ModBase, a database of annotated comparative protein structure models and associated resources.

ModBase (http://salilab.org/modbase) is a database of annotated comparative protein structure models. The models are calculated by ModPipe, an automated modeling pipeline that relies primarily on Modeller for fold assignment, sequence-structure alignment, model building and model assessment (http://salilab.org/modeller/). ModBase currently contains almost 30 million reliable models for domains in 4.7 million unique protein sequences. ModBase allows users to compute or update comparative models on demand, through an interface to the ModWeb modeling server (http://salilab.org/modweb). ModBase models are also available through the Protein Model Portal (http://www.proteinmodelportal.org/). Recently developed associated resources include the AllosMod server for modeling ligand-induced protein dynamics (http://salilab.org/allosmod), the AllosMod-FoXS server for predicting a structural ensemble that fits an SAXS profile (http://salilab.org/allosmod-foxs), the FoXSDock server for protein-protein docking filtered by an SAXS profile (http://salilab.org/foxsdock), the SAXS Merge server for automatic merging of SAXS profiles (http://salilab.org/saxsmerge) and the Pose & Rank server for scoring protein-ligand complexes (http://salilab.org/poseandrank). In this update, we also highlight two applications of ModBase: a PSI:Biology initiative to maximize the structural coverage of the human alpha-helical transmembrane proteome and a determination of structural determinants of human immunodeficiency virus-1 protease specificity.

Structure-based model of allostery predicts coupling between distant sites.

Allostery is a phenomenon that couples effector ligand binding at an allosteric site to a structural and/or dynamic change at a distant regulated site. To study an allosteric transition, we vary the size of the allosteric site and its interactions to construct a series of energy landscapes with pronounced minima corresponding to both the effector bound and unbound crystal structures. We use molecular dynamics to sample these landscapes. The degree of perturbation by the effector, modeled by the size of the allosteric site, provides an order parameter for allostery that allows us to determine how microscopic motions give rise to commonly discussed macroscopic mechanisms: (i) induced fit, (ii) population shift, and (iii) entropy driven. These mechanisms involve decreasing structural differences between the effector bound and unbound populations. A metric (ligand-induced cooperativity) can measure how cooperatively a given regulated site responds to effector binding and therefore what kind of allosteric mechanism is involved. We apply the model to three proteins with experimentally characterized transitions: (i) calmodulin-GFP Ca(2+) sensor protein, (ii) maltose binding protein, and (iii) CSL transcription factor. Remarkably, the model is able to reproduce allosteric motion and predict coupling in a manner consistent with experiment.

Electrostatic effects on funneled landscapes and structural diversity in denatured protein ensembles.

The denatured state of proteins is heterogeneous and susceptible to general hydrophobic and electrostatic forces, but to what extent does the funneled nature of protein energy landscapes play a role in the unfolded ensemble? We simulate the denatured ensemble of cytochrome c using a series of models. The models pinpoint the efficacy of incorporating energetic funnels toward the native state in contrast with models having no native structure-seeking tendency. These models also contain varying strengths of electrostatic effects and hydrophobic collapse. The simulations based on these models are compared with experimental distributions for the distances between a fluorescent donor and the heme acceptor that were extracted from time-resolved fluorescence energy transfer experiments on cytochrome c. Comparing simulations to detailed experimental data on several labeling sites allows us to quantify the dominant forces in denatured protein ensembles.

Prediction of native-state hydrogen exchange from perfectly funneled energy landscapes.

Simulations based on perfectly funneled energy landscapes often capture many of the kinetic features of protein folding. We examined whether simulations based on funneled energy functions can also describe fluctuations in native-state protein ensembles. We quantitatively compared the site-specific local stability determined from structure-based folding simulations, with hydrogen exchange protection factors measured experimentally for ubiquitin, chymotrypsin inhibitor 2, and staphylococcal nuclease. Different structural definitions for the open and closed states based on the number of native contacts for each residue, as well as the hydrogen-bonding state, or a combination of both criteria were evaluated. The predicted exchange patterns agree with the experiments under native conditions, indicating that protein topology indeed has a dominant effect on the exchange kinetics. Insights into the simplest mechanistic interpretation of the amide exchange process were thus obtained.

The folding energy landscape and free energy excitations of cytochrome c.

The covalently bound heme cofactor plays a dominant role in the folding of cytochrome c. Because of the complicated inorganic chemistry of the heme, some might consider the folding of cytochrome c to be a special case, following principles different from those used to describe the folding of proteins without cofactors. Recent investigations, however, demonstrate that common models describing folding for many proteins work well for cytochrome c when heme is explicitly introduced, generally providing results that agree with experimental observations. In this Account, we first discuss results from simple native structure-based models. These models include attractive interactions between nonadjacent residues only if they are present in the crystal structure at pH 7. Because attractive nonnative contacts are not included in native structure-based models, their energy landscapes can be described as "perfectly funneled". In other words, native structure-based models are energetically guided towards the native state and contain no energetic traps that would hinder folding. Energetic traps are denoted sources of "frustration", which cause specific transient intermediates to be populated. Native structure-based models do, however, include repulsion between residues due to excluded volume. Nonenergetic traps can therefore exist if the chain, which cannot cross over itself, must partially unfold so that folding can proceed. The ability of native structure-based models to capture this kind of motion is partly responsible for their successful predictions of folding pathways for many types of proteins. Models without frustration describe the sequence of folding events for cytochrome c well (as inferred from hydrogen-exchange experiments), thereby justifying their use as a starting point. At low pH, the experimentally observed folding sequence of cytochrome c deviates from that at pH 7 and from models with perfectly funneled energy landscapes. Here, alternate folding pathways are a result of "chemical frustration". This frustration arises because some regions of the protein are destabilized more than others due to the heterogeneous distribution of titratable residues that are protonated at low pH. Beginning with native structure-based terms, we construct more complex models by adding chemical frustration. These more complex models only modestly perturb the energy landscape, which remains, overall, well funneled. These perturbed models can accurately describe how alternative folding pathways are used at low pH. At alkaline pH, cytochrome c populates distinctly different structural ensembles. For instance, lysine residues are deprotonated and compete for the heme ligation site. The same models that can describe folding at low pH also predict well the structures and relative stabilities of intermediates populated at alkaline pH. The success of models based on funneled energy landscapes suggest that cytochrome c folding is driven primarily by native contacts. The presence of heme appears to add chemical complexity to the folding process, but it does not require fundamental modification of the general principles used to describe folding. Moreover, its added complexity provides a valuable means of probing the folding energy landscape in greater detail than is possible with simpler systems.

Origins of barriers and barrierless folding in BBL.

In contrast to classical chemical phenomenology, theory suggests that proteins may undergo downhill folding without an activation barrier under certain thermodynamic conditions. Recently, the BBL protein was proposed to fold by such a downhill scenario, but discrepancies between experimental results found in different groups argue against this. After briefly reviewing the major experimental studies of the BBL folding mechanism, we show that simulations of both coarse-grained and atomistic models can reconcile the seemingly conflicting observations.

Chemical frustration in the protein folding landscape: grand canonical ensemble simulations of cytochrome c.

A grand canonical formalism is developed to combine discrete simulations for chemically distinct species in equilibrium. Each simulation is based on a perturbed funneled landscape. The formalism is illustrated using the alkaline-induced transitions of cytochrome c as observed by FTIR spectroscopy and with various other experimental approaches. The grand canonical simulation method accounts for the acid/base chemistry of deprotonation, the inorganic chemistry of heme ligation and misligation, and the minimally frustrated folding energy landscape, thus elucidating the physics of protein folding involved with an acid/base titration of a protein. The formalism combines simulations for each of the relevant chemical species, varying by protonation and ligation states. In contrast to models based on perfectly funneled energy landscapes that contain only contacts found in the native structure, this study introduces "chemical frustration" from deprotonation and misligation that gives rise to many intermediates at alkaline pH. While the nature of these intermediates cannot be easily inferred from available experimental data, this study provides specific structural details of these intermediates, thus extending our understanding of how cytochrome c changes with an increase in pH. The results demonstrate the importance of chemical frustration for understanding biomolecular energy landscapes.

Characterization of alkaline transitions in ferricytochrome c using carbon-deuterium infrared probes.

The alkaline-induced structural transitions of ferricytochrome c have been studied intensively as a model for how changes in metal ligation contribute to protein function and folding. Previous studies have demonstrated that multiple non-native species accumulate with increasing pH. Here, we used a combination of experiments and simulations to provide a high-resolution view of the changes associated with increasing alkaline conditions. Alkaline-induced transitions were characterized under equilibrium conditions by following changes in the IR absorptions of carbon-deuterium chromophores incorporated at Leu68, Lys72, Lys73, Lys79, and Met80. The data suggest that at least four intermediates are formed as the pH is increased prior to complete unfolding of the protein. The first alkaline transition observed appears to be driven by a single deprotonation and occurs with a midpoint of pH 8.8, but surprisingly, the intermediate formed does not appear to be one of the well-characterized lysine misligates. At higher pH, second and third deprotonations, with a combined apparent midpoint pH of 10.2, induce transitions to Lys73- or Lys79-misligated species. Interestingly, the lysine misligates appear to undergo iron reduction by the coordinated amine. A transition from the lysine misligates to another intermediate, likely a hydroxide-misligated species, is associated with a fourth deprotonation and a midpoint of pH 10.7. Finally, the protein loses tertiary structure with a fifth deprotonation that occurs with a midpoint of pH 12.7. Native topology-based models with enforced misligation are employed to help understand the structures of the observed intermediates.

CryptoSite: Expanding the Druggable Proteome by Characterization and Prediction of Cryptic Binding Sites.

Many proteins have small-molecule binding pockets that are not easily detectable in the ligand-free structures. These cryptic sites require a conformational change to become apparent; a cryptic site can therefore be defined as a site that forms a pocket in a holo structure, but not in the apo structure. Because many proteins appear to lack druggable pockets, understanding and accurately identifying cryptic sites could expand the set of drug targets. Previously, cryptic sites were identified experimentally by fragment-based ligand discovery and computationally by long molecular dynamics simulations and fragment docking. Here, we begin by constructing a set of structurally defined apo-holo pairs with cryptic sites. Next, we comprehensively characterize the cryptic sites in terms of their sequence, structure, and dynamics attributes. We find that cryptic sites tend to be as conserved in evolution as traditional binding pockets but are less hydrophobic and more flexible. Relying on this characterization, we use machine learning to predict cryptic sites with relatively high accuracy (for our benchmark, the true positive and false positive rates are 73% and 29%, respectively). We then predict cryptic sites in the entire structurally characterized human proteome (11,201 structures, covering 23% of all residues in the proteome). CryptoSite increases the size of the potentially "druggable" human proteome from ~40% to ~78% of disease-associated proteins. Finally, to demonstrate the utility of our approach in practice, we experimentally validate a cryptic site in protein tyrosine phosphatase 1B using a covalent ligand and NMR spectroscopy. The CryptoSite Web server is available at http://salilab.org/cryptosite.

Mapping polymerization and allostery of hemoglobin S using point mutations.

Hemoglobin is a complex system that undergoes conformational changes in response to oxygen, allosteric effectors, mutations, and environmental changes. Here, we study allostery and polymerization of hemoglobin and its variants by application of two previously described methods: (i) AllosMod for simulating allostery dynamics given two allosterically related input structures and (ii) a machine-learning method for dynamics- and structure-based prediction of the mutation impact on allostery (Weinkam et al. J. Mol. Biol. 2013, 425, 647-661), now applicable to systems with multiple coupled binding sites, such as hemoglobin. First, we predict the relative stabilities of substates and microstates of hemoglobin, which are determined primarily by entropy within our model. Next, we predict the impact of 866 annotated mutations on hemoglobin's oxygen binding equilibrium. We then discuss a subset of 30 mutations that occur in the presence of the sickle cell mutation and whose effects on polymerization have been measured. Seven of these HbS mutations occur in three predicted druggable binding pockets that might be exploited to directly inhibit polymerization; one of these binding pockets is not apparent in the crystal structure, but only in structures generated by AllosMod. For the 30 mutations, we predict that mutation-induced conformational changes within a single tetramer tend not to significantly impact polymerization; instead, these mutations more likely impact polymerization by directly perturbing a polymerization interface. Finally, our analysis of allostery allows us to hypothesize why hemoglobin evolved to have multiple subunits and a persistent low frequency sickle cell mutation.

All-atom ensemble modeling to analyze small-angle x-ray scattering of glycosylated proteins.

The flexible and heterogeneous nature of carbohydrate chains often renders glycoproteins refractory to traditional structure determination methods. Small-angle X-ray scattering (SAXS) can be a useful tool for obtaining structural information of these systems. All-atom modeling of glycoproteins with flexible glycan chains was applied to interpret the solution SAXS data for a set of glycoproteins. For simpler systems (single glycan, with a well-defined protein structure), all-atom modeling generates models in excellent agreement with the scattering pattern and reveals the approximate spatial occupancy of the glycan chain in solution. For more complex systems (several glycan chains, or unknown protein substructure), the approach can still provide insightful models, though the orientations of glycans become poorly determined. Ab initio shape reconstructions appear to capture the global morphology of glycoproteins but in most cases offer little information about glycan spatial occupancy. The all-atom modeling methodology is available as a web server at http://salilab.org/allosmod-foxs.

Impact of mutations on the allosteric conformational equilibrium.

Allostery in a protein involves effector binding at an allosteric site that changes the structure and/or dynamics at a distant, functional site. In addition to the chemical equilibrium of ligand binding, allostery involves a conformational equilibrium between one protein substate that binds the effector and a second substate that less strongly binds the effector. We run molecular dynamics simulations using simple, smooth energy landscapes to sample specific ligand-induced conformational transitions, as defined by the effector-bound and effector-unbound protein structures. These simulations can be performed using our web server (http://salilab.org/allosmod/). We then develop a set of features to analyze the simulations and capture the relevant thermodynamic properties of the allosteric conformational equilibrium. These features are based on molecular mechanics energy functions, stereochemical effects, and structural/dynamic coupling between sites. Using a machine-learning algorithm on a data set of 10 proteins and 179 mutations, we predict both the magnitude and the sign of the allosteric conformational equilibrium shift by the mutation; the impact of a large identifiable fraction of the mutations can be predicted with an average unsigned error of 1k(B)T. With similar accuracy, we predict the mutation effects for an 11th protein that was omitted from the initial training and testing of the machine-learning algorithm. We also assess which calculated thermodynamic properties contribute most to the accuracy of the prediction.

The spectrum of biomolecular states and motions.

The universe of conformational substates of a protein molecule is huge. The complete energy landscape of proteins is, therefore, complex when studied at low temperature. Many experiments under physiological conditions commonly reveal a simpler spectrum of states. These states are individually ensembles of low temperature substates. That is, room temperature experiments probe the low free energy part of the spectrum of excitations. This paper describes how the complete landscape and the spectrum of these thermally excited motions can be related to each other. On funneled landscapes, partially folded ensembles of states are the most important excited states. Their properties and their free energy spectrum can often be predicted by native topology based models. Frustration, i.e., the conflict between inconsistent stabilizing interactions that have evolved for other purposes than optimizing folding, offers another mechanism for forming low free energy excitations. Frustration can be localized and quantified using energy landscape theory. Symmetry provides an obvious route to low free energy states in oligomeric systems, where simply repositioning parts of the molecule in ways quasi-equivalent to their relation in the native structure gives nearly degenerate energies.

A funneled energy landscape for cytochrome c directly predicts the sequential folding route inferred from hydrogen exchange experiments.

Proteins fold through a variety of mechanisms. For a given protein, folding routes largely depend on the protein's stability and its native-state geometry, because the landscape is funneled. These ideas are corroborated for cytochrome c by using a coarse-grained topology-based model with a perfect funnel landscape that includes explicit modeling of the heme. The results show the importance of the heme as a nucleation site and explain the observed hydrogen exchange patterns of cytochrome c within the context of energy landscape theory.

Building programmable jigsaw puzzles with RNA.

One challenge in supramolecular chemistry is the design of versatile, self-assembling building blocks to attain total control of arrangement of matter at a molecular level. We have achieved reliable prediction and design of the three-dimensional structure of artificial RNA building blocks to generate molecular jigsaw puzzle units called tectosquares. They can be programmed with control over their geometry, topology, directionality, and addressability to algorithmically self-assemble into a variety of complex nanoscopic fabrics with predefined periodic and aperiodic patterns and finite dimensions. This work emphasizes the modular and hierarchical characteristics of RNA by showing that small RNA structural motifs can code the precise topology of large molecular architectures. It demonstrates that fully addressable materials based on RNA can be synthesized and provides insights into self-assembly processes involving large populations of RNA molecules.