The graphics processing unit (GPU) has become a popular computational platform for molecular dynamics (MD) simulations of biomolecules. A significant speedup in the simulations of small- or medium-size systems using only a few computer nodes with a single or multiple GPUs has been reported. Because of GPU memory limitation and slow communication between GPUs on different computer nodes, it is not straightforward to accelerate MD simulations of large biological systems that contain a few million or more atoms on massively parallel supercomputers with GPUs. In this study, we develop a new scheme in our MD software, GENESIS, to reduce the total computational time on such computers. Computationally intensive real-space nonbonded interactions are computed mainly on GPUs in the scheme, while less intensive bonded interactions and communication-intensive reciprocal-space interactions are performed on CPUs. On the basis of the midpoint cell method as a domain decomposition scheme, we invent the single particle interaction list for reducing the GPU memory usage. Since total computational time is limited by the reciprocal-space computation, we utilize the RESPA multiple time-step integration and reduce the CPU resting time by assigning a subset of nonbonded interactions on CPUs as well as on GPUs when the reciprocal-space computation is skipped. We validated our GPU implementations in GENESIS on BPTI and a membrane protein, porin, by MD simulations and an alanine-tripeptide by REMD simulations. Benchmark calculations on TSUBAME supercomputer showed that an MD simulation of a million atoms system was scalable up to 256 computer nodes with GPUs.

Lipid/water interaction is essential for many biological processes. The water structure at the nonionic lipid interface remains little known, and there is no scope of a priori prediction of water orientation at nonionic interfaces, either. Here, we report our study combining advanced nonlinear spectroscopy and molecular dynamics simulation on the water orientation at the ceramide/water interface. We measured χ(2) spectrum in the OH stretch region of ceramide/isotopically diluted water interface using heterodyne-detected vibrational sum-frequency generation spectroscopy and found that the interfacial water prefers an overall hydrogen-up orientation. Molecular dynamics simulation indicates that this preferred hydrogen-up orientation of water is determined by a delicate balance between hydrogen-up and hydrogen-down orientation induced by lipid–water and intralipid hydrogen bonds. This mechanism also suggests that water orientation at neutral lipid interfaces depends highly on the chemical structure of the lipid headgroup, in contrast to the charged lipid interfaces where the net water orientation is determined solely by the charge of the lipid headgroup.

An accurate theoretical prediction of the vibrational spectrum of polypeptides remains to be a challenge due to (1) their conformational flexibility and (2) non-negligible anharmonic effects. The former makes the search for conformers that contribute to the spectrum difficult, and the latter requires an expensive, quantum mechanical calculation for both electrons and vibrations. Here, we propose a new theoretical approach, which implements an enhanced conformational sampling by the replica-exchange molecular dynamics method, a structural clustering to identify distinct conformations, and a vibrational structure calculation by the second-order vibrational quasi-degenerate perturbation theory (VQDPT2). A systematic mode-selection scheme is developed to reduce the cost of VQDPT2 and the generation of a potential energy surface by the electronic structure calculation. The proposed method is applied to a pentapeptide, SIVSF-NH2, for which the infrared spectrum has recently been measured in the gas phase with high resolution in the OH and NH stretching region. The theoretical spectrum of the lowest energy conformer is obtained with a mean absolute deviation of 11.2 cm–1 from the experimental spectrum. Furthermore, the NH stretching frequencies of the five lowest energy conformers are found to be consistent with the literature values measured for small peptides with a similar secondary structure. Therefore, the proposed method is a promising way to analyze the vibrational spectrum of polypeptides.

The cytoplasm of a cell is crowded with many different kinds of macromolecules. The macromolecular crowding affects the thermodynamics and kinetics of biological reactions in a living cell, such as protein folding, association, and diffusion. Theoretical and simulation studies using simplified models focus on the essential features of the crowding effects and provide a basis for analyzing experimental data. In most of the previous studies on the crowding effects, a uniform crowder size is assumed, which is in contrast to the inhomogeneous size distribution of macromolecules in a living cell. Here, we evaluate the free energy changes upon macromolecular association in a cell-like inhomogeneous crowding system via a theory of hard-sphere fluids and free energy calculations using Brownian dynamics trajectories. The inhomogeneous crowding model based on 41 different types of macromolecules represented by spheres with different radii mimics the physiological concentrations of macromolecules in the cytoplasm of Mycoplasma genitalium. The free energy changes of macromolecular association evaluated by the theory and simulations were in good agreement with each other. The crowder size distribution affects both specific and nonspecific molecular associations, suggesting that not only the volume fraction but also the size distribution of macromolecules are important factors for evaluating in vivo crowding effects. This study relates in vitro experiments on macromolecular crowding to in vivo crowding effects by using the theory of hard-sphere fluids with crowder-size heterogeneity.

Collective variables (CVs) are often used in molecular dynamics simulations based on enhanced sampling algorithms to investigate large conformational changes of a protein. The choice of CVs in these simulations is essential because it affects simulation results and impacts the free-energy profile, the minimum free-energy pathway (MFEP), and the transition-state structure. Here we examine how many CVs are required to capture the correct transition-state structure during the open-to-close motion of adenylate kinase using a coarse-grained model in the mean forces string method to search the MFEP. Various numbers of large amplitude principal components are tested as CVs in the simulations. The incorporation of local coordinates into CVs, which is possible in higher dimensional CV spaces, is important for capturing a reliable MFEP. The Bayesian measure proposed by Best and Hummer is sensitive to the choice of CVs, showing sharp peaks when the transition-state structure is captured. We thus evaluate the required number of CVs needed in enhanced sampling simulations for describing protein conformational changes.

Infrared (IR) and Raman spectra of a sphingomyelin (SM) bilayer have been calculated for the amide I, II and A modes and the double-bonded CC stretching mode by a weight averaged approach, based on an all-atom molecular dynamics (MD) simulation and a vibrational structure calculation. Representative structures and statistical weights of SM clusters connected by hydrogen bonds (HBs) are observed in MD trajectories. After constructing smaller fragments from the SM clusters, the vibrational spectra of the target modes were calculated by normal mode analysis with a correction for anharmonicity, using density functional theory. The final IR and Raman spectra of a SM bilayer were obtained as the weight averages over all SM clusters. The calculated Raman spectrum is in excellent agreement with a recent measurement, providing a clear assignment of the peak in question observed at 1643 cm−1 to the amide I modes of a SM bilayer. The analysis of the IR spectrum has also revealed that the amide bands are sensitive to the water content inside the membrane, since their band positions are strongly modulated by the HB between SM and water molecules. The present study suggests that the amide I band serves as a marker to identify the formation of SM clusters, and opens a new way to detect lipid rafts in the biological membrane.

Large conformational changes of multidomain proteins are difficult to simulate using all-atom molecular dynamics (MD) due to the slow time scale. We show that a simple modification of the structure-based coarse-grained (CG) model enables a stable and efficient MD simulation of those proteins. “Motion Tree”, a tree diagram that describes conformational changes between two structures in a protein, provides information on rigid structural units (domains) and the magnitudes of domain motions. In our new CG model, which we call the DoME (domain motion enhanced) model, interdomain interactions are defined as being inversely proportional to the magnitude of the domain motions in the diagram, whereas intradomain interactions are kept constant. We applied the DoME model in combination with the Go model to simulations of adenylate kinase (AdK). The results of the DoME–Go simulation are consistent with an all-atom MD simulation for 10 μs as well as known experimental data. Unlike the conventional Go model, the DoME–Go model yields stable simulation trajectories against temperature changes and conformational transitions are easily sampled despite domain rigidity. Evidently, identification of domains and their interfaces is useful approach for CG modeling of multidomain proteins.

Adenosine triphosphate (ATP) is an indispensable energy source in cells. In a wide variety of biological phenomena like glycolysis, muscle contraction/relaxation, and active ion transport, chemical energy released from ATP hydrolysis is converted to mechanical forces to bring about large-scale conformational changes in proteins. Investigation of structure–function relationships in these proteins by molecular dynamics (MD) simulations requires modeling of ATP in solution and ATP bound to proteins with accurate force-field parameters. In this study, we derived new force-field parameters for the triphosphate moiety of ATP based on the high-precision quantum calculations of methyl triphosphate. We tested our new parameters on membrane-embedded sarcoplasmic reticulum Ca2+-ATPase and four soluble proteins. The ATP-bound structure of Ca2+-ATPase remains stable during MD simulations, contrary to the outcome in shorter simulations using original parameters. Similar results were obtained with the four ATP-bound soluble proteins. The new force-field parameters were also tested by investigating the range of conformations sampled during replica-exchange MD simulations of ATP in explicit water. Modified parameters allowed a much wider range of conformational sampling compared with the bias toward extended forms with original parameters. A diverse range of structures agrees with the broad distribution of ATP conformations in proteins deposited in the Protein Data Bank. These simulations suggest that the modified parameters will be useful in studies of ATP in solution and of the many ATP-utilizing proteins.

Ordering of water structures near the surface of biological membranes has been recently extensively studied using interface-selective techniques like vibrational sum frequency generation (VSFG) spectroscopy. The detailed structures of interface water have emerged for charged lipids, but those for neutral zwitterionic lipids remain obscure. We analyze an all-atom molecular dynamics (MD) trajectory of a hydrated 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine bilayer to characterize the orientation of interface waters in different chemical environments. The structure and dynamics of interfacial waters strongly depend on both their vertical position along the bilayer normal as well as vicinal lipid charged groups. Water orientation in the vicinity of phosphate groups is opposite to that around choline groups. The results are consistent with observed VSFG spectra and demonstrate that a mosaic of water orientation structures exists on the surface of a neutral zwitterionic phospholipid bilayer, reflecting rapid water exchange and the influence of local chemical environments.Keywords: interfacial water; molecular dynamics simulation; mosaic of water orientation structures; water orientation; zwitterionic POPC/water interface;

Conformational sampling is fundamentally important for simulating complex biomolecular systems. The generalized-ensemble algorithm, especially the temperature replica-exchange molecular dynamics method (T-REMD), is one of the most powerful methods to explore structures of biomolecules such as proteins, nucleic acids, carbohydrates, and also of lipid membranes. T-REMD simulations have focused on soluble proteins rather than membrane proteins or lipid bilayers, because explicit membranes do not keep their structural integrity at high temperature. Here, we propose a new generalized-ensemble algorithm for membrane systems, which we call the surface-tension REMD method. Each replica is simulated in the NPγT ensemble, and surface tensions in a pair of replicas are exchanged at certain intervals to enhance conformational sampling of the target membrane system. We test the method on two biological membrane systems: a fully hydrated DPPC (1,2-dipalmitoyl-sn-glycero-3-phosphatidylcholine) lipid bilayer and a WALP23–POPC (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) membrane system. During these simulations, a random walk in surface tension space is realized. Large-scale lateral deformation (shrinking and stretching) of the membranes takes place in all of the replicas without collapse of the lipid bilayer structure. There is accelerated lateral diffusion of DPPC lipid molecules compared with conventional MD simulation, and a much wider range of tilt angle of the WALP23 peptide is sampled due to large deformation of the POPC lipid bilayer and through peptide–lipid interactions. Our method could be applicable to a wide variety of biological membrane systems.

The association process of NaCl in aqueous solution was studied by a combination of quantum mechanical calculations on NaCl(H2O)n (n = 1–6) clusters and quantum mechanical/effective fragment potential–molecular dynamics (QM/EFP-MD) simulations for NaCl in 292 EFP waters. The interionic hydration structures (IHSs) were topologically classified as “ring” (R), “half-bridge” (H), and “full-bridge” (F) types on the basis of the quantum mechanical calculations. Subsequent IHS analysis on QM/EFP-MD simulations revealed that the NaCl contact ion pair (CIP) mainly involved R type hydration structures while the solvent-separated ion pair (SSIP) was composed of two different groups of F-type hydration structures. Our IHS analysis also discovered H type hydration even at large separation interionic distances (∼7 Å), which is denoted as a dissociating ion pair (DIP). The analysis was able to reveal the most complete interionic structures and their reorganizations of the association process. A strong correlation between the IHSs and interionic distance suggests that not only the solvent reorganization but also the local IHS changes are equally important. Mechanistically, it is suggested that the conversion between ring-type and full-bridge hydration structures is the main rate-determining step of ion-pair association.

Tom20 is located at the outer membrane of mitochondria and functions as a receptor for the N-terminal presequence of mitochondrial-precursor proteins. Recently, three atomic structures of the Tom20-presequence complex were determined using X-ray crystallography and classified into A-, M-, and Y-poses in terms of their presequence-binding modes. Combined with biochemical and NMR data, a dynamic-equilibrium model between the three poses has been proposed. To investigate this mechanism in further detail, we performed all-atom molecular dynamics (MD) simulations and replica-exchange MD (REMD) simulations of the Tom20-presequence complex in explicit water. In the REMD simulations, one major distribution and another minor one were observed in the converged free-energy landscape at 300 K. In the major distribution, structures similar to A- and M-poses exist, whereas those similar to Y-pose are located in the minor one, suggesting that A-pose in solution is more stable than Y-pose. A k-means clustering algorithm revealed a new pose not yet obtained by X-ray crystallography. This structure has double salt bridges between Arg14′ in the presequence and Glu78 or Glu79 in Tom20 and can explain the binding affinity of the complex in previous pull-down assay experiments. Structural clustering and analyses of contacts between Tom20 and the presequence suggest smooth conformational changes from Y- to A-poses through low activation barriers. M-pose lies between Y- and A-poses as a metastable state. The REMD simulations thus provide insights into the energetics of the multiple-binding forms and help to detail the progressive conformational states in the dynamic-equilibrium model based on the experimental data.

The formation of like-ion pairs, Na+–Na+ and Cl––Cl–, in aqueous solution was studied by high-level ab initio methods, classical molecular dynamics (MD), QM/TIP5P, and QM/EFP MD (quantum mechanics/effective fragment potential molecular dynamics). Ab initio calculations on model clusters revealed that the Na+(H2O)nNa+ (n = 2–4) clusters were significantly more stabilized by bridged waters than the corresponding Cl–(H2O)nCl– clusters. QM/EFP MD simulations in solution also predicted a clear local minimum near 3.6 Å only for the Na+–Na+ pair, suggesting that Na+–Na+ pairs may be more likely to form than Cl––Cl– pairs in solution. Analysis of the hydration structures further showed that two-water bridged Na+–Na+ pairs were dominant at the local minimum. The preferred formation of Na+ like-ion pairs in solution appeared to come from significant short-range effects, in particular, charge delocalization (polarization) between the bridged oxygen p and the vacant valence Na+ orbitals.

The introduction of bisecting GlcNAc and core fucosylation in N-glycans is essential for fine functional regulation of glycoproteins. In this paper, the effect of these modifications on the conformational properties of N-glycans is examined at the atomic level by performing replica-exchange molecular dynamics (REMD) simulations. We simulate four biantennary complex-type N-glycans, namely, unmodified, two single-substituted with either bisecting GlcNAc or core fucose, and disubstituted forms. By using REMD as an enhanced sampling technique, five distinct conformers in solution, each of which is characterized by its local orientation of the Manα1-6Man glycosidic linkage, are observed for all four N-glycans. The chemical modifications significantly change their conformational equilibria. The number of major conformers is reduced from five to two and from five to four upon the introduction of bisecting GlcNAc and core fucosylation, respectively. The population change is attributed to specific inter-residue hydrogen bonds, including water-mediated ones. The experimental NMR data, including nuclear Overhauser enhancement and scalar J-coupling constants, are well reproduced taking the multiple conformers into account. Our structural model supports the concept of “conformer selection”, which emphasizes the conformational flexibility of N-glycans in protein–glycan interactions.

Protein–glycan recognition regulates a wide range of biological and pathogenic processes. Conformational diversity of glycans in solution is apparently incompatible with specific binding to their receptor proteins. One possibility is that among the different conformational states of a glycan, only one conformer is utilized for specific binding to a protein. However, the labile nature of glycans makes characterizing their conformational states a challenging issue. All-atom molecular dynamics (MD) simulations provide the atomic details of glycan structures in solution, but fairly extensive sampling is required for simulating the transitions between rotameric states. This difficulty limits application of conventional MD simulations to small fragments like di- and tri-saccharides. Replica-exchange molecular dynamics (REMD) simulation, with extensive sampling of structures in solution, provides a valuable way to identify a family of glycan conformers. This article reviews recent REMD simulations of glycans carried out by us or other research groups and provides new insights into the conformational equilibria of N-glycans and their alteration by chemical modification. We also emphasize the importance of statistical averaging over the multiple conformers of glycans for comparing simulation results with experimental observables. The results support the concept of “conformer selection” in protein–glycan recognition.

The Sec translocon, a protein-conducting channel, consists of a heterotrimeric complex (SecYEG in bacteria and Sec61αβγ in eukaryotes) that provides a pathway for secretary proteins to cross membranes, or for membrane proteins to integrate into the membrane. The Sec translocon alone is a passive channel, and association with channel partners, including the ribosome or SecA ATPase in bacteria, is needed for protein translocation. Three recently published crystal structures of SecY are considered to represent the closed (resting state), pre-open (transitional state determined with the bound Fab fragment mimicking SecA interaction), and SecA-bound forms. To elucidate mechanisms of transition between closed and pre-open forms, we performed all-atom molecular dynamics simulations for the pre-open form of Thermus thermophilus SecYE and the closed form of Methanococcus janaschii SecYEβ in explicit solvent and membranes. We found that the closed form of SecY is stable, while the pre-open form without the Fab fragment undergoes large conformational changes toward the closed form. The pre-open form of SecY with Fab remains unchanged, suggesting that the cytosolic interaction mimicking SecA binding stabilizes the pre-open form of SecY. Importantly, a lipid molecule at the lateral gate region appears to be required to maintain the pre-open form in the membrane. We propose that the conformational transition from closed to pre-open states of SecY upon association with SecA facilitates intercalation of phospholipids at the lateral gate, inducing initial entry of the positively charged signal peptide into the channel.

•The effects of protein crowder sizes on hydration are investigated by MD simulations.•Significant slow down of water diffusion in the presence of small protein crowders is observed.•Crowder size is found to be another factor to determine the effect of macromolecular crowding.We investigate the effects of protein crowder sizes on hydration structure and dynamics in macromolecular crowded systems by all-atom MD simulations. The crowded systems consisting of only small proteins showed larger total surface areas than those of large proteins at the same volume fractions. As a result, more water molecules were trapped within the hydration shells, slowing down water diffusion. The simulation results suggest that the protein crowder size is another factor to determine the effect of macromolecular crowding and to explain the experimental kinetic data of proteins and DNAs in the presence of crowding agents.

•Various enhanced conformational sampling methods and membrane models are reviewed.•REMD methods and GB models are useful for membrane–protein structure prediction.•Enhanced lipid lateral diffusion and mixing can be realized by several methods.•Large membrane proteins are still challenging targets for enhanced sampling methods.This paper reviews various enhanced conformational sampling methods and explicit/implicit solvent/membrane models, as well as their recent applications to the exploration of the structure and dynamics of membranes and membrane proteins. Molecular dynamics simulations have become an essential tool to investigate biological problems, and their success relies on proper molecular models together with efficient conformational sampling methods. The implicit representation of solvent/membrane environments is reasonable approximation to the explicit all-atom models, considering the balance between computational cost and simulation accuracy. Implicit models can be easily combined with replica-exchange molecular dynamics methods to explore a wider conformational space of a protein. Other molecular models and enhanced conformational sampling methods are also briefly discussed. As application examples, we introduce recent simulation studies of glycophorin A, phospholamban, amyloid precursor protein, and mixed lipid bilayers and discuss the accuracy and efficiency of each simulation model and method. This article is part of a Special Issue entitled: Membrane Proteins edited by J.C. Gumbart and Sergei Noskov.Figure optionsDownload full-size imageDownload high-quality image (257 K)Download as PowerPoint slide

Structural diversity of N-glycans is essential for specific binding to their receptor proteins. To gain insights into structural and dynamic aspects in atomic detail not normally accessible by experiment, we here perform extensive molecular-dynamics simulations of N-glycans in solution using the replica-exchange method. The simulations show that five distinct conformers exist in solution for the N-glycans with and without bisecting GlcNAc. Importantly, the population sizes of three of the conformers are drastically reduced upon the introduction of bisecting GlcNAc. This is caused by a local hydrogen-bond rearrangement proximal to the bisecting GlcNAc. These simulations show that an N-glycan modification like the bisecting GlcNAc selects a certain “key” (or group of “keys”) within the framework of the “bunch of keys” mechanism. Hence, the range of specific glycan-protein interactions and affinity changes need to be understood in terms of the structural diversity of glycans and the alteration of conformational equilibria by core modification.

Infrared (IR) and Raman spectra of a sphingomyelin (SM) bilayer have been calculated for the amide I, II and A modes and the double-bonded CC stretching mode by a weight averaged approach, based on an all-atom molecular dynamics (MD) simulation and a vibrational structure calculation. Representative structures and statistical weights of SM clusters connected by hydrogen bonds (HBs) are observed in MD trajectories. After constructing smaller fragments from the SM clusters, the vibrational spectra of the target modes were calculated by normal mode analysis with a correction for anharmonicity, using density functional theory. The final IR and Raman spectra of a SM bilayer were obtained as the weight averages over all SM clusters. The calculated Raman spectrum is in excellent agreement with a recent measurement, providing a clear assignment of the peak in question observed at 1643 cm−1 to the amide I modes of a SM bilayer. The analysis of the IR spectrum has also revealed that the amide bands are sensitive to the water content inside the membrane, since their band positions are strongly modulated by the HB between SM and water molecules. The present study suggests that the amide I band serves as a marker to identify the formation of SM clusters, and opens a new way to detect lipid rafts in the biological membrane.