Seminars2007 – 2011

Horse liver alcohol dehydrogenase is a homodimer, the protomer having a coenzyme-binding domain and a catalytic domain. Using all available X-ray structures and molecular dynamics simulations, the mechanism of NAD⁺-induced domain closure was investigated.[1] When the well-known loop at the domain interface is in the "open" conformation the domains are unable to close. However, when the loop was modelled to its "closed" conformation, the NAD⁺-induced domain closure from the open structure could be simulated with remarkable accuracy. Further simulations and a careful analysis of X-ray structures suggest that the loop prevents domain closure in the absence of NAD⁺, and a cooperative mechanism operates between the subunits for domain closure. This cooperative mechanism explains the role of the "switch loop" as a block to closure, as in the absence of NAD⁺ it would prevent the occurrence of an unliganded closed subunit when the other subunit closes upon NAD⁺.
The switch loop comprises a rare ProPro motif suggesting a possible role in creating a rigid arm for communicating the presence of NAD⁺ to the region that blocks domain closure. This was confirmed using a linear inverse-kinematics technique developed for loop modelling.[2]

References
[1]. S. Hayward and A. Kitao, "Molecular dynamics simulations of NAD⁺-induced domain closure in horse liver alcohol dehydrogenase", Biophysical Journal, 91: 1823-1831, 2006.
[2]. S. Hayward and A. Kitao, "Effect of end constraints on protein loop kinematics", Biophysical Journal, 98(9), 1976-1985, 2010.

Biological environments provide complex physiochemical environments due to heterogeneity and crowding. This presents challenges for fully understanding the functionally-relevant structure and dynamics of biological macromolecules in vivo from both experimental and computational studies. In this talk, novel computational approaches are presented that allow the effects of cellular environments to be included efficiently in molecular dynamics simulations. A particular focus is on multiscale methodologies including mean-field formalisms and a new transferable coarse-grained model. Such simulations were applied to the study of the conformational dynamics of membrane-interacting peptides and peptides in crowded protein environments. More specifically, conformational sampling of viral membrane peptides and the membrane-spanning peptide phospholamban are described and discussed in the context of experimental data. Furthermore, the computational sampling of melittin in crowded cellular environments is discussed.

Due to the potential applications in semiconductor industries, the surface chemical reactions particularly on semiconductor surface have gained enormous popularity and the interest is still growing.[1]　With the help of traditional organic and organometallic chemistry, a wide variety of new chemically modified silicon surfaces can be synthesized to provide fine tailoring of surface characteristics for a broad range of applications. To gain the control needed to fabricate an organic function into existing semiconductor technologies and ultimately to make new molecule-scale devices, a detailed understanding of the adsorbate surface as well as interfacial chemical reactions and their products at the atomic/molecular level is critical. To accomplish this, theoretical investigations need to play a significant role in the advance of this field. This talk begins with the theoretical methodologies adapted for surface studies and then proceeds to a consideration of the unique features of clean silicon surfaces. Then, the main focus is directed to the characteristics of surface structures and reaction mechanisms that have been theoretically accumulated for the last several years. The key understandings of surface cyloadditions, nuleophilc additions, and surface thermal decompositions shall be discussed.
[1] a) Choi, C. H.; Gordon, M. S. “Computational Materials Chemistry: Methods and Applications”, L.A. Curtiss and M.S. Gordon, Eds., Kluwer Academic Publishers, Ch. 4, pp. 125-190, 2004. “Theoretical Studies of Silicon Surface Reactions with Main Group Absorbates” b) Journal of Physical Chemistry C (2010), 114(33), 14187-14192.

tba

The free-energy analysis is essential to understand and control a chemical process in condensed phase. The current status of theoretical/computation chemistry is, however, that the free energy remains a most difficult quantity to compute. For the fast computation and molecular understanding of the free energy, a new theory of solutions is introduced and is combined with molecular simulation. This theory is called the method of energy representation, and constructs the solvation free energy as a functional of distribution functions of the solute-solvent pair interaction energy. The method of energy representation greatly expands the scope of solution theory and is amenable to such frontline topics of physical chemistry and biophysics as ionic liquid, supercritical fluid, flexible molecules with intramolecular degrees of freedom, inhomogeneous system, and quantum-mechanical/molecular-mechanical (QM/MM) system. We present a brief introduction to the distribution-function theory of solutions, and describe the method of energy representation with its performance. As an application to inhomogeneous system involving flexible species, the molecular binding into micelle and lipid membrane is analyzed by treating micelle and membrane as a mixed solvent system consisting of water and amphiphilic molecule. The free energy of protein hydration is also evaluated with explicit solvent, and the roles of excluded volume and hydrogen bonding are quantitatively discussed.

Over the last three decades, considerable efforts have been made to generalize and enhance the computational methodologies and techniques to model and simulate macromolecules of biological interest. In particular, molecular dynamics simulations have provided deeper insights into not only how they interact with the surrounding environment at the atomic level, but also the microscopic driving forces of their functions, especially when the simulations are combined with sophisticated free energy calculations. In this seminar, I would like to present the challenges of current molecular modeling and simulations in terms of their accessibility to general users including experimentalists, issues with the system size and time scale, and finally accuracy. Several case studies will be presented with some preliminary results in my research group to go possibly beyond the challenges.

The ribosome is a stochastic molecular machine responsible for protein synthesis, a process central to all organisms. During the decoding step of protein synthesis (tRNA selection), the ribosome must discriminate between correct and incorrect tRNAs. The process by which tRNAs move from the partially bound (‘A/T state’) to fully bound (‘A/A’) state is rate-limiting in the case of correct (cognate) tRNAs. This large conformational change (~70 Angstrom tRNA movement) is known as accommodation and is the subject of our study. We use several different large-scale molecular simulation techniques to study the process of accommodation. Explicit solvent targeted molecular dynamics simulations allow us to define the accommodation corridor for tRNA on the large subunit of the ribosome. All-atom structure-based simulations using a Go-like potential allow us extend our approximate timescale to hundreds of milliseconds, revealing stochastic reversible excursions of the tRNA from the A/T state towards the A/A state and back to the A/T state, consistent with similar events observed in single molecule FRET studies. These structure-based simulations are cross-validated against equilibrium explicit solvent simulations (3.2 million atoms) with one microsecond of total sampling. Reasonable agreement is obtained between structure-based simulation, explicit solvent simulation, and crystalographic B-factors. The Encanto and RoadRunner supercomputers were used.

tba

tba

Principal component analysis (PCA) of molecular dynamics (MD) simulations is a powerful tool for investigating the collective motions of biomolecules. Due to the roughness of a protein’s energy landscape, the large-amplitude collective modes resulting from PCA are typically anharmonic. Here we decompose the PCA space to sub-PCA spaces by using the expectation-maximization (EM) algorithm. Each of these sub-PCA spaces represents a conformational state and the distribution of each state is described by a multivariate Gaussian distribution. The dynamics of the protein on the PCA space is then described as a diffusion/jumping process between different conformational states. We have applied this procedure to the MD trajectories of two proteins: crambin (1 μs simulation) and profilin (440 ns simulation). The number of states predicted initially increased with the dimensionality of PCA modes, but reached a plateau for dimensions greater than five. Furthermore, conformations can be divided into states based on the Gaussian mixture models and the transition network between states can be determined. The main advantage of this EM procedure is that it is significantly fast. However the method does not guarantee to find the optimal solutions for higher dimensions. This procedure is especially useful to do conformational clustering on the low-dimensional anharmonic "essential space". In order to determine such an "essential space", we have developed a statistical protocol involving estimation of the relaxation time of each individual mode, bootstrap analysis of the independent data based on the relaxation time, and application of a statistical goodness-of-fit test on the data to quantify normality. Using a simple model system of harmonic springs, we demonstrate the application and sensitivity of this technique and its ability to correctly identify the harmonic and anharmonic modes of the system.

tba

Molecular dynamics simulations of biomolecules and other substances with complex free energy landscapes often result in trapping in free energy minima, preventing extensive structural exploration. To overcome this limitation, methods collectively referred to as extended ensemble methods have recently been widely adopted. One such powerful method is the multicanonical method. However, it has limitations: it does not allow for the specification of pressure, as it maintains constant volume, and it cannot handle phenomena that involve volume changes. Additionally, it requires the determination of weight factors before starting the simulation, which becomes increasingly difficult for large systems. To address these issues, we have recently proposed two methods. To solve the first issue, we proposed the multibaric-multithermal method [1-5]. This method enables random walks in both energy and volume spaces. We applied this method to alanine dipeptide in water and compared it with existing methods [6,7]. In conventional isothermal-isobaric simulations, structures were trapped in free energy minima, limiting the range of sampled conformations. In contrast, the multibaric-multithermal method allowed for broader structural sampling without being trapped in free energy minima. Furthermore, we examined temperature and pressure variations of these structures and calculated the differences in partial molar enthalpy and partial molar volume between them, which showed good agreement with experimental results from Raman scattering. The differences in partial molar enthalpy and partial molar volume are important physical properties that indicate how the probabilities of each structure change with temperature and pressure. However, until now, no method existed to calculate these values using molecular simulations. Our approach has made it possible to compute the partial molar enthalpy and partial molar volume differences of biomolecules through molecular simulations for the first time. To address the second issue, we proposed the partially multicanonical method [8]. In this approach, only the potential energy terms necessary for extensive structural sampling are widely sampled. This reduces the effort required to determine weight factors, leading to more efficient structural sampling. We applied this method to alanine dipeptide in water and compared it with the canonical and multicanonical methods. The results showed that the multicanonical method explored a broader range of structures than the canonical method, and the partially multicanonical method was even more efficient in structural exploration.
[1] H. Okumura and Y. Okamoto: Chem. Phys. Lett. 383 (2004) 391-396.
[2] H. Okumura and Y. Okamoto: Phys. Rev. E 70 (2004) 026702(14pages).
[3] H. Okumura and Y. Okamoto: J. Phys. Soc. Jpn. 73 (2004) 3304-3311.
[4] H. Okumura and Y. Okamoto: Chem. Phys. Lett. 391 (2004) 248-253.
[5] H. Okumura and Y. Okamoto: J. Comput. Chem. 27 (2006) 379-395.
[6] H. Okumura and Y. Okamoto: Bull. Chem. Soc. Jpn. 80 (2007)1114-1123.
[7] H. Okumura and Y. Okamoto: J. Phys. Chem. B 112 (2008),12038-12049.
[8] H. Okumura: J. Chem. Phys. 129 (2008) 124116 (9 pages).

Molecular Dynamic (MD) simulations have been successfully used to study molecular events like structural stability and molecular interaction but many of the molecular mechanism which takes place in the cell are acting on a timescale beyond the capacity of a traditional MD simulation. Such an event is the ligand unbinding from the Nuclear Receptors (NR). The NRs functions as a transcription regulator and can be activated upon ligand binding. Consequently ligand binding and unbinding constitutes a fundamental process in the regulation of genes. Even though both biochemical and structural data of NR are available, the actual mechanism of the ligand binding/unbinding remains elusive. We have performed ligand unbinding studies on NRs with modified the MD methods (1) Random Acceleration MD (RAMD) (2) and Steered MD (SMD) (3) which speed up the timescale. The results show that agonist ligand unbinding can take place from the receptor without causing major conformational changes in the receptor, while antagonist unbinding cannot. Further on ligand selectivity and method sampling were discussed.
Allosteric properties have previously been studied with covariance correlation analysis and normal mode analysis. However, these methods requires long MD simulation trajectory to detect the signal. Recently publication presented the Anisotropic Thermal Diffusion method (4) which increases the signal-noise ratio within the protein and therefore makes it possible to detect an allosteric signal. The method was used to study allosteric properties in the NR Liver X Receptor (LXR) and succeeded to map out a pathway from the AF-2 region and the ligand. The signaling pathway detected is both intra- and intermolecular and is transmitted through amino acids side chains and the backbone. Although promising results were achieved, the method contains some drawbacks which will also be discussed.
1. Carlsson, P., S. Burendahl and L. Nilsson. (2006) Unbinding of retinoic acid from the retinoic acid receptor by random expulsion molecular dynamics. Biophys J 91, 3151-61.

2. Ludemann, S. K., V. Lounnas and R. C. Wade. (2000) How do substrates enter and products exit the buried active site of cytochrome P450cam? 1. Random expulsion molecular dynamics investigation of ligand access channels and mechanisms. Journal of Molecular Biology 303, 797-811.

3. Isralewitz, B., M. Gao and K. Schulten. (2001) Steered molecular dynamics and mechanical functions of proteins. Curr Opin Struct Biol 11, 224-30.

4. Ota, N. and D. A. Agard. (2005) Intramolecular signaling pathways revealed by modeling anisotropic thermal diffusion. J Mol Biol 351, 345-54

Abst

Abst

Recent advances in the field of structural and functional bioinformatics have made it possible to propose novel hypotheses regarding protein structure, function, and interactions using computational approaches. By utilizing these methods, it is expected that enhancing general knowledge about genes and proteins potentially involved in diseases—such as the molecular functions of each protein, their interaction partners, and the pathways they belong to—will be beneficial in the field of drug discovery. In this presentation, we discuss the fundamental theory of FUGUE, a method we have developed for predicting protein three-dimensional structures from amino acid sequences, as well as its specific applications to enzymes, membrane proteins, and other cases.

The diffusion of mass and the conductivity of heat take their origin at a molecular scale in the molecular motion and in the interaction between molecules. Molecular dynamics simulation is used to investigate these properties both under equilibrium and out of equilibrium. After a short presentation of these methods and of the transport equations, I will show the results we recently obtained of the transport properties across interfaces in two cases liquid-vapour and zeolite-vapour.

Analytical fitted potential energy surfaces are a valuable tool for study of reaction dynamics and molecular spectroscopy. In full generality the surface depends on 3N-6 independent coordinates, where N is the number of nuclei, and the construction of such surfaces is a problem of high-dimensional approximation already for small systems, say of 5-9 atoms. For the effective construction of such surfaces we have found it essential to enforce strictly the property of invariance of the surface under permutation of like nuclei; we build this property into the basis of fitting functions expressed in internuclear distances. The construction of these permutationally invariant basis functions (and covariant basis functions for properties such as the dipole moment) is a nontrivial task that cannot well be carried out by hand except for the simplest systems. We rely on computational invariant theory and on the MAGMA computer algebra system to construct suitable bases for each possible molecular permutation symmetry group,and have pursued that approach for systems up to 9 and 10 atoms. Illustrative challenging recent applications in the Bowman group at Emory University include photodissociation of acetaldehyde and spectroscopy of malonaldehyde and the water trimer. In the talk I will describe the mathematical background and highlights of new applications. This work is supported by ONR and USDOE.

Membrane proteins are pharmaceutically important therapeutic targets because of their well-recognized contribution to ion transport, intracellular and intercellular signaling pathways, and critical role in cell-cell recognition.Transmembrane (TM) domains of most membrane proteins consist of helices that interact with each other via inter- and intra-helix-helix interactions as well as with nonprotein membrane constituents. In this talk,I will present the computational/theoretical studies of membrane protein folding processes as insertion, folding, tilting, and assembly of transmembrane helices. If time is allowed, I will also talk about our novel RDC restraint potential and its applications, as well as the CHARMM-GUI development project.

The multicanonical method enables a random walk in the potential energy space. In contrast, we have recently proposed a new extended ensemble method, the multibaric-multithermal method. This method realizes a random walk in both energy and volume spaces, allowing the generation of isothermal-isobaric ensembles over a wide range of temperatures and pressures. In this study, we applied the multibaric-multithermal molecular dynamics method to alanine dipeptide in water and investigated the temperature and pressure dependence of its structure [1]. The Amber parm96 and Amber parm99 force fields were used for the peptide. Based on the temperature and pressure dependence of the probability ratio of each state, we calculated the partial molar enthalpy differences and partial molar volume differences between states. Furthermore, we discuss the force field dependence of these results.
[1] H. Okumura and Y. Okamoto: Bull. Chem. Soc. Jpn. 80 (2007)1114-1123.

We present a novel approach to the calculation of coulomb and exchange contributions to the total electronic energy in Hartree-Fock and Density Functional Theory.The key numerical procedure is the Cholesky decomposition that previously has been shown to efficiently remove linear dependence in the two-electron integral matrix.The basic idea is to decompose specific matrices that enter the energy expression.In this way we obtain an auxiliary basis (cholesky basis) that is much smaller than currently used in the resolution of identity or density fitting approaches.

Activity coefficients (chemical potential) of urea solutions are calculated to explore the mechanism of its solution properties which form the basis for its well known use as a strong protein denaturant. We perform free energy simulations of urea solutions in different urea concentrations using two urea models (OPLS and KBFF models)to calculate and decompose the activity coefficients. For the case of urea, we clarify the concept of the ideal solution in different concentration scales and standard states and its effect on our subsequent analysis.The analytical form of activity coefficients depends on the concentration units and standard states.For both models studied urea displays a weak concentration dependence for excess chemical potential.However, for the OPLS force field model this results from contributions which are independent of concentration to the van der Waals and electrostatic components whereas for the KBFF model those components are nontrivial but oppose each other.The strong ideality of urea solutions in some concentration scales,implying a lack of water perturbation, is discussed in terms of recent data and ideas on the mechanism of urea denaturation of proteins. [1,2]
[1] H. Kokubo and B. M. Pettitt,Preferential Solvation in Urea Solutions at Different Concentrations: Properties from Simulation Studies,J. Phys. Chem. B 111, 5233 -5242 (2007).
[2] H. Kokubo, J. Roesgen, D. W. Bolen, and B. M. Pettitt,Molecular Basis of the Apparent Near Ideality of Urea Solutions,Biophy. J. (2007) in press.

Polytheonamide B is a 48-residue polypeptide with strong cytotoxicity, isolated and purified from the sponge Theonella swinhoei collected in Hachijojima. This peptide contains seven types of amino acids that are methylated or hydroxylated in addition to standard amino acids. Furthermore, its D- and L-amino acids are arranged alternately, forming a β6.3-helix structure within the cell membrane, which creates a cation-selective channel. Our research group aims to elucidate the channel mechanism of Polytheonamide B at the atomic level. Based on normal mode analysis, we investigated the molecular dynamics of Polytheonamide B in a vacuum. Additionally, we analyzed Gramicidin A, the only structural and functional analog of Polytheonamide B, and compared the results to discuss the channel mechanism of Polytheonamide B.

Microwaves are low-energy, low-frequency photons, yet they can efficiently and selectively heat polarizable liquid and metallic powders. In this talk, recent molecular dynamics simulation results of the heating process of water, ice and salty water will be presented,with research prospects of microwave heating of metallic powders.
参考：
マイクロ波励起・高温非平衡反応場の科学
Tanaka and Sato, J.Chem.Phys., 126, 034509 (2007).

Computer simulations are ideally suited to study the conformational dynamics of biological macromolecules. Although such methods have been widely used, two major challenges are how to reach biologically relevant time scales and how to represent the effect of biological environments accurately. Implicit solvent models can be used to address both of these issues.The application of continuum electrostatics models for aqueous solvent, dense cellular environments, and biological environments is discussed.Results from implicit solvent simulations of protein G, ubiquitin, and alanine dipeptide are compared to explicit solvent simulations and experiments to demonstrate that implicit solvent treatments can provide a high level of accuracy in dilute aqueous solvent.The conformational sampling of alanine dipeptide, poly-alanine, and melittin is studied in dense cellular environments modeled as reduced dielectric media to understand how realistic biological environments might alter the conformational preferences of biomolecules. Finally, long-time conformational sampling of the transmembrane peptide phospholamban as a function of phosphorylation is discussed to illustrate how implicit membrane models can be applied to study the relation between its dynamics and its function as a regulatory protein of SERCA, a heart muscle ATPase Ca2+ pump.