Improved conformational space annealing method to treat β-structure with the UNRES force-field and to enhance scalability of parallel implementation

Mechanical Stability of Ribonuclease A Heavily Depends on the Redox Environment

Disulfide bonds are covalent bonds that connect nonlocal fragments of proteins, and they are unique post-translational modifications of proteins. They require the oxidizing environment to be stable, which occurs for example during oxidative stress; however, in a cell the reductive environment is maintained, lowering their stability. Despite many years of research on disulfide bonds, their role in the protein life cycle is not fully understood and seems to strictly depend on a system or process in which they are involved. In this article, coarse-grained UNited RESidue (UNRES), and all-atom Assisted Model Building with Energy Refinement (AMBER) force fields were applied to run a series of steered molecular dynamics (SMD) simulations of one of the most studied, but still not fully understood, proteins─ribonuclease A (RNase A). SMD simulations were performed to study the mechanical stability of RNase A in different oxidative-reductive environments. As disulfide bonds (and any other covalent bonds) cannot break/form in any classical all-atom force field, we applied additional restraints between sulfur atoms of reduced cysteines which were able to mimic the breaking of the disulfide bonds. On the other hand, the coarse-grained UNRES force field enables us to study the breaking/formation of the disulfide bonds and control the reducing/oxidizing environment owing to the presence of the designed distance/orientation-dependent potential. This study reveals that disulfide bonds have a strong influence on the mechanical stability of RNase A only in a highly oxidative environment. However, the local stability of the secondary structure seems to play a major factor in the overall stability of the protein. Both our thermal unfolding and mechanical stretching studies show that the most stable disulfide bond is Cys65-Cys72. The breaking of disulfide bonds Cys26-Cys84 and Cys58-Cys110 is associated with large force peaks. They are structural bridges, which are mostly responsible for stabilizing the RNase A conformation, while the presence of the remaining two bonds (Cys65-Cys72 and Cys40-Cys95) is most likely connected with the enzymatic activity rather than the structural stability of RNase A in the cytoplasm. Our results prove that disulfide bonds are indeed stabilizing fragments of the proteins, but their role is strongly redox environment-dependent.

Prediction of DNA and RNA structure with the NARES-2P force field and conformational space annealing

A physics-based method for the prediction of the structures of nucleic acids, which is based on the physics-based 2-bead NARES-2P model of polynucleotides and global-optimization Conformational Space Annealing (CSA) algorithm has been proposed. The target structure is sought as the global-energy-minimum structure, which ignores the entropy component of the free energy but spares expensive multicanonical simulations necessary to find the conformational ensemble with the lowest free energy. The CSA algorithm has been modified to optimize its performance when treating both single and multi-chain nucleic acids. It was shown that the method finds the native fold for simple RNA molecules and DNA duplexes and with limited distance restraints, which can easily be obtained from the secondary-structure-prediction servers, complex RNA folds can be treated with using moderate computer resources.

Dynamic Formation and Breaking of Disulfide Bonds in Molecular Dynamics Simulations with the UNRES Force Field

Many proteins contain disulfide bonds that are usually essential for maintaining function and a stable structure. Several algorithms attempt to predict the arrangement of disulfide bonds in the context of protein structure prediction, but none can simulate the entire process of oxidative folding, including dynamic formation and breaking of disulfide bonds. In this work, a potential function developed to model disulfide bonds is coupled with the united-residue (UNRES) force field, and used in both canonical and replica exchange molecular dynamics simulations to produce complete oxidative folding pathways. The potential function is obtained by introducing a transition barrier that separates the bonded and nonbonded states of the half-cystine residues. Tests on several helical proteins show that improved predictions are obtained when dynamic disulfide-bond formation and breaking are considered. The effect of the disulfide bonds on the folding kinetics is also investigated, particularly their role in stabilizing folding intermediates, resulting in slower folding.

ASTRO-FOLD 2.0: an Enhanced Framework for Protein Structure Prediction

The three-dimensional (3-D) structure prediction of proteins, given their amino acid sequence, is addressed using the first principles-based approach ASTRO-FOLD 2.0. The key features presented are: (1) Secondary structure prediction using a novel optimization-based consensus approach, (2) β-sheet topology prediction using mixed-integer linear optimization (MILP), (3) Residue-to-residue contact prediction using a high-resolution distance-dependent force field and MILP formulation, (4) Tight dihedral angle and distance bound generation for loop residues using dihedral angle clustering and non-linear optimization (NLP), (5) 3-D structure prediction using deterministic global optimization, stochastic conformational space annealing, and the full-atomistic ECEPP/3 potential, (6) Near-native structure selection using a traveling salesman problem-based clustering approach, ICON, and (7) Improved bound generation using chemical shifts of subsets of heavy atoms, generated by SPARTA and CS23D. Computational results of ASTRO-FOLD 2.0 on 47 blind targets of the recently concluded CASP9 experiment are presented.

An improved hybrid global optimization method for protein tertiary structure prediction

First principles approaches to the protein structure prediction problem must search through an enormous conformational space to identify low-energy, near-native structures. In this paper, we describe the formulation of the tertiary structure prediction problem as a nonlinear constrained minimization problem, where the goal is to minimize the energy of a protein conformation subject to constraints on torsion angles and interatomic distances. The core of the proposed algorithm is a hybrid global optimization method that combines the benefits of the αBB deterministic global optimization approach with conformational space annealing. These global optimization techniques employ a local minimization strategy that combines torsion angle dynamics and rotamer optimization to identify and improve the selection of initial conformations and then applies a sequential quadratic programming approach to further minimize the energy of the protein conformations subject to constraints. The proposed algorithm demonstrates the ability to identify both lower energy protein structures, as well as larger ensembles of low-energy conformations.

Implementation of molecular dynamics and its extensions with the coarse-grained UNRES force field on massively parallel systems; towards millisecond-scale simulations of protein structure, dynamics, and thermodynamics

We report the implementation of our united-residue UNRES force field for simulations of protein structure and dynamics with massively parallel architectures. In addition to coarse-grained parallelism already implemented in our previous work, in which each conformation was treated by a different task, we introduce a fine-grained level in which energy and gradient evaluation are split between several tasks. The Message Passing Interface (MPI) libraries have been utilized to construct the parallel code. The parallel performance of the code has been tested on a professional Beowulf cluster (Xeon Quad Core), a Cray XT3 supercomputer, and two IBM BlueGene/P supercomputers with canonical and replica-exchange molecular dynamics. With IBM BlueGene/P, about 50 % efficiency and 120-fold speed-up of the fine-grained part was achieved for a single trajectory of a 767-residue protein with use of 256 processors/trajectory. Because of averaging over the fast degrees of freedom, UNRES provides an effective 1000-fold speed-up compared to the experimental time scale and, therefore, enables us to effectively carry out millisecond-scale simulations of proteins with 500 and more amino-acid residues in days of wall-clock time.

On potential energy models for EA-based ab initio protein structure prediction

Ab initio protein structure prediction involves determination of the three-dimensional (3D) conformation of proteins on the basis of their amino acid sequence, a potential energy (PE) model that captures the physics of the interatomic interactions, and a method to search for and identify the global minimum in the PE (or free energy) surface such as an evolutionary algorithm (EA). Many PE models have been proposed over the past three decades and more. There is currently no understanding of how the behavior of an EA is affected by the PE model used. The study reported here shows that the EA behavior can be profoundly affected: the EA performance obtained when using the ECEPP PE model is significantly worse than that obtained when using the Amber, OPLS, and CVFF PE models, and the optimal EA control parameter values for the ECEPP model also differ significantly from those associated with the other models.

Simple physics-based analytical formulas for the potentials of mean force for the interaction of amino acid side chains in water. IV. Pairs of different hydrophobic side chains

The potentials of mean force of 21 heterodimers of the molecules modeling hydrophobic amino acid side chains: ethane (for alanine), propane (for proline), isobutane (for valine), isopentane (for leucine and isoleucine), ethylbenzene (for phenylalanine), methyl propyl sulfide (for methionine), and indole (for tryptophane) were determined by umbrella-sampling molecular dynamics simulations in explicit water as functions of distance and orientation. Analytical expressions consisting of the Gay-Berne term to represent effective van der Waals interactions and the cavity term proposed in our earlier work were fitted to the potentials of mean force. The positions and depths of the contact minima and the positions and heights of the desolvation maxima, including their dependence on the orientation of the molecules, are well represented by the analytical expressions for all systems; large deviations between the MD-determined PMF and the analytical approximations are observed for pairs involving the least spheroidal solutes: ethylbenzene, indole, and methyl propyl sulfide at short distances at which the PMF is high and, consequently, these regions are rarely visited. When data from the PMF within only 10 kcal/mol above the global minimum are considered, the standard deviation between the MD-determined and the fitted PMF is from 0.25 to 0.55 kcal/mol (the relative standard deviation being from 4% to 8%); it is larger for pairs involving nonspherical solute molecules. The free energies of contact computed from the PMF surfaces are well correlated with those determined from protein-crystal data with a slope close to that relating the free energies of transfer of amino acids (from water to n-octanol) to the average contact free energies determined from protein-crystal data. These observations justify future use of the determined potentials in coarse-grained protein-folding simulations.

Computational methods in protein structure prediction

This review presents the advances in protein structure prediction from the computational methods perspective. The approaches are classified into four major categories: comparative modeling, fold recognition, first principles methods that employ database information, and first principles methods without database information. Important advances along with current limitations and challenges are presented.

Ab initio protein fold prediction using evolutionary algorithms: influence of design and control parameters on performance

True ab initio prediction of protein 3D structure requires only the protein primary structure, a physicochemical free energy model, and a search method for identifying the free energy global minimum. Various characteristics of evolutionary algorithms (EAs) mean they are in principle well suited to the latter. Studies to date have been less than encouraging, however. This is because of the limited consideration given to EA design and control parameter issues. A comprehensive study of these issues was, therefore, undertaken for ab initio protein fold prediction using a full atomistic protein model. The performance and optimal control parameter settings of twelve EA designs where first established using a 15-residue polyalanine molecule-design aspects varied include the encoding alphabet, crossover operator, and replacement strategy. It can be concluded that real encoding and multipoint crossover are superior, while both generational and steady-state replacement strategies have merits. The scaling between the optimal control parameter settings and polyalanine size was also identified for both generational and steady-state designs based on real encoding and multipoint crossover. Application of the steady-state design to met-enkephalin indicated that these scalings are potentially transferable to real proteins. Comparison of the performance of the steady state design for met-enkephalin with other ab initio methods indicates that EAs can be competitive provided the correct design and control parameter values are used.

Simple Physics-Based Analytical Formulas for the Potentials of Mean Force for the Interaction of Amino Acid Side Chains in Water. 2. Tests with Simple Spherical Systems

Simple analytical functions consisting of electrostatic, polarization, Lennard-Jones or modified Lennard-Jones, and cavity terms are proposed to express the potentials of mean force analytically for spherical particles interacting in water. The cavity term was expressed either through the molecular-surface area of the solute or by using the Gaussian-overlap model of hydrophobic hydration developed in paper 1 of this series. The analytical expressions were fitted to the potentials of mean force of a methane homodimer, heterodimers composed of a methane molecule, and an ammonium cation or a chloride anion, respectively, and dimers consisting of two chloride anions, two ammonium cations, or a chloride ion and an ammonium cation. The potentials of mean force for these dimers were determined by umbrella-sampling molecular dynamics simulations with the AMBER 7.0 force field with TIP3P water either in our earlier work or in this work. For all systems, the analytical formulas fitted the potentials of mean force very well. However, using the molecular-surface area to express the cavity term provided a good fit only when the nonbonded interactions were expressed by an all-repulsive modified Lennard-Jones potential but also resulted in non-physical values of some of the parameters. Conversely, the use of our new Gaussian-overlap-based expression for the cavity term provided a good fit to the potentials of mean force (PMFs) with Lennard-Jones nonbonded potential, and the values of all parameters were physically reasonable.

Simple Physics-Based Analytical Formulas for the Potentials of Mean Force for the Interaction of Amino Acid Side Chains in Water. 1. Approximate Expression for the Free Energy of Hydrophobic Association Based on a Gaussian-Overlap Model

A physics-based model is proposed to derive approximate analytical expressions for the cavity component of the free energy of hydrophobic association of spherical and spheroidal solutes in water. The model is based on the difference between the number and context of the water molecules in the hydration sphere of a hydrophobic dimer and of two isolated hydrophobic solutes. It is assumed that the water molecules touching the convex part of the molecular surface of the dimer and those in the hydration spheres of the monomers contribute equally to the free energy of solvation, and those touching the saddle part of the molecular surface of the dimer result in a more pronounced increase in free energy because of their more restricted mobility (entropy loss) and fewer favorable electrostatic interactions with other water molecules. The density of water in the hydration sphere around a single solute particle is approximated by the derivative of a Gaussian centered on the solute molecule with respect to its standard deviation. On the basis of this approximation, the number of water molecules in different parts of the hydration sphere of the dimer is expressed in terms of the first and the second mixed derivatives of the two Gaussians centered on the first and second solute molecules, respectively, with respect to the standard deviations of these Gaussians, and plausible analytical expressions for the cavity component of the hydrophobic-association energy of spherical and spheroidal solutes are introduced. As opposed to earlier hydration-shell models, our expressions reproduce the desolvation maxima in the potentials of mean force of pairs of nonpolar solutes in water, and their advantage over the models based on molecular-surface area is that they have continuous gradients in the coordinates of solute centers.

Simple physics-based analytical formulas for the potentials of mean force for the interaction of amino acid side chains in water. 3. Calculation and parameterization of the potentials of mean force of pairs of identical hydrophobic side chains

The potentials of mean force of homodimers of the molecules modeling hydrophobic amino acid side chains (ethane (for alanine), propane (for proline), isobutane (for valine), isopentane (for leucine and isoleucine), ethylbenzene (for phenylalanine), and methyl propyl sulfide (for methionine)) were determined by umbrella-sampling molecular dynamics simulations in explicit water as functions of distance and orientation. Analytical expressions consisting of the Gay-Berne term to represent effective van der Waals interactions and the cavity term derived in paper 1 of this series were fitted to the potentials of mean force. The positions and depths of the contact minima and the positions and heights of the desolvation maxima, including their dependence on the orientation of the molecules, were well represented by the analytical expressions for all systems, which justifies use of such potentials in coarse-grain protein-folding simulations.

Novel approach for alpha-helical topology prediction in globular proteins: generation of interhelical restraints

The protein folding problem represents one of the most challenging problems in computational biology. Distance constraints and topology predictions can be highly useful for the folding problem in reducing the conformational space that must be searched by deterministic algorithms to find a protein structure of minimum conformational energy. We present a novel optimization framework for predicting topological contacts and generating interhelical distance restraints between hydrophobic residues in alpha-helical globular proteins. It should be emphasized that since the model does not make assumptions about the form of the helices, it is applicable to all alpha-helical proteins, including helices with kinks and irregular helices. This model aims at enhancing the ASTRO-FOLD protein folding approach of Klepeis and Floudas (Journal of Computational Chemistry 2003;24:191-208), which finds the structure of global minimum conformational energy via a constrained nonlinear optimization problem. The proposed topology prediction model was evaluated on 26 alpha-helical proteins ranging from 2 to 8 helices and 35 to 159 residues, and the best identified average interhelical distances corresponding to the predicted contacts fell below 11 A in all 26 of these systems. Given the positive results of applying the model to several protein systems, the importance of interhelical hydrophobic-to-hydrophobic contacts in determining the folding of alpha-helical globular proteins is highlighted.

Modification and optimization of the united-residue (UNRES) potential energy function for canonical simulations. I. Temperature dependence of the effective energy function and tests of the optimization method with single training proteins

We report the modification and parametrization of the united-residue (UNRES) force field for energy-based protein structure prediction and protein folding simulations. We tested the approach on three training proteins separately: 1E0L (beta), 1GAB (alpha), and 1E0G (alpha + beta). Heretofore, the UNRES force field had been designed and parametrized to locate native-like structures of proteins as global minima of their effective potential energy surfaces, which largely neglected the conformational entropy because decoys composed of only lowest-energy conformations were used to optimize the force field. Recently, we developed a mesoscopic dynamics procedure for UNRES and applied it with success to simulate protein folding pathways. However, the force field turned out to be largely biased toward -helical structures in canonical simulations because the conformational entropy had been neglected in the parametrization. We applied the hierarchical optimization method, developed in our earlier work, to optimize the force field; in this method, the conformational space of a training protein is divided into levels, each corresponding to a certain degree of native-likeness. The levels are ordered according to increasing native-likeness; level 0 corresponds to structures with no native-like elements, and the highest level corresponds to the fully native-like structures. The aim of optimization is to achieve the order of the free energies of levels, decreasing as their native-likeness increases. The procedure is iterative, and decoys of the training protein(s) generated with the energy function parameters of the preceding iteration are used to optimize the force field in a current iteration. We applied the multiplexing replica-exchange molecular dynamics (MREMD) method, recently implemented in UNRES, to generate decoys; with this modification, conformational entropy is taken into account. Moreover, we optimized the free-energy gaps between levels at temperatures corresponding to a predominance of folded or unfolded structures, as well as to structures at the putative folding-transition temperature, changing the sign of the gaps at the transition temperature. This enabled us to obtain force fields characterized by a single peak in the heat capacity at the transition temperature. Furthermore, we introduced temperature dependence to the UNRES force field; this is consistent with the fact that it is a free-energy and not a potential energy function. beta

Replica Exchange and Multicanonical Algorithms with the coarse-grained UNRES force field

Three algorithms, namely a Replica Exchange method (REM), a Replica Exchange Multicanonical method (REMUCA), and Replica Exchange Multicanonical with Replica Exchange (REMUCAREM), were implemented with the coarse-grained united-residue force field (UNRES) in both Monte Carlo and Molecular Dynamics versions. The MD algorithms use the constant-temperature Berendsen thermostat, with the velocity Verlet algorithm and variable time step. The algorithms were applied to one peptide (20 residues of Alanine with free ends; ala(20)) and two small proteins, namely an α-helical protein of 46 residues (the B-domain of the staphylococal protein A; 1BDD), and an α+β-protein of 48 residues (the E. Coli Mltd Lysm Domain; 1E0G). Calculated thermodynamic averages, such as canonical average energy and heat capacity, are in good agreement among all simulations for poly-L-alanine, showing that the algorithms were implemented correctly, and that all three algorithms are equally effective for small systems. For protein A, all algorithms performed reasonably well, although some variability in the calculated results was observed whereas, for a more complicated α+β-protein (1E0G), only Replica Exchange was capable of producing reliable statistics for calculating thermodynamic quantities. Finally, from the Replica Exchange molecular dynamics results, we calculated free energy maps as functions of RMSD and radius of gyration for different temperatures. The free energy calculations show correct folding behavior for poly-L-alanine and protein A while, for 1E0G, the native structure had the lowest free energy only at very low temperatures. Hence, the entropy contribution for 1E0G is larger than that for protein A at the same temperature. A larger contribution from entropy means that there are more accessible conformations at a given temperature, making it more difficult to obtain an efficient coverage of conformational space to obtain reliable thermodynamic properties. At the same temperature, ala(20) has the smallest entropy contribution, followed by protein A, and then by 1E0G.

Molecular dynamics with the united-residue model of polypeptide chains. I. Lagrange equations of motion and tests of numerical stability in the microcanonical mode

The Lagrange formalism was implemented to derive the equations of motion for the physics-based united-residue (UNRES) force field developed in our laboratory. The C(alpha)...C(alpha) and C(alpha)...SC (SC denoting a side-chain center) virtual-bond vectors were chosen as variables. The velocity Verlet algorithm was adopted to integrate the equations of motion. Tests on the unblocked Ala(10) polypeptide showed that the algorithm is stable in short periods of time up to the time step of 1.467 fs; however, even with the shorter time step of 0.489 fs, some drift of the total energy occurs because of momentary jumps of the acceleration. These jumps are caused by numerical instability of the forces arising from the U(rot) component of UNRES that describes the energetics of side-chain-rotameric states. Test runs on the Gly(10) sequence (in which U(rot) is not present) and on the Ala(10) sequence with U(rot) replaced by a simple numerically stable harmonic potential confirmed this observation; oscillations of the total energy were observed only up to the time step of 7.335 fs, and some drift in the total energy or instability of the trajectories started to appear in long-time (2 ns and longer) trajectories only for the time step of 9.78 fs. These results demonstrate that the present U(rot) components (which are statistical potentials derived from the Protein Data Bank) must be replaced with more numerically stable functions; this work is under way in our laboratory. For the purpose of our present work, a nonsymplectic variable-time-step algorithm was introduced to reduce the energy drift for regular polypeptide sequences. The algorithm scales down the time step at a given point of a trajectory if the maximum change of acceleration exceeds a selected cutoff value. With this algorithm, the total energy is reasonably conserved up to a time step of 2.445 fs, as tested on the unblocked Ala(10) polypeptide. We also tried a symplectic multiple-time-step reversible RESPA algorithm and achieved satisfactory energy conservation for time steps up to 7.335 fs. However, at present, it appears that the reversible RESPA algorithm is several times more expensive than the variable-time-step algorithm because of the necessity to perform additional matrix multiplications. We also observed that, because Ala(10) folds and unfolds within picoseconds in the microcanonical mode, this suggests that the effective (event-based) time unit in UNRES dynamics is much larger than that of all-atom dynamics because of averaging over the fast-moving degrees of freedom in deriving the UNRES potential.

Physics-based protein-structure prediction using a hierarchical protocol based on the UNRES force field: assessment in two blind tests

Recent improvements in the protein-structure prediction method developed in our laboratory, based on the thermodynamic hypothesis, are described. The conformational space is searched extensively at the united-residue level by using our physics-based UNRES energy function and the conformational space annealing method of global optimization. The lowest-energy coarse-grained structures are then converted to an all-atom representation and energy-minimized with the ECEPP/3 force field. The procedure was assessed in two recent blind tests of protein-structure prediction. During the first blind test, we predicted large fragments of alpha and alpha+beta proteins [60-70 residues with C(alpha) rms deviation (rmsd) <6 A]. However, for alpha+beta proteins, significant topological errors occurred despite low rmsd values. In the second exercise, we predicted whole structures of five proteins (two alpha and three alpha+beta, with sizes of 53-235 residues) with remarkably good accuracy. In particular, for the genomic target TM0487 (a 102-residue alpha+beta protein from Thermotoga maritima), we predicted the complete, topologically correct structure with 7.3-A C(alpha) rmsd. So far this protein is the largest alpha+beta protein predicted based solely on the amino acid sequence and a physics-based potential-energy function and search procedure. For target T0198, a phosphate transport system regulator PhoU from T. maritima (a 235-residue mainly alpha-helical protein), we predicted the topology of the whole six-helix bundle correctly within 8 A rmsd, except the 32 C-terminal residues, most of which form a beta-hairpin. These and other examples described in this work demonstrate significant progress in physics-based protein-structure prediction.

Protein structure prediction with the UNRES force-field using Replica-Exchange Monte Carlo-with-Minimization; Comparison with MCM, CSA, and CFMC

Two current methods of global optimization are coupled to produce the Replica-Exchange method together with Monte Carlo-with-Minimization (REMCM). Its performance is compared with each separate component and with other global optimization techniques. REMCM was applied to search the conformational space of coarse grain protein systems described by the UNRES force field. The method consists of several noninteracting copies of Monte Carlo simulation, and minimization was used after every perturbation to enhance the sampling of low-energy conformations. REMCM was applied to five proteins of different topology, and the results were compared to those from other optimization methods, namely Monte Carlo-with-Minimization (MCM), Conformational Space Annealing (CSA), and Conformational Family Monte Carlo (CFMC). REMCM located global minima for four proteins faster and more consistently than either MCM or CFMC, and it converged faster than CSA for three of the five proteins tested. A performance comparison was also carried out between REMCM and the traditional Replica Exchange method (REM) for one protein, with REMCM showing a significant improvement. Moreover, because of its simplicity, REMCM was easy to implement, thereby offering an alternative to other global optimization methods used in protein structure prediction.