Suwen Zhao

The VSGB 2.0 Model: A Next Generation Energy Model for High Resolution Protein Structure Modeling

A novel energy model (VSGB 2.0) for high resolution protein structure modeling is described, which features an optimized implicit solvent model as well as physics‐based corrections for hydrogen bonding, π–π interactions, self‐contact interactions, and hydrophobic interactions. Parameters of the VSGB 2.0 model were fit to a crystallographic database of 2239 single side chain and 100 11–13 residue loop predictions. Combined with an advanced method of sampling and a robust algorithm for protonation state assignment, the VSGB 2.0 model was validated by predicting 115 super long loops up to 20 residues. Despite the dramatically increasing difficulty in reconstructing longer loops, a high accuracy was achieved: all of the lowest energy conformations have global backbone RMSDs better than 2.0 Å from the native conformations. Average global backbone RMSDs of the predictions are 0.51, 0.63, 0.70, 0.62, 0.80, 1.41, and 1.59 Å for 14, 15, 16, 17, 18, 19, and 20 residue loop predictions, respectively. When these results are corrected for possible statistical bias as explained in the text, the average global backbone RMSDs are 0.61, 0.71, 0.86, 0.62, 1.06, 1.67, and 1.59 Å. Given the precision and robustness of the calculations, we believe that the VSGB 2.0 model is suitable to tackle “real” problems, such as biological function modeling and structure‐based drug discovery. Proteins 2011;

Long loop prediction using the protein local optimization program

We have developed an improved sampling algorithm and energy model for protein loop prediction, the combination of which has yielded the first methodology capable of achieving good results for the prediction of loop backbone conformations of 11 residue length or greater. Applied to our newly constructed test suite of 104 loops ranging from 11 to 13 residues, our method obtains average/median global backbone root‐mean‐square deviations (RMSDs) to the native structure (superimposing the body of the protein, not the loop itself) of 1.00/0.62 Å for 11 residue loops, 1.15/0.60 Å for 12 residue loops, and 1.25/0.76 Å for 13 residue loops. Sampling errors are virtually eliminated, while energy errors leading to large backbone RMSDs are very infrequent compared to any previously reported efforts, including our own previous study. We attribute this success to both an improved sampling algorithm and, more critically, the inclusion of a hydrophobic term, which appears to approximately fix a major flaw in SGB solvation model that we have been employing. A discussion of these results in the context of the general question of the accuracy of continuum solvation models is presented. Proteins 2006.

Toward better refinement of comparative models: Predicting loops in inexact environments

Achieving atomic‐level accuracy in comparative protein models is limited by our ability to refine the initial, homolog‐derived model closer to the native state. Despite considerable effort, progress in developing a generalized refinement method has been limited. In contrast, methods have been described that can accurately reconstruct loop conformations in native protein structures. We hypothesize that loop refinement in homology models is much more difficult than loop reconstruction in crystal structures, in part, because side‐chain, backbone, and other structural inaccuracies surrounding the loop create a challenging sampling problem; the loop cannot be refined without simultaneously refining adjacent portions. In this work, we single out one sampling issue in an artificial but useful test set and examine how loop refinement accuracy is affected by errors in surrounding side‐chains. In 80 high‐resolution crystal structures, we first perturbed 6–12 residue loops away from the crystal conformation, and placed all protein side chains in non‐native but low energy conformations. Even these relatively small perturbations in the surroundings made the loop prediction problem much more challenging. Using a previously published loop prediction method, median backbone (N‐Cα‐C‐O) RMSDs for groups of 6, 8, 10, and 12 residue loops are 0.3/0.6/0.4/0.6 Å, respectively, on native structures and increase to 1.1/2.2/1.5/2.3 Å on the perturbed cases. We then augmented our previous loop prediction method to simultaneously optimize the rotamer states of side chains surrounding the loop. Our results show that this augmented loop prediction method can recover the native state in many perturbed structures where the previous method failed; the median RMSDs for the 6, 8, 10, and 12 residue perturbed loops improve to 0.4/0.8/1.1/1.2 Å. Finally, we highlight three comparative models from blind tests, in which our new method predicted loops closer to the native conformation than first modeled using the homolog template, a task generally understood to be difficult. Although many challenges remain in refining full comparative models to high accuracy, this work offers a methodical step toward that goal. Proteins 2008.

Discovery of new enzymes and metabolic pathways by using structure and genome context

Assigning valid functions to proteins identified in genome projects is challenging: overprediction and database annotation errors are the principal concerns. We and others are developing computation-guided strategies for functional discovery with ‘metabolite docking’ to experimentally derived or homology-based three-dimensional structures. Bacterial metabolic pathways often are encoded by ‘genome neighbourhoods’ (gene clusters and/or operons), which can provide important clues for functional assignment. We recently demonstrated the synergy of docking and pathway context by ‘predicting’ the intermediates in the glycolytic pathway in Escherichia coli. Metabolite docking to multiple binding proteins and enzymes in the same pathway increases the reliability of in silico predictions of substrate specificities because the pathway intermediates are structurally similar. Here we report that structure-guided approaches for predicting the substrate specificities of several enzymes encoded by a bacterial gene cluster allowed the correct prediction of the in vitro activity of a structurally characterized enzyme of unknown function (PDB 2PMQ), 2-epimerization of trans-4-hydroxy-l-proline betaine (tHyp-B) and cis-4-hydroxy-d-proline betaine (cHyp-B), and also the correct identification of the catabolic pathway in which Hyp-B 2-epimerase participates. The substrate-liganded pose predicted by virtual library screening (docking) was confirmed experimentally. The enzymatic activities in the predicted pathway were confirmed by in vitro assays and genetic analyses; the intermediates were identified by metabolomics; and repression of the genes encoding the pathway by high salt concentrations was established by transcriptomics, confirming the osmolyte role of tHyp-B. This study establishes the utility of structure-guided functional predictions to enable the discovery of new metabolic pathways.

Crystal structures of agonist-bound human cannabinoid receptor CB1

The cannabinoid receptor 1 (CB1) is the principal target of the psychoactive constituent of marijuana, the partial agonist Δ9-tetrahydrocannabinol (Δ9-THC). Here we report two agonist-bound crystal structures of human CB1 in complex with a tetrahydrocannabinol (AM11542) and a hexahydrocannabinol (AM841) at 2.80 Å and 2.95 Å resolution, respectively. The two CB1–agonist complexes reveal important conformational changes in the overall structure, relative to the antagonist-bound state, including a 53% reduction in the volume of the ligand-binding pocket and an increase in the surface area of the G-protein-binding region. In addition, a ‘twin toggle switch’ of Phe2003.36 and Trp3566.48 (superscripts denote Ballesteros–Weinstein numbering) is experimentally observed and appears to be essential for receptor activation. The structures reveal important insights into the activation mechanism of CB1 and provide a molecular basis for predicting the binding modes of Δ9-THC, and endogenous and synthetic cannabinoids. The plasticity of the binding pocket of CB1 seems to be a common feature among certain class A G-protein-coupled receptors. These findings should inspire the design of chemically diverse ligands with distinct pharmacological properties.

Prediction and characterization of enzymatic activities guided by sequence similarity and genome neighborhood networks

Metabolic pathways in eubacteria and archaea often are encoded by operons and/or gene clusters (genome neighborhoods) that provide important clues for assignment of both enzyme functions and metabolic pathways. We describe a bioinformatic approach (genome neighborhood network; GNN) that enables large scale prediction of the in vitro enzymatic activities and in vivo physiological functions (metabolic pathways) of uncharacterized enzymes in protein families. We demonstrate the utility of the GNN approach by predicting in vitro activities and in vivo functions in the proline racemase superfamily (PRS; InterPro IPR008794). The predictions were verified by measuring in vitro activities for 51 proteins in 12 families in the PRS that represent ∼85% of the sequences; in vitro activities of pathway enzymes, carbon/nitrogen source phenotypes, and/or transcriptomic studies confirmed the predicted pathways. The synergistic use of sequence similarity networks3 and GNNs will facilitate the discovery of the components of novel, uncharacterized metabolic pathways in sequenced genomes. DOI: http://dx.doi.org/10.7554/eLife.03275.001

Experimental strategies for functional annotation and metabolism discovery: targeted screening of solute binding proteins and unbiased panning of metabolomes.

The rate at which genome sequencing data is accruing demands enhanced methods for functional annotation and metabolism discovery. Solute binding proteins (SBPs) facilitate the transport of the first reactant in a metabolic pathway, thereby constraining the regions of chemical space and the chemistries that must be considered for pathway reconstruction. We describe high-throughput protein production and differential scanning fluorimetry platforms, which enabled the screening of 158 SBPs against a 189 component library specifically tailored for this class of proteins. Like all screening efforts, this approach is limited by the practical constraints imposed by construction of the library, i.e., we can study only those metabolites that are known to exist and which can be made in sufficient quantities for experimentation. To move beyond these inherent limitations, we illustrate the promise of crystallographic- and mass spectrometric-based approaches for the unbiased use of entire metabolomes as screening libraries. Together, our approaches identified 40 new SBP ligands, generated experiment-based annotations for 2084 SBPs in 71 isofunctional clusters, and defined numerous metabolic pathways, including novel catabolic pathways for the utilization of ethanolamine as sole nitrogen source and the use of d-Ala-d-Ala as sole carbon source. These efforts begin to define an integrated strategy for realizing the full value of amassing genome sequence data.

Assignment of polar states for protein amino acid residues using an interaction cluster decomposition algorithm and its application to high resolution protein structure modeling

We have developed a new method (Independent Cluster Decomposition Algorithm, ICDA) for creating all‐atom models of proteins given the heavy‐atom coordinates, provided by X‐ray crystallography, and the pH. In our method the ionization states of titratable residues, the crystallographic mis‐assignment of amide orientations in Asn/Gln, and the orientations of OH/SH groups are addressed under the unified framework of polar states assignment. To address the large number of combinatorial possibilities for the polar hydrogen states of the protein, we have devised a novel algorithm to decompose the system into independent interacting clusters, based on the observation of the crucial interdependence between the short range hydrogen bonding network and polar residue states, thus significantly reducing the computational complexity of the problem and making our algorithm tractable using relatively modest computational resources. We utilize an all atom protein force field (OPLS) and a Generalized Born continuum solvation model, in contrast to the various empirical force fields adopted in most previous studies. We have compared our prediction results with a few well‐documented methods in the literature (WHATIF, REDUCE). In addition, as a preliminary attempt to couple our polar state assignment method with real structure predictions, we further validate our method using single side chain prediction, which has been demonstrated to be an effective way of validating structure prediction methods without incurring sampling problems. Comparisons of single side chain prediction results after the application of our polar state prediction method with previous results with default polar state assignments indicate a significant improvement in the single side chain predictions for polar residues. Proteins 2007.

Human GLP-1 receptor transmembrane domain structure in complex with allosteric modulators

The glucagon-like peptide-1 receptor (GLP-1R) and the glucagon receptor (GCGR) are members of the secretin-like class B family of G-protein-coupled receptors (GPCRs) and have opposing physiological roles in insulin release and glucose homeostasis. The treatment of type 2 diabetes requires positive modulation of GLP-1R to inhibit glucagon secretion and stimulate insulin secretion in a glucose-dependent manner. Here we report crystal structures of the human GLP-1R transmembrane domain in complex with two different negative allosteric modulators, PF-06372222 and NNC0640, at 2.7 and 3.0 Å resolution, respectively. The structures reveal a common binding pocket for negative allosteric modulators, present in both GLP-1R and GCGR and located outside helices V–VII near the intracellular half of the receptor. The receptor is in an inactive conformation with compounds that restrict movement of the intracellular tip of helix VI, a movement that is generally associated with activation mechanisms in class A GPCRs. Molecular modelling and mutagenesis studies indicate that agonist positive allosteric modulators target the same general region, but in a distinct sub-pocket at the interface between helices V and VI, which may facilitate the formation of an intracellular binding site that enhances G-protein coupling.

Progress in Super Long Loop Prediction

Sampling errors are very common in super long loop (referring here to loops that have more than thirteen residues) prediction, simply because the sampling space is vast. We have developed a dipeptide segment sampling algorithm to solve this problem. As a first step in evaluating the performance of this algorithm, it was applied to the problem of reconstructing loops in native protein structures. With a newly constructed test set of 89 loops ranging from 14 to 17 residues, this method obtains average/median global backbone root‐mean‐square deviations (RMSDs) to the native structure (superimposing the body of the protein, not the loop itself) of 1.46/0.68 Å. Specifically, results for loops of various lengths are 1.19/0.67 Å for 36 fourteen‐residue loops, 1.55/0.75 Å for 30 fifteen‐residue loops, 1.43/0.80 Å for 14 sixteen‐residue loops, and 2.30/1.92 Å for nine seventeen‐residue loops. In the vast majority of cases, the method locates energy minima that are lower than or equal to that of the minimized native loop, thus indicating that the new sampling method is successful and rarely limits prediction accuracy. Median RMSDs are substantially lower than the averages because of a small number of outliers. The causes of these failures are examined in some detail, and some can be attributed to flaws in the energy function, such as π–π interactions are not accurately accounted for by the OPLS‐AA force field we employed in this study. By introducing a new energy model which has a superior description of π–π interactions, significantly better results were achieved for quite a few former outliers. Crystal packing is explicitly included in order to provide a fair comparison with crystal structures. Proteins 2011;.