Search

Proteins with similar structures are generally assumed to arise from similar sequences. However, there are more cases than not where this is not true. The dogma is that sequence determines structure; how, then, can very different sequences fold to the same structure? Here, we employ high temperature unfolding simulations to probe the pathways and specific interactions that direct the folding and unfolding of the SH3 domain. The SH3 metafold in the Dynameomics Database consists of 753 proteins with the same structure, but varied sequences and functions. To investigate the relationship between sequence and structure, we selected 17 targets from the SH3 metafold with high sequence variability. Six unfolding simulations were performed for each target, transition states were identified, revealing two general folding/unfolding pathways at the transition state. Transition states were also expressed as mathematical graphs of connected chemical nodes, and it was found that three positions within the structure, independent of sequence, were consistently more connected within the graph than any other nearby positions in the sequence. These positions represent a hub connecting different portions of the structure. Multiple sequence alignment and covariation analyses also revealed certain positions that were more conserved due to packing constraints and stabilizing long-range contacts. This study demonstrates that members of the SH3 domain with different sequences can unfold through two main pathways, but certain characteristics are conserved regardless of the sequence or unfolding pathway. While sequence determines structure, we show that disparate sequences can provide similar interactions that influence folding and lead to similar structures., (© 2020 The Protein Society.)

Protein folding/unfolding is a complicated process that defies high-resolution characterization by experimental methods. As an alternative, atomistic molecular dynamics simulations are now routinely employed to elucidate and magnify the accompanying conformational changes and the role of solvent in the folding process. However, the level of detail necessary to map the process at high spatial-temporal resolution provides an overwhelming amount of data. As more and better tools are developed for analysis of these large data sets and validation of the simulations, one is still left with the problem of visualizing the results in ways that provide insight into the folding/unfolding process. While viewing and interrogating static crystal structures has become commonplace, more and different approaches are required for dynamic, interconverting, unfolding, and refolding proteins. Here we review a variety of approaches, ranging from straightforward to complex and unintuitive for multiscale analysis and visualization of protein folding and unfolding., (Copyright © 2019 Elsevier Ltd. All rights reserved.)

Metamorphic proteins are biomolecules prone to adopting alternative conformations. Because of this feature, they represent ideal systems to investigate the general rules allowing primary structure to dictate protein topology. A comparative molecular dynamics study was performed on the denatured states of two proteins, sharing nearly identical amino-acid sequences (88 %) but different topologies, namely an all-α-helical bundle protein named G A 88 and an α+β-protein named G B 88. The analysis allowed successful design of and experimental validation of a site-directed mutant that promotes, at least in part, the switch in folding from G B 88 to G A 88. The mutated position, in which a glutamic acid was replaced by a glutamine, does not make any intramolecular interactions in the native state of G A 88, such that its stabilization can be explained by considering the effects on the denatured state. The results represent a direct demonstration of the role of the denatured state in sculpting native structure., (© 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.)

Molecular dynamics (MD) simulations contain considerable information with regard to the motions and fluctuations of a protein, the magnitude of which can be used to estimate conformational entropy. Here we survey conformational entropy across protein fold space using the Dynameomics database, which represents the largest existing data set of protein MD simulations for representatives of essentially all known protein folds. We provide an overview of MD-derived entropies accounting for all possible degrees of dihedral freedom on an unprecedented scale. Although different side chains might be expected to impose varying restrictions on the conformational space that the backbone can sample, we found that the backbone entropy and side chain size are not strictly coupled. An outcome of these analyses is the Dynameomics Entropy Dictionary, the contents of which have been compared with entropies derived by other theoretical approaches and experiment. As might be expected, the conformational entropies scale linearly with the number of residues, demonstrating that conformational entropy is an extensive property of proteins. The calculated conformational entropies of folding agree well with previous estimates. Detailed analysis of specific cases identifies deviations in conformational entropy from the average values that highlight how conformational entropy varies with sequence, secondary structure, and tertiary fold. Notably, α-helices have lower entropy on average than do β-sheets, and both are lower than coil regions.

The conformational propensities of amino acids are an amalgamation of sequence effects, environmental effects and underlying intrinsic behavior. Many have attempted to investigate neighboring residue effects to aid in our understanding of protein folding and improve structure prediction efforts, especially with respect to difficult to characterize states, such as disordered or unfolded states. Host-guest peptide series are a useful tool in examining the propensities of the amino acids free from the surrounding protein structure. Here, we compare the distributions of the backbone dihedral angles (φ/ψ) of the 20 proteogenic amino acids in two different sequence contexts using the AAXAA and GGXGG host-guest pentapeptide series. We further examine their intrinsic behaviors across three environmental contexts: water at 298 K, water at 498 K, and 8 M urea at 298 K. The GGXGG systems provide the intrinsic amino acid propensities devoid of any conformational context. The alanine residues in the AAXAA series enforce backbone chirality, thereby providing a model of the intrinsic behavior of amino acids in a protein chain. Our results show modest differences in φ/ψ distributions due to the steric constraints of the Ala side chains, the magnitudes of which are dependent on the denaturing conditions. One of the strongest factors modulating φ/ψ distributions was the protonation of titratable side chains, and the largest differences observed were in the amino acid propensities for the rarely sampled αL region., (© The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.)

Protein function is intimately linked to protein structure and dynamics yet experimentally determined structures frequently omit regions within a protein due to indeterminate data, which is often due protein dynamics. We propose that atomistic molecular dynamics simulations provide a diverse sampling of biologically relevant structures for these missing segments (and beyond) to improve structural modeling and structure prediction. Here we make use of the Dynameomics data warehouse, which contains simulations of representatives of essentially all known protein folds. We developed novel computational methods to efficiently identify, rank and retrieve small peptide structures, or fragments, from this database. We also created a novel data model to analyze and compare large repositories of structural data, such as contained within the Protein Data Bank and the Dynameomics data warehouse. Our evaluation compares these structural repositories for improving loop predictions and analyzes the utility of our methods and models. Using a standard set of loop structures, containing 510 loops, 30 for each loop length from 4 to 20 residues, we find that the inclusion of Dynameomics structures in fragment-based methods improves the quality of the loop predictions without being dependent on sequence homology. Depending on loop length, ∼ 25-75% of the best predictions came from the Dynameomics set, resulting in lower main chain root-mean-square deviations for all fragment lengths using the combined fragment library. We also provide specific cases where Dynameomics fragments provide better predictions for NMR loop structures than fragments from crystal structures. Online access to these fragment libraries is available at http://www.dynameomics.org/fragments., (© 2014 The Protein Society.)

D-amino acids are useful building blocks for de novo peptide design and they play a role in aging-related diseases associated with gradual protein racemization. For amino acids with achiral side chains, one should be able to presume that the conformational propensities of L- and D-amino acids are a reflection of one another due to the straightforward geometric inversion at the Cα atom. However, this presumption does not account for the directionality of the backbone dipole and the inverted propensities have never been definitively confirmed in this context. Furthermore, there is little known of how alternative side chain chirality affects the backbone conformations of isoleucine and threonine. Using a GGXGG host-guest pentapeptide system, we have completed exhaustive sampling of the conformational propensities of the D-amino acids, including D-allo-isoleucine and D-allo-threonine, using atomistic molecular dynamics simulations. Comparison of these simulations with the same systems hosting the cognate L-amino acids verifies that the intrinsic backbone conformational propensities of the D-amino acids are the inverse of their cognate L-enantiomers. Where amino acids have a chiral center in their side chain (Thr, Ile) the β-configuration affects the backbone sampling, which in turn can confer different biological properties., (© The Author 2014. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.)

Molecular dynamics simulations are a powerful theoretical tool to model the protein folding process in atomistic details under realistic conditions. Combined with a number of experimental techniques, simulations provide a detailed picture of how a protein folds or unfolds in the presence of explicit solvent and other molecular species, such as cosolvents, osmolytes, cofactors, active binding partners or inert crowding agents. The denaturing effects of temperature, pressure and external mechanical forces can also be probed. Qualitative and quantitative agreement with experiment contributes to a comprehensive molecular picture of protein states along the folding/unfolding pathway. The variety of systems examined reveals key features of the protein folding process., (Copyright © 2013. Published by Elsevier Inc.)

Membership in a protein domain database does not a domain make; a feature we realized when generating a consensus view of protein fold space with our consensus domain dictionary (CDD). This dictionary was used to select representative structures for characterization of the protein dynameome: the Dynameomics initiative. Through this endeavor we rejected a surprising 40% of the 1,695 folds in the CDD as being non-autonomous folding units. Although some of this was due to the challenges of grouping similar fold topologies, the dissonance between the cataloguing and structural qualification of protein domains remains surprising. Another potential factor is previously overlooked intrinsic disorder; predictions suggest that 40% of proteins have either local or global disorder. One thing is clear, filtering a structural database and ensuring a consistent definition for protein domains is crucial, and caution is prescribed when generalizations of globular domains are drawn from unfiltered protein domain datasets., (Copyright © 2012 WILEY Periodicals, Inc.)

Molecular dynamics simulations of protein folding or unfolding, unlike most in vitro experimental methods, are performed on a single molecule. The effects of neighboring molecules on the unfolding/folding pathway are largely ignored experimentally and simply not modeled computationally. Here, we present two all-atom, explicit solvent molecular dynamics simulations of 32 copies of the Engrailed homeodomain (EnHD), an ultrafast-folding and -unfolding protein for which the folding/unfolding pathway is well-characterized. These multimolecule simulations, in comparison with single-molecule simulations and experimental data, show that intermolecular interactions have little effect on the folding/unfolding pathway. EnHD unfolded by the same mechanism whether it was simulated in only water or also in the presence of other EnHD molecules. It populated the same native state, transition state, and folding intermediate in both simulation systems, and was in good agreement with experimental data available for each of the three states. Unfolding was slowed slightly by interactions with neighboring proteins, which were mostly hydrophobic in nature and ultimately caused the proteins to aggregate. Protein-water hydrogen bonds were also replaced with protein-protein hydrogen bonds, additionally contributing to aggregation. Despite the increase in protein-protein interactions, the protein aggregates formed in simulation did not do so at the total exclusion of water. These simulations support the use of single-molecule techniques to study protein unfolding and also provide insight into the types of interactions that occur as proteins aggregate at high temperature at an atomic level.

Graphs are rapidly becoming a powerful and ubiquitous tool for the analysis of protein structure and for event detection in dynamical protein systems. Despite their rise in popularity, however, the graph representations employed to date have shared certain features and parameters that have not been thoroughly investigated. Here, we examine and compare variations on the construction of graph nodes and graph edges. We propose a graph representation based on chemical groups of similar atoms within a protein rather than residues or secondary structure and find that even very simple analyses using this representation form a powerful event detection system with significant advantages over residue-based graph representations. We additionally compare graph edges based on probability of contact to graph edges based on contact strength and analyses of the entire graph structure to an alternative and more computationally tractable node-based analysis. We develop the simplest useful technique for analyzing protein simulations based on these comparisons and use it to shed light on the speed with which static protein structures adjust to a solvated environment at room temperature in simulation.

Background: Molecular dynamics (MD) simulations offer the ability to observe the dynamics and interactions of both whole macromolecules and individual atoms as a function of time. Taken in context with experimental data, atomic interactions from simulation provide insight into the mechanics of protein folding, dynamics, and function. The calculation of atomic interactions or contacts from an MD trajectory is computationally demanding and the work required grows exponentially with the size of the simulation system. We describe the implementation of a spatial indexing algorithm in our multi-terabyte MD simulation database that significantly reduces the run-time required for discovery of contacts. The approach is applied to the Dynameomics project data. Spatial indexing, also known as spatial hashing, is a method that divides the simulation space into regular sized bins and attributes an index to each bin. Since, the calculation of contacts is widely employed in the simulation field, we also use this as the basis for testing compression of data tables. We investigate the effects of compression of the trajectory coordinate tables with different options of data and index compression within MS SQL SERVER 2008., Results: Our implementation of spatial indexing speeds up the calculation of contacts over a 1 nanosecond (ns) simulation window by between 14% and 90% (i.e., 1.2 and 10.3 times faster). For a 'full' simulation trajectory (51 ns) spatial indexing reduces the calculation run-time between 31 and 81% (between 1.4 and 5.3 times faster). Compression resulted in reduced table sizes but resulted in no significant difference in the total execution time for neighbour discovery. The greatest compression (~36%) was achieved using page level compression on both the data and indexes., Conclusions: The spatial indexing scheme significantly decreases the time taken to calculate atomic contacts and could be applied to other multidimensional neighbor discovery problems. The speed up enables on-the-fly calculation and visualization of contacts and rapid cross simulation analysis for knowledge discovery. Using page compression for the atomic coordinate tables and indexes saves ~36% of disk space without any significant decrease in calculation time and should be considered for other non-transactional databases in MS SQL SERVER 2008.

The protein folding problem is often studied by comparing the mechanisms of proteins sharing the same structure but different sequence. The recent design of the two proteins G(A)88 and G(B)88, displaying different structures and functions while sharing 88% sequence identity (49 out of 56 amino acids), allows the unique opportunity for a complementary approach. At which stage of its folding pathway does a protein commit to a given topology? Which residues are crucial in directing folding mechanisms to a given structure? By using a combination of biophysical and computational techniques, we have characterized the folding of both G(A)88 and G(B)88. We show that, contrary to expectation, G(B)88, characterized by a native α+β fold, displays in the denatured state a content of native-like helical structure greater than G(A)88, which is all-α in its native state. Both experiments and simulations indicate that such residual structure may be tuned by changing pH. Thus, despite the high sequence identity, the folding pathways for these two proteins appear to diverge as early as in the denatured state. Our results suggest a mechanism whereby protein topology is committed very early along the folding pathway, being imprinted in the residual structure of the denatured state.

We have recently completed systematic molecular dynamics simulations of 807 different proteins representing 95% of the known autonomous protein folds in an effort we refer to as Dynameomics. Here we focus on the analysis of side chain conformations and dynamics to create a dynamic rotamer library. Overall this library is derived from 31,000 occurrences of each of 86,217 different residues, or 2.7 × 10(9) rotamers. This dynamic library has 74% overlap of rotamer distributions with rotamer libraries derived from static high-resolution crystal structures. Seventy-five percent of the residues had an assignable primary conformation, and 68% of the residues had at least one significant alternate conformation. The average correlation time for switching between rotamers ranged from 22 ps for Met to over 8 ns for Cys; this time decreased 20-fold on the surface of the protein and modestly for dihedral angles further from the main chain. Side chain S(2) axis order parameters were calculated and they correlated well with those derived from NMR relaxation experiments (R = 0.9). Relationships relating the S(2) axis order parameters to rotamer occupancy were derived. Overall the Dynameomics rotamer library offers a comprehensive depiction of side chain rotamer preferences and dynamics in solution, and more realistic distributions for dynamic proteins in solution at ambient temperature than libraries derived from crystal structures, in particular charged surface residues are better represented. Details of the rotamer library are presented here and the library itself can be downloaded at http://www.dynameomics.org., (Copyright © 2010 The Protein Society.)

The classification of protein folds is necessarily based on the structural elements that distinguish domains. Classification of protein domains consists of two problems: the partition of structures into domains and the classification of domains into sets of similar structures (or folds). Although similar topologies may arise by convergent evolution, the similarity of their respective folding pathways is unknown. The discovery and the characterization of the majority of protein folds will be followed by a similar enumeration of available protein folding pathways. Consequently, understanding the intricacies of structural domains is necessary to understanding their collective folding pathways. We review the current state of the art in the field of protein domain classification and discuss methods for the systematic and comprehensive study of protein folding across protein fold space via atomistic molecular dynamics simulation. Finally, we discuss our large-scale Dynameomics project, which includes simulations of representatives of all autonomous protein folds.

The ultrafast folding pathway of the engrailed homeodomain has been exceptionally well characterized by experiment and simulation. Helices II and III of the three-helix bundle protein form the native helix-turn-helix motif as an on-pathway intermediate within a few microseconds. The slow step is then the proper docking of the helices in approximately 15 mus. However, there is still the unexplained puzzle of why helix docking is relatively slow, which is part of the more general question as to why rearrangements of intermediates occur slowly. To address this problem, we performed 46 all-atom molecular dynamics refolding simulations in explicit water, for a total of 15 micros of simulation time. The simulations started from an intermediate state structure that was generated in an unfolding simulation at 498 K and was then quenched to folding-permissive temperatures. The protein refolded successfully in only one of the 46 simulations, and in that case the refolding pathway mirrored the unfolding pathway at high temperature. In the 45 simulations in which the protein did not fully fold, nonnative salt bridges trapped the protein, which explains why the protein folds relatively slowly from the intermediate state., (Copyright 2010 Biophysical Society. Published by Elsevier Inc. All rights reserved.)

The goal of the Dynameomics project is to perform, store, and analyze molecular dynamics simulations of representative proteins, of all known globular folds, in their native state and along their unfolding pathways. To analyze unfolding simulations, the location of the protein along the unfolding reaction coordinate (RXN) must be determined. Properties such as the fraction of native contacts and radius of gyration are often used; however, there is an issue regarding degeneracy with these properties, as native and nonnative species can overlap. Here, we used 15 physical properties of the protein to construct a multidimensional-embedded, one-dimensional RXN coordinate that faithfully captures the complex nature of unfolding. The unfolding RXN coordinates for 188 proteins (1534 simulations and 22.9 mus in explicit water) were calculated. Native, transition, intermediate, and denatured states were readily identified with the use of this RXN coordinate. A global native ensemble based on the native-state properties of the 188 proteins was created. This ensemble was shown to be effective for calculating RXN coordinates for folds outside the initial 188 targets. These RXN coordinates enable, high-throughput assignment of conformational states, which represents an important step in comparing protein properties across fold space as well as characterizing the unfolding of individual proteins., (Copyright (c) 2010 Biophysical Society. Published by Elsevier Inc. All rights reserved.)

The dynamic behavior of proteins is important for an understanding of their function and folding. We have performed molecular dynamics simulations of the native state and unfolding pathways of over 2000 protein/peptide systems (approximately 11,000 independent simulations) representing the majority of folds in globular proteins. These data are stored and organized using an innovative database approach, which can be mined to obtain both general and specific information about the dynamics and folding/unfolding of proteins, relevant subsets thereof, and individual proteins. Here we describe the project in general terms and the type of information contained in the database. Then we provide examples of mining the database for information relevant to protein folding, structure building, the effect of single-nucleotide polymorphisms, and drug design. The native state simulation data and corresponding analyses for the 100 most populated metafolds, together with related resources, are publicly accessible through http://www.dynameomics.org., (Copyright 2010 Elsevier Ltd. All rights reserved.)

The Dynameomics project aims to simulate a representative sample of all globular protein metafolds under both native and unfolding conditions. We have identified protein unfolding transition state (TS) ensembles from multiple molecular dynamics simulations of high-temperature unfolding in 183 structurally distinct proteins. These data can be used to study individual proteins and individual protein metafolds and to mine for TS structural features common across all proteins. Separating the TS structures into four different fold classes (all proteins, all-alpha, all-beta, and mixed alpha/beta and alpha +beta) resulted in no significant difference in the overall protein properties. The residues with the most contacts in the native state lost the most contacts in the TS ensemble. On average, residues beginning in an alpha-helix maintained more structure in the TS ensemble than did residues starting in beta-strands or any other conformation. The metafolds studied here represent 67% of all known protein structures, and this is, to our knowledge, the largest, most comprehensive study of the protein folding/unfolding TS ensemble to date. One might have expected broad distributions in the average global properties of the TS relative to the native state, indicating variability in the amount of structure present in the TS. Instead, the average global properties converged with low standard deviations across metafolds, suggesting that there are general rules governing the structure and properties of the TS.

Structure is only the first step in understanding the interactions and functions of proteins. In this paper, we explore the flexibility of proteins across a broad database of over 250 solvated protein molecular dynamics simulations in water for an aggregate simulation time of approximately 6 micros. These simulations are from our Dynameomics project, and these proteins represent approximately 75% of all known protein structures. We employ principal component analysis of the atomic coordinates over time to determine the primary axis and magnitude of the flexibility of each atom in a simulation. This technique gives us both a database of flexibility for many protein fold families and a compact visual representation of a particular protein's native-state conformational space, neither of which are available using experimental methods alone. These tools allow us to better understand the nature of protein motion and to describe its relationship to other structural and dynamical characteristics. In addition to reporting general properties of protein flexibility and detailing many dynamic motifs, we characterize the relationship between protein native-state flexibility and early events in thermal unfolding and show that flexibility predicts how a protein will begin to unfold. We provide evidence that fold families have conserved flexibility patterns, and family members who deviate from the conserved patterns have very low sequence identity. Finally, we examine novel aspects of highly inflexible loops that are as important to structural integrity as conventional secondary structure. These loops, which are difficult if not impossible to locate without dynamic data, may constitute new structural motifs.

Here, we compare the distributions of main chain (Phi,Psi) angles (i.e., Ramachandran maps) of the 20 naturally occurring amino acids in three contexts: (i) molecular dynamics (MD) simulations of Gly-Gly-X-Gly-Gly pentapeptides in water at 298 K with exhaustive sampling, where X = the amino acid in question; (ii) 188 independent protein simulations in water at 298 K from our Dynameomics Project; and (iii) static crystal and NMR structures from the Protein Data Bank. The GGXGG peptide series is often used as a model of the unstructured denatured state of proteins. The sampling in the peptide MD simulations is neither random nor uniform. Instead, individual amino acids show preferences for particular conformations, but the peptide is dynamic, and interconversion between conformers is facile. For a given amino acid, the (Phi,Psi) distributions in the protein simulations and the Protein Data Bank are very similar and often distinct from those in the peptide simulations. Comparison between the peptide and protein simulations shows that packing constraints, solvation, and the tendency for particular amino acids to be used for specific structural motifs can overwhelm the "intrinsic propensities" of amino acids for particular (Phi,Psi) conformations. We also compare our helical propensities with experimental consensus values using the host-guest method, which appear to be determined largely by context and not necessarily the intrinsic conformational propensities of the guest residues. These simulations represent an improved coil library free from contextual effects to better model intrinsic conformational propensities and provide a detailed view of conformations making up the "random coil" state.

The Dynameomics project is our effort to characterize the native-state dynamics and folding/unfolding pathways of representatives of all known protein folds by way of molecular dynamics simulations, as described by Beck et al. (in Protein Eng. Des. Select., the first paper in this series). The data produced by these simulations are highly multidimensional in structure and multi-terabytes in size. Both of these features present significant challenges for storage, retrieval and analysis. For optimal data modeling and flexibility, we needed a platform that supported both multidimensional indices and hierarchical relationships between related types of data and that could be integrated within our data warehouse, as described in the accompanying paper directly preceding this one. For these reasons, we have chosen On-line Analytical Processing (OLAP), a multi-dimensional analysis optimized database, as an analytical platform for these data. OLAP is a mature technology in the financial sector, but it has not been used extensively for scientific analysis. Our project is further more unusual for its focus on the multidimensional and analytical capabilities of OLAP rather than its aggregation capacities. The dimensional data model and hierarchies are very flexible. The query language is concise for complex analysis and rapid data retrieval. OLAP shows great promise for the dynamic protein analysis for bioengineering and biomedical applications. In addition, OLAP may have similar potential for other scientific and engineering applications involving large and complex datasets.

The goal of Dynameomics is to perform atomistic molecular dynamics (MD) simulations of representative proteins from all known folds in explicit water in their native state and along their thermal unfolding pathways. Here we present 188-fold representatives and their native state simulations and analyses. These 188 targets represent 67% of all the structures in the Protein Data Bank. The behavior of several specific targets is highlighted to illustrate general properties in the full dataset and to demonstrate the role of MD in understanding protein function and stability. As an example of what can be learned from mining the Dynameomics database, we identified a protein fold with heightened localized dynamics. In one member of this fold family, the motion affects the exposure of its phosphorylation site and acts as an entropy sink to offset another portion of the protein that is relatively immobile in order to present a consistent interface for protein docking. In another member of this family, a polymorphism in the highly mobile region leads to a host of disease phenotypes. We have constructed a web site to provide access to a novel hybrid relational/multidimensional database (described in the succeeding two papers) to view and interrogate simulations of the top 30 targets: http://www.dynameomics.org. The Dynameomics database, currently the largest collection of protein simulations and protein structures in the world, should also be useful for determining the rules governing protein folding and kinetic stability, which should aid in deciphering genomic information and for protein engineering and design.

Dynameomics is a project to investigate and catalog the native-state dynamics and thermal unfolding pathways of representatives of all protein folds using solvated molecular dynamics simulations, as described in the preceding paper. Here we introduce the design of the molecular dynamics data warehouse, a scalable, reliable repository that houses simulation data that vastly simplifies management and access. In the succeeding paper, we describe the development of a complementary multidimensional database. A single protein unfolding or native-state simulation can take weeks to months to complete, and produces gigabytes of coordinate and analysis data. Mining information from over 3000 completed simulations is complicated and time-consuming. Even the simplest queries involve writing intricate programs that must be built from low-level file system access primitives and include significant logic to correctly locate and parse data of interest. As a result, programs to answer questions that require data from hundreds of simulations are very difficult to write. Thus, organization and access to simulation data have been major obstacles to the discovery of new knowledge in the Dynameomics project. This repository is used internally and is the foundation of the Dynameomics portal site http://www.dynameomics.org. By organizing simulation data into a scalable, manageable and accessible form, we can begin to address substantial questions that move us closer to solving biomedical and bioengineering problems.

A properly identified transition state ensemble (TSE) in a molecular dynamics (MD) simulation can reveal a tremendous amount about how a protein folds and offer a point of comparison to experimentally derived Phi(F) values, which reflect the degree of structure in these transient states. In one such method of TSE identification, dubbed P(fold), MD simulations of individual protein structures taken from an unfolding trajectory are used to directly assess an input structure's probability of folding before unfolding, and P(fold) is, by definition, 0.5 for the TSE. Other, less computationally intensive methods, such as multidimensional scaling (MDS) of the pairwise root mean-squared deviation (RMSD) matrix of the conformations sampled in a thermal unfolding trajectory, have also been used to identify the TSE. Identification of the TSE is made from the original MD simulation without the need to run further simulations. Here we present a P(fold)-like study and describe methods for identification of the TSE through the derivation of a high fidelity, bounded, one-dimensional reaction coordinate for protein folding. These methods are applied to the engrailed homeodomain. The TSE identified by this approach is essentially identical to the TSE identified previously by MDS of the pairwise RMSD matrix. However, the cost of performing P(fold), or even our reduced P(fold)-like calculations, is at least 36,000 times greater than the MDS method.

Two independent replica-exchange molecular dynamics (REMD) simulations with an explicit water model were performed of the Trp-cage mini-protein. In the first REMD simulation, the replicas started from the native conformation, while in the second they started from a nonnative conformation. Initially, the first simulation yielded results qualitatively similar to those of two previously published REMD simulations: the protein appeared to be over-stabilized, with the predicted melting temperature 50-150K higher than the experimental value of 315K. However, as the first REMD simulation progressed, the protein unfolded at all temperatures. In our second REMD simulation, which starts from a nonnative conformation, there was no evidence of significant folding. Transitions from the unfolded to the folded state did not occur on the timescale of these simulations, despite the expected improvement in sampling of REMD over conventional molecular dynamics (MD) simulations. The combined 1.42 micros of simulation time was insufficient for REMD simulations with different starting structures to converge. Conventional MD simulations at a range of temperatures were also performed. In contrast to REMD, the conventional MD simulations provide an estimate of Tm in good agreement with experiment. Furthermore, the conventional MD is a fraction of the cost of REMD and continuous, realistic pathways of the unfolding process at atomic resolution are obtained.

The presence of a solvent-exposed alanine residue stabilizes a helix by 0.4-2 kcal.mol(-1) relative to glycine. Various factors have been suggested to account for the differences in helical propensity, from the higher conformational freedom of glycine sequences in the unfolded state to hydrophobic and van der Waals' stabilization of the alanine side chain in the helical state. We have performed all-atom molecular dynamics simulations with explicit solvent and exhaustive sampling of model peptides to address the backbone conformational entropy difference between Ala and Gly in the denatured state. The mutation of Ala to Gly leads to an increase in conformational entropy equivalent to approximately 0.4 kcal.mol(-1) in a fully flexible denatured, that is, unfolded, state. But, this energy is closely counterbalanced by the (measured) difference in free energy of transfer of the glycine and alanine side chains from the vapor phase to water so that the unfolded alanine- and glycine-containing peptides are approximately isoenergetic. The helix-stabilizing propensity of Ala relative to Gly thus mainly results from more favorable interactions of Ala in the folded helical structure. The small difference in energetics in the denatured states means that the Phi-values derived from Ala --> Gly scanning of helices are a very good measure of the extent of formation of structure in proteins with little residual structure in the denatured state.

Rarely is any solution simply solute and water. In vivo, solutes, such as proteins and nucleic acids, swim in a sea of water, salts, ions, small molecules, and lipids, not to mention other macromolecules. In vitro, virtually all solutions contain a mixture of aqueous solvents, or "cosolvents" [i.e., solvent(s) in addition to water], that can alter the dynamics, behavior, solubility, and stability of proteins and nucleic acids. We have developed models for a number of cosolvents, including the denaturant urea and the small chemical chaperone trimethylamine N-oxide (TMAO). This chapter examines the models for these two cosolvents in the context of experimental data. The direct and indirect effects of these molecules on water and protein are studied with molecular dynamics simulations. These observations and conclusions are drawn from simulations of these molecules in pure water and as a cosolvent for the protein chymotrypsin inhibitor 2. Urea-induced denaturation occurs initially through attack of the protein by water and hydration of hydrophobic protein moieties as a result of disruption of the hydrogen bonding network of water by urea. This indirect denaturing effect of urea is followed by more direct action as urea replaces some waters involved in the initial hydration of the hydrophobic core and subsequently binds to polar residues and the protein main chain to compete with the intraprotein hydrogen bonds. In the case of TMAO, we find that it encourages water-water interactions, thereby stabilizing the protein as a result of the increased penalty for the hydration of hydrophobic residues.

All atom molecular dynamics simulations have become a standard method for mapping equilibrium protein dynamics and non-equilibrium events like folding and unfolding. Here, we present detailed methods for performing such simulations. Generic protocols for minimization, solvation, simulation, and analysis derived from previous studies are also presented. As a measure of validation, our water model is compared with experiment. An example of current applications of these methods, simulations of the ultrafast folding protein Engrailed Homeodomain are presented including the experimental evidence used to verify their results. Ultrafast folders are an invaluable tool for studying protein behavior as folding and unfolding events measured by experiment occur on timescales accessible with the high-resolution molecular dynamics methods we describe. Finally, to demonstrate the prospect of these methods for folding proteins, a temperature quench simulation of a thermal unfolding intermediate of the Engrailed Homeodomain is described.

Proteins with ultra-fast folding/unfolding kinetics are excellent candidates for study by molecular dynamics. Here, we describe such simulations of a three helix bundle protein, the engrailed homeodomain (En-HD), which folds via the diffusion-collision model. The unfolding pathway of En-HD was characterized by seven simulations of the protein and 12 simulations of its helical fragments yielding over 1.1 micros of simulation time in water. Various conformational states along the unfolding pathway were identified. There is the compact native-like transition state, a U-shaped helical intermediate and an unfolded state with dynamic helical segments. Each of these states is in good agreement with experimental data. Examining these states as well as the transitions between them, we find the role of long-range tertiary contacts, specifically salt-bridges, important in the folding/unfolding pathway. In the folding direction, charged residues form long-range tertiary contacts before the hydrophobic core is formed. The formation of HII is assisted by a specific salt-bridge and by non-specific (fluctuating) tertiary contacts, which we call contact-assisted helix formation. Salt-bridges persist as the protein approaches the transition state, stabilizing HII until the hydrophobic core is formed. To complement this information, simulations of fragments of En-HD illustrate the helical propensities of the individual segments. By thermal denaturation, HII proved to be the least stable helix, unfolding in less than 450 ps at high temperature. We observed the low helical propensity of C-terminal residues from HIII in fragment simulations which, when compared to En-HD unfolding simulations, link the unraveling of HIII to the initial event that drives the unfolding of En-HD., (Copyright 2004 Elsevier Ltd.)

Here, we describe the folding/unfolding kinetics of alpha3D, a small designed three-helix bundle. Both IR temperature jump and ultrafast fluorescence mixing methods reveal a single-exponential process consistent with a minimal folding time of 3.2 +/- 1.2 micros (at approximately 50 degrees C), indicating that a protein can fold on the 1- to 5-micros time scale. Furthermore, the single-exponential nature of the relaxation indicates that the prefactor for transition state (TS)-folding models is probably >or=1 (micros)-1 for a protein of this size and topology. Molecular dynamics simulations and IR spectroscopy provide a molecular rationale for the rapid, single-exponential folding of this protein. alpha3D shows a significant bias toward local helical structure in the thermally denatured state. The molecular dynamics-simulated TS ensemble is highly heterogeneous and dynamic, allowing access to the TS via multiple pathways.

We have determined consensus protein-fold classifications on the basis of three classification methods, SCOP, CATH, and Dali. These classifications make use of different methods of defining and categorizing protein folds that lead to different views of protein-fold space. Pairwise comparisons of domains on the basis of their fold classifications show that much of the disagreement between the classification systems is due to differing domain definitions rather than assigning the same domain to different folds. However, there are significant differences in the fold assignments between the three systems. These remaining differences can be explained primarily in terms of the breadth of the fold classifications. Many structures may be defined as having one fold in one system, whereas far fewer are defined as having the analogous fold in another system. By comparing these folds for a nonredundant set of proteins, the consensus method breaks up broad fold classifications and combines restrictive fold classifications into metafolds, creating, in effect, an averaged view of fold space. This averaged view requires that the structural similarities between proteins having the same metafold be recognized by multiple classification systems. Thus, the consensus map is useful for researchers looking for fold similarities that are relatively independent of the method used to compare proteins. The 30 most populated metafolds, representing the folds of about half of a nonredundant subset of the PDB, are presented here. The full list of metafolds is presented on the Web.

Molecular dynamics simulations of the protein chymotrypsin inhibitor 2 in 8 M urea at 60 degrees C were undertaken to investigate the molecular basis of chemical denaturation. The protein unfolded rapidly under these conditions, but it retained its native structure in a control simulation in water at the same temperature. The overall process of unfolding in urea was similar to that observed in thermal denaturation simulations above the protein's T(m) of 75 degrees C. The first step in unfolding was expansion of the hydrophobic core. Then, the core was solvated by water and later by urea. The denatured structures in both urea and at high temperature contained residual native helical structure, whereas the beta-structure was completely disrupted. The average residence time for urea around hydrophilic groups was six times greater than around hydrophobic residues and in all cases greater than the corresponding water residence times. Water self-diffusion was reduced 40% in 8 M urea. Urea altered water structure and dynamics, thereby diminishing the hydrophobic effect and encouraging solvation of hydrophobic groups. In addition, through urea's weakening of water structure, water became free to compete with intraprotein interactions. Urea also interacted directly with polar residues and the peptide backbone, thereby stabilizing nonnative conformations. These simulations suggest that urea denatures proteins via both direct and indirect mechanisms.

We examined the hydration of amides of alpha(3)D, a simple, designed three-helix bundle protein. Molecular dynamics calculations show that the amide carbonyls on the surface of the protein tilt away from the helical axis to interact with solvent water, resulting in a lengthening of the hydrogen bonds on this face of the helix. Water molecules are bonded to these carbonyl groups with partial occupancy ( approximately 50%-70%), and their interaction geometries show a large variation in their hydrogen bond lengths and angles on the nsec time scale. This heterogeneity is reflected in the carbonyl stretching vibration (amide I' band) of a group of surface Ala residues. The surface-exposed amides are broad, and shift to lower frequency (reflecting strengthening of the hydrogen bonds) as the temperature is decreased. By contrast, the amide I' bands of the buried (13)C-labeled Leu residues are significantly sharper and their frequencies are consistent with the formation of strong hydrogen bonds, independent of temperature. The rates of hydrogen-deuterium exchange and the proton NMR chemical shifts of the helical amide groups also depend on environment. The partial occupancy of the hydration sites on the surface of helices suggests that the interaction is relatively weak, on the order of thermal energy at room temperature. One unexpected feature that emerged from the dynamics calculations was that a Thr side chain subtly disrupted the helical geometry 4-7 residues N-terminal in sequence, which was reflected in the proton chemical shifts and the rates of amide proton exchange for several amides that engage in a mixed 3(10)/alpha/pi-helical conformation.

The structure and dynamics of the water hydrating peptides and proteins are examined here at atomic resolution via molecular dynamics simulations. Detailed solvation density and residence time data for all 20 L-amino acids in an end-capped AXA tripeptide motif are presented. In addition, the solvation of the protein chymotrypsin inhibitor 2 is investigated as a point of comparison. Residues on the surface of proteins are not isolated; they interact both locally and non-locally in sequence space, and comparison of the solvation properties of each amino acid in both the peptide and protein allow us to distinguish inherent solvation properties from context-dependent perturbations due to neighboring residues. This work moves beyond traditional radial distribution functions and presents graphical representations of preferential solvation and orientation of water by side chains and the main chain. The combination of 0.3 micros of simulation data improves the statistical sampling over previous studies and reveals the significance of bridging water molecules that stabilize and mediate side chain-side chain, side chain-main chain and main chain-main chain interactions at the solvation interface.

All-atom molecular dynamics simulations of proteins in solvent are now able to realistically map the protein-unfolding pathway. The agreement with experiments probing both folding and unfolding suggests that these simulated unfolding events also shed light on folding. The simulations have produced detailed models of protein folding transition, intermediate, and denatured states that are in both qualitative and quantitative agreement with experiment. The various studies presented here highlight how such simulations both complement and extend experiment.

Experiment and simulation are now conspiring to give atomic-level descriptions of protein folding relevant to folding, misfolding, trafficking, and degradation in the cell. We are on the threshold of predicting those protein folding events using simulation that has been carefully benchmarked by experiment.

Trimethylamine n-oxide (TMAO) is a naturally occurring osmolyte that stabilizes proteins and offsets the destabilizing effects of urea. To investigate the molecular mechanism of these effects, we have studied the thermodynamics of interaction between TMAO and protein functional groups. The solubilities of a homologous series of cyclic dipeptides were measured by differential refractive index and the dissolution heats were determined calorimetrically as a function of TMAO concentration at 25 degrees C. The transfer free energy of the amide unit (-CONH-) from water to 1 M TMAO is large and positive, indicating an unfavorable interaction between the TMAO solution and the amide unit. This unfavorable interaction is enthalpic in origin. The interaction between TMAO and apolar groups is slightly favorable. The transfer free energy of apolar groups from water to TMAO consists of favorable enthalpic and unfavorable entropic contributions. This is in contrast to the contributions for the interaction between urea and apolar groups. Molecular dynamics simulations were performed to provide a structural framework for the interpretation of these results. The simulations show enhancement of water structure by TMAO in the form of a slight increase in the number of hydrogen bonds per water molecule, stronger water hydrogen bonds, and long-range spatial ordering of the solvent. These findings suggest that TMAO stabilizes proteins via enhancement of water structure, such that interactions with the amide unit are discouraged.