energies obtained from these simulations are overestimated compared to experimental results. Significant differences are observed between the two protein-DNA systems and between the two force fields, which are explained by different propensities to form inter- and intramolecular contacts. The number of protein-DNA contacts increases with increasing temperature, in agreement with the experimentally known temperature dependence of enthalpies of binding. However, conclusions about the temperature dependence of the standard binding free energies cannot be made with confidence, as the differences among the values are on the order of statistical uncertainty.

Absolute Binding Free Energy Calculation and Design of a Subnanomolar Inhibitor of Phosphodiesterase-10

Accurate prediction of absolute protein-ligand binding free energy could considerably enhance the success rate of structure-based drug design but is extremely challenging and time-consuming. Free energy perturbation (FEP) has been proven reliable but is limited to prediction of relative binding free energies of similar ligands (with only minor structural differences) in binding with a same drug target in practical drug design applications. Herein, a Gaussian algorithm-enhanced FEP (GA-FEP) protocol has been developed to enhance the FEP simulation performance, enabling to efficiently carry out the FEP simulations on vanishing the whole ligand and, thus, predict the absolute binding free energies (ABFEs). Using the GA-FEP protocol, the FEP simulations for the ABFE calculation (denoted as GA-FEP/ABFE) can achieve a satisfactory accuracy for both structurally similar and diverse ligands in a dataset of more than 100 receptor-ligand systems. Further, our GA-FEP/ABFE-guided lead optimization against phosphodiesterase-10 led to the discovery of a subnanomolar inhibitor (IC₅₀ = 0.87 nM, ∼2000-fold improvement in potency) with cocrystal confirmation.

Solvation Free Energy Calculations with Quantum Mechanics/Molecular Mechanics and Machine Learning Models

For exploration of chemical and biological systems, the combined quantum mechanics and molecular mechanics (QM/MM) and machine learning (ML) models have been developed recently to achieve high accuracy and efficiency for molecular dynamics (MD) simulations. Despite its success on reaction free energy calculations, how to identify new configurations on insufficiently sampled regions during MD and how to update the current ML models with the growing database on the fly are both very important but still challenging. In this article, we apply the QM/MM ML method to solvation free energy calculations and address these two challenges. We employ three approaches to detect new data points and introduce the gradient boosting algorithm to reoptimize efficiently the ML model during ML-based MD sampling. The solvation free energy calculations on several typical organic molecules demonstrate that our developed method provides a systematic, robust, and efficient way to explore new chemistry using ML-based QM/MM MD simulations.

Toward Learned Chemical Perception of Force Field Typing Rules

Molecular mechanics force fields define how the energy and forces in a molecular system are computed from its atomic positions, thus enabling the study of such systems through computational methods like molecular dynamics and Monte Carlo simulations. Despite progress toward automated force field parametrization, considerable human expertise is required to develop or extend force fields. In particular, human input has long been required to define atom types, which encode chemically unique environments that determine which parameters will be assigned. However, relying on humans to establish atom types is suboptimal. Human-created atom types are often developed without statistical justification, leading to over- or under-fitting of data. Human-created types are also difficult to extend in a systematic and consistent manner when new chemistries must be modeled or new data becomes available. Finally, human effort is not scalable when force fields must be generated for new (bio)polymers, compound classes, or materials. To remedy these deficiencies, our long-term goal is to replace human specification of atom types with an automated approach, based on rigorous statistics and driven by experimental and/or quantum chemical reference data. In this work, we describe novel methods that automate the discovery of appropriate chemical perception: SMARTY allows for the creation of atom types, while SMIRKY goes further by automating the creation of fragment (nonbonded, bonds, angles, and torsions) types. These approaches enable the creation of move sets in atom or fragment type space, which are used within a Monte Carlo optimization approach. We demonstrate the power of these new methods by automating the rediscovery of human defined atom types (SMARTY) or fragment types (SMIRKY) in existing small molecule force fields. We assess these approaches using several molecular data sets, including one which covers a diverse subset of the DrugBank database.

Towards operando computational modeling in heterogeneous catalysis

An increased synergy between experimental and theoretical investigations in heterogeneous catalysis has become apparent during the last decade. Experimental work has extended from ultra-high vacuum and low temperature towards operando conditions. These developments have motivated the computational community to move from standard descriptive computational models, based on inspection of the potential energy surface at 0 K and low reactant concentrations (0 K/UHV model), to more realistic conditions. The transition from 0 K/UHV to operando models has been backed by significant developments in computer hardware and software over the past few decades. New methodological developments, designed to overcome part of the gap between 0 K/UHV and operando conditions, include (i) global optimization techniques, (ii) ab initio constrained thermodynamics, (iii) biased molecular dynamics, (iv) microkinetic models of reaction networks and (v) machine learning approaches. The importance of the transition is highlighted by discussing how the molecular level picture of catalytic sites and the associated reaction mechanisms changes when the chemical environment, pressure and temperature effects are correctly accounted for in molecular simulations. It is the purpose of this review to discuss each method on an equal footing, and to draw connections between methods, particularly where they may be applied in combination.

Reproducibility of Free Energy Calculations across Different Molecular Simulation Software Packages

Alchemical free energy calculations are an increasingly important modern simulation technique to calculate free energy changes on binding or solvation. Contemporary molecular simulation software such as AMBER, CHARMM, GROMACS, and SOMD include support for the method. Implementation details vary among those codes, but users expect reliability and reproducibility, i.e., for a given molecular model and set of force field parameters, comparable free energy differences should be obtained within statistical bounds regardless of the code used. Relative alchemical free energy (RAFE) simulation is increasingly used to support molecule discovery projects, yet the reproducibility of the methodology has been less well tested than its absolute counterpart. Here we present RAFE calculations of hydration free energies for a set of small organic molecules and demonstrate that free energies can be reproduced to within about 0.2 kcal/mol with the aforementioned codes. Absolute alchemical free energy simulations have been carried out as a reference. Achieving this level of reproducibility requires considerable attention to detail and package-specific simulation protocols, and no universally applicable protocol emerges. The benchmarks and protocols reported here should be useful for the community to validate new and future versions of software for free energy calculations.

Using correlated motions to determine sufficient sampling times for molecular dynamics

Here we present a time-dependent correlation method that provides insight into how long a system takes to grow into its equal-time (Pearson) correlation. We also show a usage of an extant time-lagged correlation method that indicates the time for parts of a system to become decorrelated, relative to equal-time correlation. Given a completed simulation (or set of simulations), these tools estimate (i) how long of a simulation of the same system would be sufficient to observe the same correlated motions, (ii) if patterns of observed correlated motions indicate events beyond the timescale of the simulation, and (iii) how long of a simulation is needed to observe these longer timescale events. We view this method as a decision-support tool that will aid researchers in determining necessary sampling times. In principle, this tool is extendable to any multidimensional time series data with a notion of correlated fluctuations; however, here we limit our discussion to data from molecular-dynamics simulations.

Relative Ligand-Binding Free Energies Calculated from Multiple Short QM/MM MD Simulations

We have devised a new efficient approach to compute combined quantum mechanical (QM) and molecular mechanical (MM, i.e. QM/MM) ligand-binding relative free energies. Our method employs the reference-potential approach with free-energy perturbation both at the MM level (between the two ligands) and from MM to QM/MM (for each ligand). To ensure that converged results are obtained for the MM → QM/MM perturbations, explicit QM/MM molecular dynamics (MD) simulations are performed with two intermediate mixed states. To speed up the calculations, we utilize the fact that the phase space can be extensively sampled at the MM level. Therefore, we run many short QM/MM MD simulations started from snapshots of the MM simulations, instead of a single long simulation. As a test case, we study the binding of nine cyclic carboxylate ligands to the octa-acid deep cavitand. Only the ligand is in the QM system, treated with the semiempirical PM6-DH+ method. We show that for eight of the ligands, we obtain well converged results with short MD simulations (1-15 ps). However, in one case, the convergence is slower (∼50 ps) owing to a mismatch between the conformational preferences of the MM and QM/MM potentials. We test the effect of initial minimization, the need of equilibration, and how many independent simulations are needed to reach a certain precision. The results show that the present approach is about four times faster than using standard MM → QM/MM free-energy perturbations with the same accuracy and precision.

The Role of the Active Site Flap in Streptavidin/Biotin Complex Formation

Obtaining a detailed description of how active site flap motion affects substrate or ligand binding will advance structure-based drug design (SBDD) efforts on systems including the kinases, HSP90, HIV protease, ureases, etc. Through this understanding, we will be able to design better inhibitors and better proteins that have desired functions. Herein we address this issue by generating the relevant configurational states of a protein flap on the molecular energy landscape using an approach we call MT_Flex-b and then following this with a procedure to estimate the free energy associated with the motion of the flap region. To illustrate our overall workflow, we explored the free energy changes in the streptavidin/biotin system upon introducing conformational flexibility in loop_3-4 in the biotin unbound ( apo) and bound ( holo) state. The free energy surfaces were created using the Movable Type free energy method, and for further validation, we compared them to potential of mean force (PMF) generated free energy surfaces using MD simulations employing the FF99SBILDN and FF14SB force fields. We also estimated the free energy thermodynamic cycle using an ensemble of closed-like and open-like end states for the ligand unbound and bound states and estimated the binding free energy to be approximately -16.2 kcal/mol (experimental -18.3 kcal/mol). The good agreement between MT_Flex-b in combination with the MT method with experiment and MD simulations supports the effectiveness of our strategy in obtaining unique insights into the motions in proteins that can then be used in a range of biological and biomedical applications.

A GPU-Accelerated Parameter Interpolation Thermodynamic Integration Free Energy Method

There has been a resurgence of interest in free energy methods motivated by the performance enhancements offered by molecular dynamics (MD) software written for specialized hardware, such as graphics processing units (GPUs). In this work, we exploit the properties of a parameter-interpolated thermodynamic integration (PI-TI) method to connect states by their molecular mechanical (MM) parameter values. This pathway is shown to be better behaved for Mg2+ → Ca2+ transformations than traditional linear alchemical pathways (with and without soft-core potentials). The PI-TI method has the practical advantage that no modification of the MD code is required to propagate the dynamics, and unlike with linear alchemical mixing, only one electrostatic evaluation is needed (e.g., single call to particle-mesh Ewald) leading to better performance. In the case of AMBER, this enables all the performance benefits of GPU-acceleration to be realized, in addition to unlocking the full spectrum of features available within the MD software, such as Hamiltonian replica exchange (HREM). The TI derivative evaluation can be accomplished efficiently in a post-processing step by reanalyzing the statistically independent trajectory frames in parallel for high throughput. We also show how one can evaluate the particle mesh Ewald contribution to the TI derivative evaluation without needing to perform two reciprocal space calculations. We apply the PI-TI method with HREM on GPUs in AMBER to predict p Ka values in double stranded RNA molecules and make comparison with experiments. Convergence to under 0.25 units for these systems required 100 ns or more of sampling per window and coupling of windows with HREM. We find that MM charges derived from ab initio QM/MM fragment calculations improve the agreement between calculation and experimental results.

The Photosystem II Subunit S under Stress

Nonphotochemical quenching is the protective mechanism against overexcitation of photosystem II, triggered by excess ΔpH in photosynthetic membranes. The light-harvesting complexes (LHCs), the de-epoxidation of violaxanthin to zeaxanthin, and the photosystem II subunit S (PsbS) work in synergy for an optimized multilevel response. Understanding the fine details of this synergy has proven challenging to scientific research. Here, we employ large-scale, all-atom molecular simulations and beyond experimental insight, we proceed a step further in identifying the PsbS dynamics that could possibly be associated with this synergy. For the first time, to our knowledge, we probe the distinct behavior of PsbS under ΔpH that probes the details of the potential dimer-to-monomer transition, and in a violaxanthin/zeaxanthin-rich membrane, at an all-atom resolution. We propose that the lumen-exposed residues, threonine 162 and glutamic acid 173, form stabilizing hydrogen bonds between the PsbS monomers only at high lumen pH, whereas at low pH (excess ΔpH) this interaction is lost, and leads to higher flexibility of the protein and potentially to the dimer-to-monomer transition. Lastly, we discuss how conformational changes under the presence of ΔpH/zeaxanthin are related to the PsbS role in the current nonphotochemical quenching model in the literature. For the latter, we probe a PsbS-monomeric LHCII association. The association is proposed to potentially alter the monomeric LHCII sensitivity to ΔpH by changing the pKa values of interacting LHCII residues. This serves as an example where protonation-ligation events enhance protein-protein interactions fundamental to many life processes.

Computation of Hemagglutinin Free Energy Difference by the Confinement Method

Hemagglutinin (HA) mediates membrane fusion, a crucial step during influenza virus cell entry. How many HAs are needed for this process is still subject to debate. To aid in this discussion, the confinement free energy method was used to calculate the conformational free energy difference between the extended intermediate and postfusion state of HA. Special care was taken to comply with the general guidelines for free energy calculations, thereby obtaining convergence and demonstrating reliability of the results. The energy that one HA trimer contributes to fusion was found to be 34.2 ± 3.4kBT, similar to the known contributions from other fusion proteins. Although computationally expensive, the technique used is a promising tool for the further energetic characterization of fusion protein mechanisms. Knowledge of the energetic contributions per protein, and of conserved residues that are crucial for fusion, aids in the development of fusion inhibitors for antiviral drugs.

Relative Binding Free Energy Calculations in Drug Discovery: Recent Advances and Practical Considerations

Accurate in silico prediction of protein-ligand binding affinities has been a primary objective of structure-based drug design for decades due to the putative value it would bring to the drug discovery process. However, computational methods have historically failed to deliver value in real-world drug discovery applications due to a variety of scientific, technical, and practical challenges. Recently, a family of approaches commonly referred to as relative binding free energy (RBFE) calculations, which rely on physics-based molecular simulations and statistical mechanics, have shown promise in reliably generating accurate predictions in the context of drug discovery projects. This advance arises from accumulating developments in the underlying scientific methods (decades of research on force fields and sampling algorithms) coupled with vast increases in computational resources (graphics processing units and cloud infrastructures). Mounting evidence from retrospective validation studies, blind challenge predictions, and prospective applications suggests that RBFE simulations can now predict the affinity differences for congeneric ligands with sufficient accuracy and throughput to deliver considerable value in hit-to-lead and lead optimization efforts. Here, we present an overview of current RBFE implementations, highlighting recent advances and remaining challenges, along with examples that emphasize practical considerations for obtaining reliable RBFE results. We focus specifically on relative binding free energies because the calculations are less computationally intensive than absolute binding free energy (ABFE) calculations and map directly onto the hit-to-lead and lead optimization processes, where the prediction of relative binding energies between a reference molecule and new ideas (virtual molecules) can be used to prioritize molecules for synthesis. We describe the critical aspects of running RBFE calculations, from both theoretical and applied perspectives, using a combination of retrospective literature examples and prospective studies from drug discovery projects. This work is intended to provide a contemporary overview of the scientific, technical, and practical issues associated with running relative binding free energy simulations, with a focus on real-world drug discovery applications. We offer guidelines for improving the accuracy of RBFE simulations, especially for challenging cases, and emphasize unresolved issues that could be improved by further research in the field.

Free energy calculations on the stability of the 14-3-3ζ protein

Mutations of cysteine are often introduced to e.g. avoid formation of non-physiological inter-molecular disulfide bridges in in-vitro experiments, or to maintain specificity in labeling experiments. Alanine or serine is typically preferred, which usually do not alter the overall protein stability, when the original cysteine was surface exposed. However, selecting the optimal mutation for cysteines in the hydrophobic core of the protein is more challenging. In this work, the stability of selected Cys mutants of 14-3-3ζ was predicted by free-energy calculations and the obtained data were compared with experimentally determined stabilities. Both the computational predictions as well as the experimental validation point at a significant destabilization of mutants C94A and C94S. This destabilization could be attributed to the formation of hydrophobic cavities and a polar solvation of a hydrophilic side chain. A L12E, M78K double mutant was further studied in terms of its reduced dimerization propensity. In contrast to naïve expectations, this double mutant did not lead to the formation of strong salt bridges, which was rationalized in terms of a preferred solvation of the ionic species. Again, experiments agreed with the calculations by confirming the monomerization of the double mutants. Overall, the simulation data is in good agreement with experiments and offers additional insight into the stability and dimerization of this important family of regulatory proteins.

Efficient Construction of Free Energy Profiles of Breathing Metal-Organic Frameworks Using Advanced Molecular Dynamics Simulations

In order to reliably predict and understand the breathing behavior of highly flexible metal–organic frameworks from thermodynamic considerations, an accurate estimation of the free energy difference between their different metastable states is a prerequisite. Herein, a variety of free energy estimation methods are thoroughly tested for their ability to construct the free energy profile as a function of the unit cell volume of MIL-53(Al). The methods comprise free energy perturbation, thermodynamic integration, umbrella sampling, metadynamics, and variationally enhanced sampling. A series of molecular dynamics simulations have been performed in the frame of each of the five methods to describe structural transformations in flexible materials with the volume as the collective variable, which offers a unique opportunity to assess their computational efficiency. Subsequently, the most efficient method, umbrella sampling, is used to construct an accurate free energy profile at different temperatures for MIL-53(Al) from first principles at the PBE+D3(BJ) level of theory. This study yields insight into the importance of the different aspects such as entropy contributions and anharmonic contributions on the resulting free energy profile. As such, this thorough study provides unparalleled insight in the thermodynamics of the large structural deformations of flexible materials.

Synthesis and anti-HCV activity of a series of β-d-2'-deoxy-2'-dibromo nucleosides and their corresponding phosphoramidate prodrugs

Several β-d-2'-deoxy-2'-substituted nucleoside analogs have displayed potent and selective anti-HCV activities and some of them have reached human clinical trials. In that regard, we report herein the synthesis of a series of 2'-deoxy,2'-dibromo substituted U, C, G and A nucleosides 10a-d and their corresponding phosphoramidate prodrugs 13a-d. The synthesized nucleosides 10a-d and prodrugs 13a-d were evaluated for their inhibitory activity against HCV as well as cellular toxicity. The results showed that the most potent compound was prodrug 13a, which exhibited micromolar inhibitory activity (EC₅₀ = 1.5 ± 0.8 µM) with no observed toxicity. In addition, molecular modeling and free energy perturbation calculations for the 5'-triphosphate formed from 13a and related 2'-modified nucleotides are discussed.

Sequence statistics of tertiary structural motifs reflect protein stability

The Protein Data Bank (PDB) has been a key resource for learning general rules of sequence-structure relationships in proteins. Quantitative insights have been gained by defining geometric descriptors of structure (e.g., distances, dihedral angles, solvent exposure, etc.) and observing their distributions and sequence preferences. Here we argue that as the PDB continues to grow, it may become unnecessary to reduce structure into a set of elementary descriptors. Instead, it could be possible to deduce quantitative sequence-structure relationships in the context of precisely-defined complex structural motifs by mining the PDB for closely matching backbone geometries. To validate this idea, we turned to the the task of predicting changes in protein stability upon amino-acid substitution—a difficult problem of broad significance. We defined non-contiguous tertiary motifs (TERMs) around a protein site of interest and extracted sequence preferences from ensembles of closely-matching substructures in the PDB to predict mutational stability changes at the site, ΔΔG_m. We demonstrate that these ensemble statistics predict ΔΔG_m on par with state-of-the-art statistical and machine-learning methods on large thermodynamic datasets, and outperform these, along with a leading structure-based modeling approach, when tested in the context of unbiased diverse mutations. Further, we show that the performance of the TERM-based method is directly related to the amount of available relevant structural data, automatically improving with the growing PDB. This enables a means of estimating prediction accuracy. Our results clearly demonstrate that: 1) statistics of non-contiguous structural motifs in the PDB encode fundamental sequence-structure relationships related to protein thermodynamic stability, and 2) the PDB is now large enough that such statistics are already useful in practice, with their accuracy expected to continue increasing as the database grows. These observations suggest new ways of using structural data towards addressing problems of computational structural biology.

Efficient Round-Trip Time Optimization for Replica-Exchange Enveloping Distribution Sampling (RE-EDS)

Replica-exchange enveloping distribution sampling (RE-EDS) allows the efficient estimation of free-energy differences between multiple end-states from a single molecular dynamics (MD) simulation. In EDS, a reference state is sampled, which can be tuned by two types of parameters, i.e., smoothness parameters(s) and energy offsets, such that all end-states are sufficiently sampled. However, the choice of these parameters is not trivial. Replica exchange (RE) or parallel tempering is a widely applied technique to enhance sampling. By combining EDS with the RE technique, the parameter choice problem could be simplified and the challenge shifted toward an optimal distribution of the replicas in the smoothness-parameter space. The choice of a certain replica distribution can alter the sampling efficiency significantly. In this work, global round-trip time optimization (GRTO) algorithms are tested for the use in RE-EDS simulations. In addition, a local round-trip time optimization (LRTO) algorithm is proposed for systems with slowly adapting environments, where a reliable estimate for the round-trip time is challenging to obtain. The optimization algorithms were applied to RE-EDS simulations of a system of nine small-molecule inhibitors of phenylethanolamine N-methyltransferase (PNMT). The energy offsets were determined using our recently proposed parallel energy-offset (PEOE) estimation scheme. While the multistate GRTO algorithm yielded the best replica distribution for the ligands in water, the multistate LRTO algorithm was found to be the method of choice for the ligands in complex with PNMT. With this, the 36 alchemical free-energy differences between the nine ligands were calculated successfully from a single RE-EDS simulation 10 ns in length. Thus, RE-EDS presents an efficient method for the estimation of relative binding free energies.

Predicting Binding Free Energies: Frontiers and Benchmarks

Binding free energy calculations based on molecular simulations provide predicted affinities for biomolecular complexes. These calculations begin with a detailed description of a system, including its chemical composition and the interactions among its components. Simulations of the system are then used to compute thermodynamic information, such as binding affinities. Because of their promise for guiding molecular design, these calculations have recently begun to see widespread applications in early-stage drug discovery. However, many hurdles remain in making them a robust and reliable tool. In this review, we highlight key challenges of these calculations, discuss some examples of these challenges, and call for the designation of standard community benchmark test systems that will help the research community generate and evaluate progress. In our view, progress will require careful assessment and evaluation of new methods, force fields, and modeling innovations on well-characterized benchmark systems, and we lay out our vision for how this can be achieved.

A single-walker approach for studying quasi-nonergodic systems

The jump-walking Monte-Carlo algorithm is revisited and updated to study the equilibrium properties of systems exhibiting quasi-nonergodicity. It is designed for a single processing thread as opposed to currently predominant algorithms for large parallel processing systems. The updated algorithm is tested on the Ising model and applied to the lattice-gas model for sorption in aerogel at low temperatures, when dynamics of the system is critically slowed down. It is demonstrated that the updated jump-walking simulations are able to produce equilibrium isotherms which are typically hidden by the hysteresis effect characteristic of the standard single-flip simulations.

Replica exchange enveloping distribution sampling (RE-EDS): A robust method to estimate multiple free-energy differences from a single simulation

In molecular dynamics (MD) simulations, free-energy differences are often calculated using free energy perturbation or thermodynamic integration (TI) methods. However, both techniques are only suited to calculate free-energy differences between two end states. Enveloping distribution sampling (EDS) presents an attractive alternative that allows to calculate multiple free-energy differences in a single simulation. In EDS, a reference state is simulated which "envelopes" the end states. The challenge of this methodology is the determination of optimal reference-state parameters to ensure equal sampling of all end states. Currently, the automatic determination of the reference-state parameters for multiple end states is an unsolved issue that limits the application of the methodology. To resolve this, we have generalised the replica-exchange EDS (RE-EDS) approach, introduced by Lee et al. [J. Chem. Theory Comput. 10, 2738 (2014)] for constant-pH MD simulations. By exchanging configurations between replicas with different reference-state parameters, the complexity of the parameter-choice problem can be substantially reduced. A new robust scheme to estimate the reference-state parameters from a short initial RE-EDS simulation with default parameters was developed, which allowed the calculation of 36 free-energy differences between nine small-molecule inhibitors of phenylethanolamine N-methyltransferase from a single simulation. The resulting free-energy differences were in excellent agreement with values obtained previously by TI and two-state EDS simulations.