Toxicity testing in the 21st century: implications for human health risk assessment

Use of Tox21 screening data to profile PFAS bioactivities on nuclear receptors, cellular stress pathways, and cytochrome p450 enzymes

Per- and poly-fluoroalkyl substances (PFAS) are synthetic chemicals widely used in commercial products. PFAS are a global concern due to their persistence in the environment and extensive associations with adverse health outcomes. While legacy PFAS have been extensively studied, many non-legacy PFAS lack sufficient toxicity information. In this study, we first analyzed the bioactivity of PFAS using Tox21 screening data surveying more than 75 assay endpoints (e.g., nuclear receptors, stress response, and metabolism) to understand the toxicity of non-legacy PFAS and investigate potential new targets of PFAS. From the Tox21 screening data analysis, we confirmed several known PFAS targets/pathways and identified several potential novel targets/pathways of PFAS. To confirm the effect of PFAS on these novel targets/pathways, we conducted several cell- and enzyme-based assays in the follow-up studies. We found PFAS inhibited cytochromes P450s (CYPs), especially CYP2C9 with IC50 values of < 1 µM. Considering PFAS affected other targets/pathways at > 10 µM, PFAS have a higher affinity to CYP2C9. This PFAS-CYP2C9 interaction was further investigated using molecular docking analysis. The result suggested that PFAS directly bind to the active sites of CYP2C9. These findings have important implications to understand the mechanism of PFAS action and toxicity.

Determination of the binding affinities of OPEs to integrin αvβ3 and elucidation of the underlying mechanisms via a competitive binding assay, pharmacophore modeling, molecular docking and QSAR modeling

Organophosphate esters (OPEs) can cause adverse biological effects through binding to integrin αvβ3. However, few studies have focused on the binding activity and mechanism of OPEs to integrin αvβ3. Herein, a comprehensive investigation of the mechanisms by which OPEs bind to integrin αvβ3 and determination of the binding affinity were conducted by in vitro and in silico approaches: competitive binding assay as well as pharmacophore, molecular docking and QSAR modeling. The results showed that all 18 OPEs exhibited binding activities to integrin αvβ3; moreover, hydrogen bonds were identified as crucial intermolecular interactions. In addition, essential factors, including the -P = O structure of OPEs, key amino acid residues and suitable cavity volume of integrin αvβ3, were identified to contribute to the formation of hydrogen bonds. Moreover, aryl-OPEs exhibited a lower binding activity with integrin αvβ3 than halogenated- and alkyl-OPEs. Ultimately, the QSAR model constructed in this study was effectively used to predict the binding affinity of OPEs to integrin αvβ3, and the results suggest that some OPEs might pose potential risks in aquatic environments. The results of this study comprehensively elucidated the binding mechanism of OPEs to integrin αvβ3, and supported the environmental risk management of these emerging pollutants.

High-Throughput Screening to Advance In Vitro Toxicology: Accomplishments, Challenges, and Future Directions

Traditionally, chemical toxicity is determined by in vivo animal studies, which are low throughput, expensive, and sometimes fail to predict compound toxicity in humans. Due to the increasing number of chemicals in use and the high rate of drug candidate failure due to toxicity, it is imperative to develop in vitro, high-throughput screening methods to determine toxicity. The Tox21 program, a unique research consortium of federal public health agencies, was established to address and identify toxicity concerns in a high-throughput, concentration-responsive manner using a battery of in vitro assays. In this article, we review the advancements in high-throughput robotic screening methodology and informatics processes to enable the generation of toxicological data, and their impact on the field; further, we discuss the future of assessing environmental toxicity utilizing efficient and scalable methods that better represent the corresponding biological and toxicodynamic processes in humans.

Recent Advances in Toxicity Prediction: Applications of Deep Graph Learning

The development of new drugs is time-consuming and expensive, and as such, accurately predicting the potential toxicity of a drug candidate is crucial in ensuring its safety and efficacy. Recently, deep graph learning has become prevalent in this field due to its computational power and cost efficiency. Many novel deep graph learning methods aid toxicity prediction and further prompt drug development. This review aims to connect fundamental knowledge with burgeoning deep graph learning methods. We first summarize the essential components of deep graph learning models for toxicity prediction, including molecular descriptors, molecular representations, evaluation metrics, validation methods, and data sets. Furthermore, based on various graph-related representations of molecules, we introduce several representative studies and methods for toxicity prediction from the perspective of GNN architectures and graph pretrained models. Compared to other types of models, deep graph models not only advance in higher accuracy and efficiency but also provide more intuitive insights, which is significant in the development of model interpretation and generalization ability. The graph pretrained models are emerging as they can extract prominent features from large-scale unlabeled molecular graph data and improve the performance of downstream toxicity prediction tasks. We hope this survey can serve as a handbook for individuals interested in exploring deep graph learning for toxicity prediction.

In vitro profiling of pesticides within the Tox21 10K compound library for bioactivity and potential toxicity

Pesticides include a diverse class of toxic chemicals, often having numerous modes of actions when used in agriculture against targeted organisms to control insect infestation, halt unwanted vegetation, and prevent the spread of disease. In this study, the in vitro assay activity of pesticides within the Tox21 10K compound library were examined. The assays in which pesticides showed significantly more activities than non-pesticide chemicals revealed potential targets and mechanisms of action for pesticides. Furthermore, pesticides that showed promiscuous activity against many targets and cytotoxicity were identified, which warrant further toxicological evaluation. Several pesticides were shown to require metabolic activation, demonstrating the importance of introducing metabolic capacity to in vitro assays. Overall, the activity profiles of pesticides highlighted in this study can contribute to the knowledge gaps surrounding pesticide mechanisms and to the better understanding of the on- and off-target organismal effects of pesticides.

Toxicity Tests for Chemical Pesticide Registration: Requirement Differences among the United States, the European Union, Japan, and China?

Pesticide registration is a scientific, legal, and administrative process that checks if a pesticide is safe and effective for its intended use before it can be used. The toxicity test is a key part of pesticide registration, which includes human health and ecological effect testing. Different countries adopt their own toxicity test criteria for pesticide registration guidelines. However, these differences, which may help accelerate the progress of pesticide registration and reduce the number of animals used, are yet to be explored and compared. Herein, we outlined the details and compared the differences between the toxicity tests in the United States, the European Union, Japan, and China. Some differences lie in the types and waiver policy, while others are in new approach methodologies (NAMs). On the basis of these differences, there is great potential for the optimization of NAMs during the toxicity tests. It is expected that this perspective can contribute to developing and adopting NAMs.

Accurate clinical toxicity prediction using multi-task deep neural nets and contrastive molecular explanations

Explainable machine learning for molecular toxicity prediction is a promising approach for efficient drug development and chemical safety. A predictive ML model of toxicity can reduce experimental cost and time while mitigating ethical concerns by significantly reducing animal and clinical testing. Herein, we use a deep learning framework for simultaneously modeling in vitro, in vivo, and clinical toxicity data. Two different molecular input representations are used; Morgan fingerprints and pre-trained SMILES embeddings. A multi-task deep learning model accurately predicts toxicity for all endpoints, including clinical, as indicated by the area under the Receiver Operator Characteristic curve and balanced accuracy. In particular, pre-trained molecular SMILES embeddings as input to the multi-task model improved clinical toxicity predictions compared to existing models in MoleculeNet benchmark. Additionally, our multitask approach is comprehensive in the sense that it is comparable to state-of-the-art approaches for specific endpoints in in vitro, in vivo and clinical platforms. Through both the multi-task model and transfer learning, we were able to indicate the minimal need of in vivo data for clinical toxicity predictions. To provide confidence and explain the model's predictions, we adapt a post-hoc contrastive explanation method that returns pertinent positive and negative features, which correspond well to known mutagenic and reactive toxicophores, such as unsubstituted bonded heteroatoms, aromatic amines, and Michael receptors. Furthermore, toxicophore recovery by pertinent feature analysis captures more of the in vitro (53%) and in vivo (56%), rather than of the clinical (8%), endpoints, and indeed uncovers a preference in known toxicophore data towards in vitro and in vivo experimental data. To our knowledge, this is the first contrastive explanation, using both present and absent substructures, for predictions of clinical and in vivo molecular toxicity.

Targeted Per- and Polyfluoroalkyl substances (PFAS) assessments for high throughput screening: Analytical and testing considerations to inform a PFAS stock quality evaluation framework

Per- and polyfluoroalkyl substances (PFAS) represent a large chemical class lacking hazard, toxicokinetic, and exposure information. To accelerate PFAS hazard evaluation, new approach methodologies (NAMs) comprised of in vitro high-throughput toxicity screening, toxicokinetic data, and computational modeling are being employed in read across strategies to evaluate the larger PFAS landscape. A critical consideration to ensure robust evaluations is a parallel assessment of the quality of the screening stock solutions, where dimethyl sulfoxide (DMSO) is often the diluent of choice. Challenged by the lack of commercially available reference standards for many of the selected PFAS and reliance on mass spectrometry approaches for such an evaluation, we developed a high-throughput framework to evaluate the quality of screening stocks for 205 PFAS selected for these NAM efforts. Using mass spectrometry coupled with either liquid or gas chromatography, a quality scoring system was developed that incorporated observations during mass spectral examination to provide a simple pass or fail notation. Informational flags were used to further describe findings regarding parent analyte presence through accurate mass identification, evidence of contaminants and/or degradation, or further describe characteristics such as isomer presence. Across the PFAS-DMSO stocks tested, 148 unique PFAS received passing quality scores to allow for further in vitro testing whereas 57 received a failing score primarily due to detection issues or confounding effects of DMSO. Principle component analysis indicated vapor pressure and Henry's Law Constant as top indicators for a failed quality score for those analyzed by gas chromatography. Three PFAS in the hexafluoropropylene oxide family failed due to degradation in DMSO. As the PFAS evaluated spanned over 20 different structural categories, additional commentary describes analytical observations across specific groups related to PFAS stock composition, detection, stability, and methodologic considerations that will be useful for informing future analytical assessment and downstream HTS efforts. The high-throughput stock quality scoring workflow presented holds value as a tool to evaluate chemical presence and quality efficiently and for informing data inclusion in PFAS or other NAM screening efforts.

A geospatial modeling approach to quantifying the risk of exposure to environmental chemical mixtures via a common molecular target

In the real world, individuals are exposed to chemicals from sources that vary over space and time. However, traditional risk assessments based on in vivo animal studies typically use a chemical-by-chemical approach and apical disease endpoints. New approach methodologies (NAMs) in toxicology, such as in vitro high-throughput (HTS) assays generated in Tox21 and ToxCast, can more readily provide mechanistic chemical hazard information for chemicals with no existing data than in vivo methods. In this paper, we establish a workflow to assess the joint action of 41 modeled ambient chemical exposures in the air from the USA-wide National Air Toxics Assessment by integrating human exposures with hazard data from curated HTS (cHTS) assays to identify counties where exposure to the local chemical mixture may perturb a common biological target. We exemplify this proof-of-concept using CYP1A1 mRNA up-regulation. We first estimate internal exposure and then convert the inhaled concentration to a steady state plasma concentration using physiologically based toxicokinetic modeling parameterized with county-specific information on ages and body weights. We then use the estimated blood plasma concentration and the concentration-response curve from the in vitro cHTS assay to determine the chemical-specific effects of the mixture components. Three mixture modeling methods were used to estimate the joint effect from exposure to the chemical mixture on the activity levels, which were geospatially mapped. Finally, a Monte Carlo uncertainty analysis was performed to quantify the influence of each parameter on the combined effects. This workflow demonstrates how NAMs can be used to predict early-stage biological perturbations that can lead to adverse health outcomes that result from exposure to chemical mixtures. As a result, this work will advance mixture risk assessment and other early events in the effects of chemicals.

Epidemiology meets toxicogenomics: Mining toxicologic evidence in support of an untargeted analysis of pesticides exposure and Parkinson's disease

BACKGROUND

Pesticides have been widely used in agriculture for more than half a century. However, with thousands currently in use, most have not been adequately assessed for influence Parkinson's disease (PD).

OBJECTIVES

Here we aimed to assess biologic plausibility of 70 pesticides implicated with PD through an agnostic pesticide-wide association study using a data mining approach linking toxicology and toxicogenomics databases.

METHODS

We linked the 70 targeted pesticides to quantitative high-throughput screening assay findings from the Toxicology in the 21st Century (Tox21) program and pesticide-related genetic/disease information with the Comparative Toxicogenomics Database (CTD). We used the CTD to determine networks of genes each pesticide has been linked to and assess enrichment of relevant gene ontology (GO) annotations. With Tox21, we evaluated pesticide induced activity on a series of 43 nuclear receptor and stress response assays and two cytotoxicity assays.

RESULTS

Overall, 59 % of the 70 pesticides had chemical-gene networks including at least one PD gene/gene product. In total, 41 % of the pesticides had chemical-gene networks enriched for ≥ 1 high-priority PD GO terms. For instance, 23 pesticides had chemical-gene networks enriched for response to oxidative stress, 21 for regulation of neuron death, and twelve for autophagy, including copper sulfate, endosulfan and chlorpyrifos. Of the pesticides tested against the Tox21 assays, 79 % showed activity on ≥ 1 assay and 11 were toxic to the two human cell lines. The set of PD-associated pesticides showed more activity than expected on assays testing for xenobiotic homeostasis, mitochondrial membrane permeability, and genotoxic stress.

CONCLUSIONS

Overall, cross-database queries allowed us to connect a targeted set of pesticides implicated in PD via epidemiology to specific biologic targets relevant to PD etiology. This knowledge can be used to help prioritize targets for future experimental studies and improve our understanding of the role of pesticides in PD etiology.

Prediction of drug-induced liver injury and cardiotoxicity using chemical structure and in vitro assay data

Drug-induced liver injury (DILI) and cardiotoxicity (DICT) are major adverse effects triggered by many clinically important drugs. To provide an alternative to in vivo toxicity testing, the U.S. Tox21 consortium has screened a collection of ∼10K compounds, including drugs in clinical use, against >70 cell-based assays in a quantitative high-throughput screening (qHTS) format. In this study, we compiled reference compound lists for DILI and DICT and compared the potential of Tox21 assay data with chemical structure information in building prediction models for human in vivo hepatotoxicity and cardiotoxicity. Models were built with four different machine learning algorithms (e.g., Random Forest, Naïve Bayes, eXtreme Gradient Boosting, and Support Vector Machine) and model performance was evaluated by calculating the area under the receiver operating characteristic curve (AUC-ROC). Chemical structure-based models showed reasonable predictive power for DILI (best AUC-ROC = 0.75 ± 0.03) and DICT (best AUC-ROC = 0.83 ± 0.03), while Tox21 assay data alone only showed better than random performance. DILI and DICT prediction models built using a combination of assay data and chemical structure information did not have a positive impact on model performance. The suboptimal predictive performance of the assay data is likely due to insufficient coverage of an adequately predictive number of toxicity mechanisms. The Tox21 consortium is currently expanding coverage of biological response space with additional assays that probe toxicologically important targets and under-represented pathways that may improve the prediction of in vivo toxicity such as DILI and DICT.

Confirmation of high-throughput screening data and novel mechanistic insights into FXR-xenobiotic interactions by orthogonal assays

Toxicology in the 21st Century (Tox21) is a federal collaboration employing a high-throughput robotic screening system to test 10,000 environmental chemicals. One of the primary goals of the program is prioritizing toxicity evaluations through in vitro high-throughput screening (HTS) assays for large numbers of chemicals already in commercial use for which little or no toxicity data is available. Within the Tox21 screening program, disruption in nuclear receptor (NR) signaling represents a particular area of interest. Given the role of NR's in modulating a wide range of biological processes, alterations of their activity can have profound biological impacts. Farnesoid X receptor (FXR) is a member of the nuclear receptor superfamily that has demonstrated importance in bile acid homeostasis, glucose metabolism, lipid homeostasis and hepatic regeneration. In this study, we re-evaluated 24 FXR agonists and antagonists identified through Tox21 using select orthogonal assays. In transient transactivation assays, 7/8 putative agonists and 4/4 putative inactive compounds were confirmed. Likewise, we confirmed 9/12 antagonists tested. Using a mammalian two hybrid approach we demonstrate that both FXR agonists and antagonists facilitate FXRα-coregulator interactions suggesting that differential coregulator recruitment may mediate activation/repression of FXRα mediated transcription. Additionally, we tested the ability of select FXR agonists and antagonists to facilitate hepatic transcription of FXR gene targets Shp and Bsep in a teleost (Medaka) model. Through application of in vitro cell-based assays, in silico modeling and in vivo gene expressions, we demonstrated the molecular complexity of FXR:ligand interactions and confirmed the ability of diverse ligands to modulate FXRα, facilitate differential coregulator recruitment and activate/repress receptor-mediated transcription. Overall, we suggest a multiplicative approach to assessment of nuclear receptor function may facilitate a greater understanding of the biological and mechanistic complexities of nuclear receptor activities and further our ability to interpret broad HTS outcomes.

Advances in computational methods along the exposure to toxicological response paradigm

Human health risk assessment is a function of chemical toxicity, bioavailability to reach target biological tissues, and potential environmental exposure. These factors are complicated by many physiological, biochemical, physical and lifestyle factors. Furthermore, chemical health risk assessment is challenging in view of the large, and continually increasing, number of chemicals found in the environment. These challenges highlight the need to prioritize resources for the efficient and timely assessment of those environmental chemicals that pose greatest health risks. Computational methods, either predictive or investigative, are designed to assist in this prioritization in view of the lack of cost prohibitive in vivo experimental data. Computational methods provide specific and focused toxicity information using in vitro high throughput screening (HTS) assays. Information from the HTS assays can be converted to in vivo estimates of chemical levels in blood or target tissue, which in turn are converted to in vivo dose estimates that can be compared to exposure levels of the screened chemicals. This manuscript provides a review for the landscape of computational methods developed and used at the U.S. Environmental Protection Agency (EPA) highlighting their potentials and challenges.

Identifying Protein Features and Pathways Responsible for Toxicity Using Machine Learning and Tox21: Implications for Predictive Toxicology

Humans are exposed to numerous compounds daily, some of which have adverse effects on health. Computational approaches for modeling toxicological data in conjunction with machine learning algorithms have gained popularity over the last few years. Machine learning approaches have been used to predict toxicity-related biological activities using chemical structure descriptors. However, toxicity-related proteomic features have not been fully investigated. In this study, we construct a computational pipeline using machine learning models for predicting the most important protein features responsible for the toxicity of compounds taken from the Tox21 dataset that is implemented within the multiscale Computational Analysis of Novel Drug Opportunities (CANDO) therapeutic discovery platform. Tox21 is a highly imbalanced dataset consisting of twelve in vitro assays, seven from the nuclear receptor (NR) signaling pathway and five from the stress response (SR) pathway, for more than 10,000 compounds. For the machine learning model, we employed a random forest with the combination of Synthetic Minority Oversampling Technique (SMOTE) and the Edited Nearest Neighbor (ENN) method (SMOTE+ENN), which is a resampling method to balance the activity class distribution. Within the NR and SR pathways, the activity of the aryl hydrocarbon receptor (NR-AhR) and the mitochondrial membrane potential (SR-MMP) were two of the top-performing twelve toxicity endpoints with AUCROCs of 0.90 and 0.92, respectively. The top extracted features for evaluating compound toxicity were analyzed for enrichment to highlight the implicated biological pathways and proteins. We validated our enrichment results for the activity of the AhR using a thorough literature search. Our case study showed that the selected enriched pathways and proteins from our computational pipeline are not only correlated with AhR toxicity but also form a cascading upstream/downstream arrangement. Our work elucidates significant relationships between protein and compound interactions computed using CANDO and the associated biological pathways to which the proteins belong for twelve toxicity endpoints. This novel study uses machine learning not only to predict and understand toxicity but also elucidates therapeutic mechanisms at a proteomic level for a variety of toxicity endpoints.

Identification of environmental chemicals that activate p53 signaling after in vitro metabolic activation

Currently, approximately 80,000 chemicals are used in commerce. Most have little-to-no toxicity information. The U.S. Toxicology in the 21st Century (Tox21) program has conducted a battery of in vitro assays using a quantitative high-throughput screening (qHTS) platform to gain toxicity information on environmental chemicals. Due to technical challenges, standard methods for providing xenobiotic metabolism could not be applied to qHTS assays. To address this limitation, we screened the Tox21 10,000-compound (10K) library, with concentrations ranging from 2.8 nM to 92 µM, using a p53 beta-lactamase reporter gene assay (p53-bla) alone or with rat liver microsomes (RLM) or human liver microsomes (HLM) supplemented with NADPH, to identify compounds that induce p53 signaling after biotransformation. Two hundred and seventy-eight compounds were identified as active under any of these three conditions. Of these 278 compounds, 73 gave more potent responses in the p53-bla assay with RLM, and 2 were more potent in the p53-bla assay with HLM compared with the responses they generated in the p53-bla assay without microsomes. To confirm the role of metabolism in the differential responses, we re-tested these 75 compounds in the absence of NADPH or with heat-attenuated microsomes. Forty-four compounds treated with RLM, but none with HLM, became less potent under these conditions, confirming the role of RLM in metabolic activation. Further evidence of biotransformation was obtained by measuring the half-life of the parent compounds in the presence of microsomes. Together, the data support the use of RLM in qHTS for identifying chemicals requiring biotransformation to induce biological responses.

A Quantitative High-Throughput Screening Data Analysis Pipeline for Activity Profiling

The U.S. Tox21 program has developed in vitro assays to test large collections of environmental chemicals in a quantitative high-throughput screening (qHTS) format, using triplicate 15-dose titrations to generate over 100 million data points to date. Counterscreens are also employed to minimize interferences from non-target-specific assay artifacts, such as compound autofluorescence and cytotoxicity. New data analysis approaches are needed to integrate these data and characterize the activities observed from these assays. Here, we describe a complete analysis pipeline that evaluates these qHTS data for technical quality in terms of signal reproducibility. We integrate signals from repeated assay runs, primary readouts and counterscreens to produce a final call on on-target compound activity.

Acute Oral, Subacute, and Developmental Toxicity Profiling of Naphthalene 2-Yl, 2-Chloro, 5-Nitrobenzoate: Assessment Based on Stress Response, Toxicity, and Adverse Outcome Pathways

The U.S. National Research Council (NRC) introduced new approaches to report toxicity studies. The NRC vision is to explore the toxicity pathways leading to the adverse effects in intact organisms by the exposure of the chemicals. This study examines the toxicity profiling of the naphthalene-2-yl 2-chloro-5-dinitrobenzoate (SF5) by adopting the vision of NRC that moves from traditional animal studies to the cellular pathways. Acute, subacute, and developmental toxicity studies were assayed according to the Organization for Economic Cooperation and Development (OECD) guidelines. The stress response pathway, toxicity pathway, and adverse effects outcome parameters were analyzed by using their standard protocols. The results showed that the acute toxicity study increases the liver enzyme levels. In a subacute toxicity study, alkaline phosphatase (ALP) levels were raised in both male and female animals. SF5 significantly increases the normal sperm count in the male animals corresponding to a decrease in the abnormality count. Developmental toxicity showed the normal skeletal and morphological parameters, except little hydrocephalus was observed in developmental toxicity. Doses of 20 mg/kg in males and 4 mg/kg in females showed decreased glutathione (GSH) levels in the kidney and liver. MDA levels were also increased in the kidney and liver. However, histopathological studies did not show any cellular change in these organs. No statistical difference was observed in histamine levels, testosterone, nuclear factor erythroid two-related factor-2 (Nrf2), and nuclear factor-kappa B (NF-κB), which showed no initiation of the stress response, toxicity, and adverse effect pathways. Immunomodulation was observed at low doses in subacute toxicity studies. It was concluded that SF5 did not produce abrupt and high-toxicity levels in organs and biochemical parameters. So, it is safe for further studies.

Multitask CapsNet: An Imbalanced Data Deep Learning Method for Predicting Toxicants

Drug development has a high failure rate, with safety properties constituting a considerable challenge. To reduce risk, in silico tools, including various machine learning methods, have been applied for toxicity prediction. However, these approaches often confront a serious problem: the training data sets are usually biased (imbalanced positive and negative samples), which would result in model training difficulty and unsatisfactory prediction accuracy. Multitask networks obtained significantly better predictive accuracies than single-task methods, and capsule neural networks showed excellent performance in sparse data sets in previous studies. In this study, we developed a new multitask framework based on a capsule neural network (multitask CapsNet) to measure 12 different toxic effects simultaneously. We found that multitask CapsNet excelled in toxicity prediction and outperformed many other computational approaches using the multitask strategy. Only after training on biased data sets did multitask CapsNet achieve significantly improved prediction accuracy on the Tox21 Data Challenge, which gave the largest ratio of highest accuracy (8/12) among compared models. Our model gave a prediction accuracy of 96.6% for the target NR.PPAR.gamma, whose ratio of negative to positive samples was up to 36:1. These results suggested that multitask CapsNet could overcome the bias problems and would provide a novel, accurate, and efficient approach for predicting the toxicities of compounds.

Comparison of Acute Effects of Neurotoxic Compounds on Network Activity in Human and Rodent Neural Cultures

Assessment of neuroactive effects of chemicals in cell-based assays remains challenging as complex functional tissue is required for biologically relevant readouts. Recent in vitro models using rodent primary neural cultures grown on multielectrode arrays allow quantitative measurements of neural network activity suitable for neurotoxicity screening. However, robust systems for testing effects on network function in human neural models are still lacking. The increasing number of differentiation protocols for generating neurons from human-induced pluripotent stem cells (hiPSCs) holds great potential to overcome the unavailability of human primary tissue and expedite cell-based assays. Yet, the variability in neuronal activity, prolonged ontogeny and rather immature stage of most neuronal cells derived by standard differentiation techniques greatly limit their utility for screening neurotoxic effects on human neural networks. Here, we used excitatory and inhibitory neurons, separately generated by direct reprogramming from hiPSCs, together with primary human astrocytes to establish highly functional cultures with defined cell ratios. Such neuron/glia cocultures exhibited pronounced neuronal activity and robust formation of synchronized network activity on multielectrode arrays, albeit with noticeable delay compared with primary rat cortical cultures. We further investigated acute changes of network activity in human neuron/glia cocultures and rat primary cortical cultures in response to compounds with known adverse neuroactive effects, including gamma amino butyric acid receptor antagonists and multiple pesticides. Importantly, we observed largely corresponding concentration-dependent effects on multiple neural network activity metrics using both neural culture types. These results demonstrate the utility of directly converted neuronal cells from hiPSCs for functional neurotoxicity screening of environmental chemicals.

Exploration of xenobiotic metabolism within cell lines used for Tox21 chemical screening

The Tox21 Program has investigated thousands of chemicals with high-throughput screening assays using cell-based assays to link thousands of chemicals to individual molecular targets/pathways. However, these systems have been widely criticized for their suspected lack of 'metabolic competence' to bioactivate or detoxify chemical exposures. In this study, 9 cell line backgrounds used in Tox21 assays (i.e., HepG2, HEK293, Hela, HCT116, ME180, CHO-K1, GH3.TRE-Luc, C3H10T1/2 and MCF7) were evaluated via metabolite formation rates, along with metabolic clearance and metabolite profiling for HepG2, HEK293, and MCF-7aroERE, in comparison to pooled donor (50) suspensions of primary human hepatocytes (PHHs). Using prototype clinical drug substrates for CYP1A2, CYP2B6, and CYP3A4/5, extremely low-to-undetectable CYP450 metabolism was observed (24 h), and consistent with their purported 'lack' of metabolic competence. However, for Phase II metabolizing enzymes and metabolic clearance, surprisingly proficient metabolism was observed for bisphenol AF, bisphenol S, and 7-hydroxycoumarin. Here, comparatively low glucuronidation relative to sulfation was observed in contrast to equivalent levels in PHHs. Overall, while a lack of CYP450 metabolism was confirmed in this benchmarking effort, Tox21 cell lines were not 'incompetent' for xenobiotic metabolism, and displayed surprisingly high proficiency for sulfation that rivaled PHHs. These findings have implications for the interpretation of Tox21 assay data, and establish a framework for evaluating of 'metabolic competence' with in vitro models.

Fine-tuning of a generative neural network for designing multi-target compounds

Exploring the origin of multi-target activity of small molecules and designing new multi-target compounds are highly topical issues in pharmaceutical research. We have investigated the ability of a generative neural network to create multi-target compounds. Data sets of experimentally confirmed multi-target, single-target, and consistently inactive compounds were extracted from public screening data considering positive and negative assay results. These data sets were used to fine-tune the REINVENT generative model via transfer learning to systematically recognize multi-target compounds, distinguish them from single-target or inactive compounds, and construct new multi-target compounds. During fine-tuning, the model showed a clear tendency to increasingly generate multi-target compounds and structural analogs. Our findings indicate that generative models can be adopted for de novo multi-target compound design.

Deep Learning Model for Identifying Critical Structural Motifs in Potential Endocrine Disruptors

This paper aims to identify structural motifs within a molecule that contribute the most toward a chemical being an endocrine disruptor. We have developed a deep neural network-based toolkit toward this aim. The trained model can virtually assess a synthetic chemical's potential to be an endocrine disruptor using machine-readable molecular representation, simplified molecular input line entry system (SMILES). Our proposed toolkit is a multilabel or multioutput classification model that combines both convolution and long short-term memory (LSTM) architectures. The toolkit leverages the advantages of an active learning-based framework that combines multiple sources of data. Class activation maps (CAMs) generated from the feature-extraction layers can identify the structural alerts and the chemical environment that determines the specificity of the structural alerts.

Predicting mechanism of action of cellular perturbations with pathway activity signatures

MOTIVATION

Misregulation of signaling pathway activity is etiologic for many human diseases, and modulating activity of signaling pathways is often the preferred therapeutic strategy. Understanding the mechanism of action (MOA) of bioactive chemicals in terms of targeted signaling pathways is the essential first step in evaluating their therapeutic potential. Changes in signaling pathway activity are often not reflected in changes in expression of pathway genes which makes MOA inferences from transcriptional signatures (TSeses) a difficult problem.

RESULTS

We developed a new computational method for implicating pathway targets of bioactive chemicals and other cellular perturbations by integrated analysis of pathway network topology, the Library of Integrated Network-based Cellular Signature TSes of genetic perturbations of pathway genes and the TS of the perturbation. Our methodology accurately predicts signaling pathways targeted by the perturbation when current pathway analysis approaches utilizing only the TS of the perturbation fail.

AVAILABILITY AND IMPLEMENTATION

Open source R package paslincs is available at https://github.com/uc-bd2k/paslincs.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Pharmacological Actions of Carbamate Insecticides at Mammalian Melatonin Receptors

Integrated in silico chemical clustering and melatonin receptor molecular modeling combined with in vitro 2-[125I]-iodomelatonin competition binding were used to identify carbamate insecticides with affinity for human melatonin receptor 1 (hMT1) and human melatonin receptor 2 (hMT2). Saturation and kinetic binding studies with 2-[125I]-iodomelatonin revealed lead carbamates (carbaryl, fenobucarb, bendiocarb, carbofuran) to be orthosteric ligands with antagonist apparent efficacy at hMT1 and agonist apparent efficacy at hMT2 Furthermore, using quantitative receptor autoradiography in coronal brain slices from C3H/HeN mice, carbaryl, fenobucarb, and bendiocarb competed for 2-[125I]-iodomelatonin binding in the suprachiasmatic nucleus (SCN), paraventricular nucleus of the thalamus (PVT), and pars tuberalis (PT) with affinities similar to those determined for the hMT1 receptor. Carbaryl (10 mg/kg i.p.) administered in vivo also competed ex vivo for 2-[125I]-iodomelatonin binding to the SCN, PVT, and PT, demonstrating the ability to reach brain melatonin receptors in C3H/HeN mice. Furthermore, the same dose of carbaryl given to C3H/HeN mice in constant dark for three consecutive days at subjective dusk (circadian time 10) phase-advanced circadian activity rhythms (mean = 0.91 hours) similar to melatonin (mean = 1.12 hours) when compared with vehicle (mean = 0.04 hours). Carbaryl-mediated phase shift of overt circadian activity rhythm onset is likely mediated via interactions with SCN melatonin receptors. Based on the pharmacological actions of carbaryl and other carbamate insecticides at melatonin receptors, exposure may modulate time-of-day information conveyed to the master biologic clock relevant to adverse health outcomes. SIGNIFICANCE STATEMENT: In silico chemical clustering and molecular modeling in conjunction with in vitro bioassays identified several carbamate insecticides (i.e., carbaryl, carbofuran, fenobucarb, bendiocarb) as pharmacologically active orthosteric melatonin receptor 1 and 2 ligands. This work further demonstrated that carbaryl competes for melatonin receptor binding in the master biological clock (suprachiasmatic nucleus) and phase-advances overt circadian activity rhythms in C3H/HeN mice, supporting the relevance of circadian effects when interpreting toxicological findings related to carbamate insecticide exposure.

Application of cell-based biological bioassays for health risk assessment of PM2.5 exposure in three megacities, China

The determination of PM_2.5-induced biological response is essential for understanding the adverse health risk associated with PM_2.5 exposure. In this study, we conducted cell-based bioassays to measure the toxic effects of PM_2.5 exposure, including cytotoxicity, oxidative stress, genotoxicity and inflammatory response. The concentration-response relationship was analyzed by benchmark dose (BMD) modeling and the BMDL₁₀ was used to estimate the biological potency of PM_2.5 exposure. PM_2.5 samples were collected from three typical megacities of China (Beijing, BJ; Wuhan, WH; Guangzhou, GZ) in typical seasons (winter and summer). The total PM, water-soluble fractions (WSF), and organic extracts (OE) were prepared and subjected to examination of toxic effects. The biological potencies for cytotoxicity, oxidative stress and genotoxicity were generally higher in winter samples, while the inflammatory potency of PM_2.5 was higher in summer samples. The relative health risk (RHR) was determined by integration of the biological potencies and the cumulative exposure level, and the ranks of RHR were BJ-W > WH-W > BJ-S > WH-S > GZ-W > GZ-S. Notably, we note that different PM_2.5 compositions were associated with distinct biological effects, and the health effects distribution of PM_2.5 varied in regions and seasons. These findings demonstrate that the approach of integrated cell-based bioassays could be used for the evaluation of health effects of PM_2.5 exposure.

Computational Study of Drugs Targeting Nuclear Receptors

Endocrine-disrupting chemicals have been shown to interfere with the endocrine system function at the level of hormone synthesis, transport, metabolism, binding, action, and elimination. They are associated with several health problems in humans: obesity, diabetes mellitus, infertility, impaired thyroid and neuroendocrine functions, neurodevelopmental problems, and cancer are among them. As drugs are chemicals humans can be frequently exposed to for longer periods of time, special emphasis should be put on their endocrine-disrupting potential. In this study, we conducted a screen of 1046 US-approved and marketed small-molecule drugs (molecular weight between 60 and 600) for estimating their endocrine-disrupting properties. Binding affinity to 12 nuclear receptors was assessed with a molecular-docking program, Endocrine Disruptome. We identified 130 drugs with a high binding affinity to a nuclear receptor that is not their pharmacological target. In a subset of drugs with predicted high binding affinities to a nuclear receptor with Endocrine Disruptome, the positive predictive value was 0.66 when evaluated with in silico results obtained with another molecular docking program, VirtualToxLab, and 0.32 when evaluated with in vitro results from the Tox21 database. Computational screening was proven useful in prioritizing drugs for in vitro testing. We suggest that the novel interactions of drugs with nuclear receptors predicted here are further investigated.

Predictive Models for Human Organ Toxicity Based on In Vitro Bioactivity Data and Chemical Structure

Traditional toxicity testing reliant on animal models is costly and low throughput, posing a significant challenge with the increasing numbers of chemicals that humans are exposed to in the environment. The purpose of this investigation was to build optimal prediction models for various human in vivo/organ-level toxicity end points (extracted from ChemIDPlus) using chemical structure and Tox21 in vitro quantitative high-throughput screening (qHTS) bioactivity assay data. Several supervised machine learning algorithms were applied to model 14 human toxicity end points pertaining to vascular, kidney, ureter and bladder, and liver organ systems. Three metrics were used to evaluate model performance: area under the receiver operating characteristic curve (AUC-ROC), balanced accuracy (BA), and Matthews correlation coefficient (MCC). The top four models, with AUC-ROC values >0.8, were derived for endocrine (0.90 ± 0.00), musculoskeletal (0.88 ± 0.02), peripheral nerve and sensation (0.85 ± 0.01), and brain and coverings (0.83 ± 0.02) toxicities, whereas the best model AUC-ROC values were >0.7 for the remaining 10 toxicities. Model performance was found to be dependent on the specific data set, model type, and feature selection method used. In addition, chemical structure and assay data showed different levels of contribution to the prediction of different toxicity end points. Although in vitro assay data, when combined with chemical structure, slightly improved the predictive accuracy for most end points (11 out of 14), a noteworthy finding was the near equal success of the structure-only models, which do not require Tox21 qHTS screening data, and the relatively poor performance of assay-only models. Thus, the top-performing structure-only models from this study could be applied for hazard screening of large sets of chemicals for potential human toxicity, whereas the largest assay contributions to models (i.e., cellular targets) could be used, along with the top-contributing structural features, to provide insight into toxicity mechanisms.

High-Throughput Screening to Predict Chemical-Assay Interference

The U.S. federal consortium on toxicology in the 21^st century (Tox21) produces quantitative, high-throughput screening (HTS) data on thousands of chemicals across a wide range of assays covering critical biological targets and cellular pathways. Many of these assays, and those used in other in vitro screening programs, rely on luciferase and fluorescence-based readouts that can be susceptible to signal interference by certain chemical structures resulting in false positive outcomes. Included in the Tox21 portfolio are assays specifically designed to measure interference in the form of luciferase inhibition and autofluorescence via multiple wavelengths (red, blue, and green) and under various conditions (cell-free and cell-based, two cell types). Out of 8,305 chemicals tested in the Tox21 interference assays, percent actives ranged from 0.5% (red autofluorescence) to 9.9% (luciferase inhibition). Self-organizing maps and hierarchical clustering were used to relate chemical structural clusters to interference activity profiles. Multiple machine learning algorithms were applied to predict assay interference based on molecular descriptors and chemical properties. The best performing predictive models (accuracies of ~80%) have been included in a web-based tool called InterPred that will allow users to predict the likelihood of assay interference for any new chemical structure and thus increase confidence in HTS data by decreasing false positive testing results.

Improving natural product research translation: From source to clinical trial

While great interest in health effects of natural product (NP) including dietary supplements and foods persists, promising preclinical NP research is not consistently translating into actionable clinical trial (CT) outcomes. Generally considered the gold standard for assessing safety and efficacy, CTs, especially phase III CTs, are costly and require rigorous planning to optimize the value of the information obtained. More effective bridging from NP research to CT was the goal of a September, 2018 transdisciplinary workshop. Participants emphasized that replicability and likelihood of successful translation depend on rigor in experimental design, interpretation, and reporting across the continuum of NP research. Discussions spanned good practices for NP characterization and quality control; use and interpretation of models (computational through in vivo) with strong clinical predictive validity; controls for experimental artefacts, especially for in vitro interrogation of bioactivity and mechanisms of action; rigorous assessment and interpretation of prior research; transparency in all reporting; and prioritization of research questions. Natural product clinical trials prioritized based on rigorous, convergent supporting data and current public health needs are most likely to be informative and ultimately affect public health. Thoughtful, coordinated implementation of these practices should enhance the knowledge gained from future NP research.

Bioactivity Signatures of Drugs vs. Environmental Chemicals Revealed by Tox21 High-Throughput Screening Assays

Humans are exposed to tens of thousands of chemicals over the course of a lifetime, yet there remains inadequate data on the potential harmful effects of these substances on human health. Using quantitative high-throughput screening (qHTS), we can test these compounds for potential toxicity in a more efficient and cost-effective way compared to traditional animal studies. Tox21 has developed a library of ~10,000 chemicals (Tox21 10K) comprising one-third approved and investigational drugs and two-thirds environmental chemicals. In this study, the Tox21 10K was screened in a qHTS format against a panel of 70 cell-based assays with 213 readouts covering a broad range of biological pathways. Activity profiles were compared with chemical structure to assess their ability to differentiate drugs from environmental chemicals, and structure was found to be a better predictor of which chemicals are likely to be drugs. Drugs and environmental chemicals were further analyzed for diversity in structure and biological response space and showed distinguishable, but not distinct, responses in the Tox21 assays. Inclusion of other targets and pathways to further diversify the biological response space covered by these assays is likely required to better evaluate the safety profile of drugs and environmental chemicals to prioritize for in-depth toxicological studies.

Toxicity of chlorinated and ozonated wastewater effluents probed by genetically modified bioluminescent bacteria and cyanobacteria Spirulina sp

Chlorination and ozonation of various waters may be associated with the formation of toxic disinfection byproducts (DBPs) and cause health risks to humans. Monitoring the toxicity of chlorinated and ozonated water and identification of different toxicity mechanisms are therefore required. This study is one of its kind to examine the toxic effects of chlorinated and ozonated wastewater effluents on three genetically modified bioluminescent bacteria, in comparison to the naturally isolated cyanobacteria, Spirulina strains as test systems. Three different secondary wastewater effluents were collected from treatment plants, chlorinated using sodium hypochlorite (at 1 and 10 mg L^-1 of chlorine) or treated using 3-4 mg L^-1 of ozone at different contact times. As compared to cyanobacterial Spirulina sp., the genetically modified bacteria enhancing bioluminescence at the presence of stress agents demonstrated greater sensitivity to the toxicity induction and have also provided mechanism-specific responses associated with genotoxicity, cytotoxicity and reactive oxygen species (ROS) generation in wastewater effluents. Effects of effluent chlorination time and chlorine concentration revealed by means of bioluminescent bacteria suggest the formation of genotoxic and cytotoxic DBPs followed with their possible disappearance at longer times. Ozonation could degrade genotoxic compounds in some effluents, but the cytotoxic potential of wastewater effluents may certainly increase with ozonation time. No induction of ROS-related toxicity was detected in either chlorinated or ozonated wastewater effluents. UV absorbance- and fluorescence emission-based spectroscopic characteristics may be variously correlated with changes in genotoxicity in ozonated effluents, however, no associations were obtained in chlorinated wastewater effluents. The bacterial response to the developed mechanism-specific toxicity differs among wastewater effluents, reflecting variability in effluent compositions.

The development of a cell-based model for the assessment of carcinogenic potential upon long-term PM2.5 exposure

To assess the carcinogenic potential of PM2.5 exposure, we developed a cell-based experimental protocol to examine the cell transformation activity of PM2.5 samples from different regions in China. The seasonal ambient PM2.5 samples were collected from three megacities, Beijing (BJ), Wuhan (WH), and Guangzhou (GZ), from November 2016 to October 2017. The mean concentrations of PM2.5 were much higher in the winter season (BJ: 109.64 μg/m³, WH: 79.99 μg/m³, GZ: 49.99 μg/m³) than that in summer season (BJ: 42.40 μg/m³, WH: 25.82 μg/m³, GZ: 19.82 μg/m³). The organic extracts (OE) of PM2.5 samples from combined summer (S) (June, July, August) or winter (W) (November, December, January) seasons were subjected to characterization of chemical components. We treated human bronchial epithelial (HBE) cells expressing CYP1A1 (HBE-1A1) with PM2.5 samples at doses ranging from 0 to 100 μg/mL (0, 1.563, 3.125, 6.25, 12.5, 25, 50, 100 μg/mL) and determined the phenotype of malignant cell transformation. A dose-response relationship was analyzed by benchmark dose (BMD) modeling, and the potential were indicated by BMDL₁₀. The order of the carcinogenic risk of seasonal PM2.5 samples from high to low was BJ-W, WH-W, GZ-W, WH-S, BJ-S, and GZ-S. Notably, we found that the alteration in the lung cancer-related biomarkers, KRAS, PTEN, p53, c-Myc, PCNA, pAKT/AKT, and pERK/ERK was congruent with the activity of cell transformation and the content of specific components of polycyclic aromatic hydrocarbon (PAHs) bound to PM2.5. Taken together, we have successfully developed a cell-based alternative model for the evaluation of potent carcinogenicity upon long-term PM2.5 exposure.

Limited Chemical Structural Diversity Found to Modulate Thyroid Hormone Receptor in the Tox21 Chemical Library

BACKGROUND

Thyroid hormone receptors (TRs) are critical endocrine receptors that regulate a multitude of processes in adult and developing organisms, and thyroid hormone disruption is of high concern for neurodevelopmental and reproductive toxicities in particular. To date, only a small number of chemical classes have been identified as possible TR modulators, and the receptors appear highly selective with respect to the ligand structural diversity. Thus, the question of whether TRs are an important screening target for protection of human and wildlife health remains.

OBJECTIVE

Our goal was to evaluate the hypothesis that there is limited structural diversity among environmentally relevant chemicals capable of modulating TR activity via the collaborative interagency Tox21 project.

METHODS

We screened the Tox21 chemical library (8,305 unique structures) in a quantitative high-throughput, cell-based reporter gene assay for TR agonist or antagonist activity. Active compounds were further characterized using additional orthogonal assays, including mammalian one-hybrid assays, coactivator recruitment assays, and a high-throughput, fluorescent imaging, nuclear receptor translocation assay.

RESULTS

Known agonist reference chemicals were readily identified in the TR transactivation assay, but only a single novel, direct agonist was found, the pharmaceutical betamipron. Indirect activation of TR through activation of its heterodimer partner, the retinoid-X-receptor (RXR), was also readily detected by confirmation in an RXR agonist assay. Identifying antagonists with high confidence was a challenge with the presence of significant confounding cytotoxicity and other, non-TR-specific mechanisms common to the transactivation assays. Only three pharmaceuticals-mefenamic acid, diclazuril, and risarestat-were confirmed as antagonists.

DISCUSSION

The results support limited structural diversity for direct ligand effects on TR and imply that other potential target sites in the thyroid hormone axis should be a greater priority for bioactivity screening for thyroid axis disruptors. https://doi.org/10.1289/EHP5314.

A novel hiPSC-derived system for hematoendothelial and myeloid blood toxicity screens identifies compounds promoting and inhibiting endothelial-to-hematopoietic transition

The exposure to toxic environmental and pharmaceutical substances can pose a long-term risk to human's health. In this study, we sought to investigate the potential of our recently developed method for induction of myeloid hematoendothelial and blood cells by overexpression of two transcription factors, GATA2 and ETV2, in human induced pluripotent stem cells (hiPSCs) for toxicity screening. For the primary screen in a high-throughput format, we selected twenty-two chemicals with various degrees of cytotoxicity available from the NIEHS National Toxicology Program (Tox21). The compounds were applied during the endothelial-to-hematopoietic transition and to differentiated myeloid progenitors growing in suspension. The system was capable of identifying compounds with both inhibitory and favorable effects on hematopoietic network, changes in expression of hematopoietic markers, and mitochondrial and cytotoxicity. The findings were confirmed and further investigated by secondary screens, colony forming cell assay, and gene expression profiling. The hematoendothelial toxicity of 5-fluorouracil, berberine chloride, and benzo(a)pyrene is characterized by the inhibition of cell division and a shift of hematopoietic programming to non-hemogenic endothelial and mesenchymal fates. This study demonstrates the feasibility of transcription factor (TF)-based differentiation systems to monitor endothelial and hematotoxicity and serves as an informative platform for screening myelosuppressive or stimulatory drugs and mechanistic studies of their action.

Integrated In Vivo Genotoxicity Assessment of Procarbazine Hydrochloride Demonstrates Induction of Pig-a and LacZ Mutations, and Micronuclei, in MutaMouse Hematopoietic Cells

Procarbazine hydrochloride (PCH) is a DNA-reactive hematopoietic carcinogen with potent and well-characterized clastogenic activity. However, there is a paucity of in vivo mutagenesis data for PCH, and in vitro assays often fail to detect the genotoxic effects of PCH due to the complexity of its metabolic activation. We comprehensively evaluated the in vivo genotoxicity of PCH on hematopoietic cells of male MutaMouse transgenic rodents using a study design that facilitated assessments of micronuclei and Pig-a mutation in circulating erythrocytes, and lacZ mutant frequencies in bone marrow. Mice were orally exposed to PCH (0, 6.25, 12.5, and 25 mg/kg/day) for 28 consecutive days. Blood samples collected 2 days after cessation of treatment exhibited significant dose-related induction of micronuclei in both immature and mature erythrocytes. Bone marrow and blood collected 3 and 70 days after cessation of treatment also showed significantly elevated mutant frequencies in both the lacZ and Pig-a assays even at the lowest dose tested. PCH-induced lacZ and Pig-a (immature and mature erythrocytes) mutant frequencies were highly correlated, with R² values ≥0.956, with the exception of lacZ vs. Pig-a mutants in mature erythrocytes at the 70-day time point (R² = 0.902). These results show that PCH is genotoxic in vivo and demonstrate that the complex metabolism and resulting genotoxicity of PCH is best evaluated in intact animal models. Our results further support the concept that multiple biomarkers of genotoxicity, especially hematopoietic cell genotoxicity, can be readily combined into one study provided that adequate attention is given to manifestation times. Environ. Mol. Mutagen. 60:505-512, 2019. © 2018 Her Majesty the Queen in Right of Canada.

Identifying Compounds with Genotoxicity Potential Using Tox21 High-Throughput Screening Assays

Genotoxicity is a critical component of a comprehensive toxicological profile. The Tox21 Program used five quantitative high-throughput screening (qHTS) assays measuring some aspect of DNA damage/repair to provide information on the genotoxic potential of over 10 000 compounds. Included were assays detecting activation of p53, increases in the DNA repair protein ATAD5, phosphorylation of H2AX, and enhanced cytotoxicity in DT40 cells deficient in DNA-repair proteins REV3 or KU70/RAD54. Each assay measures a distinct component of the DNA damage response signaling network; >70% of active compounds were detected in only one of the five assays. When qHTS results were compared with results from three standard genotoxicity assays (bacterial mutation, in vitro chromosomal aberration, and in vivo micronucleus), a maximum of 40% of known, direct-acting genotoxicants were active in one or more of the qHTS genotoxicity assays, indicating low sensitivity. This suggests that these qHTS assays cannot in their current form be used to replace traditional genotoxicity assays. However, despite the low sensitivity, ranking chemicals by potency of response in the qHTS assays revealed an enrichment for genotoxicants up to 12-fold compared with random selection, when allowing a 1% false positive rate. This finding indicates these qHTS assays can be used to prioritize chemicals for further investigation, allowing resources to focus on compounds most likely to induce genotoxic effects. To refine this prioritization process, models for predicting the genotoxicity potential of chemicals that were active in Tox21 genotoxicity assays were constructed using all Tox21 assay data, yielding a prediction accuracy up to 0.83. Data from qHTS assays related to stress-response pathway signaling (including genotoxicity) were the most informative for model construction. By using the results from qHTS genotoxicity assays, predictions from models based on qHTS data, and predictions from commercial bacterial mutagenicity QSAR models, we prioritized Tox21 chemicals for genotoxicity characterization.

Development of a Zebrafish S1500+ Sentinel Gene Set for High-Throughput Transcriptomics

Sentinel gene sets have been developed with the purpose of maximizing the information from targeted transcriptomic platforms. We recently described the development of an S1500+ sentinel gene set, which was built for the human transcriptome, utilizing a data- and knowledge-driven hybrid approach to select a small subset of genes that optimally capture transcriptional diversity, correlation with other genes based on large-scale expression profiling, and known pathway annotation within the human genome. While this detailed bioinformatics approach for gene selection can in principle be applied to other species, the reliability of the resulting gene set depends on availability of a large body of transcriptomics data. For the model organism zebrafish, we aimed to create a similar sentinel gene set (Zf S1500+ gene set); however, there is insufficient standardized expression data in the public domain to train the gene correlation model. Therefore, our strategy was to use human-zebrafish ortholog mapping of the human S1500+ genes and nominations from experts in the zebrafish scientific community. In this study, we present the bioinformatics curation and refinement process to produce the final Zf S1500+ gene set, explore whole transcriptome extrapolation using this gene set, and assess pathway-level inference. This gene set will add value to targeted high-throughput transcriptomics in zebrafish for toxicogenomic screening and other research domains.

The Next Generation Blueprint of Computational Toxicology at the U.S. Environmental Protection Agency

The U.S. Environmental Protection Agency (EPA) is faced with the challenge of efficiently and credibly evaluating chemical safety often with limited or no available toxicity data. The expanding number of chemicals found in commerce and the environment, coupled with time and resource requirements for traditional toxicity testing and exposure characterization, continue to underscore the need for new approaches. In 2005, EPA charted a new course to address this challenge by embracing computational toxicology (CompTox) and investing in the technologies and capabilities to push the field forward. The return on this investment has been demonstrated through results and applications across a range of human and environmental health problems, as well as initial application to regulatory decision-making within programs such as the EPA's Endocrine Disruptor Screening Program. The CompTox initiative at EPA is more than a decade old. This manuscript presents a blueprint to guide the strategic and operational direction over the next 5 years. The primary goal is to obtain broader acceptance of the CompTox approaches for application to higher tier regulatory decisions, such as chemical assessments. To achieve this goal, the blueprint expands and refines the use of high-throughput and computational modeling approaches to transform the components in chemical risk assessment, while systematically addressing key challenges that have hindered progress. In addition, the blueprint outlines additional investments in cross-cutting efforts to characterize uncertainty and variability, develop software and information technology tools, provide outreach and training, and establish scientific confidence for application to different public health and environmental regulatory decisions.

The NCATS BioPlanet - An Integrated Platform for Exploring the Universe of Cellular Signaling Pathways for Toxicology, Systems Biology, and Chemical Genomics

Chemical genomics aims to comprehensively define, and ultimately predict, the effects of small molecule compounds on biological systems. Chemical activity profiling approaches must consider chemical effects on all pathways operative in mammalian cells. To enable a strategic and maximally efficient chemical profiling of pathway space, we have created the NCATS BioPlanet, a comprehensive integrated pathway resource that incorporates the universe of 1,658 human pathways sourced from publicly available, manually curated sources, which have been subjected to thorough redundancy and consistency cross-evaluation. BioPlanet supports interactive browsing, retrieval, and analysis of pathways, exploration of pathway connections, and pathway search by gene targets, category, and availability of corresponding bioactivity assay, as well as visualization of pathways on a 3-dimensional globe, in which the distance between any two pathways is proportional to their degree of gene component overlap. Using this resource, we propose a strategy to identify a minimal set of 362 biological assays that can interrogate the universe of human pathways. The NCATS BioPlanet is a public resource, which will be continually expanded and updated, for systems biology, toxicology, and chemical genomics, available at http://tripod.nih.gov/bioplanet/.

Improving the Utility of the Tox21 Dataset by Deep Metadata Annotations and Constructing Reusable Benchmarked Chemical Reference Signatures

The Toxicology in the 21st Century (Tox21) project seeks to develop and test methods for high-throughput examination of the effect certain chemical compounds have on biological systems. Although primary and toxicity assay data were readily available for multiple reporter gene modified cell lines, extensive annotation and curation was required to improve these datasets with respect to how FAIR (Findable, Accessible, Interoperable, and Reusable) they are. In this study, we fully annotated the Tox21 published data with relevant and accepted controlled vocabularies. After removing unreliable data points, we aggregated the results and created three sets of signatures reflecting activity in the reporter gene assays, cytotoxicity, and selective reporter gene activity, respectively. We benchmarked these signatures using the chemical structures of the tested compounds and obtained generally high receiver operating characteristic (ROC) scores, suggesting good quality and utility of these signatures and the underlying data. We analyzed the results to identify promiscuous individual compounds and chemotypes for the three signature categories and interpreted the results to illustrate the utility and re-usability of the datasets. With this study, we aimed to demonstrate the importance of data standards in reporting screening results and high-quality annotations to enable re-use and interpretation of these data. To improve the data with respect to all FAIR criteria, all assay annotations, cleaned and aggregate datasets, and signatures were made available as standardized dataset packages (Aggregated Tox21 bioactivity data, 2019).

Optimization of a Deep-Learning Method Based on the Classification of Images Generated by Parameterized Deep Snap a Novel Molecular-Image-Input Technique for Quantitative Structure-Activity Relationship (QSAR) Analysis

Numerous chemical compounds are distributed around the world and may affect the homeostasis of the endocrine system by disrupting the normal functions of hormone receptors. Although the risks associated with these compounds have been evaluated by acute toxicity testing in mammalian models, the chronic toxicity of many chemicals remains due to high cost of the compounds and the testing, etc. However, computational approaches may be promising alternatives and reduce these evaluations. Recently, deep learning (DL) has been shown to be promising prediction models with high accuracy for recognition of images, speech, signals, and videos since it greatly benefits from large datasets. Recently, a novel DL-based technique called DeepSnap was developed to conduct QSAR analysis using three-dimensional images of chemical structures. It can be used to predict the potential toxicity of many different chemicals to various receptors without extraction of descriptors. DeepSnap has been shown to have a very high capacity in tests using Tox21 quantitative qHTP datasets. Numerous parameters must be adjusted to use the DeepSnap method but they have not been optimized. In this study, the effects of these parameters on the performance of the DL prediction model were evaluated in terms of the loss in validation as an indicator for evaluating the performance of the DL using the toxicity information in the Tox21 qHTP database. The relations of the parameters of DeepSnap such as (1) number of molecules per SDF split into (2) zoom factor percentage, (3) atom size for van der waals percentage, (4) bond radius, (5) minimum bond distance, and (6) bond tolerance, with the validation loss following quadratic function curves, which suggests that optimal thresholds exist to attain the best performance with these prediction models. Using the parameter values set with the best performance, the prediction model of chemical compounds for CAR agonist was built using 64 images, at 105° angle, with AUC of 0.791. Thus, based on these parameters, the proposed DeepSnap-DL approach will be highly reliable and beneficial to establish models to assess the risk associated with various chemicals.

Visual and Semantic Enrichment of Analytical Chemistry Literature Searches by Combining Text Mining and Computational Chemistry

The open-access scientific literature contains a wealth of information for meaningful text mining. However, this information is not always easy to retrieve. This technical note addresses the problem by a new flexible method combining in a single workflow existing resources for literature searches, text mining, and large-scale prediction of physicochemical and biological properties. The results are visualized as virtual mass spectra, chromatograms, or images in styles new to text mining but familiar to analytical chemistry. The method is demonstrated on comparisons of analytical-chemistry techniques and semantically enriched searches for proteins and their activities, but it may also be of general utility in experimental design, drug discovery, chemical syntheses, business intelligence, and historical studies. The method is realized in shareable scientific workflows using only freely available data, services, and software that scale to millions of publications and named chemical entities in the literature.

In Silico Prediction of Endocrine Disrupting Chemicals Using Single-Label and Multilabel Models

Endocrine disruption (ED) has become a serious public health issue and also poses a significant threat to the ecosystem. Due to complex mechanisms of ED, traditional in silico models focusing on only one mechanism are insufficient for detection of endocrine disrupting chemicals (EDCs), let alone offering an overview of possible action mechanisms for a known EDC. To remove these limitations, in this study both single-label and multilabel models were constructed across six ED targets, namely, AR (androgen receptor), ER (estrogen receptor alpha), TR (thyroid receptor), GR (glucocorticoid receptor), PPARg (peroxisome proliferator-activated receptor gamma), and aromatase. Two machine learning methods were used to build the single-label models, with multiple random under-sampling combining voting classification to overcome the challenge of data imbalance. Four methods were explored to construct the multilabel models that can predict the interaction of one EDC against multiple targets simultaneously. The single-label models of all the six targets have achieved reasonable performance with balanced accuracy (BA) values from 0.742 to 0.816. Each top single-label model was then joined to predict the multilabel test set with BA values from 0.586 to 0.711. The multilabel models could offer a significant boost over the single-label baselines with BA values for the multilabel test set from 0.659 to 0.832. Therefore, we concluded that single-label models could be employed for identification of potential EDCs, while multilabel ones are preferable for prediction of possible mechanisms of known EDCs.

Screening Strategies and Methods for Better Off-Target Liability Prediction and Identification of Small-Molecule Pharmaceuticals

Pharmaceutical discovery and development is a long and expensive process that, unfortunately, still results in a low success rate, with drug safety continuing to be a major impedance. Improved safety screening strategies and methods are needed to more effectively fill this critical gap. Recent advances in informatics are now making it possible to manage bigger data sets and integrate multiple sources of screening data in a manner that can potentially improve the selection of higher-quality drug candidates. Integrated screening paradigms have become the norm in Pharma, both in discovery screening and in the identification of off-target toxicity mechanisms during later-stage development. Furthermore, advances in computational methods are making in silico screens more relevant and suggest that they may represent a feasible option for augmenting the current screening paradigm. This paper outlines several fundamental methods of the current drug screening processes across Pharma and emerging techniques/technologies that promise to improve molecule selection. In addition, the authors discuss integrated screening strategies and provide examples of advanced screening paradigms.

Uncertainty quantification in ToxCast high throughput screening

High throughput screening (HTS) projects like the U.S. Environmental Protection Agency's ToxCast program are required to address the large and rapidly increasing number of chemicals for which we have little to no toxicity measurements. Concentration-response parameters such as potency and efficacy are extracted from HTS data using nonlinear regression, and models and analyses built from these parameters are used to predict in vivo and in vitro toxicity of thousands of chemicals. How these predictions are impacted by uncertainties that stem from parameter estimation and propagated through the models and analyses has not been well explored. While data size and complexity makes uncertainty quantification computationally expensive for HTS datasets, continued advancements in computational resources have allowed these computational challenges to be met. This study uses nonparametric bootstrap resampling to calculate uncertainties in concentration-response parameters from a variety of HTS assays. Using the ToxCast estrogen receptor model for bioactivity as a case study, we highlight how these uncertainties can be propagated through models to quantify the uncertainty in model outputs. Uncertainty quantification in model outputs is used to identify potential false positives and false negatives and to determine the distribution of model values around semi-arbitrary activity cutoffs, increasing confidence in model predictions. At the individual chemical-assay level, curves with high variability are flagged for manual inspection or retesting, focusing subject-matter-expert time on results that need further input. This work improves the confidence of predictions made using HTS data, increasing the ability to use this data in risk assessment.

Comprehensive Analyses and Prioritization of Tox21 10K Chemicals Affecting Mitochondrial Function by in-Depth Mechanistic Studies

BACKGROUND

A central challenge in toxicity testing is the large number of chemicals in commerce that lack toxicological assessment. In response, the Tox21 program is re-focusing toxicity testing from animal studies to less expensive and higher throughput in vitro methods using target/pathway-specific, mechanism-driven assays.

OBJECTIVES

Our objective was to use an in-depth mechanistic study approach to prioritize and characterize the chemicals affecting mitochondrial function.

METHODS

We used a tiered testing approach to prioritize for more extensive testing 622 compounds identified from a primary, quantitative high-throughput screen of 8,300 unique small molecules, including drugs and industrial chemicals, as potential mitochondrial toxicants by their ability to significantly decrease the mitochondrial membrane potential (MMP). Based on results from secondary MMP assays in HepG2 cells and rat hepatocytes, 34 compounds were selected for testing in tertiary assays that included formation of reactive oxygen species (ROS), upregulation of p53 and nuclear erythroid 2-related factor 2/antioxidant response element (Nrf2/ARE), mitochondrial oxygen consumption, cellular Parkin translocation, and larval development and ATP status in the nematode Caenorhabditis elegans.

RESULTS

A group of known mitochondrial complex inhibitors (e.g., rotenone) and uncouplers (e.g., chlorfenapyr), as well as potential novel complex inhibitors and uncouplers, were detected. From this study, we identified four not well-characterized potential mitochondrial toxicants (lasalocid, picoxystrobin, pinacyanol, and triclocarban) that merit additional in vivo characterization.

CONCLUSIONS

The tier-based approach for identifying and mechanistically characterizing mitochondrial toxicants can potentially reduce animal use in toxicological testing. https://doi.org/10.1289/EHP2589.

A Workflow for Identifying Metabolically Active Chemicals to Complement in vitro Toxicity Screening

The new paradigm of toxicity testing approaches involves rapid screening of thousands of chemicals across hundreds of biological targets through use of in vitro assays. Such assays may lead to false negatives when the complex metabolic processes that render a chemical bioactive in a living system are unable to be replicated in an in vitro environment. In the current study, a workflow is presented for complementing in vitro testing results with in silico and in vitro techniques to identify inactive parents that may produce active metabolites. A case study applying this workflow involved investigating the influence of metabolism for over 1,400 chemicals considered inactive across18 in vitro assays related to the estrogen receptor (ER) pathway. Over 7,500 first-generation and second-generation metabolites were generated for these in vitro inactive chemicals using an in silico software program. Next, a consensus model comprised of four individual quantitative structure activity relationship (QSAR) models was used to predict ER-binding activity for each of the metabolites. Binding activity was predicted for ~8-10% of metabolites in each generation, with these metabolites linked to 259 in vitro inactive parent chemicals. Metabolites were enriched in substructures consisting of alcohol, aromatic, and phenol bonds relative to their inactive parent chemicals, suggesting these features are potentially favorable for ER-binding. The workflow presented here can be used to identify parent chemicals that can be potentially bioactive, to aid confidence in high throughput risk screening.

Transcriptomic and phenotypic profiling in developing zebrafish exposed to thyroid hormone receptor agonists

There continues to be a need to develop in vivo high-throughput screening (HTS) and computational methods to screen chemicals for interaction with the estrogen, androgen, and thyroid pathways and as complements to in vitro HTS assays. This study explored the utility of an embryonic zebrafish HTS approach to identify and classify endocrine bioactivity using phenotypically-anchored transcriptome profiling. Transcriptome analysis was conducted on zebrafish embryos exposed to 25 estrogen-, androgen-, or thyroid-active chemicals at concentrations that elicited adverse malformations or mortality at 120 h post-fertilization in 80% of animals exposed. Analysis of the top 1000 significant differentially expressed transcripts and developmental toxicity profiles across all treatments identified a unique transcriptional and phenotypic signature for thyroid hormone receptor agonists. This unique signature has the potential to be used as a tiered in vivo HTS and may aid in identifying chemicals that interact with the thyroid hormone receptor.

The US Federal Tox21 Program: A strategic and operational plan for continued leadership

The traditional approaches to toxicity testing have posed multiple challenges for evaluating the safety of commercial chemicals, pesticides, food additives/contaminants, and medical products.The challenges include number of chemicals that need to be tested, time and resource intensive nature of traditional toxicity tests, and unexpected adverse effects that occur in pharmaceutical clinical trials despite the extensive toxicological testing.Over a decade ago, the U.S. Environmental Protection Agency (EPA), National Toxicology Program (NTP), National Center for Advancing Translational Sciences (NCATS), and the Food and Drug Administration (FDA) formed a federal consortium for "Toxicology in the 21st Century" (Tox21) with a focus on developing and evaluating in vitro high-throughput screening (HTS) methods for hazard identification and providing mechanistic insights.The Tox21 consortium generated data on thousands of pharmaceuticals and datapoor chemicals, developed better understanding of the limits and applications of in vitro methods, and enabled incorporation of HTS data into regulatory decisions. To more broadly address the challenges in toxicology, Tox21 has developed a new strategic and operational plan that expands the focus of its research activities. The new focus areas include developing an expanded portfolio of alternative test systems, addressing technical limitations of in vitrotest systems, curating legacy in vivo toxicity testing data, establishing scientific confidence in the in vitrotest systems, and refining alternative methods for characterizing pharmacokinetics and in vitro assay disposition.The new Tox21 strategic and operational plan addresses key challenges to advance toxicology testing and will benefit both the organizations involved and the toxicology community.