Docking and Scoring with Target-Specific Pose Classifier Succeeds in Native-Like Pose Identification But Not Binding Affinity Prediction in the CSAR 2014 Benchmark Exercise.

The CSAR 2014 exercise provided an important benchmark for testing current approaches for pose identification and ligand ranking using three X-ray characterized proteins: Factor Xa (FXa), Spleen Tyrosine Kinase (SYK), and tRNA Methyltransferase (TRMD). In Phase 1 of the exercise, we employed Glide and MedusaDock docking software, both individually and in combination, with the special target-specific pose classifier trained to discriminate native-like from decoy poses. All approaches succeeded in the accurate detection of native and native-like poses. We then used Glide SP and MedusaScore scoring functions individually and in combination with the pose-scoring approach to predict relative binding affinities of the congeneric series of ligands in Phase 2 of the exercise. Similar to other participants in the CSAR 2014 exercise, we found that our models showed modest prediction accuracy. Quantitative structure-activity relationship (QSAR) models developed for the FXa ligands using available bioactivity data from ChEMBL showed relatively low prediction accuracy for the CSAR 2014 ligands of the same target. Interestingly, QSAR models built with CSAR data only yielded Spearman correlation coefficients as high as ρ = 0.69 for FXa and ρ = 0.79 for SYK based on 5-fold cross-validation. Virtual screening of the DUD library using the FXa structure was successful in discriminating between active compounds and decoys in spite of poor ranking accuracy of the underlying scoring functions. Our results suggest that two of the three common tasks associated with molecular docking, i.e., native-like pose identification and virtual screening, but not binding affinity prediction, could be accomplished successfully for the CSAR 2014 challenge data set.

Cell-based and in silico evidence against quercetin and structurally-related flavonols as activators of vitamin D receptor.

It has been reported that quercetin is an activator of rat vitamin D receptor (rVDR). However, the conclusion was based on experiments performed without all the appropriate control groups, raising the possibility of a false-positive finding. Furthermore, distinct differences exist in the chemical structures of quercetin and 1α,25-dihydroxyvitamin D3, which is a prototypic agonist of VDR. Therefore, we investigated systematically whether quercetin and other flavonols are agonists of rVDR, mouse VDR (mVDR), or human VDR (hVDR). Quercetin, 3-hydroxyflavone, galangin, datiscetin, kaempferol, morin, isorhamnetin, tamarixetin, myricetin, and syringetin did not activate rVDR, mVDR, or hVDR in HEK-293 and HepG2 cells transfected with the corresponding receptor expression plasmid and either the secreted phosphoprotein 1 (Spp1) or cytochrome P450 24A1 (CYP24A1) reporter plasmid, when compared to the respective empty vector control group transfected with one or the other reporter plasmid and treated with one of the flavonols. Control analysis indicated that lithocholic acid and 1α,25-dihydroxyvitamin D3, but not rifampicin, activated rVDR, mVDR, and hVDR. As shown in transfected HEK293 and HepG2 cells, the flavonols did not influence hVDR ligand binding domain transactivation, steroid receptor coactivator-1 recruitment, or hVDR target gene expression (transient receptor potential cation channel 6 and CYP24A1) in hVDR-expressing Caco-2 or LS180 cells. The cumulative data from the cell-based experiments were corroborated by results obtained from molecular docking analysis. In conclusion, quercetin, 3-hydroxyflavone, galangin, datiscetin, kaempferol, morin, isorhamnetin, tamarixetin, myricetin, and syringetin are not agonists of rVDR, mVDR, or hVDR, as judged by cell-based and in silico evidence.

CERAPP: Collaborative Estrogen Receptor Activity Prediction Project.

BACKGROUND: Humans are exposed to thousands of man-made chemicals in the environment. Some chemicals mimic natural endocrine hormones and, thus, have the potential to be endocrine disruptors. Most of these chemicals have never been tested for their ability to interact with the estrogen receptor (ER). Risk assessors need tools to prioritize chemicals for evaluation in costly in vivo tests, for instance, within the U.S. EPA Endocrine Disruptor Screening Program. OBJECTIVES: We describe a large-scale modeling project called CERAPP (Collaborative Estrogen Receptor Activity Prediction Project) and demonstrate the efficacy of using predictive computational models trained on high-throughput screening data to evaluate thousands of chemicals for ER-related activity and prioritize them for further testing. METHODS: CERAPP combined multiple models developed in collaboration with 17 groups in the United States and Europe to predict ER activity of a common set of 32,464 chemical structures. Quantitative structure-activity relationship models and docking approaches were employed, mostly using a common training set of 1,677 chemical structures provided by the U.S. EPA, to build a total of 40 categorical and 8 continuous models for binding, agonist, and antagonist ER activity. All predictions were evaluated on a set of 7,522 chemicals curated from the literature. To overcome the limitations of single models, a consensus was built by weighting models on scores based on their evaluated accuracies. RESULTS: Individual model scores ranged from 0.69 to 0.85, showing high prediction reliabilities. Out of the 32,464 chemicals, the consensus model predicted 4,001 chemicals (12.3%) as high priority actives and 6,742 potential actives (20.8%) to be considered for further testing. CONCLUSION: This project demonstrated the possibility to screen large libraries of chemicals using a consensus of different in silico approaches. This concept will be applied in future projects related to other end points. CITATION: Mansouri K, Abdelaziz A, Rybacka A, Roncaglioni A, Tropsha A, Varnek A, Zakharov A, Worth A, Richard AM, Grulke CM, Trisciuzzi D, Fourches D, Horvath D, Benfenati E, Muratov E, Wedebye EB, Grisoni F, Mangiatordi GF, Incisivo GM, Hong H, Ng HW, Tetko IV, Balabin I, Kancherla J, Shen J, Burton J, Nicklaus M, Cassotti M, Nikolov NG, Nicolotti O, Andersson PL, Zang Q, Politi R, Beger RD, Todeschini R, Huang R, Farag S, Rosenberg SA, Slavov S, Hu X, Judson RS. 2016. CERAPP: Collaborative Estrogen Receptor Activity Prediction Project. Environ Health Perspect 124:1023-1033; http://dx.doi.org/10.1289/ehp.1510267.

Alarms about structural alerts.

Structural alerts are widely accepted in chemical toxicology and regulatory decision support as a simple and transparent means to flag potential chemical hazards or group compounds into categories for read-across. However, there has been a growing concern that alerts disproportionally flag too many chemicals as toxic, which questions their reliability as toxicity markers. Conversely, the rigorously developed and properly validated statistical QSAR models can accurately and reliably predict the toxicity of a chemical; however, their use in regulatory toxicology has been hampered by the lack of transparency and interpretability. We demonstrate that contrary to the common perception of QSAR models as "black boxes" they can be used to identify statistically significant chemical substructures (QSAR-based alerts) that influence toxicity. We show through several case studies, however, that the mere presence of structural alerts in a chemical, irrespective of the derivation method (expert-based or QSAR-based), should be perceived only as hypotheses of possible toxicological effect. We propose a new approach that synergistically integrates structural alerts and rigorously validated QSAR models for a more transparent and accurate safety assessment of new chemicals.

Target-specific native/decoy pose classifier improves the accuracy of ligand ranking in the CSAR 2013 benchmark.

As part of the CSAR 2013 benchmark exercise, we have implemented a hybrid docking and scoring workflow to rank 10 steroid ligands of an engineered digoxigenin-binding protein. Schrödinger's Glide docking software was used to generate poses for each steroid ligand and rank them according to both standard docking precision (SP) and extra docking precision (XP) scoring functions. The unique component of our approach was the use of a target-specific pose classifier trained to discriminate nativelike from decoy poses. To build the classifier, a single cognate ligand with a known native pose (PDB code 4J8T) was docked multiple times into its target protein, and the generated poses were divided into two classes (nativelike and decoy) using a root-mean-square deviation threshold of 2 Å. All of the poses were characterized by the MCT-Tess descriptors of the protein-ligand interface, and random forest (RF) models were trained to discriminate the two classes of poses on the basis of their descriptors. The consensus pose classifier was then applied to the Glide-generated poses of each CSAR ligand in order to filter out those poses predicted as decoys and rerank the remaining ones using both XP and SP scoring functions. The best-scoring pose for each ligand following this filtering step was used for final ligand ranking. Overall, the ranking accuracy for the 10 ligands evaluated by the Spearman correlation coefficient was 0.64 for SP and 0.52 for XP but reached 0.75 for SP/RF consensus scoring (ranked third in the CSAR 2013 benchmark exercise). This study reconfirms that target-specific pose scoring models are capable of enhancing the reliability of structure-based molecular docking by discarding decoy poses.

Prediction of binding affinity and efficacy of thyroid hormone receptor ligands using QSAR and structure-based modeling methods.

The thyroid hormone receptor (THR) is an important member of the nuclear receptor family that can be activated by endocrine disrupting chemicals (EDC). Quantitative Structure-Activity Relationship (QSAR) models have been developed to facilitate the prioritization of THR-mediated EDC for the experimental validation. The largest database of binding affinities available at the time of the study for ligand binding domain (LBD) of THRß was assembled to generate both continuous and classification QSAR models with an external accuracy of R(2)=0.55 and CCR=0.76, respectively. In addition, for the first time a QSAR model was developed to predict binding affinities of antagonists inhibiting the interaction of coactivators with the AF-2 domain of THRß (R(2)=0.70). Furthermore, molecular docking studies were performed for a set of THRß ligands (57 agonists and 15 antagonists of LBD, 210 antagonists of the AF-2 domain, supplemented by putative decoys/non-binders) using several THRß structures retrieved from the Protein Data Bank. We found that two agonist-bound THRß conformations could effectively discriminate their corresponding ligands from presumed non-binders. Moreover, one of the agonist conformations could discriminate agonists from antagonists. Finally, we have conducted virtual screening of a chemical library compiled by the EPA as part of the Tox21 program to identify potential THRß-mediated EDCs using both QSAR models and docking. We concluded that the library is unlikely to have any EDC that would bind to the THRß. Models developed in this study can be employed either to identify environmental chemicals interacting with the THR or, conversely, to eliminate the THR-mediated mechanism of action for chemicals of concern.

Diversity in the mechanisms of cosolute action on biomolecular processes.

Numerous cellular cosolutes significantly impact the way that proteins and other biomacromolecules act and interact. We have followed the thermodynamic effect of several cosolute classes, including polymers, cellular osmolytes, and inorganic salts, on the stability of biomolecular folding and complexation. By comparing changes in free energy, enthalpy, and entropy upon cosolutes addition for these processes, we identify several thermodynamically distinct mechanisms. Surprisingly, even while many cosolutes display similar scaling of the change in stabilizing free energy with their concentration, a breakdown of this free energy into enthalpic and entropic contributions distinguishes different families of cosolutes. We discuss how these "thermodynamic fingerprints" can direct towards possible underlying mechanisms that govern the cosolute effect.

Crowding alone cannot account for cosolute effect on amyloid aggregation.

Amyloid fiber formation is a specific form of protein aggregation, often resulting from the misfolding of native proteins. Aimed at modeling the crowded environment of the cell, recent experiments showed a reduction in fibrillation halftimes for amyloid-forming peptides in the presence of cosolutes that are preferentially excluded from proteins and peptides. The effect of excluded cosolutes has previously been attributed to the large volume excluded by such inert cellular solutes, sometimes termed "macromolecular crowding". Here, we studied a model peptide that can fold to a stable monomeric ß-hairpin conformation, but under certain solution conditions aggregates in the form of amyloid fibrils. Using Circular Dichroism spectroscopy (CD), we found that, in the presence of polyols and polyethylene glycols acting as excluded cosolutes, the monomeric ß-hairpin conformation was stabilized with respect to the unfolded state. Stabilization free energy was linear with cosolute concentration, and grew with molecular volume, as would also be predicted by crowding models. After initiating the aggregation process with a pH jump, fibrillation in the presence and absence of cosolutes was followed by ThT fluorescence, transmission electron microscopy, and CD spectroscopy. Polyols (glycerol and sorbitol) increased the lag time for fibril formation and elevated the amount of aggregated peptide at equilibrium, in a cosolute size and concentration dependent manner. However, fibrillation rates remained almost unaffected by a wide range of molecular weights of soluble polyethylene glycols. Our results highlight the importance of other forces beyond the excluded volume interactions responsible for crowding that may contribute to the cosolute effects acting on amyloid formation.

Unraveling the Molecular Mechanism of Enthalpy Driven Peptide Folding by Polyol Osmolytes.

Many polyols and carbohydrates serve in different organisms as protective osmolytes that help to stabilize proteins in their native, functional state, even under a variety of environmental stresses. However, despite their important role, much of the molecular mechanism by which these osmolytes exert their action remains elusive. We have recently shown experimentally that, although polyols and carbohydrates are excluded from protein and peptide interfaces, as also expected for the known entropic "crowding" mechanism, the osmolyte folding action can in fact primarily be enthalpic in nature. To follow this newly resolved enthalpically driven stabilization mechanism, we report here on molecular dynamics simulations of a model peptide that can fold in solution into a ß-hairpin. In agreement with experiments, our simulations indicate that sorbitol, a representative polyol, promotes peptide folding by preferential exclusion. At the molecular level, simulations further show that peptide stabilization can be explained by sorbitol's perturbation of the solution hydrogen bonding network in the peptide first hydration shells. Consequently, fewer hydrogen bonds between peptide and solvating water are lost upon folding, and additional internal peptide hydrogen bonds are formed in the presence of sorbitol, while internal peptide and water-associated hydrogen bonds are strengthened, resulting in stabilization of the peptide folded state. We further find that changes in water orientational entropy are reduced upon folding in sorbitol solution, reflecting the struggle of water molecules to maintain optimal hydrogen bonding in the presence of competing polyols. By providing first molecular underpinnings for enthalpically driven osmolyte stabilization of peptides and proteins, this mechanism should allow a better understanding of the variety of physical forces by which protective osmolytes act in biologically realistic solutions.

Enthalpically driven peptide stabilization by protective osmolytes.

We determined the added enthalpic and entropic contributions of protective osmolytes to the folding of a model peptide into its native beta-hairpin state. In contrast to entropically driven steric "crowding", this study shows that sugars and polyols can act as protective osmolytes by primarily diminishing the unfavourable enthalpic contribution to folding.

The impact of polyols on water structure in solution: a computational study.

Using molecular dynamics simulations, we study the effect of polyalcohols on water structuring in concentrated solutions, comparing six different polyols that vary in the number of hydroxyl groups and internal structure. For all polyols, we find that the hydrogen bond network order, as assessed by changes in the tetrahedral order parameter, is distorted in the binary solutions as compared with that of pure water and depends on the number of hydroxyl groups and the polyol conformation. While the total number of hydrogen bonds is only slightly reduced relative to that found in pure water, we find that hydrogen bonds that form with polyols tend to be less linear than hydrogen bonds formed between water molecules. We suggest that this reflects the competition between water and polyol molecules for hydrogen bonding with surrounding waters and offer a link between this competition and the resulting disorder that follows polyol solvation. The conclusions of this study should help shed light on the action that polyols can have as stabilizers to other macromolecules such as proteins in solution.

Computational design of calmodulin mutants with up to 900-fold increase in binding specificity.

Calmodulin (CaM) is a ubiquitous second messenger protein that regulates a variety of structurally and functionally diverse targets in response to changes in Ca(2+) concentration. CaM-dependent protein kinase II (CaMKII) and calcineurin (CaN) are the prominent CaM targets that play an opposing role in many cellular functions including synaptic regulation. Since CaMKII and CaN compete for the available Ca(2+)/CaM, the differential affinity of these enzymes for CaM is crucial for achieving a balance in Ca(2+) signaling. We used the computational protein design approach to modify CaM binding specificity for these two targets. Starting from the X-ray structure of CaM in complex with the CaM-binding domain of CaMKII, we optimized CaM interactions with CaMKII by introducing mutations into the CaM sequence. CaM optimization was performed with a protein design program, ORBIT, using a modified energy function that emphasized intermolecular interactions in the sequence selection procedure. Several CaM variants were experimentally constructed and tested for binding to the CaMKII and CaN peptides using the surface plasmon resonance technique. Most of our CaM mutants demonstrated small increase in affinity for the CaMKII peptide and substantial decrease in affinity for the CaN peptide compared to that of wild-type CaM. Our best CaM design exhibited an about 900-fold increase in binding specificity towards the CaMKII peptide, becoming the highest specificity switch achieved in any protein-protein interface through the computational protein design approach. Our results show that computational redesign of protein-protein interfaces becomes a reliable method for altering protein binding affinity and specificity.