Search

TWIK-related potassium channel 1 (TREK1), a two-pore-domain mammalian potassium (K+) channel, regulates the resting potential across cell membranes, presenting a promising therapeutic target for neuropathy treatment. The gating of this channel converges in the conformation of the narrowest part of the pore: the selectivity filter (SF). Various hypotheses explain TREK1 gating modulation, including the dynamics of loops connecting the SF with transmembrane helices and the stability of hydrogen bond (HB) networks adjacent to the SF. Recently, two small molecules (Q6F and Q5F) were reported as activators that affect TREK1 by increasing its open probability in single-channel current measurements. Here, using molecular dynamics simulations, we investigate the effect of these ligands on the previously proposed modulation mechanisms of TREK1 gating compared to the apo channel. Our findings reveal that loop dynamics at the upper region of the SF exhibit only a weak correlation with permeation events/nonpermeation periods, whereas the HB network behind the SF appears more correlated. These nonpermeation periods arise from both distinct mechanisms: a C-type inactivation (resulting from dilation at the top of the SF), which has been described previously, and a carbonyl flipping in an SF binding site. We find that, besides the prevention of C-type inactivation in the channel, the ligands increase the probability of permeation by modulating the dynamics of the carbonyl flipping, influenced by a threonine residue at the bottom of the SF. These results offer insights for rational ligand design to optimize the gating modulation of TREK1 and related K+channels.

The Coulomb interactions in molecular simulations are inherently approximated due to the finite size of the molecular box sizes amenable to current-day compute power. Several methods exist for treating long-range electrostatic interactions, yet these approaches are subject to various finite-size-related artifacts. Lattice-sum methods are frequently used to approximate long-range interactions; however, these approaches also suffer from artifacts which become particularly pronounced for free-energy calculations that involve charge changes. The artifacts, however, also affect the sampling when plain simulations are performed, leading to a biased ensemble. Here, we investigate two previously described model systems to determine if artifacts continue to play a role when overall neutral boxes are considered, in the context of both free-energy calculations and sampling. We find that ensuring that no net-charge changes take place, while maintaining a neutral simulation box, may be sufficient provided that the simulation boxes are large enough. Addition of salt to the solution (when appropriate) can further alleviate the remaining artifacts in the sampling or the calculated free-energy differences. We provide practical guidelines to avoid finite-size artifacts.

Potassium channels are responsible for the selective yet efficient permeation of potassium ions across cell membranes. Despite many available high-resolution structures of potassium channels, those conformations inform only on static information on the ion permeation processes. Here, we use molecular dynamics simulations and Markov state models to obtain dynamical details of ion permeation. The permeation cycles, expressed in terms of selectivity filter occupancy and representing ion permeation events, are illustrated. We show that the direct knock-on permeation represents the dominant permeation mechanism over a wide range of potassium concentrations, temperatures, and membrane voltages for the pore of MthK. Direct knock-on is also observed in other potassium channels with a highly conserved selectivity filter, demonstrating the robustness of the permeation mechanism. Lastly, we investigate the charge strength dependence of permeation cycles. Our results shed light on the underlying permeation details, which are valuable in studying conduction mechanisms in potassium channels.

Drug discovery can be thought of as a search for a needle in a haystack: searching through a large chemical space for the most active compounds. Computational techniques can narrow the search space for experimental follow up, but even they become unaffordable when evaluating large numbers of molecules. Therefore, machine learning (ML) strategies are being developed as computationally cheaper complementary techniques for navigating and triaging large chemical libraries. Here, we explore how an active learning protocol can be combined with first-principles based alchemical free energy calculations to identify high affinity phosphodiesterase 2 (PDE2) inhibitors. We first calibrate the procedure using a set of experimentally characterized PDE2 binders. The optimized protocol is then used prospectively on a large chemical library to navigate toward potent inhibitors. In the active learning cycle, at every iteration a small fraction of compounds is probed by alchemical calculations and the obtained affinities are used to train ML models. With successive rounds, high affinity binders are identified by explicitly evaluating only a small subset of compounds in a large chemical library, thus providing an efficient protocol that robustly identifies a large fraction of true positives.

We assess costs and efficiency of state-of-the-art high-performance cloud computing and compare the results to traditional on-premises compute clusters. Our use case is atomistic simulations carried out with the GROMACS molecular dynamics (MD) toolkit with a particular focus on alchemical protein–ligand binding free energy calculations. We set up a compute cluster in the Amazon Web Services (AWS) cloud that incorporates various different instances with Intel, AMD, and ARM CPUs, some with GPU acceleration. Using representative biomolecular simulation systems, we benchmark how GROMACS performs on individual instances and across multiple instances. Thereby we assess which instances deliver the highest performance and which are the most cost-efficient ones for our use case. We find that, in terms of total costs, including hardware, personnel, room, energy, and cooling, producing MD trajectories in the cloud can be about as cost-efficient as an on-premises cluster given that optimal cloud instances are chosen. Further, we find that high-throughput ligand-screening can be accelerated dramatically by using global cloud resources. For a ligand screening study consisting of 19 872 independent simulations or ∼200 μs of combined simulation trajectory, we made use of diverse hardware available in the cloud at the time of the study. The computations scaled-up to reach peak performance using more than 4 000 instances, 140 000 cores, and 3 000 GPUs simultaneously. Our simulation ensemble finished in about 2 days in the cloud, while weeks would be required to complete the task on a typical on-premises cluster consisting of several hundred nodes.

The flow of ions across cell membranes facilitated by ion channels is an important function for all living cells. Despite the huge amount of structural data provided by crystallography, elucidating the exact interactions between the selectivity filter atoms and bound ions is challenging. Here, we detect bound 15N-labeled ammonium ions as a mimic for potassium ions in ion channels using solid-state NMR under near-native conditions. The non-selective ion channel NaK showed two ammonium peaks corresponding to its two ion binding sites, while its potassium-selective mutant NaK2K that has a signature potassium-selective selectivity filter with four ion binding sites gave rise to four ammonium peaks. Ions bound in specific ion binding sites were identified based on magnetization transfer between the ions and carbon atoms in the selectivity filters. Magnetization transfer between bound ions and water molecules revealed that only one out of four ions in the selectivity filter of NaK2K is in close contact with water, which is in agreement with the direct knock-on ion conduction mechanism where ions are conducted through the channel by means of direct interactions without water molecules in between. Interestingly, the potassium-selective ion channels investigated here (NaK2K and, additionally, KcsA-Kv1.3) showed remarkably different chemical shifts for their bound ions, despite having identical amino acid sequences and crystal structures of their selectivity filters. Molecular dynamics simulations show similar ion binding and conduction behavior between ammonium and potassium ions and identify the origin of the differences between the investigated potassium channels.

Biliverdin IXβ reductase B (BLVRB) has recently been proposed as a novel therapeutic target for thrombocytopenia through its reactive oxygen species (ROS)-associated mechanism. Thus, we aim at repurposing drugs as new inhibitors of BLVRB. Based on IC50(<5 μM), we have identified 20 compounds out of 1496 compounds from the Food and Drug Administration (FDA)-approved library and have clearly mapped their binding sites to the active site. Furthermore, we show the detailed BLVRB-binding modes and thermodynamic properties (ΔH, ΔS, and KD) with nuclear magnetic resonance (NMR) and isothermal titration calorimetry together with complex structures of eight water-soluble compounds. We anticipate that the results will serve as a novel platform for further in-depth studies on BLVRB effects for related functions such as ROS accumulation and megakaryocyte differentiation, and ultimately treatments of platelet disorders.

Binding free energy calculations have become increasingly valuable to drive decision making in drug discovery projects. However, among other issues, inadequate sampling can reduce accuracy, limiting the value of the technique. In this paper, we apply absolute binding free energy calculations to ligands binding to T4 lysozyme L99A and HSP90 using equilibrium and nonequilibrium approaches. We highlight sampling problems encountered in these systems, such as slow side chain rearrangements and slow changes of water placement upon ligand binding. These same types of challenges are also likely to show up in other protein–ligand systems, and we propose some strategies to diagnose and test for such problems in alchemical free energy calculations. We also explore similarities and differences in how the equilibrium and the nonequilibrium approaches handle these problems. Our results show the large amount of work still to be done to make free energy calculations robust and reliable and provide insight for future research in this area.

In the numerous molecular recognition and catalytic processes across biochemistry involving adenosine triphosphate (ATP), the common bioactive form is its magnesium chelate, ATP·Mg2+. In aqueous solution, two chelation geometries predominate, distinguished by bidentate and tridentate Mg2+–phosphate coordination. These are approximately isoenergetic but separated by a high energy barrier. Force field-based atomistic simulation studies of this complex require an accurate representation of its structure and energetics. Here we focused on the energetics of ATP·Mg2+coordination. Applying an enhanced sampling scheme to circumvent prohibitively slow sampling of transitions between coordination modes, we observed striking contradictions between Amber and CHARMM force field descriptions, most prominently in opposing predictions of the favored coordination mode. Through further configurational free energy calculations, conducted against a diverse set of ATP·Mg2+–protein complex structures to supplement otherwise limited experimental data, we quantified systematic biases for each force field. The force field calculations were strongly predictive of experimentally observed coordination modes, enabling additive corrections to the coordination free energy that deliver close agreement with experiment. We reassessed the applicability of the thus corrected force field descriptions of ATP·Mg2+for biomolecular simulation and observed that, while the CHARMM parameters display an erroneous preference for overextended triphosphate configurations that will affect many common biomolecular simulation applications involving ATP, the force field energy landscapes broadly agree with experimental measurements of solution geometry and the distribution of ATP·Mg2+structures found in the Protein Data Bank. Our force field evaluation and correction approach, based on maximizing consistency with the large and heterogeneous collection of structural information encoded in the PDB, should be broadly applicable to many other systems.

Mechanosensitive channels sense mechanical forces in cell membranes and underlie many biological sensing processes1–3. However, how exactly they sense mechanical force remains under investigation4. The bacterial mechanosensitive channel of small conductance, MscS, is one of the most extensively studied mechanosensitive channels4–8, but how it is regulated by membrane tension remains unclear, even though the structures are known for its open and closed states9–11. Here we used cryo-electron microscopy to determine the structure of MscS in different membrane environments, including one that mimics a membrane under tension. We present the structures of MscS in the subconducting and desensitized states, and demonstrate that the conformation of MscS in a lipid bilayer in the open state is dynamic. Several associated lipids have distinct roles in MscS mechanosensation. Pore lipids are necessary to prevent ion conduction in the closed state. Gatekeeper lipids stabilize the closed conformation and dissociate with membrane tension, allowing the channel to open. Pocket lipids in a solvent-exposed pocket between subunits are pulled out under sustained tension, allowing the channel to transition to the subconducting state and then to the desensitized state. Our results provide a mechanistic underpinning and expand on the ‘force-from-lipids’ model for MscS mechanosensation4,11.

The natural antivitamin 2'-methoxy-thiamine (MTh) is implicated in the suppression of microbial growth. However, its mode of action and enzyme-selective inhibition mechanism have remained elusive. Intriguingly, MTh inhibits some thiamine diphosphate (ThDP) enzymes, while being coenzymatically active in others. Here we report the strong inhibition of Escherichia colitransketolase activity by MTh and unravel its mode of action and the structural basis thereof. The unique 2'-methoxy group of MTh diphosphate (MThDP) clashes with a canonical glutamate required for cofactor activation in ThDP-dependent enzymes. This glutamate is forced into a stable, anticatalytic low-barrier hydrogen bond with a neighboring glutamate, disrupting cofactor activation. Molecular dynamics simulations of transketolases and other ThDP enzymes identify active-site flexibility and the topology of the cofactor-binding locale as key determinants for enzyme-selective inhibition. Human enzymes either retain enzymatic activity with MThDP or preferentially bind authentic ThDP over MThDP, while core bacterial metabolic enzymes are inhibited, demonstrating therapeutic potential.

Sequence-specific DNA binding transcription factors play an essential role in the transcriptional regulation of all organisms. The development of reliable in silico methods to predict the binding affinity landscapes of transcription factors thus promises to provide rapid screening of transcription factor specificities and, at the same time, yield valuable insight into the atomistic details of the interactions driving those specificities. Recent literature has reported highly discrepant results on the current ability of state-of-the-art atomistic molecular dynamics simulations to reproduce experimental binding free energy landscapes for transcription factors. Here, we resolve one important discrepancy by noting that in the case of alchemical free energy calculations involving base pair mutations, a common convention used in improving end point convergence of mixed potentials in fact can lead to erroneous results. The underlying cause for inaccurate double free energy difference estimates is specific to the particular implementation of the alchemical transformation protocol. Using the Gromacs simulation package, which is not affected by this issue, we obtain free energy landscapes in agreement with the experimental measurements; equivalent results are obtained for a small set of test cases with a modified version of the AMBER package. Our findings provide a consistent and optimistic outlook on the current state of prediction of protein−DNA binding free energy interactions using molecular dynamics simulations and an important precaution for appropriate end point handling in a broad range of free energy calculations.

In drug discovery, the in silico prediction of binding affinity is one of the major means to prioritize compounds for synthesis. Alchemical relative binding free energy (RBFE) calculations based on molecular dynamics (MD) simulations are nowadays a popular approach for the accurate affinity ranking of compounds. MD simulations rely on empirical force field parameters, which strongly influence the accuracy of the predicted affinities. Here, we evaluate the ability of six different small-molecule force fields to predict experimental protein–ligand binding affinities in RBFE calculations on a set of 598 ligands and 22 protein targets. The public force fields OpenFF Parsley and Sage, GAFF, and CGenFF show comparable accuracy, while OPLS3e is significantly more accurate. However, a consensus approach using Sage, GAFF, and CGenFF leads to accuracy comparable to OPLS3e. While Parsley and Sage are performing comparably based on aggregated statistics across the whole dataset, there are differences in terms of outliers. Analysis of the force field reveals that improved parameters lead to significant improvement in the accuracy of affinity predictions on subsets of the dataset involving those parameters. Lower accuracy can not only be attributed to the force field parameters but is also dependent on input preparation and sampling convergence of the calculations. Especially large perturbations and nonconverged simulations lead to less accurate predictions. The input structures, Gromacs force field files, as well as the analysis Python notebooks are available on GitHub.

Resistance to small molecule drugs often emerges in cancer cells, viruses, and bacteria as a result of the evolutionary pressure exerted by the therapy. Protein mutations that directly impair drug binding are frequently involved in resistance, and the ability to anticipate these mutations would be beneficial in drug development and clinical practice. Here, we evaluate the ability of three distinct computational methods to predict ligand binding affinity changes upon protein mutation for the cancer target Abl kinase. These structure-based approaches rely on first-principle statistical mechanics, mixed physics- and knowledge-based potentials, and machine learning, and were able to estimate binding affinity changes and identify resistant mutations with remarkable accuracy. We expect that these complementary approaches will enable the routine prediction of resistance-causing mutations in a variety of other target proteins.

Antimicrobial peptides (AMPs) carry great potential as new antibiotics against “superbugs.” Dermcidin (DCD), a broad-spectrum AMP in human sweat, has been recently crystallized in its oligomeric state and showed channel-like properties. In this work, we performed multiscale molecular dynamics simulations to study how the membrane composition influences the behavior of a transmembrane pore formed by the DCD oligomer in the hope of revealing the origin of the membrane selectivity of this AMP toward bacteria. Our results indicate that bilayers composed of various lipids (DMPC, DPPC, and DSPC) with different thicknesses result in different orientations of the DCD oligomer when embedded in lipid bilayers. The thicker the bilayer, the less tilted the channel. Cholesterol makes the bilayers more rigid and thicker, which also affects the orientation of the channel. Furthermore, we observed that the predicted conductance of the channel from computational electrophysiology simulations is related to its orientation in the lipid bilayer: the larger the tilt, the larger the conductance. Our results indicate that the membrane composition has a significant influence on the activity of the DCD channel, with thicker, cholesterol-rich membranes showing lower conductance than that of thinner membranes.

NMR relaxation dispersion methods provide a holistic way to observe microsecond time-scale protein backbone motion both in solution and in the solid state. Different nuclei (1H and 15N) and different relaxation dispersion techniques (Bloch–McConnell and near-rotary-resonance) give complementary information about the amplitudes and time scales of the conformational dynamics and provide comprehensive insights into the mechanistic details of the structural rearrangements. In this paper, we exemplify the benefits of the combination of various solution- and solid-state relaxation dispersion methods on a microcrystalline protein (α-spectrin SH3 domain), for which we are able to identify and model the functionally relevant conformational rearrangements around the ligand recognition loop occurring on multiple microsecond time scales. The observed loop motions suggest that the SH3 domain exists in a binding-competent conformation in dynamic equilibrium with a sterically impaired ground-state conformation both in solution and in crystalline form. This inherent plasticity between the interconverting macrostates is compatible with a conformational-preselection model and provides new insights into the recognition mechanisms of SH3 domains.

We introduce a computational toolset, named GROmaρs, to obtain and compare time-averaged density maps from molecular dynamics simulations. GROmaρs efficiently computes density maps by fast multi-Gaussian spreading of atomic densities onto a three-dimensional grid. It complements existing map-based tools by enabling spatial inspection of atomic average localization during the simulations. Most importantly, it allows the comparison between computed and reference maps (e.g., experimental) through calculation of difference maps and local and time-resolved global correlation. These comparison operations proved useful to quantitatively contrast perturbed and control simulation data sets and to examine how much biomolecular systems resemble both synthetic and experimental density maps. This was especially advantageous for multimolecule systems in which standard comparisons like RMSDs are difficult to compute. In addition, GROmaρs incorporates absolute and relative spatial free-energy estimates to provide an energetic picture of atomistic localization. This is an open-source GROMACS-based toolset, thus allowing for static or dynamic selection of atoms or even coarse-grained beads for the density calculation. Furthermore, masking of regions was implemented to speed up calculations and to facilitate the comparison with experimental maps. Beyond map comparison, GROmaρs provides a straightforward method to detect solvent cavities and average charge distribution in biomolecular systems. We employed all these functionalities to inspect the localization of lipid and water molecules in aquaporin systems, the binding of cholesterol to the G protein coupled chemokine receptor type 4, and the identification of permeation pathways through the dermicidin antimicrobial channel. Based on these examples, we anticipate a high applicability of GROmaρs for the analysis of molecular dynamics simulations and their comparison with experimentally determined densities.

An atomistically detailed picture of protein folding at interfaces can effectively be obtained by comparing interface-sensitive spectroscopic techniques to molecular simulations. Here, we present an extensive evaluation of the capability of contemporary force fields to model protein folding at air–water interfaces with a general scheme for sampling and reweighting theoretical conformational ensembles of interfacial peptides. Force field combinations of CHARMM22*/TIP3P and AMBER99SB*-ILDN/SPC/E were found to reproduce experimental observations best.

The design of proteins with novel ligand-binding functions holds great potential for application in biomedicine and biotechnology. However, our ability to engineer ligand-binding proteins is still limited, and current approaches rely primarily on experimentation. Computation could reduce the cost of the development process and would allow rigorous testing of our understanding of the principles governing molecular recognition. While computational methods have proven successful in the early stages of the discovery process, optimization approaches that can quantitatively predict ligand affinity changes upon protein mutation are still lacking. Here, we assess the ability of free energy calculations based on first-principles statistical mechanics, as well as the latest Rosetta protocols, to quantitatively predict such affinity changes on a challenging set of 134 mutations. After evaluating different protocols with computational efficiency in mind, we investigate the performance of different force fields. We show that both the free energy calculations and Rosetta are able to quantitatively predict changes in ligand binding affinity upon protein mutations, yet the best predictions are the result of combining the estimates of both methods. These closely match the experimentally determined ΔΔGvalues, with a root-mean-square error of 1.2 kcal/mol for the full benchmark set and of 0.8 kcal/mol for a subset of protein systems providing the most reproducible results. The currently achievable accuracy offers the prospect of being able to employ computation for the optimization of ligand-binding proteins as well as the prediction of drug resistance.

A large number of proteins assemble as homooligomers. These homooligomers accomplish their function either symmetrically or asymmetrically. If asymmetry is prevalent in a structure ensemble, the asymmetric motion will occur in any of the subunits. Many computational analysis tools implicitly use ensemble averages to determine protein motions, e.g., principle component analysis. Therefore, taken together, this approach results in a loss of the asymmetric signal and a false symmetric output, rendering it impossible to analyze asymmetric motions with available tools. A first step toward understanding asymmetric systems is the quantification of asymmetry. Only a few tools exist to calculate asymmetry quantitatively, such as the continuous symmetry measure (CSM). In this study, we present an extension of CSM delivering additional information about the subunit contributions to the overall asymmetry. Furthermore, we introduce an algorithm termed the functional asymmetry measure (FAME). FAME is based on an algorithm that predicts functionally relevant motions of a protein (PLS-FMA) and thus allows calculating asymmetry in relation to protein function. To validate our developed algorithm, we applied it to two different potassium channels, TREK-2 and KcsA, as well as to the unfolding mechanism of the carrier protein Transthyretin. For both potassium channel systems, an artificial asymmetric motion was introduced to benchmark the algorithm in addition to demonstrate the interpretation potential of the results. Therefore, the degree of overall as well as subunit based asymmetry for KcsA was quantified using CSM as the provided extension requires more than two subunits. The functional modes of asymmetric TREK-2 motions were recovered and their asymmetry was quantified using FAME as a dimeric protein is the simplest application. FAME was further used to study the asymmetry of the unfolding pathway of Transthyretin. We show the ability of both algorithms to correctly predict asymmetry. The tools are available online and can be applied to most homooligomeric systems.

Despite a large number of antiretroviral drugs targeting HIV-1 protease for inhibition, mutations in this protein during the course of patient treatment can render them inefficient. This emerging resistance inspired numerous computational studies of the HIV-1 protease aimed at predicting the effect of mutations on drug binding in terms of free binding energy ΔG, as well as in mechanistic terms. In this study, we analyze ten different protease-inhibitor complexes carrying major resistance-associated mutations (RAMs) G48V, I50V, and L90M using molecular dynamics simulations. We demonstrate that alchemical free energy calculations can consistently predict the effect of mutations on drug binding. By explicitly probing different protonation states of the catalytic aspartic dyad, we reveal the importance of the correct choice of protonation state for the accuracy of the result. We also provide insight into how different mutations affect drug binding in their specific ways, with the unifying theme of how all of them affect the crucial drug binding regions of the protease.

The mechanism of mechanosensitive gating of ion channels underlies many physiological processes, including the sensations of touch, hearing, and pain perception. TREK-2 is the best-studied mechanosensitive member of the two-pore domain potassium channel family. Apart from pressure sensing, it responds to a diverse range of stimuli. Two states, termed “up” and “down,” are known from x-ray structural crystallographic studies and have been suggested to differ in conductance. However, the structural details of the gating behavior are largely unknown. In this work, we used molecular dynamics simulations to study the conductance of the states as well as the effect of mechanical membrane stretch on the channel. We find that the down state is less conductive than the up state. The introduction of membrane stretch in the simulations shifts the state of the channel toward an up configuration, independent of the starting configuration, and also increases its conductance. The correlation of the selectivity filter state and the conductance supports a model in which the selectivity filter gates by a carbonyl flip. This gate is stabilized by the pore helices. We suggest a modulation of these helices by an interface to the transmembrane helices. Membrane pressure changes the conformation of the transmembrane helices directly and consequently also influences the channel conductance.

Nucleotide-sequence-dependent interactions between proteins and DNA are responsible for a wide range of gene regulatory functions. Accurate and generalizable methods to evaluate the strength of protein–DNA binding have long been sought. While numerous computational approaches have been developed, most of them require fitting parameters to experimental data to a certain degree, e.g., machine learning algorithms or knowledge-based statistical potentials. Molecular-dynamics-based free energy calculations offer a robust, system-independent, first-principles-based method to calculate free energy differences upon nucleotide mutation. We present an automated procedure to set up alchemical MD-based calculations to evaluate free energy changes occurring as the result of a nucleotide mutation in DNA. We used these methods to perform a large-scale mutation scan comprising 397 nucleotide mutation cases in 16 protein–DNA complexes. The obtained prediction accuracy reaches 5.6 kJ/mol average unsigned deviation from experiment with a correlation coefficient of 0.57 with respect to the experimentally measured free energies. Overall, the first-principles-based approach performed on par with the molecular modeling approaches Rosetta and FoldX. Subsequently, we utilized the MD-based free energy calculations to construct protein–DNA binding profiles for the zinc finger protein Zif268. The calculation results compare remarkably well with the experimentally determined binding profiles. The software automating the structure and topology setup for alchemical calculations is a part of the pmx package; the utilities have also been made available online at http://pmx.mpibpc.mpg.de/dna_webserver.html.

Electrical cell signaling requires adjustment of ion channel, receptor, or transporter function in response to changes in membrane potential. For the majority of such membrane proteins, the molecular details of voltage sensing remain insufficiently understood. Here, we present a molecular dynamics simulation-based method to determine the underlying charge movement across the membrane—the gating charge—by measuring electrical capacitor properties of membrane-embedded proteins. We illustrate the approach by calculating the charge transfer upon membrane insertion of the HIV gp41 fusion peptide, and validate the method on two prototypical voltage-dependent proteins, the Kv1.2 K+channel and the voltage sensor of the Ciona intestinalisvoltage-sensitive phosphatase, against experimental data. We then use the gating charge analysis to study how the T1 domain modifies voltage sensing in Kv1.2 channels and to investigate the voltage dependence of the initial binding of two Na+ions in Na+-coupled glutamate transporters. Our simulation approach quantifies various mechanisms of voltage sensing, enables direct comparison with experiments, and supports mechanistic interpretation of voltage sensitivity by fractional amino acid contributions.

Compared to other aquaporins (AQPs), lens-specific AQP0 is a poor water channel, and its permeability was reported to be pH-dependent. To date, most water conduction studies on AQP0 were performed on protein expressed in Xenopusoocytes, and the results may therefore also reflect effects introduced by the oocytes themselves. Experiments with purified AQP0 reconstituted into liposomes are challenging because the water permeability of AQP0 is only slightly higher than that of pure lipid bilayers. By reconstituting high amounts of AQP0 and using high concentrations of cholesterol to reduce the permeability of the lipid bilayer, we improved the signal-to-noise ratio of water permeability measurements on AQP0 proteoliposomes. Our measurements show that mutation of two pore-lining tyrosine residues, Tyr-23 and Tyr-149 in sheep AQP0, to the corresponding residues in the high-permeability water channel AQP1 have additive effects and together increase the water permeability of AQP0 40-fold to a level comparable to that of AQP1. Molecular dynamics simulations qualitatively support these experimental findings and suggest that mutation of Tyr-23 changes the pore profile at the gate formed by residue Arg-187.

The stability, solubility, and function of a protein depend on both its net charge and the protonation states of its individual residues. pKais a measure of the tendency for a given residue to (de)protonate at a specific pH. Although pKavalues can be resolved experimentally, theory and computation provide a compelling alternative. To this end, we assess the applicability of a nonequilibrium (NEQ) alchemical free energy method to the problem of pKaprediction. On a data set of 144 residues that span 13 proteins, we report an average unsigned error of 0.77 ± 0.09, 0.69 ± 0.09, and 0.52 ± 0.04 pKfor aspartate, glutamate, and lysine, respectively. This is comparable to current state-of-the-art predictors and the accuracy recently reached using free energy perturbation methods (e.g., FEP+). Moreover, we demonstrate that our open-source, pmx-based approach can accurately resolve the pKavalues of coupled residues and observe a substantial performance disparity associated with the lysine partial charges in Amber14SB/Amber99SB*-ILDN, for which an underused fix already exists.

The voltage-dependent anion channel 1 (VDAC-1) is an important protein of the outer mitochondrial membrane that transports energy metabolites and is involved in apoptosis. The available structures of VDAC proteins show a wide β-stranded barrel pore, with its N-terminal α-helix (N-α) bound to its interior. Electrophysiology experiments revealed that voltage, its polarity, and membrane composition modulate VDAC currents. Experiments with VDAC-1 mutants identified amino acids that regulate the gating process. However, the mechanisms for how these factors regulate VDAC-1, and which changes they trigger in the channel, are still unknown. In this study, molecular dynamics simulations and single-channel experiments of VDAC-1 show agreement for the current-voltage relationships of an “open” channel and they also show several subconducting transient states that are more cation selective in the simulations. We observed voltage-dependent asymmetric distortions of the VDAC-1 barrel and the displacement of particular charged amino acids. We constructed conformational models of the protein voltage response and the pore changes that consistently explain the protein conformations observed at opposite voltage polarities, either in phosphatidylethanolamine or phosphatidylcholine membranes. The submicrosecond VDAC-1 voltage response shows intrinsic structural changes that explain the role of key gating amino acids and support some of the current gating hypotheses. These voltage-dependent protein changes include asymmetric barrel distortion, its interaction with the membrane, and significant displacement of N-αamino acids.

Intrinsically disordered proteins (IDPs) are notoriously challenging to study both experimentally and computationally. The structure of IDPs cannot be described by a single conformation but must instead be described as an ensemble of interconverting conformations. Atomistic simulations are increasingly used to obtain such IDP conformational ensembles. Here, we have compared the IDP ensembles generated by eight all-atom empirical force fields against primary small-angle X-ray scattering (SAXS) and NMR data. Ensembles obtained with different force fields exhibit marked differences in chain dimensions, hydrogen bonding, and secondary structure content. These differences are unexpectedly large: changing the force field is found to have a stronger effect on secondary structure content than changing the entire peptide sequence. The CHARMM 22* ensemble performs best in this force field comparison: it has the lowest error in chemical shifts and J-couplings and agrees well with the SAXS data. A high population of left-handed α-helix is present in the CHARMM 36 ensemble, which is inconsistent with measured scalar couplings. To eliminate inadequate sampling as a reason for differences between force fields, extensive simulations were carried out (0.964 ms in total); the remaining small sampling uncertainty is shown to be much smaller than the observed differences. Our findings highlight how IDPs, with their rugged energy landscapes, are highly sensitive test systems that are capable of revealing force field deficiencies and, therefore, contributing to force field development.

Conformational selection is an established mechanism in molecular recognition. Despite its power to explain binding events, it is hardly used in protein/ligand design to modulate molecular recognition. Here, we explore the opportunities and limitations of design by conformational selection. Using appropriate thermodynamic cycles, our approach predicts the effects of a conformational shift on binding affinity and also allows one to disentangle the effects induced by a conformational shift from other effects influencing the binding affinity. The method is assessed and applied to explain the contribution of a conformational shift on the binding affinity of six ubiquitin mutants showing different conformational shifts in six different complexes.

The all-atom additive CHARMM36 protein force field is widely used in molecular modeling and simulations. We present its refinement, CHARMM36m (http://mackerell.umaryland.edu/charmm_ff.shtml), with improved accuracy in generating polypeptide backbone conformational ensembles for intrinsically disordered peptides and proteins.

Ewald summation, which has become the de factostandard for computing electrostatic interactions in biomolecular simulations, formally requires that the simulation box is neutral. For non-neutral systems, the Ewald algorithm implicitly introduces a uniform background charge distribution that effectively neutralizes the simulation box. Because a uniform distribution of counter charges typically deviates from the spatial distribution of counterions in real systems, artifacts may arise, in particular in systems with an inhomogeneous dielectric constant. Here, we derive an analytical expression for the effect of using an implicit background charge instead of explicit counterions, on the chemical potential of ions in heterogeneous systems, which (i) provides a quantitative criterium for deciding if the background charge offers an acceptable trade-off between artifacts arising from sampling problems and artifacts arising from the homogeneous background charge distribution, and (ii) can be used to correct this artifact in certain cases. Our model quantifies the artifact in terms of the differencein charge density between the non-neutral system with a uniform neutralizing background charge and the real neutral system with a physically correct distribution of explicit counterions. We show that for inhomogeneous systems, such as proteins and membranes in water, the artifact manifests itself by an overstabilization of ions inside the lower dielectric by tens to even hundreds kilojoules per mole. We have tested the accuracy of our model in molecular dynamics simulations and found that the error in the calculated free energy for moving a test charge from water into hexadecane, at different net charges of the system and different simulation box sizes, is correctly predicted by the model. The calculations further confirm that the incorrect distribution of counter charges in the simulation box is solely responsible for the errors in the PMFs.

Human aquaporin-1 (hAQP1) is a water channel found in many tissues and potentially involved in several human pathologies. Selective inhibitors of hAQP1 are discussed as novel treatment opportunities for glaucoma, brain edema, inflammatory pain, and certain types of cancer. However, only very few potent and chemically attractive blockers have been reported to date. In this study we present three novel hAQP1 blockers that have been identified by virtual screening and inhibit water flux through hAQP1 in Xenopus laevisoocyte swelling assays at low micromolar concentrations. The newly discovered compounds display no chemical similarity to hitherto known hAQP1 blockers and bind at the extracellular entrance of the channel, close to the ar/R selectivity filter. Futhermore, mutagenesis studies showed that Lys36, which is not conserved among the hAQP family, is crucially involved in binding and renders the discovered compounds suitable as leads for the development of selective hAQP1 inhibitors.

Analysis of the internal dynamics of a biological molecule requires the successful removal of overall translation and rotation. Particularly for flexible or intrinsically disordered peptides, this is a challenging task due to the absence of a well-defined reference structure that could be used for superpositioning. In this work, we started the analysis with a widely known formulation of an objective for the problem of superimposing a set of multiple molecules as variance minimization over an ensemble. A negative effect of this superpositioning method is the introduction of ambiguous rotations, where different rotation matrices may be applied to structurally similar molecules. We developed two algorithms to resolve the suboptimal rotations. The first approach minimizes the variance together with the distance of a structure to a preceding molecule in the ensemble. The second algorithm seeks for minimal variance together with the distance to the nearest neighbors of each structure. The newly developed methods were applied to molecular-dynamics trajectories and normal-mode ensembles of the Aβpeptide, RS peptide, and lysozyme. These new (to our knowledge) superpositioning methods combine the benefits of variance and distance between nearest-neighbor(s) minimization, providing a solution for the analysis of intrinsic motions of flexible molecules and resolving ambiguous rotations.

Water permeability through single-file channels is affected by intrinsic factors such as their size and polarity and by external determinants like their lipid environment in the membrane. Previous computational studies revealed that the obstruction of the channel by lipid headgroups can be long-lived, in the range of nanoseconds, and that pore-length-matching membrane mimetics could speed up water permeability. To test the hypothesis of lipid-channel interactions modulating channel permeability, we designed different gramicidin A derivatives with attached acyl chains. By combining extensive molecular-dynamics simulations and single-channel water permeation measurements, we show that by tuning lipid-channel interactions, these modifications reduce the presence of lipid headgroups in the pore, which leads to a clear and selective increase in their water permeability.

We introduce an approach based on the recently introduced functional mode analysis to identify collective modes of internal dynamics that maximally correlate to an external order parameter of functional interest. Input structural data can be either experimentally determined structure ensembles or simulated ensembles, such as molecular dynamics trajectories. Partial least-squares regression is shown to yield a robust solution to the multidimensional optimization problem, with a minimal and controllable risk of overfitting, as shown by extensive cross-validation. Several examples illustrate that the partial least-squares-based functional mode analysis successfully reveals the collective dynamics underlying the fluctuations in selected functional order parameters. Applications to T4 lysozyme, the Trp-cage, the aquaporin channels Aqy1 and hAQP1, and the CLC-ec1 chloride antiporter are presented in which the active site geometry, the hydrophobic solvent-accessible surface, channel gating dynamics, water permeability (pf), and a dihedral angle are defined as functional order parameters. The Aqy1 case reveals a gating mechanism that connects the inner channel gating residues with the protein surface, thereby providing an explanation of how the membrane may affect the channel. hAQP1 shows how the pfcorrelates with structural changes around the aromatic/arginine region of the pore. The CLC-ec1 application shows how local motions of the gating Glu148couple to a collective motion that affects ion affinity in the pore.

The fields of rational drug design and protein engineering benefit from accurate free energy calculations based on molecular dynamics simulations. A thermodynamic integration scheme is often used to calculate changes in the free energy of a system by integrating the change of the system’s Hamiltonian with respect to a coupling parameter. These methods exploit nonphysical pathways over thermodynamic cycles involving particle introduction and annihilation. Such alchemical transitions require the modification of the classical nonbonded potential energy terms by applying soft-core potential functions to avoid singularity points. In this work, we propose a novel formulation for a soft-core potential to be applied in nonequilibrium free energy calculations that alleviates singularities, numerical instabilities, and additional minima in the potential energy for all combinations of nonbonded interactions at all intermediate alchemical states. The method was validated by application to (a) the free energy calculations of a closed thermodynamic cycle, (b) the mutation influence on protein thermostability, (c) calculations of small ligand solvation free energies, and (d) the estimation of binding free energies of trypsin inhibitors. The results show that the novel soft-core function provides a robust and accurate general purpose solution to alchemical free energy calculations.

Mit vereinten Kräften: Die Lebensdauer der Umwandlung von Grundzuständen eines Proteinensembles konnte nur in einem Zeitfenster, das von 4 ns bis 50 μs reichte, bestimmt werden. Durch Kombination temperaturabhängiger kernmagnetischer Relaxationsdispersionsexperimente (RD) und dielektrischer Relaxationsspektroskopie (DR) in Lösung wurde bei 309 K für die Lebensdauer nun ein Wert von (10±9) μs bestimmt.

A good combination: The lifetime of interconversion among the ground states of a protein ensemble representation could only be assigned to a time window that is four orders of magnitude large and ranges from 4 ns to 50 μs. By combining temperature‐dependent NMR relaxation dispersion (RD) experiments and dielectric relaxation (DR) spectroscopy in solution, the lifetime was now identified to a value of (10±9) μs at 309 K.

Presently, most simulations of ion channel function rely upon nonatomistic Brownian dynamics calculations, indirect interpretation of energy maps, or application of external electric fields. We present a computational method to directly simulate ion flux through membrane channels based on biologically realistic electrochemical gradients. In close analogy to single-channel electrophysiology, physiologically and experimentally relevant timescales are achieved. We apply our method to the bacterial channel PorB from pathogenic Neisseria meningitidis, which, during Neisserial infection, inserts into the mitochondrial membrane of target cells and elicits apoptosis by dissipating the membrane potential. We show that our method accurately predicts ion conductance and selectivity and elucidates ion conduction mechanisms in great detail. Handles for overcoming channel-related antibiotic resistance are identified.

The Weighted Histogram Analysis Method (WHAM) is a standard technique used to compute potentials of mean force (PMFs) from a set of umbrella sampling simulations. Here, we present a new WHAM implementation, termed g_wham, which is distributed freely with the GROMACS molecular simulation suite. g_wham estimates statistical errors using the technique of bootstrap analysis. Three bootstrap methods are supported: (i) bootstrapping new trajectories based on the umbrella histograms, (ii) bootstrapping of complete histograms, and (iii) Bayesian bootstrapping of complete histograms, that is, bootstrapping via the assignment of random weights to the histograms. Because methods ii and iii consider only complete histograms as independent data points, these methods do not require the accurate calculation of autocorrelation times. We demonstrate that, given sufficient sampling, bootstrapping new trajectories allows for an accurate error estimate. In the presence of long autocorrelations, however, (Bayesian) bootstrapping of complete histograms yields a more reliable error estimate, whereas bootstrapping of new trajectories may underestimate the error. In addition, we emphasize that the incorporation of autocorrelations into WHAM reduces the bias from limited sampling, in particular, when computing periodic PMFs in inhomogeneous systems such as solvated lipid membranes or protein channels.

Protein dynamics on the atomic level and on the microsecond timescale has recently become accessible from both computation and experiment. To validate molecular dynamics (MD) at the submicrosecond timescale against experiment we present microsecond MD simulations in 10 different force-field configurations for two globular proteins, ubiquitin and the gb3 domain of protein G, for which extensive NMR data is available. We find that the reproduction of the measured NMR data strongly depends on the chosen force field and electrostatics treatment. Generally, particle-mesh Ewald outperforms cut-off and reaction-field approaches. A comparison to measured J-couplings across hydrogen bonds suggests that there is room for improvement in the force-field description of hydrogen bonds in most modern force fields. Our results show that with current force fields, simulations beyond hundreds of nanoseconds run an increased risk of undergoing transitions to nonnative conformational states or will persist within states of high free energy for too long, thus skewing the obtained population frequencies. Only for the AMBER99sb force field have such transitions not been observed. Thus, our results have significance for the interpretation of data obtained with long MD simulations, for the selection of force fields for MD studies and for force-field development. We hope that this comprehensive benchmark based on NMR data applied to many popular MD force fields will serve as a useful resource to the MD community. Finally, we find that for gb3, the force-field AMBER99sb reaches comparable accuracy in back-calculated residual dipolar couplings and J-couplings across hydrogen bonds to ensembles obtained by refinement against NMR data.

Thermal stability of proteins is crucial for both biotechnological and therapeutic applications. Rational protein engineering therefore frequently aims at increasing thermal stability by introducing stabilizing mutations. The accurate prediction of the thermodynamic consequences caused by mutations, however, is highly challenging as thermal stability changes are caused by alterations in the free energy of folding. Growing computational power, however, increasingly allows us to use alchemical free energy simulations, such as free energy perturbation or thermodynamic integration, to calculate free energy differences with relatively high accuracy. In this article, we present an automated protocol for setting up alchemical free energy calculations for mutations of naturally occurring amino acids (except for proline) that allows an unprecedented, automated screening of large mutant libraries. To validate the developed protocol, we calculated thermodynamic stability differences for 109 mutations in the microbial Ribonuclease Barnase. The obtained quantitative agreement with experimental data illustrates the potential of the approach in protein engineering and design.

We present a systematic study directed toward the secondary structure propensity and sampling behavior in peptide folding simulations with eight different molecular dynamics force-field variants in explicit solvent. We report on the combinational result of force field, water model, and electrostatic interaction schemes and compare to available experimental characterization of five studied model peptides in terms of reproduced structure and dynamics. The total simulation time exceeded 18 μs and included simulations that started from both folded and extended conformations. Despite remaining sampling issues, a number of distinct trends in the folding behavior of the peptides emerged. Pronounced differences in the propensity of finding prominent secondary structure motifs in the different applied force fields suggest that problems point in particular to the balance of the relative stabilities of helical and extended conformations.

Naturally occurring pores show a variety of polarities and sizes that are presumably directly linked to their biological function. Many biological channels are selective toward permeants similar or smaller in size than water molecules, and therefore their pores operate in the regime of single-file water pores. Intrinsic factors affecting water permeability through such pores include the channel-membrane match, the structural stability of the channel, the channel geometry and channel-water affinity. We present an extensive molecular dynamics study on the role of the channel geometry and polarity on the water osmotic and diffusive permeability coefficients. We show that the polarity of the naturally occurring peptidic channels is close to optimal for water permeation, and that the water mobility for a wide range of channel polarities is essentially length independent. By systematically varying the geometry and polarity of model hydrophilic pores, based on the fold of gramicidin A, the water density, occupancy, and permeability are studied. Our focus is on the characterization of the transition between different permeation regimes in terms of the structure of water in the pores, the average pore occupancy and the dynamics of the permeating water molecules. We show that a general relationship between osmotic and diffusive water permeability coefficients in the single-file regime accounts for the time averaged pore occupancy, and that the dynamics of the permeating water molecules through narrow non single file channels effectively behaves like independent single-file columns.

Nowadays, drug design projects benefit from highly accurate protein–ligand binding free energy predictions based on molecular dynamics simulations. While such calculations have been computationally expensive in the past, we now demonstrate that workflows built on open source software packages can efficiently leverage pre-exascale computing resources to screen hundreds of compounds in a matter of days. We report our results of free energy calculations on a large set of pharmaceutically relevant targets assembled to reflect industrial drug discovery projects.