CERAPP: Collaborative Estrogen Receptor Activity Prediction Project

Hepatic toxicity prediction of bisphenol analogs by machine learning strategy

Toxicological studies have demonstrated the hepatic toxicity of several bisphenol analogs (BPs), a prevalent type of endocrine disruptor. The development of Adverse Outcome Pathway (AOP) has substantially contributed to the rapid risk assessment for human health. However, the lack of in vitro and in vivo data for the emerging BPs has limited the hazard assessment of these synthetic chemicals. Here, we aimed to develop a new strategy to rapidly predict BPs' hepatotoxicity using network analysis coupled with machine learning models. Considering the structural and functional similarities shared by BPs with Bisphenol A (BPA), we first integrated hepatic disease related genes from multiple databases into BPA-Gene-Phenotype-hepatic toxicity network and subjected it to the computational AOP (cAOP). Through cAOP network and conventional machine learning approaches, we scored the hepatotoxicity of 20 emerging BPs and provided new insights into how BPs' structure features contributed to biologic functions with limited experimental data. Additionally, we assessed the interactions between emerging BPs and ESR1 using molecular docking and proposed an AOP framework wherein ESR1 was a molecular initiating event. Overall, our study provides a computational approach to predict the hepatotoxicity of emerging BPs.

Navigating the complexities of docking tools with nicotinic receptors and acetylcholine binding proteins in the realm of neonicotinoids

Molecular docking, pivotal in predicting small-molecule ligand binding modes, struggles with accurately identifying binding conformations and affinities. This is particularly true for neonicotinoids, insecticides whose impacts on ecosystems require precise molecular interaction modeling. This study scrutinizes the effectiveness of prominent docking software (Ledock, ADFR, Autodock Vina, CDOCKER) in simulating interactions of environmental chemicals, especially neonicotinoid-like molecules with nicotinic acetylcholine receptors (nAChRs) and acetylcholine binding proteins (AChBPs). We aimed to assess the accuracy and reliability of these tools in reproducing crystallographic data, focusing on semi-flexible and flexible docking approaches. Our analysis identified Ledock as the most accurate in semi-flexible docking, while Autodock Vina with Vinardo scoring function proved most reliable. However, no software consistently excelled in both accuracy and reliability. Additionally, our evaluation revealed that none of the tools could establish a clear correlation between docking scores and experimental dissociation constants (Kd) for neonicotinoid-like compounds. In contrast, a strong correlation was found with drug-like compounds, bringing to light a bias in considered software towards pharmaceuticals, thus limiting their applicability to environmental chemicals. The comparison between semi-flexible and flexible docking revealed that the increased computational complexity of the latter did not result in enhanced accuracy. In fact, the higher computational cost of flexible docking with its lack of enhanced predictive accuracy, rendered this approach useless for this class of compounds. Conclusively, our findings emphasize the need for continued development of docking methodologies, particularly for environmental chemicals. This study not only illuminates current software capabilities but also underscores the urgency for advancements in computational molecular docking as it is a relevant tool to environmental sciences.

Grouping strategies for assessing and managing persistent and mobile substances

Background

Persistent, mobile and toxic (PMT), or very persistent and very mobile (vPvM) substances are a wide class of chemicals that are recalcitrant to degradation, easily transported, and potentially harmful to humans and the environment. Due to their persistence and mobility, these substances are often widespread in the environment once emitted, particularly in water resources, causing increased challenges during water treatment processes. Some PMT/vPvM substances such as GenX and perfluorobutane sulfonic acid have been identified as substances of very high concern (SVHCs) under the European Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) regulation. With hundreds to thousands of potential PMT/vPvM substances yet to be assessed and managed, effective and efficient approaches that avoid a case-by-case assessment and prevent regrettable substitution are necessary to achieve the European Union's zero-pollution goal for a non-toxic environment by 2050.

Main

Substance grouping has helped global regulation of some highly hazardous chemicals, e.g., through the Montreal Protocol and the Stockholm Convention. This article explores the potential of grouping strategies for identifying, assessing and managing PMT/vPvM substances. The aim is to facilitate early identification of lesser-known or new substances that potentially meet PMT/vPvM criteria, prompt additional testing, avoid regrettable use or substitution, and integrate into existing risk management strategies. Thus, this article provides an overview of PMT/vPvM substances and reviews the definition of PMT/vPvM criteria and various lists of PMT/vPvM substances available. It covers the current definition of groups, compares the use of substance grouping for hazard assessment and regulation, and discusses the advantages and disadvantages of grouping substances for regulation. The article then explores strategies for grouping PMT/vPvM substances, including read-across, structural similarity and commonly retained moieties, as well as the potential application of these strategies using cheminformatics to predict P, M and T properties for selected examples.

Conclusions

Effective substance grouping can accelerate the assessment and management of PMT/vPvM substances, especially for substances that lack information. Advances to read-across methods and cheminformatics tools are needed to support efficient and effective chemical management, preventing broad entry of hazardous chemicals into the global market and favouring safer and more sustainable alternatives.

A Data-Driven Computational Framework for Assessing the Risk of Placental Exposure to Environmental Chemicals

A computational framework based on placental gene networks was proposed in this work to improve the accuracy of the placental exposure risk assessment of environmental compounds. The framework quantitatively characterizes the ability of compounds to cross the placental barrier by systematically considering the interaction and pathway-level information on multiple placental transporters. As a result, probability scores were generated for 307 compounds crossing the placental barrier based on this framework. These scores were then used to categorize the compounds into different levels of transplacental transport range, creating a gradient partition. These probability scores not only facilitated a more intuitive understanding of a compound's ability to cross the placental barrier but also provided valuable information for predicting potential placental disruptors. Compounds with probability scores greater than 90% were considered to have significant transplacental transport potential, whereas those with probability scores less than 80% were classified as unlikely to cross the placental barrier. Furthermore, external validation set results showed that the probability score could accurately predict the compounds known to cross the placental barrier. In conclusion, the computational framework proposed in this study enhances the intuitive understanding of the ability of compounds to cross the placental barrier and opens up new avenues for assessing the placental exposure risk of compounds.

Two-Stage Machine Learning-Based Approach to Predict Points of Departure for Human Noncancer and Developmental/Reproductive Effects

Chemical points of departure (PODs) for critical health effects are crucial for evaluating and managing human health risks and impacts from exposure. However, PODs are unavailable for most chemicals in commerce due to a lack of in vivo toxicity data. We therefore developed a two-stage machine learning (ML) framework to predict human-equivalent PODs for oral exposure to organic chemicals based on chemical structure. Utilizing ML-based predictions for structural/physical/chemical/toxicological properties from OPERA 2.9 as features (Stage 1), ML models using random forest regression were trained with human-equivalent PODs derived from in vivo data sets for general noncancer effects (n = 1,791) and reproductive/developmental effects (n = 2,228), with robust cross-validation for feature selection and estimating generalization errors (Stage 2). These two-stage models accurately predicted PODs for both effect categories with cross-validation-based root-mean-squared errors less than an order of magnitude. We then applied one or both models to 34,046 chemicals expected to be in the environment, revealing several thousand chemicals of moderate concern and several hundred chemicals of high concern for health effects at estimated median population exposure levels. Further application can expand by orders of magnitude the coverage of organic chemicals that can be evaluated for their human health risks and impacts.

Evaluation of in silico model predictions for mammalian acute oral toxicity and regulatory application in pesticide hazard and risk assessment

The United States Environmental Protection Agency (USEPA) uses the lethal dose 50% (LD50) value from in vivo rat acute oral toxicity studies for pesticide product label precautionary statements and environmental risk assessment (RA). The Collaborative Acute Toxicity Modeling Suite (CATMoS) is a quantitative structure-activity relationship (QSAR)-based in silico approach to predict rat acute oral toxicity that has the potential to reduce animal use when registering a new pesticide technical grade active ingredient (TGAI). This analysis compared LD50 values predicted by CATMoS to empirical values from in vivo studies for the TGAIs of 177 conventional pesticides. The accuracy and reliability of the model predictions were assessed relative to the empirical data in terms of USEPA acute oral toxicity categories and discrete LD50 values for each chemical. CATMoS was most reliable at placing pesticide TGAIs in acute toxicity categories III (>500-5000 mg/kg) and IV (>5000 mg/kg), with 88% categorical concordance for 165 chemicals with empirical in vivo LD50 values ≥ 500 mg/kg. When considering an LD50 for RA, CATMoS predictions of 2000 mg/kg and higher were found to agree with empirical values from limit tests (i.e., single, high-dose tests) or definitive results over 2000 mg/kg with few exceptions.

Chemistry domain of applicability evaluation against existing estrogen receptor high-throughput assay-based activity models

Introduction

The U. S. Environmental Protection Agency's Endocrine Disruptor Screening Program (EDSP) Tier 1 assays are used to screen for potential endocrine system-disrupting chemicals. A model integrating data from 16 high-throughput screening assays to predict estrogen receptor (ER) agonism has been proposed as an alternative to some low-throughput Tier 1 assays. Later work demonstrated that as few as four assays could replicate the ER agonism predictions from the full model with 98% sensitivity and 92% specificity. The current study utilized chemical clustering to illustrate the coverage of the EDSP Universe of Chemicals (UoC) tested in the existing ER pathway models and to investigate the utility of chemical clustering to evaluate the screening approach using an existing 4-assay model as a test case. Although the full original assay battery is no longer available, the demonstrated contribution of chemical clustering is broadly applicable to assay sets, chemical inventories, and models, and the data analysis used can also be applied to future evaluation of minimal assay models for consideration in screening.

Methods

Chemical structures were collected for 6,947 substances via the CompTox Chemicals Dashboard from the over 10,000 UoC and grouped based on structural similarity, generating 826 chemical clusters. Of the 1,812 substances run in the original ER model, 1,730 substances had a single, clearly defined structure. The ER model chemicals with a clearly defined structure that were not present in the EDSP UoC were assigned to chemical clusters using a k-nearest neighbors approach, resulting in 557 EDSP UoC clusters containing at least one ER model chemical.

Results and Discussion

Performance of an existing 4-assay model in comparison with the existing full ER agonist model was analyzed as related to chemical clustering. This was a case study, and a similar analysis can be performed with any subset model in which the same chemicals (or subset of chemicals) are screened. Of the 365 clusters containing >1 ER model chemical, 321 did not have any chemicals predicted to be agonists by the full ER agonist model. The best 4-assay subset ER agonist model disagreed with the full ER agonist model by predicting agonist activity for 122 chemicals from 91 of the 321 clusters. There were 44 clusters with at least two chemicals and at least one agonist based upon the full ER agonist model, which allowed accuracy predictions on a per-cluster basis. The accuracy of the best 4-assay subset ER agonist model ranged from 50% to 100% across these 44 clusters, with 32 clusters having accuracy ≥90%. Overall, the best 4-assay subset ER agonist model resulted in 122 false-positive and only 2 false-negative predictions compared with the full ER agonist model. Most false positives (89) were active in only two of the four assays, whereas all but 11 true positive chemicals were active in at least three assays. False positive chemicals also tended to have lower area under the curve (AUC) values, with 110 out of 122 false positives having an AUC value below 0.214, which is lower than 75% of the positives as predicted by the full ER agonist model. Many false positives demonstrated borderline activity. The median AUC value for the 122 false positives from the best 4-assay subset ER agonist model was 0.138, whereas the threshold for an active prediction is 0.1.

Conclusion

Our results show that the existing 4-assay model performs well across a range of structurally diverse chemicals. Although this is a descriptive analysis of previous results, several concepts can be applied to any screening model used in the future. First, the clustering of the chemicals provides a means of ensuring that future screening evaluations consider the broad chemical space represented by the EDSP UoC. The clusters can also assist in prioritizing future chemicals for screening in specific clusters based on the activity of known chemicals in those clusters. The clustering approach can be useful in providing a framework to evaluate which portions of the EDSP UoC chemical space are reliably covered by in silico and in vitro approaches and where predictions from either method alone or both methods combined are most reliable. The lessons learned from this case study can be easily applied to future evaluations of model applicability and screening to evaluate future datasets.

Deus Ex Machina? The Rise of Artificial Intelligence in Toxicology

Artificial intelligence (AI) is rising rapidly, driven by big data, complex algorithms, and computing resources. Current research presented at the American Chemical Society Fall 2023 Meeting demonstrates AI to be a valuable predictive and supporting tool across all facets of toxicology.

Free and open-source QSAR-ready workflow for automated standardization of chemical structures in support of QSAR modeling

The rapid increase of publicly available chemical structures and associated experimental data presents a valuable opportunity to build robust QSAR models for applications in different fields. However, the common concern is the quality of both the chemical structure information and associated experimental data. This is especially true when those data are collected from multiple sources as chemical substance mappings can contain many duplicate structures and molecular inconsistencies. Such issues can impact the resulting molecular descriptors and their mappings to experimental data and, subsequently, the quality of the derived models in terms of accuracy, repeatability, and reliability. Herein we describe the development of an automated workflow to standardize chemical structures according to a set of standard rules and generate two and/or three-dimensional "QSAR-ready" forms prior to the calculation of molecular descriptors. The workflow was designed in the KNIME workflow environment and consists of three high-level steps. First, a structure encoding is read, and then the resulting in-memory representation is cross-referenced with any existing identifiers for consistency. Finally, the structure is standardized using a series of operations including desalting, stripping of stereochemistry (for two-dimensional structures), standardization of tautomers and nitro groups, valence correction, neutralization when possible, and then removal of duplicates. This workflow was initially developed to support collaborative modeling QSAR projects to ensure consistency of the results from the different participants. It was then updated and generalized for other modeling applications. This included modification of the "QSAR-ready" workflow to generate "MS-ready structures" to support the generation of substance mappings and searches for software applications related to non-targeted analysis mass spectrometry. Both QSAR and MS-ready workflows are freely available in KNIME, via standalone versions on GitHub, and as docker container resources for the scientific community. Scientific contribution: This work pioneers an automated workflow in KNIME, systematically standardizing chemical structures to ensure their readiness for QSAR modeling and broader scientific applications. By addressing data quality concerns through desalting, stereochemistry stripping, and normalization, it optimizes molecular descriptors' accuracy and reliability. The freely available resources in KNIME, GitHub, and docker containers democratize access, benefiting collaborative research and advancing diverse modeling endeavors in chemistry and mass spectrometry.

Protecting Human and Animal Health: The Road from Animal Models to New Approach Methods

Animals and animal models have been invaluable for our current understanding of human and animal biology, including physiology, pharmacology, biochemistry, and disease pathology. However, there are increasing concerns with continued use of animals in basic biomedical, pharmacological, and regulatory research to provide safety assessments for drugs and chemicals. There are concerns that animals do not provide sufficient information on toxicity and/or efficacy to protect the target population, so scientists are utilizing the principles of replacement, reduction, and refinement (the 3Rs) and increasing the development and application of new approach methods (NAMs). NAMs are any technology, methodology, approach, or assay used to understand the effects and mechanisms of drugs or chemicals, with specific focus on applying the 3Rs. Although progress has been made in several areas with NAMs, complete replacement of animal models with NAMs is not yet attainable. The road to NAMs requires additional development, increased use, and, for regulatory decision making, usually formal validation. Moreover, it is likely that replacement of animal models with NAMs will require multiple assays to ensure sufficient biologic coverage. The purpose of this manuscript is to provide a balanced view of the current state of the use of animal models and NAMs as approaches to development, safety, efficacy, and toxicity testing of drugs and chemicals. Animals do not provide all needed information nor do NAMs, but each can elucidate key pieces of the puzzle of human and animal biology and contribute to the goal of protecting human and animal health. SIGNIFICANCE STATEMENT: Data from traditional animal studies have predominantly been used to inform human health safety and efficacy. Although it is unlikely that all animal studies will be able to be replaced, with the continued advancement in new approach methods (NAMs), it is possible that sometime in the future, NAMs will likely be an important component by which the discovery, efficacy, and toxicity testing of drugs and chemicals is conducted and regulatory decisions are made.

Screening of the Antagonistic Activity of Potential Bisphenol A Alternatives toward the Androgen Receptor Using Machine Learning and Molecular Dynamics Simulation

Over the past few decades, extensive research has indicated that exposure to bisphenol A (BPA) increases the health risks in humans. Toxicological studies have demonstrated that BPA can bind to the androgen receptor (AR), resulting in endocrine-disrupting effects. In recent investigations, many alternatives to BPA have been detected in various environmental media as major pollutants. However, related experimental evaluations of BPA alternatives have not been systematically implemented for the assessment of chemical safety and the effects of structural characteristics on the antagonistic activity of the AR. To promote the green development of BPA alternatives, high-throughput toxicological screening is fundamental for prioritizing chemical tests. Therefore, we proposed a hybrid deep learning architecture that combines molecular descriptors and molecular graphs to predict AR antagonistic activity. Compared to previous models, this hybrid architecture can extract substantial chemical information from various molecular representations to improve the model's generalization ability for BPA alternatives. Our predictions suggest that lignin-derivable bisguaiacols, as alternatives to BPA, are likely to be nonantagonist for AR compared to bisphenol analogues. Additionally, molecular dynamics (MD) simulations identified the dihydrotestosterone-bound pocket, rather than the surface, as the major binding site of bisphenol analogues. The conformational changes of key helix H12 from an agonistic to an antagonistic conformation can be evaluated qualitatively by accelerated MD simulations to explain the underlying mechanism. Overall, our computational study is helpful for toxicological screening of BPA alternatives and the design of environmentally friendly BPA alternatives.

Computational modeling approaches for developing a synergistic effect prediction model of estrogen agonistic activity

The concerns regarding the potential health threats caused by estrogenic endocrine-disrupting chemicals (EDCs) and their mixtures manufactured by the chemical industry are increasing worldwide. Conventional experimental tests for understanding the estrogenic activity of mixtures are expensive and time-consuming. Although non-testing methods using computational modeling approaches have been developed to reduce the number of traditional tests, they are unsuitable for predicting synergistic effects because current prediction models consider only a single chemical. Thus, the development of predictive models is essential for predicting the mixture toxicity, including chemical interactions. However, selecting suitable computational modeling approaches to develop a high-performance prediction model requires considerable time and effort. In this study, we provide a suitable computational approach to develop a predictive model for the synergistic effects of estrogenic activity. We collected datasets on mixture toxicity based on the synergistic effect of estrogen agonistic activity in binary mixtures. Using the model deviation ratio approach, we classified the labels of the binary mixtures as synergistic or non-synergistic effects. We assessed five molecular descriptors, four machine learning-based algorithms, and a deep learning-based algorithm to provide a suitable computational modeling approach. Compared with other modeling approaches, the prediction model using the deep learning-based algorithm and chemical-protein network descriptors exhibited the best performance in predicting the synergistic effects. In conclusion, we developed a new high-performance binary classification model using a deep neural network and chemical-protein network-based descriptors. The developed model will be helpful for the preliminary screening of the synergistic effects of binary mixtures during the development process of chemical products.

Development and application of consensus in silico models for advancing high-throughput toxicological predictions

Computational toxicology models have been successfully implemented to prioritize and screen chemicals. There are numerous in silico (quantitative) structure-activity relationship ([Q]SAR) models for the prediction of a range of human-relevant toxicological endpoints, but for a given endpoint and chemical, not all predictions are identical due to differences in their training sets, algorithms, and methodology. This poses an issue for high-throughput screening of a large chemical inventory as it necessitates several models to cover diverse chemistries but will then generate data conflicts. To address this challenge, we developed a consensus modeling strategy to combine predictions obtained from different existing in silico (Q)SAR models into a single predictive value while also expanding chemical space coverage. This study developed consensus models for nine toxicological endpoints relating to estrogen receptor (ER) and androgen receptor (AR) interactions (i.e., binding, agonism, and antagonism) and genotoxicity (i.e., bacterial mutation, in vitro chromosomal aberration, and in vivo micronucleus). Consensus models were created by combining different (Q)SAR models using various weighting schemes. As a multi-objective optimization problem, there is no single best consensus model, and therefore, Pareto fronts were determined for each endpoint to identify the consensus models that optimize the multiple-criterion decisions simultaneously. Accordingly, this work presents sets of solutions for each endpoint that contain the optimal combination, regardless of the trade-off, with the results demonstrating that the consensus models improved both the predictive power and chemical space coverage. These solutions were further analyzed to find trends between the best consensus models and their components. Here, we demonstrate the development of a flexible and adaptable approach for in silico consensus modeling and its application across nine toxicological endpoints related to ER activity, AR activity, and genotoxicity. These consensus models are developed to be integrated into a larger multi-tier NAM-based framework to prioritize chemicals for further investigation and support the transition to a non-animal approach to risk assessment in Canada.

Application of the Key Characteristics Framework to Identify Potential Breast Carcinogens Using Publicly Available in Vivo, in Vitro, and in Silico Data

BACKGROUND

Chemicals that induce mammary tumors in rodents or activate estrogen or progesterone signaling are likely to increase breast cancer (BC) risk. Identifying chemicals with these activities can prompt steps to protect human health.

OBJECTIVES

We compiled data on rodent tumors, endocrine activity, and genotoxicity to assess the key characteristics (KCs) of rodent mammary carcinogens (MCs), and to identify other chemicals that exhibit these effects and may therefore increase BC risk.

METHODS

Using authoritative databases, including International Agency for Research on Cancer (IARC) Monographs and the US Environmental Protection's (EPA) ToxCast, we selected chemicals that induce mammary tumors in rodents, stimulate estradiol or progesterone synthesis, or activate the estrogen receptor (ER) in vitro. We classified these chemicals by their genotoxicity and strength of endocrine activity and calculated the overrepresentation (enrichment) of these KCs among MCs. Finally, we evaluated whether these KCs predict whether a chemical is likely to induce mammary tumors.

RESULTS

We identified 279 MCs and an additional 642 chemicals that stimulate estrogen or progesterone signaling. MCs were significantly enriched for steroidogenicity, ER agonism, and genotoxicity, supporting the use of these KCs to predict whether a chemical is likely to induce rodent mammary tumors and, by inference, increase BC risk. More MCs were steroidogens than ER agonists, and many increased both estradiol and progesterone. Enrichment among MCs was greater for strong endocrine activity vs. weak or inactive, with a significant trend.

DISCUSSION

We identified hundreds of compounds that have biological activities that could increase BC risk and demonstrated that these activities are enriched among MCs. We argue that many of these should not be considered low hazard without investigating their ability to affect the breast, and chemicals with the strongest evidence can be targeted for exposure reduction. We describe ways to strengthen hazard identification, including improved assessments for mammary effects, developing assays for more KCs, and more comprehensive chemical testing. https://doi.org/10.1289/EHP13233.

Interpretable machine learning for the identification of estrogen receptor agonists, antagonists, and binders

An abnormal hormonal activity or exposure to endocrine-disrupting chemicals (EDCs) can cause endocrine system malfunction. Among the many interactions EDCs can affect is the disruption of estrogen signalling, which can lead to adverse health effects such as cancer, osteoporosis, neurodegenerative diseases, cardiovascular disease, insulin resistance, and obesity. Knowing which chemical can act as an EDC is a significant advantage and a practical necessity. New Approach Methodologies (NAM) computational models offer a quick and cost-effective solution for preliminary hazard assessment of chemicals without animal testing. Therefore, a machine learning approach was used to investigate the relationships between estrogen receptor (ER) activity and chemical structure to identify chemicals that can interact with ER. For this purpose, the consolidated in vitro assay data from ToxCast/Tox21 projects was used for developing Random Forest classification models for ER binding, agonists, and antagonists. The overall classification prediction accuracy reaches up to 82%, depending on whether the model predicted agonists, antagonists, or compounds that bind to the active site. Given the imbalance in endocrine disruption data, the derived models are good candidates for deprioritising chemicals and reducing animal testing. The interpretation of theoretical molecular descriptors of the models was consistent with the molecular interactions known in the ligand binding pocket. The estimated class probabilities enabled the analysis of the applicability domain of the developed models and the assessment of the predictions' reliability, followed by the guidelines for interpreting prediction results. The models are openly accessible and useable at QsarDB.org (http://dx.doi.org/10.15152/QDB.259) according to the FAIR (Findable, Accessible, Interoperable, Reusable) principles.

Evaluation of new alternative methods for the identification of estrogenic, androgenic and steroidogenic effects: a comparative in vitro/in silico study

A suite of in vitro assays and in silico models were evaluated to identify which best detected the endocrine-disrupting (ED) potential of 10 test chemicals according to their estrogenic, androgenic and steroidogenic (EAS) potential compared to the outcomes from ToxCast. In vitro methods included receptor-binding, CALUX transactivation, H295R steroidogenesis, aromatase activity inhibition and the Yeast oestrogen (YES) and Yeast androgen screen (YAS) assays. The impact of metabolism was also evaluated. The YES/YAS assays exhibited a high sensitivity for ER effects and, despite some challenges in predicting AR effects, is a good initial screening assay. Results from receptor-binding and CALUX assays generally correlated and were in accordance with classifications based on ToxCast assays. ER agonism and AR antagonism of benzyl butyl phthalate were abolished when CALUX assays included liver S9. In silico final calls were mostly in agreement with the in vitro assays, and predicted ER and AR effects well. The efficiency of the in silico models (reflecting applicability domains or inconclusive results) was 43-100%. The percentage of correct calls for ER (50-100%), AR (57-100%) and aromatase (33-100%) effects when compared to the final ToxCast call covered a wide range from highly reliable to less reliable models. In conclusion, Danish (Q)SAR, Opera, ADMET Lab LBD and ProToxII models demonstrated the best overall performance for ER and AR effects. These can be combined with the YES/YAS assays in an initial screen of chemicals in the early tiers of an NGRA to inform on the MoA and the design of mechanistic in vitro assays used later in the assessment. Inhibition of aromatase was best predicted by the Vega, AdmetLab and ProToxII models. Other mechanisms and exposure should be considered when making a conclusion with respect to ED effects.

Where developmental toxicity meets explainable artificial intelligence: state-of-the-art and perspectives

INTRODUCTION

The application of Artificial Intelligence (AI) to predictive toxicology is rapidly increasing, particularly aiming to develop non-testing methods that effectively address ethical concerns and reduce economic costs. In this context, Developmental Toxicity (Dev Tox) stands as a key human health endpoint, especially significant for safeguarding maternal and child well-being.

AREAS COVERED

This review outlines the existing methods employed in Dev Tox predictions and underscores the benefits of utilizing New Approach Methodologies (NAMs), specifically focusing on eXplainable Artificial Intelligence (XAI), which proves highly efficient in constructing reliable and transparent models aligned with recommendations from international regulatory bodies.

EXPERT OPINION

The limited availability of high-quality data and the absence of dependable Dev Tox methodologies render XAI an appealing avenue for systematically developing interpretable and transparent models, which hold immense potential for both scientific evaluations and regulatory decision-making.

A Graph Neural Network Model with a Transparent Decision-Making Process Defines the Applicability Domain for Environmental Estrogen Screening

The application of deep learning (DL) models for screening environmental estrogens (EEs) for the sound management of chemicals has garnered significant attention. However, the currently available DL model for screening EEs lacks both a transparent decision-making process and effective applicability domain (AD) characterization, making the reliability of its prediction results uncertain and limiting its practical applications. To address this issue, a graph neural network (GNN) model was developed to screen EEs, achieving accuracy rates of 88.9% and 92.5% on the internal and external test sets, respectively. The decision-making process of the GNN model was explored through the network-like similarity graphs (NSGs) based on the model features (FT). We discovered that the accuracy of the predictions is dependent on the feature distribution of compounds in NSGs. An AD characterization method called ADFT was proposed, which excludes predictions falling outside of the model's prediction range, leading to a 15% improvement in the F1 score of the GNN model. The GNN model with the AD method may serve as an efficient tool for screening EEs, identifying 800 potential EEs in the Inventory of Existing Chemical Substances of China. Additionally, this study offers new insights into comprehending the decision-making process of DL models.

Review of machine learning and deep learning models for toxicity prediction

The ever-increasing number of chemicals has raised public concerns due to their adverse effects on human health and the environment. To protect public health and the environment, it is critical to assess the toxicity of these chemicals. Traditional in vitro and in vivo toxicity assays are complicated, costly, and time-consuming and may face ethical issues. These constraints raise the need for alternative methods for assessing the toxicity of chemicals. Recently, due to the advancement of machine learning algorithms and the increase in computational power, many toxicity prediction models have been developed using various machine learning and deep learning algorithms such as support vector machine, random forest, k-nearest neighbors, ensemble learning, and deep neural network. This review summarizes the machine learning- and deep learning-based toxicity prediction models developed in recent years. Support vector machine and random forest are the most popular machine learning algorithms, and hepatotoxicity, cardiotoxicity, and carcinogenicity are the frequently modeled toxicity endpoints in predictive toxicology. It is known that datasets impact model performance. The quality of datasets used in the development of toxicity prediction models using machine learning and deep learning is vital to the performance of the developed models. The different toxicity assignments for the same chemicals among different datasets of the same type of toxicity have been observed, indicating benchmarking datasets is needed for developing reliable toxicity prediction models using machine learning and deep learning algorithms. This review provides insights into current machine learning models in predictive toxicology, which are expected to promote the development and application of toxicity prediction models in the future.

Utilizing Heteroatom Types and Numbers from Extensive Ligand Libraries to Develop Novel hERG Blocker QSAR Models Using Machine Learning-Based Classifiers

The human ether-à-go-go-related gene (hERG) channel plays a crucial role in membrane repolarization. Any disruptions in its function can lead to severe cardiovascular disorders such as long QT syndrome (LQTS), which increases the risk of serious cardiovascular problems such as tachyarrhythmia and sudden cardiac death. Drug-induced LQTS is a significant concern and has resulted in drug withdrawals from the market in the past. The main objective of this study is to pinpoint crucial heteroatoms present in ligands that initiate interactions leading to the effective blocking of the hERG channel. To achieve this aim, ligand-based quantitative structure-activity relationships (QSAR) models were constructed using extensive ligand libraries, considering the heteroatom types and numbers, and their associated hERG channel blockage pIC50 values. Machine learning-assisted QSAR models were developed to analyze the key structural components influencing compound activity. Among the various methods, the KPLS method proved to be the most efficient, allowing the construction of models based on eight distinct fingerprints. The study delved into investigating the influence of heteroatoms on the activity of hERG blockers, revealing their significant role. Furthermore, by quantifying the effect of heteroatom types and numbers on ligand activity at the hERG channel, six compound pairs were selected for molecular docking. Subsequent molecular dynamics simulations and per residue MM/GBSA calculations were performed to comprehensively analyze the interactions of the selected pair compounds.

Comparison between endocrine activity assessed using ToxCast/Tox21 database and human plasma concentration of sunscreen active ingredients/UV filters

Sunscreen products are composed of ultraviolet (UV) filters and formulated to reduce exposure to sunlight thereby lessening skin damage. Concerns have been raised regarding the toxicity and potential endocrine disrupting (ED) effects of UV filters. The ToxCast/Tox21 program, that is, CompTox, is a high-throughput in vitro screening database of chemicals that identify adverse outcome pathways, key events, and ED potential of chemicals. Using the ToxCast/Tox21 database, octisalate, homosalate, octocrylene, oxybenzone, octinoxate, and avobenzone, 6 commonly used organic UV filters, were found to have been evaluated. These UV filters showed low potency in these bioassays with most activity detected above the range of the cytotoxic burst. The pathways that were most affected were the cell cycle and the nuclear receptor pathways. Most activity was observed in liver and kidney-based bioassays. These organic filters and their metabolites showed relatively weak ED activity when tested in bioassays measuring estrogen receptor (ER), androgen receptor (AR), thyroid receptor, and steroidogenesis activity. Except for oxybenzone, all activity in the endocrine assays occurred at concentrations greater than the cytotoxic burst. Moreover, except for oxybenzone, plasma concentrations (Cmax) measured in humans were at least 100× lower than bioactive (AC50/ACC) concentrations that produced a response in ToxCast/Tox21 assays. These data are consistent with in vivo animal/human studies showing weak or negligible endocrine activity. In sum, when considered as part of a weight-of-evidence assessment and compared with measured plasma concentrations, the results show these organic UV filters have low intrinsic biological activity and risk of toxicity including endocrine disruption in humans.

Priority list of potential endocrine-disrupting chemicals in food chemical contaminants: a docking study and in vitro/epidemiological evidence integration

Diet is an important exposure route of endocrine-disrupting chemicals (EDCs), but many unfiltered potential EDCs remain in food. The in silico prediction of EDCs is a popular method for preliminary screening. Potential EDCs in food were screened using Endocrine Disruptome, an open-source platform for inverse docking, to predict the binding probabilities of 587 food chemical contaminants with 18 human nuclear hormone receptor (NHR) conformations. In total, 25 contaminants were bound to multiple NHRs such as oestrogen receptor α/β and androgen receptor. These 25 compounds mainly include pesticides and per- and polyfluoroalkyl substances (PFASs). The prediction results were validated with the in vitro data. The structural features and the crucial amino acid residues of the four NHRs were also validated based on previous literature. The findings indicate that the screening has good prediction efficiency. In addition, the epidemic evidence about endocrine interference of PFASs in food on children was further validated through this screening. This study provides preliminary screening results for EDCs in food and a priority list for in vitro and in vivo research.

High Throughput Read-Across for Screening a Large Inventory of Related Structures by Balancing Artificial Intelligence/Machine Learning and Human Knowledge

Read-across is an in silico method applied in chemical risk assessment for data-poor chemicals. The read-across outcomes for repeated-dose toxicity end points include the no-observed-adverse-effect level (NOAEL) and estimated uncertainty for a particular category of effects. We have previously developed a new paradigm for estimating NOAELs based on chemoinformatics analysis and experimental study qualities from selected analogues, not relying on quantitative structure-activity relationships (QSARs) or rule-based SAR systems, which are not well-suited to end points for which the underpinning data are weakly grounded in specific chemical-biological interactions. The central hypothesis of this approach is that similar compounds have similar toxicity profiles and, hence, similar NOAEL values. Analogue quality (AQ) quantifies the suitability of an analogue candidate for reading across to the target by considering similarity from structure, physicochemical, ADME (absorption, distribution, metabolism, excretion), and biological perspectives. Biological similarity is based on experimental data; assay vectors derived from aggregations of ToxCast/Tox21 data are used to derive machine learning (ML) hybrid rules that serve as biological fingerprints to capture target-analogue similarity relevant to specific effects of interest, for example, hormone receptors (ER/AR/THR). Once one or more analogues have been qualified for read-across, a decision theory approach is used to estimate confidence bounds for the NOAEL of the target. The confidence interval is dramatically narrowed when analogues are constrained to biologically related profiles. Although this read-across process works well for a single target with several analogues, it can become unmanageable when, for example, screening multiple targets (e.g., virtual screening library) or handling a parent compound having numerous metabolites. To this end, we have established a digitalized framework to enable the assessment of a large number of substances, while still allowing for human decisions for filtering and prioritization. This workflow was developed and validated through a use case of a large set of bisphenols and their metabolites.

ProfhEX: AI-based platform for small molecules liability profiling

Off-target drug interactions are a major reason for candidate failure in the drug discovery process. Anticipating potential drug's adverse effects in the early stages is necessary to minimize health risks to patients, animal testing, and economical costs. With the constantly increasing size of virtual screening libraries, AI-driven methods can be exploited as first-tier screening tools to provide liability estimation for drug candidates. In this work we present ProfhEX, an AI-driven suite of 46 OECD-compliant machine learning models that can profile small molecules on 7 relevant liability groups: cardiovascular, central nervous system, gastrointestinal, endocrine, renal, pulmonary and immune system toxicities. Experimental affinity data was collected from public and commercial data sources. The entire chemical space comprised 289'202 activity data for a total of 210'116 unique compounds, spanning over 46 targets with dataset sizes ranging from 819 to 18896. Gradient boosting and random forest algorithms were initially employed and ensembled for the selection of a champion model. Models were validated according to the OECD principles, including robust internal (cross validation, bootstrap, y-scrambling) and external validation. Champion models achieved an average Pearson correlation coefficient of 0.84 (SD of 0.05), an R² determination coefficient of 0.68 (SD = 0.1) and a root mean squared error of 0.69 (SD of 0.08). All liability groups showed good hit-detection power with an average enrichment factor at 5% of 13.1 (SD of 4.5) and AUC of 0.92 (SD of 0.05). Benchmarking against already existing tools demonstrated the predictive power of ProfhEX models for large-scale liability profiling. This platform will be further expanded with the inclusion of new targets and through complementary modelling approaches, such as structure and pharmacophore-based models. ProfhEX is freely accessible at the following address: https://profhex.exscalate.eu/ .

Predictive Models for Compound Binding to Androgen and Estrogen Receptors Based on Counter-Propagation Artificial Neural Networks

Endocrine-disrupting chemicals (EDCs) are exogenous substances that interfere with the normal function of the human endocrine system. These chemicals can affect specific nuclear receptors, such as androgen receptors (ARs) or estrogen receptors (ER) α and β, which play a crucial role in regulating complex physiological processes in humans. It is now more crucial than ever to identify EDCs and reduce exposure to them. For screening and prioritizing chemicals for further experimentation, the use of artificial neural networks (ANN), which allow the modeling of complicated, nonlinear relationships, is most appropriate. We developed six models that predict the binding of a compound to ARs, ERα, or ERβ as agonists or antagonists, using counter-propagation artificial neural networks (CPANN). Models were trained on a dataset of structurally diverse compounds, and activity data were obtained from the CompTox Chemicals Dashboard. Leave-one-out (LOO) tests were performed to validate the models. The results showed that the models had excellent performance with prediction accuracy ranging from 94% to 100%. Therefore, the models can predict the binding affinity of an unknown compound to the selected nuclear receptor based solely on its chemical structure. As such, they represent important alternatives for the safety prioritization of chemicals.

New Approach Methodologies for the Endocrine Activity Toolbox: Environmental Assessment for Fish and Amphibians

Multiple in vivo test guidelines focusing on the estrogen, androgen, thyroid, and steroidogenesis pathways have been developed and validated for mammals, amphibians, or fish. However, these tests are resource-intensive and often use a large number of laboratory animals. Developing alternatives for in vivo tests is consistent with the replacement, reduction, and refinement principles for animal welfare considerations, which are supported by increasing mandates to move toward an "animal-free" testing paradigm worldwide. New approach methodologies (NAMs) hold great promise to identify molecular, cellular, and tissue changes that can be used to predict effects reliably and more efficiently at the individual level (and potentially on populations) while reducing the number of animals used in (eco)toxicological testing for endocrine disruption. In a collaborative effort, experts from government, academia, and industry met in 2020 to discuss the current challenges of testing for endocrine activity assessment for fish and amphibians. Continuing this cross-sector initiative, our review focuses on the current state of the science regarding the use of NAMs to identify chemical-induced endocrine effects. The present study highlights the challenges of using NAMs for safety assessment and what work is needed to reduce their uncertainties and increase their acceptance in regulatory processes. We have reviewed the current NAMs available for endocrine activity assessment including in silico, in vitro, and eleutheroembryo models. New approach methodologies can be integrated as part of a weight-of-evidence approach for hazard or risk assessment using the adverse outcome pathway framework. The development and utilization of NAMs not only allows for replacement, reduction, and refinement of animal testing but can also provide robust and fit-for-purpose methods to identify chemicals acting via endocrine mechanisms. Environ Toxicol Chem 2023;42:757-777. © 2023 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.

A method for in vitro data and structure curation to optimize for QSAR modelling of minimum absolute potency levels and a comparative use case

Large screening programs such as the US Tox21 are releasing experimental in vitro results for many endpoints of relevance for human health. In (Q)SAR modelling, it is essential to clearly define the endpoint (OECD QSAR Validation Principle 1) and extract the most robust data points according to the definition. We have developed a comprehensive data curation procedure to interpret in vitro experimental data sets for (Q)SAR development, with modules for selecting actives according to quality of curve fittings, magnitude of activity and 'absolute' potency cut-offs, requiring non-cytotoxicity at activity concentration; extracting only very robust inactives; selecting only substances tested in high purity; and accounting for assay signal interference. A structure curation procedure with uniform representation of tautomeric classes of substances is also developed. The detailed method and a use case of modelling Tox21 data for an estrogen receptor α agonism assay with and without use of the method is presented.

Practical application of the interim internal threshold of toxicological concern (iTTC): a case study based on clinical data

We present a case study that provides a practical step-by-step example of how the internal Threshold of Toxicological Concern (iTTC) can be used as a tool to refine a TTC-based assessment for dermal exposures to consumer products. The case study uses a theoretical scenario where there are no systemic toxicity data for the case study chemicals (avobenzone, oxybenzone, octocrylene, homosalate, octisalate, octinoxate, and ecamsule). Human dermal pharmacokinetic data following single and repeat dermal exposure to products containing the case study chemicals were obtained from data published by the US FDA. The clinical studies utilized an application procedure that followed maximal use conditions (product applied as 2 mg/cm² to 75% of the body surface area, 4 times a day). The case study chemicals were first reviewed to determine if they were in the applicability domain of the iTTC, and then, the human plasma concentrations were compared to an iTTC limit of 1 µM. When assessed under maximum usage, the external exposure of all chemicals exceeded the external dose TTC limits. By contrast, the internal exposure to all chemicals, except oxybenzone, was an order of magnitude lower than the 1 µM interim iTTC threshold. This work highlights the importance of understanding internal exposure relative to external dose and how the iTTC can be a valuable tool for assessing low-level internal exposures; additionally, the work demonstrates how to use an iTTC, and highlights considerations and refinement opportunities for the approach.

Racial and social disparities in Ventura County, California related to agricultural pesticide applications and toxicity

Application of agricultural pesticides poses health concerns for farmworkers and for local communities due to pesticide drift from spraying or fumigation, pesticide volatilization into the air, contamination of household dust, as well as direct exposure for people who work in agriculture and their families. In this analysis of pesticide use records for Ventura County, California (USA) from 2016 to 2018, we identified the most prevalent toxicological effects of the pesticides applied. We also developed a cumulative toxicity index that incorporates specific toxicity endpoints for individual pesticides, the severity and strength of association for each endpoint, and the reliability of the data sources. Combining the toxicity index for each pesticide with the pounds applied within each square mile section in Ventura County, we calculated the total toxicity-weighted pesticide use and identified pesticides associated with higher potential risk to health. Analysis of U.S. Census data for Ventura County found a greater percentage of Hispanic/Latino, African American and Asian community members in township sections with a greater volume of pesticides applied and higher toxicity-weighted pesticide use. Similarly, areas with limited economic and social resources had elevated pesticide application overall and elevated toxicity-weighted pesticide use. The combination of toxicological and demographic analyses presented in this study provides information that can support the development of policies to protect public health from excessive exposure to pesticides and better environmental health protection for socially vulnerable populations.

Predicting Endocrine Disruption Using Conformal Prediction - A Prioritization Strategy to Identify Hazardous Chemicals with Confidence

Receptor-mediated molecular initiating events (MIEs) and their relevance in endocrine activity (EA) have been highlighted in literature. More than 15 receptors have been associated with neurodevelopmental adversity and metabolic disruption. MIEs describe chemical interactions with defined biological outcomes, a relationship that could be described with quantitative structure-activity relationship (QSAR) models. QSAR uncertainty can be assessed using the conformal prediction (CP) framework, which provides similarity (i.e., nonconformity) scores relative to the defined classes per prediction. CP calibration can indirectly mitigate data imbalance during model development, and the nonconformity scores serve as intrinsic measures of chemical applicability domain assessment during screening. The focus of this work was to propose an in silico predictive strategy for EA. First, 23 QSAR models for MIEs associated with EA were developed using high-throughput data for 14 receptors. To handle the data imbalance, five protocols were compared, and CP provided the most balanced class definition. Second, the developed QSAR models were applied to a large data set (∼55,000 chemicals), comprising chemicals representative of potential risk for human exposure. Using CP, it was possible to assess the uncertainty of the screening results and identify model strengths and out of domain chemicals. Last, two clustering methods, t-distributed stochastic neighbor embedding and Tanimoto similarity, were used to identify compounds with potential EA using known endocrine disruptors as reference. The cluster overlap between methods produced 23 chemicals with suspected or demonstrated EA potential. The presented models could be utilized for first-tier screening and identification of compounds with potential biological activity across the studied MIEs.

Navigating Transcriptomic Connectivity Mapping Workflows to Link Chemicals with Bioactivities

Screening new compounds for potential bioactivities against cellular targets is vital for drug discovery and chemical safety. Transcriptomics offers an efficient approach for assessing global gene expression changes, but interpreting chemical mechanisms from these data is often challenging. Connectivity mapping is a potential data-driven avenue for linking chemicals to mechanisms based on the observation that many biological processes are associated with unique gene expression signatures (gene signatures). However, mining the effects of a chemical on gene signatures for biological mechanisms is challenging because transcriptomic data contain thousands of noisy genes. New connectivity mapping approaches seeking to distinguish signal from noise continue to be developed, spurred by the promise of discovering chemical mechanisms, new drugs, and disease targets from burgeoning transcriptomic data. Here, we analyze these approaches in terms of diverse transcriptomic technologies, public databases, gene signatures, pattern-matching algorithms, and statistical evaluation criteria. To navigate the complexity of connectivity mapping, we propose a harmonized scheme to coherently organize and compare published workflows. We first standardize concepts underlying transcriptomic profiles and gene signatures based on various transcriptomic technologies such as microarrays, RNA-Seq, and L1000 and discuss the widely used data sources such as Gene Expression Omnibus, ArrayExpress, and MSigDB. Next, we generalize connectivity mapping as a pattern-matching task for finding similarity between a query (e.g., transcriptomic profile for new chemical) and a reference (e.g., gene signature of known target). Published pattern-matching approaches fall into two main categories: vector-based use metrics like correlation, Jaccard index, etc., and aggregation-based use parametric and nonparametric statistics (e.g., gene set enrichment analysis). The statistical methods for evaluating the performance of different approaches are described, along with comparisons reported in the literature on benchmark transcriptomic data sets. Lastly, we review connectivity mapping applications in toxicology and offer guidance on evaluating chemical-induced toxicity with concentration-response transcriptomic data. In addition to serving as a high-level guide and tutorial for understanding and implementing connectivity mapping workflows, we hope this review will stimulate new algorithms for evaluating chemical safety and drug discovery using transcriptomic data.

Evaluating structure-based activity in a high-throughput assay for steroid biosynthesis

Data from a high-throughput human adrenocortical carcinoma assay (HT-H295R) for steroid hormone biosynthesis are available for >2000 chemicals in single concentration and 654 chemicals in multi-concentration (mc). Previously, a metric describing the effect size of a chemical on the biosynthesis of 11 hormones was derived using mc data referred to as the maximum mean Mahalanobis distance (maxmMd). However, mc HT-H295R assay data remain unavailable for many chemicals. This work leverages existing HT-H295R assay data by constructing structure-activity relationships to make predictions for data-poor chemicals, including: (1) identification of individual structural descriptors, known as ToxPrint chemotypes, associated with increased odds of affecting estrogen or androgen synthesis; (2) a random forest (RF) classifier using physicochemical property descriptors to predict HT-H295R maxmMd binary (positive or negative) outcomes; and, (3) a local approach to predict maxmMd binary outcomes using nearest neighbors (NNs) based on two types of chemical fingerprints (chemotype or Morgan). Individual chemotypes demonstrated high specificity (85-98%) for modulators of estrogen and androgen synthesis but with low sensitivity. The best RF model for maxmMd classification included 13 predicted physicochemical descriptors, yielding a balanced accuracy (BA) of 71% with only modest improvement when hundreds of structural features were added. The best two NN models for binary maxmMd prediction demonstrated BAs of 85 and 81% using chemotype and Morgan fingerprints, respectively. Using an external test set of 6302 chemicals (lacking HT-H295R data), 1241 were identified as putative estrogen and androgen modulators. Combined results across the three classification models (global RF model and two local NN models) predict that 1033 of the 6302 chemicals would be more likely to affect HT-H295R bioactivity. Together, these in silico approaches can efficiently prioritize thousands of untested chemicals for screening to further evaluate their effects on steroid biosynthesis.

Estimating provisional margins of exposure for data-poor chemicals using high-throughput computational methods

Current computational technologies hold promise for prioritizing the testing of the thousands of chemicals in commerce. Here, a case study is presented demonstrating comparative risk-prioritization approaches based on the ratio of surrogate hazard and exposure data, called margins of exposure (MoEs). Exposures were estimated using a U.S. EPA’s ExpoCast predictive model (SEEM3) results and estimates of bioactivity were predicted using: 1) Oral equivalent doses (OEDs) derived from U.S. EPA’s ToxCast high-throughput screening program, together with in vitro to in vivo extrapolation and 2) thresholds of toxicological concern (TTCs) determined using a structure-based decision-tree using the Toxtree open source software. To ground-truth these computational approaches, we compared the MoEs based on predicted noncancer TTC and OED values to those derived using the traditional method of deriving points of departure from no-observed adverse effect levels (NOAELs) from in vivo oral exposures in rodents. TTC-based MoEs were lower than NOAEL-based MoEs for 520 out of 522 (99.6%) compounds in this smaller overlapping dataset, but were relatively well correlated with the same (r² = 0.59). TTC-based MoEs were also lower than OED-based MoEs for 590 (83.2%) of the 709 evaluated chemicals, indicating that TTCs may serve as a conservative surrogate in the absence of chemical-specific experimental data. The TTC-based MoE prioritization process was then applied to over 45,000 curated environmental chemical structures as a proof-of-concept for high-throughput prioritization using TTC-based MoEs. This study demonstrates the utility of exploiting existing computational methods at the pre-assessment phase of a tiered risk-based approach to quickly, and conservatively, prioritize thousands of untested chemicals for further study.

Novel machine learning models to predict endocrine disruption activity for high-throughput chemical screening

An area of ongoing concern in toxicology and chemical risk assessment is endocrine disrupting chemicals (EDCs). However, thousands of legacy chemicals lack the toxicity testing required to assess their respective EDC potential, and this is where computational toxicology can play a crucial role. The US (United States) Environmental Protection Agency (EPA) has run two programs, the Collaborative Estrogen Receptor Activity Project (CERAPP) and the Collaborative Modeling Project for Receptor Activity (CoMPARA) which aim to predict estrogen and androgen activity, respectively. The US EPA solicited research groups from around the world to provide endocrine receptor activity Qualitative (or Quantitative) Structure Activity Relationship ([Q]SAR) models and then combined them to create consensus models for different toxicity endpoints. Random Forest (RF) models were developed to cover a broader range of substances with high predictive capabilities using large datasets from CERAPP and CoMPARA for estrogen and androgen activity, respectively. By utilizing simple descriptors from open-source software and large training datasets, RF models were created to expand the domain of applicability for predicting endocrine disrupting activity and help in the screening and prioritization of extensive chemical inventories. In addition, RFs were trained to conservatively predict the activity, meaning models are more likely to make false-positive predictions to minimize the number of False Negatives. This work presents twelve binary and multi-class RF models to predict binding, agonism, and antagonism for estrogen and androgen receptors. The RF models were found to have high predictive capabilities compared to other in silico modes, with some models reaching balanced accuracies of 93% while having coverage of 89%. These models are intended to be incorporated into evolving priority-setting workflows and integrated strategies to support the screening and selection of chemicals for further testing and assessment by identifying potential endocrine-disrupting substances.

Development, Validation, and Application of a Human Reproductive Toxicity Prediction Model Based on Adverse Outcome Pathway

A growing number of environmental contaminants have been proved to have reproductive toxicity to males and females. However, the unclear toxicological mechanism of reproductive toxicants limits the development of virtual screening methods. By consolidating androgen (AR)-/estrogen receptors (ERs)-mediated adverse outcome pathways (AOPs) with more than 8000 chemical substances, we uncovered relationships between chemical features, a series of pathway-related effects, and reproductive apical outcomes─changes in sex organ weights. An AOP-based computational model named RepTox was developed and evaluated to predict and characterize chemicals' reproductive toxicity for males and females. Results showed that RepTox has three outstanding advantages. (I) Compared with the traditional models (37 and 81% accuracy, respectively), AOP significantly improved the predictive robustness of RepTox (96.3% accuracy). (II) Compared with the application domain (AD) of models based on small in vivo datasets, AOP expanded the ADs of RepTox by 1.65-fold for male and 3.77-fold for female, respectively. (III) RepTox implied that hydrophobicity, cyclopentanol substructure, and several topological indices (e.g., hydrogen-bond acceptors) were important, unbiased features associated with reproductive toxicants. Finally, RepTox was applied to the inventory of existing chemical substances of China and identified 2100 and 7281 potential toxicants to the male and female reproductive systems, respectively.

Ligand-based prediction of hERG-mediated cardiotoxicity based on the integration of different machine learning techniques

Drug-induced cardiotoxicity is a common side effect of drugs in clinical use or under postmarket surveillance and is commonly due to off-target interactions with the cardiac human-ether-a-go-go-related (hERG) potassium channel. Therefore, prioritizing drug candidates based on their hERG blocking potential is a mandatory step in the early preclinical stage of a drug discovery program. Herein, we trained and properly validated 30 ligand-based classifiers of hERG-related cardiotoxicity based on 7,963 curated compounds extracted by the freely accessible repository ChEMBL (version 25). Different machine learning algorithms were tested, namely, random forest, K-nearest neighbors, gradient boosting, extreme gradient boosting, multilayer perceptron, and support vector machine. The application of 1) the best practices for data curation, 2) the feature selection method VSURF, and 3) the synthetic minority oversampling technique (SMOTE) to properly handle the unbalanced data, allowed for the development of highly predictive models (BAMAX = 0.91, AUCMAX = 0.95). Remarkably, the undertaken temporal validation approach not only supported the predictivity of the herein presented classifiers but also suggested their ability to outperform those models commonly used in the literature. From a more methodological point of view, the study put forward a new computational workflow, freely available in the GitHub repository (https://github.com/PDelre93/hERG-QSAR), as valuable for building highly predictive models of hERG-mediated cardiotoxicity.

Profiling prediction of nuclear receptor modulators with multi-task deep learning methods: toward the virtual screening

Nuclear receptors (NRs) are ligand-activated transcription factors, which constitute one of the most important targets for drug discovery. Current computational strategies mainly focus on a single target, and the transfer of learned knowledge among NRs was not considered yet. Herein we proposed a novel computational framework named NR-Profiler for prediction of potential NR modulators with high affinity and specificity. First, we built a comprehensive NR data set including 42 684 interactions to connect 42 NRs and 31 033 compounds. Then, we used multi-task deep neural network and multi-task graph convolutional neural network architectures to construct multi-task multi-classification models. To improve the predictive capability and robustness, we built a consensus model with an area under the receiver operating characteristic curve (AUC) = 0.883. Compared with conventional machine learning and structure-based approaches, the consensus model showed better performance in external validation. Using this consensus model, we demonstrated the practical value of NR-Profiler in virtual screening for NRs. In addition, we designed a selectivity score to quantitatively measure the specificity of NR modulators. Finally, we developed a freely available standalone software for users to make profiling predictions for their compounds of interest. In summary, our NR-Profiler provides a useful tool for NR-profiling prediction and is expected to facilitate NR-based drug discovery.

A framework for establishing scientific confidence in new approach methodologies

Robust and efficient processes are needed to establish scientific confidence in new approach methodologies (NAMs) if they are to be considered for regulatory applications. NAMs need to be fit for purpose, reliable and, for the assessment of human health effects, provide information relevant to human biology. They must also be independently reviewed and transparently communicated. Ideally, NAM developers should communicate with stakeholders such as regulators and industry to identify the question(s), and specified purpose that the NAM is intended to address, and the context in which it will be used. Assessment of the biological relevance of the NAM should focus on its alignment with human biology, mechanistic understanding, and ability to provide information that leads to health protective decisions, rather than solely comparing NAM-based chemical testing results with those from traditional animal test methods. However, when NAM results are compared to historical animal test results, the variability observed within animal test method results should be used to inform performance benchmarks. Building on previous efforts, this paper proposes a framework comprising five essential elements to establish scientific confidence in NAMs for regulatory use: fitness for purpose, human biological relevance, technical characterization, data integrity and transparency, and independent review. Universal uptake of this framework would facilitate the timely development and use of NAMs by the international community. While this paper focuses on NAMs for assessing human health effects of pesticides and industrial chemicals, many of the suggested elements are expected to apply to other types of chemicals and to ecotoxicological effect assessments.

A Population-Based Human In Vitro Approach to Quantify Inter-Individual Variability in Responses to Chemical Mixtures

Human cell-based population-wide in vitro models have been proposed as a strategy to derive chemical-specific estimates of inter-individual variability; however, the utility of this approach has not yet been tested for cumulative exposures in mixtures. This study aimed to test defined mixtures and their individual components and determine whether adverse effects of the mixtures were likely to be more variable in a population than those of the individual chemicals. The in vitro model comprised 146 human lymphoblastoid cell lines from four diverse subpopulations of European and African descent. Cells were exposed, in concentration−response, to 42 chemicals from diverse classes of environmental pollutants; in addition, eight defined mixtures were prepared from these chemicals using several exposure- or hazard-based scenarios. Points of departure for cytotoxicity were derived using Bayesian concentration−response modeling and population variability was quantified in the form of a toxicodynamic variability factor (TDVF). We found that 28 chemicals and all mixtures exhibited concentration−response cytotoxicity, enabling calculation of the TDVF. The median TDVF across test substances, for both individual chemicals or defined mixtures, ranged from a default assumption (101/2) of toxicodynamic variability in human population to >10. The data also provide a proof of principle for single-variant genome-wide association mapping for toxicity of the chemicals and mixtures, although replication would be necessary due to statistical power limitations with the current sample size. This study demonstrates the feasibility of using a set of human lymphoblastoid cell lines as an in vitro model to quantify the extent of inter-individual variability in hazardous properties of both individual chemicals and mixtures. The data show that population variability of the mixtures is unlikely to exceed that of the most variable component, and that similarity in genome-wide associations among components may be used to accrue additional evidence for grouping of constituents in a mixture for cumulative assessments.

Orthogonal assay for validation of Tox21 PPARγ data and applicability to in silico prediction model

High-throughput screening data from the Tox21 database is used for prioritizing hazardous chemicals and building in silico-based toxicity prediction models. One of the Tox21 dataset, peroxisome proliferator-activated receptor-gamma (PPARγ), a nuclear receptor superfamily, identified various endpoints in HEK293 cells. PPARγ mediates various toxic effects when its receptors are activated or inhibited by ligands such as thiazolidinedione and GW9662. In this study, an orthogonal assay was constructed to verify the effectiveness of the Tox21 PPARγ data, and the effect of highly reliable data on in silico model construction was investigated. The orthogonal assay was a reporter gene assay based on the PPARγ ligand binding domain in CV-1 cells. Only 39% of agonists and 55% of antagonists had similar responses in CV-1 and HEK293 cells. Thus, the effectiveness of Tox21 data on PPARγ may vary depending on the cell line. However, in silico PLS-DA analysis with only high-reliability data (i.e., the same response in both cell lines), yielded more accurate prediction of the activity of potential chemical ligands, despite the small number of samples. Thus, obtaining reliable chemical screening data for PPARγ through orthogonal analysis, even for only limited chemicals, supports the construction of highly predictive in silico models with improved screening efficiency.

Are preserved coastal water bodies in Spanish Mediterranean basin impacted by human activity? Water quality evaluation using chemical and biological analyses

The Spanish Mediterranean basin is particularly susceptible to climate change and human activities, making it vulnerable to the influence of anthropogenic contaminants. Therefore, conducting comprehensive and exhaustive water quality assessment in relevant water bodies of this basin is pivotal. In this work, surface water samples from coastal lagoons or estuaries were collected across the Spanish Mediterranean coastline and subjected to target and suspect screening of 1,585 organic micropollutants by liquid chromatography coupled to ion mobility separation and high resolution mass spectrometry. In total, 91 organic micropollutants could be confirmed and 5 were tentatively identified, with pharmaceuticals and pesticides being the most prevalent groups of chemicals. Chemical analysis data was compared with data on bioanalysis of those samples (recurrent aryl hydrocarbon receptor (AhR) activation, and estrogenic receptor (ER) inhibition in wetland samples affected by wastewater streams). The number of identified organic contaminants containing aromatic rings could explain the AhR activation observed. For the ER antagonistic effects, predictions on estrogenic inhibition potency for the detected compounds were used to explain the activities observed. The integration of chemical analysis with bioanalytical observations allowed a comprehensive overview of the quality of the water bodies under study.

Safety assessment of natural products in Malaysia: current practices, challenges, and new strategies

The belief that natural products are inherently safe is a primary reason for consumers to choose traditional medicines and herbal supplements for health maintenance and disease prevention. Unfortunately, some natural products on the market have been found to contain toxic compounds, such as heavy metals and microbes, as well as banned ingredients such as aristolochic acids. It shows that the existing regulatory system is inadequate and highlights the importance of thorough safety evaluations. In Malaysia, the National Pharmaceutical Regulatory Agency is responsible for the regulatory control of medicinal products and cosmetics, including natural products. For registration purpose, the safety of natural products is primarily determined through the review of documents, including monographs, research articles and scientific reports. One of the main factors hampering safety evaluations of natural products is the lack of toxicological data from animal studies. However, international regulatory agencies such as the European Food Safety Authority and the United States Food and Drug Administration are beginning to accept data obtained using alternative strategies such as non-animal predictive toxicological tools. Our paper discusses the use of state-of-the-art techniques, including chemometrics, in silico modelling and omics technologies and their applications to the safety assessments of natural products.

Predicting Prenatal Developmental Toxicity Based On the Combination of Chemical Structures and Biological Data

For hazard identification, classification, and labeling purposes, animal testing guidelines are required by law to evaluate the developmental toxicity potential of new and existing chemical products. However, guideline developmental toxicity studies are costly, time-consuming, and require many laboratory animals. Computational modeling has emerged as a promising, animal-sparing, and cost-effective method for evaluating the developmental toxicity potential of chemicals, such as endocrine disruptors, without the use of animals. We aimed to develop a predictive and explainable computational model for developmental toxicants. To this end, a comprehensive dataset of 1244 chemicals with developmental toxicity classifications was curated from public repositories and literature sources. Data from 2140 toxicological high-throughput screening assays were extracted from PubChem and the ToxCast program for this dataset and combined with information about 834 chemical fragments to group assays based on their chemical-mechanistic relationships. This effort revealed two assay clusters containing 83 and 76 assays, respectively, with high positive predictive rates for developmental toxicants identified with animal testing guidelines (PPV = 72.4 and 77.3% during cross-validation). These two assay clusters can be used as developmental toxicity models and were applied to predict new chemicals for external validation. This study provides a new strategy for constructing alternative chemical developmental toxicity evaluations that can be replicated for other toxicity modeling studies.

Current status and future directions for a neurotoxicity hazard assessment framework that integrates in silico approaches

Neurotoxicology is the study of adverse effects on the structure or function of the developing or mature adult nervous system following exposure to chemical, biological, or physical agents. The development of more informative alternative methods to assess developmental (DNT) and adult (NT) neurotoxicity induced by xenobiotics is critically needed. The use of such alternative methods including in silico approaches that predict DNT or NT from chemical structure (e.g., statistical-based and expert rule-based systems) is ideally based on a comprehensive understanding of the relevant biological mechanisms. This paper discusses known mechanisms alongside the current state of the art in DNT/NT testing. In silico approaches available today that support the assessment of neurotoxicity based on knowledge of chemical structure are reviewed, and a conceptual framework for the integration of in silico methods with experimental information is presented. Establishing this framework is essential for the development of protocols, namely standardized approaches, to ensure that assessments of NT and DNT based on chemical structures are generated in a transparent, consistent, and defendable manner.

Considerations for Improving Metabolism Predictions for In Vitro to In Vivo Extrapolation

High-throughput (HT) in vitro to in vivo extrapolation (IVIVE) is an integral component in new approach method (NAM)-based risk assessment paradigms, for rapidly translating in vitro toxicity assay results into the context of in vivo exposure. When coupled with rapid exposure predictions, HT-IVIVE supports the use of HT in vitro assays for risk-based chemical prioritization. However, the reliability of prioritization based on HT bioactivity data and HT-IVIVE can be limited as the domain of applicability of current HT-IVIVE is generally restricted to intrinsic clearance measured primarily in pharmaceutical compounds. Further, current approaches only consider parent chemical toxicity. These limitations occur because current state-of-the-art HT prediction tools for clearance and metabolite kinetics do not provide reliable data to support HT-IVIVE. This paper discusses current challenges in implementation of IVIVE for prioritization and risk assessment and recommends a path forward for addressing the most pressing needs and expanding the utility of IVIVE.

Predicting the binding of small molecules to nuclear receptors using machine learning

Nuclear receptors (NRs) are important biological targets of endocrine-disrupting chemicals (EDCs). Identifying chemicals that can act as EDCs and modulate the function of NRs is difficult because of the time and cost of in vitro and in vivo screening to determine the potential hazards of the 100 000s of chemicals that humans are exposed to. Hence, there is a need for computational approaches to prioritize chemicals for biological testing. Machine learning (ML) techniques are alternative methods that can quickly screen millions of chemicals and identify those that may be an EDC. Computational models of chemical binding to multiple NRs have begun to emerge. Recently, a Nuclear Receptor Activity (NuRA) dataset, describing experimentally derived small-molecule activity against various NRs has been created. We have used the NuRA dataset to develop an ensemble of ML-based models to predict the agonism, antagonism, binding and effector binding of small molecules to nine different human NRs. We defined the applicability domain of the ML models as a measure of Tanimoto similarity to the molecules in the training set, which enhanced the performance of the developed classifiers. We further developed a user-friendly web server named 'NR-ToxPred' to predict the binding of chemicals to the nine NRs using the best-performing models for each receptor. This web server is freely accessible at http://nr-toxpred.cchem.berkeley.edu. Users can upload individual chemicals using Simplified Molecular-Input Line-Entry System, CAS numbers or sketch the molecule in the provided space to predict the compound's activity against the different NRs and predict the binding mode for each.

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.

Machine learning models on chemical inhibitors of mitochondrial electron transport chain

Chemicals can induce adverse effects in humans by inhibiting mitochondrial electron transport chain (ETC) such as disrupting mitochondrial membrane potential, enhancing oxidative stress and causing some diseases. Thus, identifying ETC inhibitors (ETCi) is important to chemical risk assessment and protecting the public health. However, it is not feasible to identify all ETCi with experimental methods. Quantitative structure-activity relationship (QSAR) modeling is a promising method to rapidly and effectively identify ETCi. In this study, QSAR models for predicting ETCi were developed using machine learning methods. A clustering-based under-sampling (CBUS) method was developed to handle the imbalance issue in training sets. Structure-activity landscapes were generated and analyzed for training sets generated by the CBUS method. The consensus QSAR models constructed with CBUS achieved satisfactory performances (balanced accuracy = 0.852) in 100 iterations of five-fold cross validations, indicating the models can effectively classify ETCi. The classification model was further employed to screen chemicals in the Inventory of Existing Chemical Substances of China and 13 chemicals were identified as ETCi. Fifteen structural alerts for ETCi were identified in this study. These results demonstrated that the model and structural alerts are useful to screen ETCi.

Quality Control for Single Cell Imaging Analytics Using Endocrine Disruptor-Induced Changes in Estrogen Receptor Expression

BACKGROUND

Diverse toxicants and mixtures that affect hormone responsive cells [endocrine disrupting chemicals (EDCs)] are highly pervasive in the environment and are directly linked to human disease. They often target the nuclear receptor family of transcription factors modulating their levels and activity. Many high-throughput assays have been developed to query such toxicants; however, single-cell analysis of EDC effects on endogenous receptors has been missing, in part due to the lack of quality control metrics to reproducibly measure cell-to-cell variability in responses.

OBJECTIVE

We began by developing single-cell imaging and informatic workflows to query whether the single cell distribution of the estrogen receptor-α (ER), used as a model system, can be used to measure effects of EDCs in a sensitive and reproducible manner.

METHODS

We used high-throughput microscopy, coupled with image analytics to measure changes in single cell ER nuclear levels on treatment with ∼100 toxicants, over a large number of biological and technical replicates.

RESULTS

We developed a two-tiered quality control pipeline for single cell analysis and tested it against a large set of biological replicates, and toxicants from the EPA and Agency for Toxic Substances and Disease Registry lists. We also identified a subset of potentially novel EDCs that were active only on the endogenous ER level and activity as measured by single molecule RNA fluorescence in situ hybridization (RNA FISH).

DISCUSSION

We demonstrated that the distribution of ER levels per cell, and the changes upon chemical challenges were remarkably stable features; and importantly, these features could be used for quality control and identification of endocrine disruptor toxicants with high sensitivity. When coupled with orthogonal assays, ER single cell distribution is a valuable resource for high-throughput screening of environmental toxicants. https://doi.org/10.1289/EHP9297.