ToxCast Chemical Landscape: Paving the Road to 21st Century Toxicology

Chemical effects on neural network activity: Comparison of acute versus network formation exposure in microelectrode array assays

New approach methodologies (NAMs) can address information gaps on potential neurotoxicity or developmental neurotoxicity hazard for data-poor chemicals. Two assays have been previously developed using microelectrode arrays (MEA), a technology which measures neural activity. The MEA acute network function assay (AcN) uses dissociated rat cortical cells cultured at postnatal day 0 and evaluates network activity during a 40-minute chemical exposure on day in vitro (DIV)13 or 15. In contrast, the MEA network formation assay (NFA) uses a developmental exposure paradigm spanning DIV0 through DIV12. Measures of network activity over time at DIV5, 7, 9, and 12 in the NFA are reduced to an estimated area under the curve to facilitate concentration-response evaluation. Here, we evaluated the hypothesis that chemicals with effects in the AcN also perturb the NFA by examining quantitative and qualitative concordance between assays. Out of 243 chemicals screened in both assays, we observed 70.3% concordance between the AcN and NFA after eliminating activity inferred to be cytotoxic (selective activity), with the majority of discordance explained by chemicals that altered selective activity in the AcN but not NFA. The NFA detected more active chemicals when evaluating activity associated with cytotoxicity. Median potency values were lower in the NFA compared to the AcN, but within-chemical potency values were not uniformly lower in the NFA than the AcN. Lastly, the AcN and NFA captured unique bioactivity fingerprints; the AcN was more informative for identifying chemicals with a shared mode of action, while the NFA provided information relevant to developmental exposure. Taken together, this analysis provides a rationale for using both approaches for chemical evaluation with consideration of the context of use, such as screening/ prioritization, hazard identification, or to address questions regarding biological mechanism or function.

Interpreting Neural Network Models for Toxicity Prediction by Extracting Learned Chemical Features

Neural network models have become a popular machine-learning technique for the toxicity prediction of chemicals. However, due to their complex structure, it is difficult to understand predictions made by these models which limits confidence. Current techniques to tackle this problem such as SHAP or integrated gradients provide insights by attributing importance to the input features of individual compounds. While these methods have produced promising results in some cases, they do not shed light on how representations of compounds are transformed in hidden layers, which constitute how neural networks learn. We present a novel technique to interpret neural networks which identifies chemical substructures in training data found to be responsible for the activation of hidden neurons. For individual test compounds, the importance of hidden neurons is determined, and the associated substructures are leveraged to explain the model prediction. Using structural alerts for mutagenicity from the Derek Nexus expert system as ground truth, we demonstrate the validity of the approach and show that model explanations are competitive with and complementary to explanations obtained from an established feature attribution method.

Exploring a mechanism-based approach for the identification of endocrine disruptors using Adverse Outcome Pathways (AOPs) and New Approach Methodologies (NAMs) : A perfluorooctane sulfonic acid case study

Endocrine disruptors (EDs) pose a serious threat to human health and the environment and require a comprehensive evaluation to be identified. The identification of EDs require a substantial amount of data, both in vitro and in vivo, due to the current scientific criteria in the EU. At the same time, the EU strives to reduce animal testing due to concerns regarding animal welfare and sensitivity of animal studies to adequately detect adverse effects relevant for human health. Perfluorooctane sulfonic acid (PFOS) is a persistent organic pollutant that is suspected to be an ED based on academic research, however it is not identified as such from a regulatory perspective. It has previously been shown that PFOS has the potential to cause neurotoxicity as well as affect the thyroid system, and it is known that specific thyroid hormone levels are critical in the development of the brain during. In this work, the aim was to evaluate a mechanism-based approach to identify ED properties of PFOS based on the Adverse Outcome Pathway (AOP) framework and using New Approach Methods (NAMs), by comparing this approach to an ED assessment based on the currently available guidance document. An AOP network (AOPN) was generated for the thyroid modality, and AOPs leading to developmental neurotoxicity (DNT) were identified. A literature search and screening process based on the AOPN, and systematic review methodology, was performed, followed by a rigorous Weight-of-Evidence (WoE) assessment. Evidence was mapped back onto the AOPN used for the literature search, to identify possible endocrine Modes-of-Action (MoAs) for PFOS and data gaps in the two assessments. It could be concluded that PFOS fulfils the criteria for ED classification in the standard ED assessment, but not in the mechanism-based assessment. The need for quantitative information, such as quantitative AOPs, for the mechanism-based approach is discussed. The possibility of a directly neurotoxic alternative MoA was also highlighted based on available in vitro data. Opportunities and challenges with implementing AOPs and NAMs into the regulatory assessment of EDs, and assessing hazard in the Next Generation Risk Assessment, is discussed. This case study exploring the mechanism-based approach to ED identification represents an important step toward more accurate and predictive assessment of EDs based on AOPs and NAMs, and to the Next Generation Risk Assessment (NGRA) concept.

Pervasive environmental chemicals impair oligodendrocyte development

Exposure to environmental chemicals can impair neurodevelopment, and oligodendrocytes may be particularly vulnerable, as their development extends from gestation into adulthood. However, few environmental chemicals have been assessed for potential risks to oligodendrocytes. Here, using a high-throughput developmental screen in cultured cells, we identified environmental chemicals in two classes that disrupt oligodendrocyte development through distinct mechanisms. Quaternary compounds, ubiquitous in disinfecting agents and personal care products, were potently and selectively cytotoxic to developing oligodendrocytes, whereas organophosphate flame retardants, commonly found in household items such as furniture and electronics, prematurely arrested oligodendrocyte maturation. Chemicals from each class impaired oligodendrocyte development postnatally in mice and in a human 3D organoid model of prenatal cortical development. Analysis of epidemiological data showed that adverse neurodevelopmental outcomes were associated with childhood exposure to the top organophosphate flame retardant identified by our screen. This work identifies toxicological vulnerabilities for oligodendrocyte development and highlights the need for deeper scrutiny of these compounds' impacts on human health.

Wildlife ecological risk assessment in the 21st century: Promising technologies to assess toxicological effects

Despite advances in toxicity testing and the development of new approach methodologies (NAMs) for hazard assessment, the ecological risk assessment (ERA) framework for terrestrial wildlife (i.e., air-breathing amphibians, reptiles, birds, and mammals) has remained unchanged for decades. While survival, growth, and reproductive endpoints derived from whole-animal toxicity tests are central to hazard assessment, nonstandard measures of biological effects at multiple levels of biological organization (e.g., molecular, cellular, tissue, organ, organism, population, community, ecosystem) have the potential to enhance the relevance of prospective and retrospective wildlife ERAs. Other factors (e.g., indirect effects of contaminants on food supplies and infectious disease processes) are influenced by toxicants at individual, population, and community levels, and need to be factored into chemically based risk assessments to enhance the "eco" component of ERAs. Regulatory and logistical challenges often relegate such nonstandard endpoints and indirect effects to postregistration evaluations of pesticides and industrial chemicals and contaminated site evaluations. While NAMs are being developed, to date, their applications in ERAs focused on wildlife have been limited. No single magic tool or model will address all uncertainties in hazard assessment. Modernizing wildlife ERAs will likely entail combinations of laboratory- and field-derived data at multiple levels of biological organization, knowledge collection solutions (e.g., systematic review, adverse outcome pathway frameworks), and inferential methods that facilitate integrations and risk estimations focused on species, populations, interspecific extrapolations, and ecosystem services modeling, with less dependence on whole-animal data and simple hazard ratios. Integr Environ Assess Manag 2024;20:725-748. © 2023 His Majesty the King in Right of Canada and The Authors. Integrated Environmental Assessment and Management published by Wiley Periodicals LLC on behalf of Society of Environmental Toxicology & Chemistry (SETAC). Reproduced with the permission of the Minister of Environment and Climate Change Canada. This article has been contributed to by US Government employees and their work is in the public domain in the USA.

Health of wild fish exposed to contaminants of emerging concern in freshwater ecosystems utilized by a Minnesota Tribal community

Fish serve as indicators of exposure to contaminants of emerging concern (CECs)-chemicals such as pharmaceuticals, hormones, and personal care products-which are often designed to impact vertebrates. To investigate fish health and CECs in situ, we evaluated the health of wild fish exposed to CECs in waterbodies across northeastern Minnesota with varying anthropogenic pressures and CEC exposures: waterbodies with no human development along their shorelines, those with development, and those directly receiving treated wastewater effluent. Then, we compared three approaches to evaluate the health of fish exposed to CECs in their natural environment: a refined fish health assessment index, a histopathological index, and high-throughput (ToxCast) in vitro assays. Lastly, we mapped adverse outcome pathways (AOPs) associated with identified ToxCast assays to determine potential impacts across levels of biological organization within the aquatic system. These approaches were applied to subsistence fish collected from the Grand Portage Indian Reservation and 1854 Ceded Territory in 2017 and 2019. Overall, 24 CECs were detected in fish tissues, with all but one of the sites having at least one detection. The combined implementation of these tools revealed that subsistence fish exposed to CECs had histological and macroscopic tissue and organ abnormalities, although a direct causal link could not be established. The health of fish in undeveloped sites was as poor, or sometimes poorer, than fish in developed and wastewater effluent-impacted sites based on gross and histologic tissue lesions. Adverse outcome pathways revealed potential hazardous pathways of individual CECs to fish. A better understanding of how the health of wild fish harvested for consumption is affected by CECs may help prioritize risk management research efforts and can ultimately be used to guide fishery management and public health decisions. Integr Environ Assess Manag 2024;20:846-863. © 2023 The Authors. Integrated Environmental Assessment and Management published by Wiley Periodicals LLC on behalf of Society of Environmental Toxicology & Chemistry (SETAC).

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.

Fast and effective molecular property prediction with transferability map

Effective transfer learning for molecular property prediction has shown considerable strength in addressing insufficient labeled molecules. Many existing methods either disregard the quantitative relationship between source and target properties, risking negative transfer, or require intensive training on target tasks. To quantify transferability concerning task-relatedness, we propose Principal Gradient-based Measurement (PGM) for transferring molecular property prediction ability. First, we design an optimization-free scheme to calculate a principal gradient for approximating the direction of model optimization on a molecular property prediction dataset. We have analyzed the close connection between the principal gradient and model optimization through mathematical proof. PGM measures the transferability as the distance between the principal gradient obtained from the source dataset and that derived from the target dataset. Then, we perform PGM on various molecular property prediction datasets to build a quantitative transferability map for source dataset selection. Finally, we evaluate PGM on multiple combinations of transfer learning tasks across 12 benchmark molecular property prediction datasets and demonstrate that it can serve as fast and effective guidance to improve the performance of a target task. This work contributes to more efficient discovery of drugs, materials, and catalysts by offering a task-relatedness quantification prior to transfer learning and understanding the relationship between chemical properties.

Uncertainty Quantification and Interpretability for Clinical Trial Approval Prediction

Background: Clinical trial is a crucial step in the development of a new therapy (e.g., medication) and is remarkably expensive and time-consuming. Forecasting the approval of clinical trials accurately would enable us to circumvent trials destined to fail, thereby allowing us to allocate more resources to therapies with better chances. However, existing approval prediction algorithms did not quantify the uncertainty and provide interpretability, limiting their usage in real-world clinical trial management. Methods: This paper quantifies uncertainty and improves interpretability in clinical trial approval predictions. We devised a selective classification approach and integrated it with the Hierarchical Interaction Network, the state-of-the-art clinical trial prediction model. Selective classification, encompassing a spectrum of methods for uncertainty quantification, empowers the model to withhold decision-making in the face of samples marked by ambiguity or low confidence. This approach not only amplifies the accuracy of predictions for the instances it chooses to classify but also notably enhances the model's interpretability. Results: Comprehensive experiments demonstrate that incorporating uncertainty markedly enhances the model's performance. Specifically, the proposed method achieved 32.37%, 21.43%, and 13.27% relative improvement on area under the precision-recall curve over the base model (Hierarchical Interaction Network) in phase I, II, and III trial approval predictions, respectively. For phase III trials, our method reaches 0.9022 area under the precision-recall curve scores. In addition, we show a case study of interpretability that helps domain experts to understand model's outcome. The code is publicly available at https://github.com/Vincent-1125/Uncertainty-Quantification-on-Clinical-Trial-Outcome-Prediction. Conclusion: Our approach not only measures model uncertainty but also greatly improves interpretability and performance for clinical trial approval prediction.

Pre-training molecular representation model with spatial geometry for property prediction

AI-enhanced bioinformatics and cheminformatics pivots on generating increasingly descriptive and generalized molecular representation. Accurate prediction of molecular properties needs a comprehensive description of molecular geometry. We design a novel Graph Isomorphic Network (GIN) based model integrating a three-level network structure with a dual-level pre-training approach that aligns the characteristics of molecules. In our Spatial Molecular Pre-training (SMPT) Model, the network can learn implicit geometric information in layers from lower to higher according to the dimension. Extensive evaluations against established baseline models validate the enhanced efficacy of SMPT, with notable accomplishments in classification tasks. These results emphasize the importance of spatial geometric information in molecular representation modeling and demonstrate the potential of SMPT as a valuable tool for property prediction.

Mapping drug biology to disease genetics to discover drug impacts on the human phenome

Motivation

Medications can have unexpected effects on disease, including not only harmful drug side effects, but also beneficial drug repurposing. These effects on disease may result from hidden influences of drugs on disease gene networks. Then, discovering how biological effects of drugs relate to disease biology can both provide insight into the mechanism of latent drug effects, and can help predict new effects.

Results

Here, we develop Draphnet, a model that integrates molecular data on 429 drugs and gene associations of nearly 200 common phenotypes to learn a network that explains drug effects on disease in terms of these molecular signals. We present evidence that our method can both predict drug effects, and can provide insight into the biology of unexpected drug effects on disease. Using Draphnet to map a drug's known molecular effects to downstream effects on the disease genome, we put forward disease genes impacted by drugs, and we suggest a new grouping of drugs based on shared effects on the disease genome. Our approach has multiple applications, including predicting drug uses and learning drug biology, with implications for personalized medicine.

Availability and implementation

Code to reproduce the analysis is available at https://github.com/RDMelamed/drug-phenome.

Phthalates and uterine disorders

Humans are ubiquitously exposed to environmental endocrine disrupting chemicals such as phthalates. Phthalates can migrate out of products and enter the human body through ingestion, inhalation, or dermal application, can have potential estrogenic/antiestrogenic and/or androgenic/antiandrogenic activity, and are involved in many diseases. As a female reproductive organ that is regulated by hormones such as estrogen, progesterone and androgen, the uterus can develop several disorders such as leiomyoma, endometriosis and abnormal bleeding. In this review, we summarize the hormone-like activities of phthalates, in vitro studies of endometrial cells exposed to phthalates, epigenetic modifications in the uterus induced by phthalate exposure, and associations between phthalate exposure and uterine disorders such as leiomyoma and endometriosis. Moreover, we also discuss the current research gaps in understanding the relationship between phthalate exposure and uterine disorders.

Indoor Air Sources of Outdoor Air Pollution: Health Consequences, Policy, and Recommendations: An Official American Thoracic Society Workshop Report

Indoor sources of air pollution worsen indoor and outdoor air quality. Thus, identifying and reducing indoor pollutant sources would decrease both indoor and outdoor air pollution, benefit public health, and help address the climate crisis. As outdoor sources come under regulatory control, unregulated indoor sources become a rising percentage of the problem. This American Thoracic Society workshop was convened in 2022 to evaluate this increasing proportion of indoor contributions to outdoor air quality. The workshop was conducted by physicians and scientists, including atmospheric and aerosol scientists, environmental engineers, toxicologists, epidemiologists, regulatory policy experts, and pediatric and adult pulmonologists. Presentations and discussion sessions were centered on 1) the generation and migration of pollutants from indoors to outdoors, 2) the sources and circumstances representing the greatest threat, and 3) effective remedies to reduce the health burden of indoor sources of air pollution. The scope of the workshop was residential and commercial sources of indoor air pollution in the United States. Topics included wood burning, natural gas, cooking, evaporative volatile organic compounds, source apportionment, and regulatory policy. The workshop concluded that indoor sources of air pollution are significant contributors to outdoor air quality and that source control and filtration are the most effective measures to reduce indoor contributions to outdoor air. Interventions should prioritize environmental justice: Households of lower socioeconomic status have higher concentrations of indoor air pollutants from both indoor and outdoor sources. We identify research priorities, potential health benefits, and mitigation actions to consider (e.g., switching from natural gas to electric stoves and transitioning to scent-free consumer products). The workshop committee emphasizes the benefits of combustion-free homes and businesses and recommends economic, legislative, and education strategies aimed at achieving this goal.

Hazard and risk characterization of 56 structurally diverse PFAS using a targeted battery of broad coverage assays using six human cell types

Per- and poly-fluoroalkyl substances (PFAS) are extensively used in commerce leading to their prevalence in the environment. Due to their chemical stability, PFAS are considered to be persistent and bioaccumulative; they are frequently detected in both the environment and humans. Because of this, PFAS as a class (composed of hundreds to thousands of chemicals) are contaminants of very high concern. Little information is available for the vast majority of PFAS, and regulatory agencies lack safety data to determine whether exposure limits or restrictions are needed. Cell-based assays are a pragmatic approach to inform decision-makers on potential health hazards; therefore, we hypothesized that a targeted battery of human in vitro assays can be used to determine whether there are structure-bioactivity relationships for PFAS, and to characterize potential risks by comparing bioactivity (points of departure) to exposure estimates. We tested 56 PFAS from 8 structure-based subclasses in concentration response (0.1-100 μM) using six human cell types selected from target organs with suggested adverse effects of PFAS - human induced pluripotent stem cell (iPSC)-derived hepatocytes, neurons, and cardiomyocytes, primary human hepatocytes, endothelial and HepG2 cells. While many compounds were without effect; certain PFAS demonstrated cell-specific activity highlighting the necessity of using a compendium of in vitro models to identify potential hazards. No class-specific groupings were evident except for some chain length- and structure-related trends. In addition, margins of exposure (MOE) were derived using empirical and predicted exposure data. Conservative MOE calculations showed that most tested PFAS had a MOE in the 1-100 range; ∼20% of PFAS had MOE<1, providing tiered priorities for further studies. Overall, we show that a compendium of human cell-based models can be used to derive bioactivity estimates for a range of PFAS, enabling comparisons with human biomonitoring data. Furthermore, we emphasize that establishing structure-bioactivity relationships may be challenging for the tested PFAS.

ViCEKb: Vitiligo-linked Chemical Exposome Knowledgebase

Vitiligo is a complex disease wherein the environmental factors, in conjunction with the underlying genetic predispositions, trigger the autoimmune destruction of melanocytes, ultimately leading to depigmented patches on the skin. While genetic factors have been extensively studied, the knowledge on environmental triggers remains sparse and less understood. To address this knowledge gap, we present the first comprehensive knowledgebase of vitiligo-triggering chemicals namely, Vitiligo-linked Chemical Exposome Knowledgebase (ViCEKb). ViCEKb involves an extensive and systematic manual effort in curation of published literature and subsequent compilation of 113 unique chemical triggers of vitiligo. ViCEKb standardizes various chemical information, and categorizes the chemicals based on their evidences and sources of exposure. Importantly, ViCEKb contains a wide range of metrics necessary for different toxicological evaluations. Notably, we observed that ViCEKb chemicals are present in a variety of consumer products. For instance, Propyl gallate is present as a fragrance substance in various household products, and Flutamide is used in medication to treat prostate cancer. These two chemicals have the highest level of evidence in ViCEKb, but are not regulated for their skin sensitizing effects. Furthermore, an extensive cheminformatics-based investigation revealed that ViCEKb chemical space is structurally diverse and comprises unique chemical scaffolds in comparison with skin specific regulatory lists. For example, Neomycin and 2,3,5-Triglycidyl-4-aminophenol have unique chemical scaffolds and the highest level of evidence in ViCEKb, but are not regulated for their skin sensitizing effects. Finally, a transcriptomics-based analysis of ViCEKb chemical perturbations in skin cell samples highlighted the commonality in their linked biological processes. Overall, we present the first comprehensive effort in compilation and exploration of various chemical triggers of vitiligo. We believe such a resource will enable in deciphering the complex etiology of vitiligo and aid in the characterization of human chemical exposome. ViCEKb is freely available for academic research at: https://cb.imsc.res.in/vicekb.

Creation of Polymer Datasets with Targeted Backbones for Screening of High-Performance Membranes for Gas Separation

A simple approach was developed to computationally construct a polymer dataset by combining simplified molecular-input line-entry system (SMILES) strings of a targeted polymer backbone and a variety of molecular fragments. This method was used to create 14 polymer datasets by combining seven polymer backbones and molecules from two large molecular datasets (MOSES and QM9). Polymer backbones that were studied include four polydimethylsiloxane (PDMS) based backbones, poly(ethylene oxide) (PEO), poly(allyl glycidyl ether) (PAGE), and polyphosphazene (PPZ). The generated polymer datasets can be used for various cheminformatics tasks, including high-throughput screening for gas permeability and selectivity. This study utilized machine learning (ML) models to screen the polymers for CO2/CH4 and CO2/N2 gas separation using membranes. Several polymers of interest were identified. The results highlight that employing an ML model fitted to polymer selectivities leads to higher accuracy in predicting polymer selectivity compared to using the ratio of predicted permeabilities.

Suspect Screening Analysis of Pooled Human Serum Samples Using GC × GC/TOF-MS

Humans interact with thousands of chemicals. This study aims to identify substances of emerging concern and in need of human health risk evaluations. Sixteen pooled human serum samples were constructed from 25 individual samples each from the National Institute of Environmental Health Sciences' Clinical Research Unit. Samples were analyzed using gas chromatography (GC) × GC/time-of-flight (TOF)-mass spectrometry (MS) in a suspect screening analysis, with follow-up confirmation analysis of 19 substances. A standard reference material blood sample was also analyzed through the confirmation process for comparison. The pools were stratified by sex (female and male) and by age (≤45 and >45). Publicly available information on potential exposure sources was aggregated to annotate presence in serum as either endogenous, food/nutrient, drug, commerce, or contaminant. Of the 544 unique substances tentatively identified by spectral matching, 472 were identified in females, while only 271 were identified in males. Surprisingly, 273 of the identified substances were found only in females. It is known that behavior and near-field environments can drive exposures, and this work demonstrates the existence of exposure sources uniquely relevant to females.

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.

SG-ATT: A Sequence Graph Cross-Attention Representation Architecture for Molecular Property Prediction

Existing formats based on the simplified molecular input line entry system (SMILES) encoding and molecular graph structure are designed to encode the complete semantic and structural information of molecules. However, the physicochemical properties of molecules are complex, and a single encoding of molecular features from SMILES sequences or molecular graph structures cannot adequately represent molecular information. Aiming to address this problem, this study proposes a sequence graph cross-attention (SG-ATT) representation architecture for a molecular property prediction model to efficiently use domain knowledge to enhance molecular graph feature encoding and combine the features of molecular SMILES sequences. The SG-ATT fuses the two-dimensional molecular features so that the current model input molecular information contains molecular structure information and semantic information. The SG-ATT was tested on nine molecular property prediction tasks. Among them, the biggest SG-ATT model performance improvement was 4.5% on the BACE dataset, and the average model performance improvement was 1.83% on the full dataset. Additionally, specific model interpretability studies were conducted to showcase the performance of the SG-ATT model on different datasets. In-depth analysis was provided through case studies of in vitro validation. Finally, network tools for molecular property prediction were developed for the use of researchers.

Exploring the effects of experimental parameters and data modeling approaches on in vitro transcriptomic point-of-departure estimates

Multiple new approach methods (NAMs) are being developed to rapidly screen large numbers of chemicals to aid in hazard evaluation and risk assessments. High-throughput transcriptomics (HTTr) in human cell lines has been proposed as a first-tier screening approach for determining the types of bioactivity a chemical can cause (activation of specific targets vs. generalized cell stress) and for calculating transcriptional points of departure (tPODs) based on changes in gene expression. In the present study, we examine a range of computational methods to calculate tPODs from HTTr data, using six data sets in which MCF7 cells cultured in two different media formulations were treated with a panel of 44 chemicals for 3 different exposure durations (6, 12, 24 hr). The tPOD calculation methods use data at the level of individual genes and gene set signatures, and compare data processed using the ToxCast Pipeline 2 (tcplfit2), BMDExpress and PLIER (Pathway Level Information ExtractoR). Methods were evaluated by comparing to in vitro PODs from a validated set of high-throughput screening (HTS) assays for a set of estrogenic compounds. Key findings include: (1) for a given chemical and set of experimental conditions, tPODs calculated by different methods can vary by several orders of magnitude; (2) tPODs are at least as sensitive to computational methods as to experimental conditions; (3) in comparison to an external reference set of PODs, some methods give generally higher values, principally PLIER and BMDExpress; and (4) the tPODs from HTTr in this one cell type are mostly higher than the overall PODs from a broad battery of targeted in vitro ToxCast assays, reflecting the need to test chemicals in multiple cell types and readout technologies for in vitro hazard screening.

Evidence of transboundary movement of chemicals from Mexico to the U.S. in Tijuana River Estuary sediments

The Tijuana River Estuary (TRE) has been a public health hazard and point of contention between the United States and Mexico for decades, with sources of pollution on both sides of the border. The goal of our study is to determine the presence and dynamics of chemical contamination in the TRE. We sampled sediment from four TRE locations in the U.S. during stable dry conditions and immediately after a wet weather period. Organic chemicals were initially screened with non-targeted analysis using gas chromatography high-resolution mass spectrometry (GC/HRMS) that tentatively identified 6978 chemicals in the NIST 20 database. These tentative identifications were filtered using the USEPA CompTox database to guide quantitative targeted analysis at detection limits below 1 ng/g dry weight sediment. Quantitative targeted analysis of 152 organic pollutants and 18 inorganic elements via GC/HRMS revealed generally higher concentrations of contaminants in dry weather sediments compared to wet weather sediments. The highest concentrations of all chemical classes were detected at the site closest to the U.S.-Mexico border, followed by an urban area near Imperial Beach, California, U.S. All sites exhibited a mixture of petrogenic and pyrogenic polycyclic aromatic hydrocarbons (PAHs). Current-use pesticides were dominated by pyrethroid insecticides and the thiocarbamate herbicide s-Ethyl dipropylthiocarbamate (EPTC), while the U.S.-banned organochlorine pesticides were dominated by chlordanes, dieldrin, and dichlorodiphenyltrichloroethane (DDT) and its degradation byproducts. Polychlorinated biphenyl (PCB) concentrations were greatest at the site closest to the U.S.-Mexico border but in the low nanogram-per-gram range. Phthalates were only found at the same site, with relatively high concentrations of bis(2-ethylhexyl) phthalate. This study provides positive identification and quantitative concentrations for organic pollutants in TRE sediments. Our data suggest that there are multiple sources of chemical contamination in the estuary, including possible transboundary movement of pollutants from Mexico.

Deep learning algorithms applied to computational chemistry

Recently, there has been a significant increase in the use of deep learning techniques in the molecular sciences, which have shown high performance on datasets and the ability to generalize across data. However, no model has achieved perfect performance in solving all problems, and the pros and cons of each approach remain unclear to those new to the field. Therefore, this paper aims to review deep learning algorithms that have been applied to solve molecular challenges in computational chemistry. We proposed a comprehensive categorization that encompasses two primary approaches; conventional deep learning and geometric deep learning models. This classification takes into account the distinct techniques employed by the algorithms within each approach. We present an up-to-date analysis of these algorithms, emphasizing their key features and open issues. This includes details of input descriptors, datasets used, open-source code availability, task solutions, and actual research applications, focusing on general applications rather than specific ones such as drug discovery. Furthermore, our report discusses trends and future directions in molecular algorithm design, including the input descriptors used for each deep learning model, GPU usage, training and forward processing time, model parameters, the most commonly used datasets, libraries, and optimization schemes. This information aids in identifying the most suitable algorithms for a given task. It also serves as a reference for the datasets and input data frequently used for each algorithm technique. In addition, it provides insights into the benefits and open issues of each technique, and supports the development of novel computational chemistry systems.

Cheminformatics Analysis of the Multitarget Structure-Activity Landscape of Environmental Chemicals Binding to Human Endocrine Receptors

In human exposome, environmental chemicals can target and disrupt different endocrine axes, ultimately leading to several endocrine disorders. Such chemicals, termed endocrine disrupting chemicals, can promiscuously bind to different endocrine receptors and lead to varying biological end points. Thus, understanding the complexity of molecule-receptor binding of environmental chemicals can aid in the development of robust toxicity predictors. Toward this, the ToxCast project has generated the largest resource on the chemical-receptor activity data for environmental chemicals that were screened across various endocrine receptors. However, the heterogeneity in the multitarget structure-activity landscape of such chemicals is not yet explored. In this study, we systematically curated the chemicals targeting eight human endocrine receptors, their activity values, and biological end points from the ToxCast chemical library. We employed dual-activity difference and triple-activity difference maps to identify single-, dual-, and triple-target cliffs across different target combinations. We annotated the identified activity cliffs through the matched molecular pair (MMP)-based approach and observed that a small fraction of activity cliffs form MMPs. Further, we structurally classified the activity cliffs and observed that R-group cliffs form the highest fraction among the cliffs identified in various target combinations. Finally, we leveraged the mechanism of action (MOA) annotations to analyze structure-mechanism relationships and identified strong MOA-cliffs and weak MOA-cliffs, for each of the eight endocrine receptors. Overall, insights from this first study analyzing the structure-activity landscape of environmental chemicals targeting multiple human endocrine receptors will likely contribute toward the development of better toxicity prediction models for characterizing the human chemical exposome.

Changes in chemical occurrence, concentration, and bioactivity in the Colorado River before and after replacement of the Moab, Utah wastewater treatment plant

Long-term (2010-19) water-quality monitoring on the Colorado River downstream from Moab Utah indicated the persistent presence of Bioactive Chemicals (BC), such as pesticides and pharmaceuticals. This stream reach near Canyonlands National Park provides critical habitat for federally endangered species. The Moab wastewater treatment plant (WWTP) outfall discharges to the Colorado River and is the nearest potential point-source to this reach. The original WWTP was replaced in 2018. In 2016-19, a study was completed to determine if the new plant reduced BC input to the Colorado River at, and downstream from, the outfall. Water samples were collected before and after the plant replacement at sites upstream and downstream from the outfall. Samples were analyzed for as many as 243 pesticides, 109 pharmaceuticals, 20 hormones, 51 wastewater indicator chemicals, 20 metals, and 8 nutrients. BC concentrations, hazard quotients (HQs), and exposure activity ratios (EARs) were used to identify and prioritize contaminants for their potential to have adverse biological effects on the health of native and endangered wildlife. There were 22 BC with HQs >1, mostly metals and hormones; and 23 BC with EARs >0.1, mostly hormones and pharmaceuticals. Most high HQs or EARs were associated with samples collected at the WWTP outfall site prior to its replacement. Discharge from the new plant had reduced concentrations of nutrients, hormones, pharmaceuticals, and other BC. For example, all 16 of the hormones detected at the WWTP outfall site had maximum concentrations in samples collected prior to the WWTP replacement. The WWTP replacement had less effect on instream concentrations of metals and pesticides, BC whose sources are less directly tied to domestic wastewater. Study results indicate that improved WWTP technology can create substantial reductions in concentrations of non-regulated BC such as pharmaceuticals, in addition to regulated contaminants such as nutrients.

Identification of pararosaniline as a modifier of RNA splicing in Caenorhabditis elegans

Posttranscriptional splicing of premessenger RNA (mRNA) is an evolutionarily conserved eukaryotic process for producing mature mRNA that is translated into proteins. Accurate splicing is necessary for normal growth and development, and aberrant splicing is increasingly evident in various human pathologies. To study environmental factors that influence RNA splicing, we employed a fluorescent Caenorhabditis elegans in vivo splicing reporter as a biomarker for splicing fidelity to screen against the US EPA ToxCast chemical library. We identified pararosaniline hydrochloride as a strong modifier of RNA splicing. Through gene expression analysis, we found that pararosaniline activates the oxidative stress response and alters the expression of key RNA splicing regulator genes. Physiological assays show that pararosaniline is deleterious to C. elegans development, reproduction, and aging. Through a targeted RNAi screen, we found that inhibiting protein translation can reverse pararosaniline's effect on the splicing reporter and provide significant protection against long-term pararosaniline toxicity. Together, this study reveals a new chemical modifier of RNA splicing and describes translation inhibition as a genetic mechanism to provide resistance.

Untargeted Metabolomics Analyses and Contaminant Chemistry of Dreissenid Mussels at the Maumee River Area of Concern in the Great Lakes

Bivalves serve as an ideal ecological indicator; hence, their use by the NOAA Mussel Watch Program to monitor environmental health. This study aimed to expand the baseline knowledge of using metabolic end points in environmental monitoring by investigating the dreissenid mussel metabolome in the field. Dreissenids were caged at four locations along the Maumee River for 30 days. The mussel metabolome was measured using nuclear magnetic resonance spectroscopy, and mussel tissue chemical contaminants were analyzed using gas or liquid chromatography coupled with mass spectrometry. All Maumee River sites had a distinct mussel metabolome compared to the reference site and revealed changes in the energy metabolism and amino acids. Data also highlighted the importance of considering seasonality or handling effects on the metabolome at the time of sampling. The furthest upstream site presented a specific mussel tissue chemical signature of pesticides (atrazine and metolachlor), while a downstream site, located at Toledo's wastewater treatment plant, was characterized by polycyclic aromatic hydrocarbons and other organic contaminants. Further research into the dreissenid mussel's natural metabolic cycle and metabolic response to specific anthropogenic stressors is necessary before successful implementation of metabolomics in a biomonitoring program.

Identification of neurotoxicology (NT)/developmental neurotoxicology (DNT) adverse outcome pathways and key event linkages with in vitro DNT screening assays

There is a need to assess compounds reliably and quickly for neurotoxicity (NT) and developmental neurotoxicity (DNT). Adverse outcome pathways (AOPs) enable the mapping of molecular events to an apical endpoint in a chemical agnostic manner and have begun to be applied in NT and DNT testing frameworks. We assessed the status of NT/DNT AOPs in the AOP-Wiki (ca. 2/1/23; https://aopwiki.org/), to characterize the state of AOP development, identify strengths and knowledge gaps, elucidate areas for improvement, and describe areas for future focus. AOPs in the Wiki database were assessed for inclusion of NT/DNT molecular events and endpoints, AOP development and endorsement, as well as the linkages of key neurodevelopmental processes with in vitro new approach methods (NAMs). This review found that 41 AOPs have been proposed detailing NT/DNT, of which eight were endorsed by working parties in OECD. Further, this review determined that learning and memory is included as an adverse outcome in eight NT/DNT AOPS, often without distinction regarding the varying forms of learning and memory, regional specification, temporal dynamics, or acquisition mechanisms involved. There is also an overlap with key events (KEs) and in vitro NAMs, which synaptogenesis appeared as a common process. Overall, progress on NT/DNT AOPs could be expanded, adding in modes of action that are missing, improvement in defining apical endpoints, as well as utilizing NAMs further to develop AOPs and identify gaps in current knowledge.

Exogenous Chemicals Impact Virus Receptor Gene Transcription: Insights from Deep Learning

Despite the fact that coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has been disrupting human life and health worldwide since the outbreak in late 2019, the impact of exogenous substance exposure on the viral infection remains unclear. It is well-known that, during viral infection, organism receptors play a significant role in mediating the entry of viruses to enter host cells. A major receptor of SARS-CoV-2 is the angiotensin-converting enzyme 2 (ACE2). This study proposes a deep learning model based on the graph convolutional network (GCN) that enables, for the first time, the prediction of exogenous substances that affect the transcriptional expression of the ACE2 gene. It outperforms other machine learning models, achieving an area under receiver operating characteristic curve (AUROC) of 0.712 and 0.703 on the validation and internal test set, respectively. In addition, quantitative polymerase chain reaction (qPCR) experiments provided additional supporting evidence for indoor air pollutants identified by the GCN model. More broadly, the proposed methodology can be applied to predict the effect of environmental chemicals on the gene transcription of other virus receptors as well. In contrast to typical deep learning models that are of black box nature, we further highlight the interpretability of the proposed GCN model and how it facilitates deeper understanding of gene change at the structural level.

ELIXIR and Toxicology: a community in development

Toxicology has been an active research field for many decades, with academic, industrial and government involvement. Modern omics and computational approaches are changing the field, from merely disease-specific observational models into target-specific predictive models. Traditionally, toxicology has strong links with other fields such as biology, chemistry, pharmacology and medicine. With the rise of synthetic and new engineered materials, alongside ongoing prioritisation needs in chemical risk assessment for existing chemicals, early predictive evaluations are becoming of utmost importance to both scientific and regulatory purposes. ELIXIR is an intergovernmental organisation that brings together life science resources from across Europe. To coordinate the linkage of various life science efforts around modern predictive toxicology, the establishment of a new ELIXIR Community is seen as instrumental. In the past few years, joint efforts, building on incidental overlap, have been piloted in the context of ELIXIR. For example, the EU-ToxRisk, diXa, HeCaToS, transQST, and the nanotoxicology community have worked with the ELIXIR TeSS, Bioschemas, and Compute Platforms and activities. In 2018, a core group of interested parties wrote a proposal, outlining a sketch of what this new ELIXIR Toxicology Community would look like. A recent workshop (held September 30th to October 1st, 2020) extended this into an ELIXIR Toxicology roadmap and a shortlist of limited investment-high gain collaborations to give body to this new community. This Whitepaper outlines the results of these efforts and defines our vision of the ELIXIR Toxicology Community and how it complements other ELIXIR activities.

The regulation of endocrine-disrupting chemicals to minimize their impact on health

Endocrine-disrupting chemicals (EDCs) are substances generated by human industrial activities that are detrimental to human health through their effects on the endocrine system. The global societal and economic burden posed by EDCs is substantial. Poorly defined or unenforced policies can increase human exposure to EDCs, thereby contributing to human disease, disability and economic damage. Researchers have shown that policies and interventions implemented at both individual and government levels have the potential to reduce exposure to EDCs. This Review describes a set of evidence-based policy actions to manage, minimize or even eliminate the widespread use of these chemicals and better protect human health and society. A number of specific challenges exist: defining, identifying and prioritizing EDCs; considering the non-linear or non-monotonic properties of EDCs; accounting for EDC exposure effects that are latent and do not appear until later in life; and updating testing paradigms to reflect 'real-world' mixtures of chemicals and cumulative exposure. A sound strategy also requires partnering with health-care providers to integrate strategies to prevent EDC exposure in clinical care. Critical next steps include addressing EDCs within global policy frameworks by integrating EDC exposure prevention into emerging climate policy.

A vision for safer food contact materials: Public health concerns as drivers for improved testing

Food contact materials (FCMs) and food contact articles are ubiquitous in today's globalized food system. Chemicals migrate from FCMs into foodstuffs, so called food contact chemicals (FCCs), but current regulatory requirements do not sufficiently protect public health from hazardous FCCs because only individual substances used to make FCMs are tested and mostly only for genotoxicity while endocrine disruption and other hazard properties are disregarded. Indeed, FCMs are a known source of a wide range of hazardous chemicals, and they likely contribute to highly prevalent non-communicable diseases. FCMs can also include non-intentionally added substances (NIAS), which often are unknown and therefore not subject to risk assessment. To address these important shortcomings, we outline how the safety of FCMs may be improved by (1) testing the overall migrate, including (unknown) NIAS, of finished food contact articles, and (2) expanding toxicological testing beyond genotoxicity to multiple endpoints associated with non-communicable diseases relevant to human health. To identify mechanistic endpoints for testing, we group chronic health outcomes associated with chemical exposure into Six Clusters of Disease (SCOD) and we propose that finished food contact articles should be tested for their impacts on these SCOD. Research should focus on developing robust, relevant, and sensitive in-vitro assays based on mechanistic information linked to the SCOD, e.g., through Adverse Outcome Pathways (AOPs) or Key Characteristics of Toxicants. Implementing this vision will improve prevention of chronic diseases that are associated with hazardous chemical exposures, including from FCMs.

Critical Assessment of the Chemical Space Covered by LC-HRMS Non-Targeted Analysis

Non-targeted analysis (NTA) has emerged as a valuable approach for the comprehensive monitoring of chemicals of emerging concern (CECs) in the exposome. The NTA approach can theoretically identify compounds with diverse physicochemical properties and sources. Even though they are generic and have a wide scope, non-targeted analysis methods have been shown to have limitations in terms of their coverage of the chemical space, as the number of identified chemicals in each sample is very low (e.g., ≤5%). Investigating the chemical space that is covered by each NTA assay is crucial for understanding the limitations and challenges associated with the workflow, from the experimental methods to the data acquisition and data processing techniques. In this review, we examined recent NTA studies published between 2017 and 2023 that employed liquid chromatography-high-resolution mass spectrometry. The parameters used in each study were documented, and the reported chemicals at confidence levels 1 and 2 were retrieved. The chosen experimental setups and the quality of the reporting were critically evaluated and discussed. Our findings reveal that only around 2% of the estimated chemical space was covered by the NTA studies investigated for this review. Little to no trend was found between the experimental setup and the observed coverage due to the generic and wide scope of the NTA studies. The limited coverage of the chemical space by the reviewed NTA studies highlights the necessity for a more comprehensive approach in the experimental and data processing setups in order to enable the exploration of a broader range of chemical space, with the ultimate goal of protecting human and environmental health. Recommendations for further exploring a wider range of the chemical space are given.

Alternatives to Fish Acute Whole Effluent Toxicity (WET) Testing: Predictability of RTgill-W1 Cells and Fathead Minnow Embryos with Actual Wastewater Samples

Toxicity assays using fish cells and embryos continue to gain momentum as a more ethical and informative alternative to fish acute toxicity testing. The goal of our study was to test the accuracy of RTgill-W1 cells and the fathead minnow (Pimephales promelas) embryos to predict actual whole effluent toxicity (WET) in the fathead minnow larvae. The three models were compared concurrently using samples of various origins and treatment types. Additionally, the toxicity of reference toxicants (Cd, Cu, NH3-N, 3,4-dichloraniline, and benzalkonium chloride) spiked into a nontoxic wastewater was compared. The toxicity of reference toxicants was tested in isosmotic and hypoosmotic exposure media in RTgill-W1 cells. Of the 28 wastewater samples, 14 induced a toxic response in fish larvae. Embryos predicted 11 of the 14 wastewater samples toxic to the larvae, whereas RTgill-W1 cells predicted the toxicity of all 14 toxic samples to the larvae. In addition, embryos and RTgill-W1 cells predicted toxicity in two and six additional samples, respectively, that were nontoxic to larvae. Exposures in hypoosmotic medium significantly increased sensitivity of RTgill-W1 cells to all reference toxicants, excluding benzalkonium chloride, compared to exposures in isosmotic medium and showed toxicity levels similar to that in larvae. Thus, hypoosmotic exposure medium should be considered for aquatic toxicity testing applications. Overall, both gill cell and embryo models predicted toxicity in the majority of wastewater samples toxic to larvae and demonstrated their applicability for regulatory WET testing.

Extended study on atomic featurization in graph neural networks for molecular property prediction

Graph neural networks have recently become a standard method for analyzing chemical compounds. In the field of molecular property prediction, the emphasis is now on designing new model architectures, and the importance of atom featurization is oftentimes belittled. When contrasting two graph neural networks, the use of different representations possibly leads to incorrect attribution of the results solely to the network architecture. To better understand this issue, we compare multiple atom representations by evaluating them on the prediction of free energy, solubility, and metabolic stability using graph convolutional networks. We discover that the choice of atom representation has a significant impact on model performance and that the optimal subset of features is task-specific. Additional experiments involving more sophisticated architectures, including graph transformers, support these findings. Moreover, we demonstrate that some commonly used atom features, such as the number of neighbors or the number of hydrogens, can be easily predicted using only information about bonds and atom type, yet their explicit inclusion in the representation has a positive impact on model performance. Finally, we explain the predictions of the best-performing models to better understand how they utilize the available atomic features.

Deep autoencoder-based behavioral pattern recognition outperforms standard statistical methods in high-dimensional zebrafish studies

Zebrafish have become an essential tool in screening for developmental neurotoxic chemicals and their molecular targets. The success of zebrafish as a screening model is partially due to their physical characteristics including their relatively simple nervous system, rapid development, experimental tractability, and genetic diversity combined with technical advantages that allow for the generation of large amounts of high-dimensional behavioral data. These data are complex and require advanced machine learning and statistical techniques to comprehensively analyze and capture spatiotemporal responses. To accomplish this goal, we have trained semi-supervised deep autoencoders using behavior data from unexposed larval zebrafish to extract quintessential "normal" behavior. Following training, our network was evaluated using data from larvae shown to have significant changes in behavior (using a traditional statistical framework) following exposure to toxicants that include nanomaterials, aromatics, per- and polyfluoroalkyl substances (PFAS), and other environmental contaminants. Further, our model identified new chemicals (Perfluoro-n-octadecanoic acid, 8-Chloroperfluorooctylphosphonic acid, and Nonafluoropentanamide) as capable of inducing abnormal behavior at multiple chemical-concentrations pairs not captured using distance moved alone. Leveraging this deep learning model will allow for better characterization of the different exposure-induced behavioral phenotypes, facilitate improved genetic and neurobehavioral analysis in mechanistic determination studies and provide a robust framework for analyzing complex behaviors found in higher-order model systems.

Physiological variables in machine learning QSARs allow for both cross-chemical and cross-species predictions

A major challenge in ecological risk assessment is estimating chemical-induced effects across taxa without species-specific testing. Where ecotoxicological data may be more challenging to gather, information on species physiology is more available for a broad range of taxa. Physiology is known to drive species sensitivity but understanding about the relative contribution of specific underlying processes is still elusive. Consequently, there remains a need to understand which physiological processes lead to differences in species sensitivity. The objective of our study was to utilize existing knowledge about organismal physiology to both understand and predict differences in species sensitivity. Machine learning models were trained to predict chemical- and species-specific endpoints as a function of both chemical fingerprints/descriptors and physiological properties represented by dynamic energy budget (DEB) parameters. We found that random forest models were able to predict chemical- and species-specific endpoints, and that DEB parameters were relatively important in the models, particularly for invertebrates. Our approach illuminates how physiological properties may drive species sensitivity, which will allow more realistic predictions of effects across species without the need for additional animal testing.

Identification and Prioritization of Environmental Organic Pollutants: From an Analytical and Toxicological Perspective

Exposure to environmental organic pollutants has triggered significant ecological impacts and adverse health outcomes, which have been received substantial and increasing attention. The contribution of unidentified chemical components is considered as the most significant knowledge gap in understanding the combined effects of pollutant mixtures. To address this issue, remarkable analytical breakthroughs have recently been made. In this review, the basic principles on recognition of environmental organic pollutants are overviewed. Complementary analytical methodologies (i.e., quantitative structure-activity relationship prediction, mass spectrometric nontarget screening, and effect-directed analysis) and experimental platforms are briefly described. The stages of technique development and/or essential parts of the analytical workflow for each of the methodologies are then reviewed. Finally, plausible technique paths and applications of the future nontarget screening methods, interdisciplinary techniques for achieving toxicant identification, and burgeoning strategies on risk assessment of chemical cocktails are discussed.

Revolutionizing Medicinal Chemistry: The Application of Artificial Intelligence (AI) in Early Drug Discovery

Artificial intelligence (AI) has permeated various sectors, including the pharmaceutical industry and research, where it has been utilized to efficiently identify new chemical entities with desirable properties. The application of AI algorithms to drug discovery presents both remarkable opportunities and challenges. This review article focuses on the transformative role of AI in medicinal chemistry. We delve into the applications of machine learning and deep learning techniques in drug screening and design, discussing their potential to expedite the early drug discovery process. In particular, we provide a comprehensive overview of the use of AI algorithms in predicting protein structures, drug-target interactions, and molecular properties such as drug toxicity. While AI has accelerated the drug discovery process, data quality issues and technological constraints remain challenges. Nonetheless, new relationships and methods have been unveiled, demonstrating AI's expanding potential in predicting and understanding drug interactions and properties. For its full potential to be realized, interdisciplinary collaboration is essential. This review underscores AI's growing influence on the future trajectory of medicinal chemistry and stresses the importance of ongoing synergies between computational and domain experts.

Integrating Concentration-Dependent Toxicity Data and Toxicokinetics To Inform Hepatotoxicity Response Pathways

Failure of animal models to predict hepatotoxicity in humans has created a push to develop biological pathway-based alternatives, such as those that use in vitro assays. Public screening programs (e.g., ToxCast/Tox21 programs) have tested thousands of chemicals using in vitro high-throughput screening (HTS) assays. Developing pathway-based models for simple biological pathways, such as endocrine disruption, has proven successful, but development remains a challenge for complex toxicities like hepatotoxicity, due to the many biological events involved. To this goal, we aimed to develop a computational strategy for developing pathway-based models for complex toxicities. Using a database of 2171 chemicals with human hepatotoxicity classifications, we identified 157 out of 1600+ ToxCast/Tox21 HTS assays to be associated with human hepatotoxicity. Then, a computational framework was used to group these assays by biological target or mechanisms into 52 key event (KE) models of hepatotoxicity. KE model output is a KE score summarizing chemical potency against a hepatotoxicity-relevant biological target or mechanism. Grouping hepatotoxic chemicals based on the chemical structure revealed chemical classes with high KE scores plausibly informing their hepatotoxicity mechanisms. Using KE scores and supervised learning to predict in vivo hepatotoxicity, including toxicokinetic information, improved the predictive performance. This new approach can be a universal computational toxicology strategy for various chemical toxicity evaluations.

Computational Construction of Toxicant Signaling Networks

Humans and animals are regularly exposed to compounds that may have adverse effects on health. The Toxicity Forecaster (ToxCast) program was developed to use high throughput screening assays to quickly screen chemicals by measuring their effects on many biological end points. Many of these assays test for effects on cellular receptors and transcription factors (TFs), under the assumption that a toxicant may perturb normal signaling pathways in the cell. We hypothesized that we could reconstruct the intermediate proteins in these pathways that may be directly or indirectly affected by the toxicant, potentially revealing important physiological processes not yet tested for many chemicals. We integrate data from ToxCast with a human protein interactome to build toxicant signaling networks that contain physical and signaling protein interactions that may be affected as a result of toxicant exposure. To build these networks, we developed the EdgeLinker algorithm, which efficiently finds short paths in the interactome that connect the receptors to TFs for each toxicant. We performed multiple evaluations and found evidence suggesting that these signaling networks capture biologically relevant effects of toxicants. To aid in dissemination and interpretation, interactive visualizations of these networks are available at http://graphspace.org.

Analysis of structure-activity and structure-mechanism relationships among thyroid stimulating hormone receptor binding chemicals by leveraging the ToxCast library

The thyroid stimulating hormone receptor (TSHR) is crucial in thyroid hormone production in humans, and dysregulation in TSHR activation can lead to adverse health effects such as hypothyroidism and Graves' disease. Further, animal studies have shown that binding of endocrine disrupting chemicals (EDCs) with TSHR can lead to developmental toxicity. Hence, several such chemicals have been screened for their adverse physiological effects in human cell lines via high-throughput assays in the ToxCast project. The invaluable data generated by the ToxCast project has enabled the development of toxicity predictors, but they can be limited in their predictive ability due to the heterogeneity in structure-activity relationships among chemicals. Here, we systematically investigated the heterogeneity in structure-activity as well as structure-mechanism relationships among the TSHR binding chemicals from ToxCast. By employing a structure-activity similarity (SAS) map, we identified 79 activity cliffs among 509 chemicals in TSHR agonist dataset and 69 activity cliffs among 650 chemicals in the TSHR antagonist dataset. Further, by using the matched molecular pair (MMP) approach, we find that the resultant activity cliffs (MMP-cliffs) are a subset of activity cliffs identified via the SAS map approach. Subsequently, by leveraging ToxCast mechanism of action (MOA) annotations for chemicals common to both TSHR agonist and TSHR antagonist datasets, we identified 3 chemical pairs as strong MOA-cliffs and 19 chemical pairs as weak MOA-cliffs. In conclusion, the insights from this systematic investigation of the TSHR binding chemicals are likely to inform ongoing efforts towards development of better predictive toxicity models for characterization of the chemical exposome.

Prioritization of risks posed by synthetic chemicals manufactured in China toward humans and the environment via persistence, bioaccumulation, mobility and toxicity properties

Over a third of the global chemical production and sales occurred in China, which make effective assessment and management for chemicals produced by China's chemical industry essential not just for China but for the world. Here, we systematical assessed the persistence (P), bioaccumulation (B), mobility (M) and toxicity (T) potency properties for the chemicals listed in Inventory of Existing Chemical Substances of China (IECSC) via experimental data retrieved from large scale databases and in silico data generated with well-established models. Potential PBT, PMT and PB&MT substances were identified. High risk potentials were highlighted for groups of synthetic intermediates, raw materials, as well as a series of biocides. The potential PBT and PMT synthetic intermediates and/or raw materials unique to the IECSC were dominated with organofluorines, for example, the intermediates used as electronic light-emitting materials. Meanwhile, the biocides unique to the IECSC were mainly organochlorines. Some conventional classes of insecticides, such as organochlorines and pyrethroids, were classified as being of high concern. We further identified a group of PB&MT substances that were considered to be both "bioaccumulative" and "mobile". Their properties and common substructures for several major clusters were characterized. The present results prioritized groups of substances with high potentials to cause adverse effects to the environment and humans, many of which have not yet been fully recognized.

Per- and polyfluoroalkyl substances (PFAS) in United States tapwater: Comparison of underserved private-well and public-supply exposures and associated health implications

Drinking-water quality is a rising concern in the United States (US), emphasizing the need to broadly assess exposures and potential health effects at the point-of-use. Drinking-water exposures to per- and poly-fluoroalkyl substances (PFAS) are a national concern, however, there is limited information on PFAS in residential tapwater at the point-of-use, especially from private-wells. We conducted a national reconnaissance to compare human PFAS exposures in unregulated private-well and regulated public-supply tapwater. Tapwater from 716 locations (269 private-wells; 447 public supply) across the US was collected during 2016-2021 including three locations where temporal sampling was conducted. Concentrations of PFAS were assessed by three laboratories and compared with land-use and potential-source metrics to explore drivers of contamination. The number of individual PFAS observed ranged from 1 to 9 (median: 2) with corresponding cumulative concentrations (sum of detected PFAS) ranging from 0.348 to 346 ng/L. Seventeen PFAS were observed at least once with PFBS, PFHxS and PFOA observed most frequently in approximately 15% of the samples. Across the US, PFAS profiles and estimated median cumulative concentrations were similar among private wells and public-supply tapwater. We estimate that at least one PFAS could be detected in about 45% of US drinking-water samples. These detection probabilities varied spatially with limited temporal variation in concentrations/numbers of PFAS detected. Benchmark screening approaches indicated potential human exposure risk was dominated by PFOA and PFOS, when detected. Potential source and land-use information was related to cumulative PFAS concentrations, and the number of PFAS detected; however, corresponding relations with specific PFAS were limited likely due to low detection frequencies and higher detection limits. Information generated supports the need for further assessments of cumulative health risks of PFAS as a class and in combination with other co-occurring contaminants, particularly in unmonitored private-wells where information is limited or not available.

Enhancing drug property prediction with dual-channel transfer learning based on molecular fragment

BACKGROUND

Accurate prediction of molecular property holds significance in contemporary drug discovery and medical research. Recent advances in AI-driven molecular property prediction have shown promising results. Due to the costly annotation of in vitro and in vivo experiments, transfer learning paradigm has been gaining momentum in extracting general self-supervised information to facilitate neural network learning. However, prior pretraining strategies have overlooked the necessity of explicitly incorporating domain knowledge, especially the molecular fragments, into model design, resulting in the under-exploration of the molecular semantic space.

RESULTS

We propose an effective model with FRagment-based dual-channEL pretraining (FREL). Equipped with molecular fragments, FREL comprehensively employs masked autoencoder and contrastive learning to learn intra- and inter-molecule agreement, respectively. We further conduct extensive experiments on ten public datasets to demonstrate its superiority over state-of-the-art models. Further investigations and interpretations manifest the underlying relationship between molecular representations and molecular properties.

CONCLUSIONS

Our proposed model FREL achieves state-of-the-art performance on the benchmark datasets, emphasizing the importance of incorporating molecular fragments into model design. The expressiveness of learned molecular representations is also investigated by visualization and correlation analysis. Case studies indicate that the learned molecular representations better capture the drug property variation and fragment semantics.

Investigation on the potential adverse outcome pathway of the sensitive endpoint for nephrotoxicity induced by gardenia yellow based on an integrated strategy using bioinformatics analysis and in vitro testing validation

To explore the potential the adverse outcome pathway of Gardenia Yellow (GY)-induced sensitive endpoint for nephrotoxicity, an integrated strategy was applied in the present study. Using bioinformatic analysis, based on the constructed Protein-protein interaction networks, Gene Ontology function and Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis on the core target network were performed to illustrate the potential gene targets and signal pathways. Then, the most enriched pathway was validated with Cell counting kit-8 assays and Western blot analysis in embryonic kidney epithelial 293 cell models. According to the findings, GY may interact with 321 targets related to the endpoint. The five targets on the top ranking in the PPI network were STAT3, SRC, HRAS, AKT1, EP300. Among them, PI3K/Akt was the most enriched pathway. In vitro testing showed that GY exerted a proliferative effect on the cell variability in a dose-dependent manner. GY at concentration of 1000 μg/ml and stimulation for 30 min can significantly enhance the expression of phosphorylated Akt. Thus, after the quantitative weight of evidence evaluation, Akt phosphorylation induced PI3K/Akt activation was speculated as a molecular initiating event leading to a proliferative and inflammatory response in renal tubular epithelial cells.

Augmenting Polymer Datasets by Iterative Rearrangement

One of the biggest obstacles to successful polymer property prediction is an effective representation that accurately captures the sequence of repeat units in a polymer. Motivated by the success of data augmentation in computer vision and natural language processing, we explore augmenting polymer data by iteratively rearranging the molecular representation while preserving the correct connectivity, revealing additional substructural information that is not present in a single representation. We evaluate the effects of this technique on the performance of machine learning models trained on three polymer datasets and compare them to common molecular representations. Data augmentation does not yield significant improvements in machine learning property prediction performance compared to equivalent (non-augmented) representations. In datasets where the target property is primarily influenced by the polymer sequence rather than experimental parameters, this data augmentation technique provides molecular embedding with more information to improve property prediction accuracy.

Exploring the potential of ToxCast™ data for mechanism-based prioritization of chemicals in regulatory context: Case study with priority existing chemicals (PECs) under K-REACH

Recent studies have highlighted the potential of ToxCast™ database to mechanism-based prioritization of chemicals. To explore the applicability of ToxCast data in the context of regulatory inventory chemicals, we screened 510 priority existing chemicals (PECs) regulated under the Act on the Registration and Evaluation of Chemical Substances (K-REACH) using ToxCast bioassays. In our analysis, a hit-call data matrix containing 298984 chemical-gene interactions was computed for 949 bioassays with the intended target genes, which enabled the identification of the putative toxicity mechanisms. Based on the reactivity to the chemicals, we analyzed 412 bioassays whose intended target gene families were cytochrome P450, oxidoreductase, transporter, nuclear receptor, steroid hormone, and DNA-binding. We also identified 141 chemicals based on their reactivity in the bioassays. These chemicals are mainly in consumer products including colorants, preservatives, air fresheners, and detergents. Our analysis revealed that in vitro bioactivities were involved in the relevant mechanisms inducing in vivo toxicity; however, this was not sufficient to predict more hazardous chemicals. Overall, the current results point to a potential and limitation in using ToxCast data for chemical prioritization in regulatory context in the absence of suitable in vivo data.

Integrated data-driven and experimental approaches to accelerate lead optimization targeting SARS-CoV-2 main protease

Identification of potential therapeutic candidates can be expedited by integrating computational modeling with domain aware machine learning (ML) models followed by experimental validation in an iterative manner. Generative deep learning models can generate thousands of new candidates, however, their physiochemical and biochemical properties are typically not fully optimized. Using our recently developed deep learning models and a scaffold as a starting point, we generated tens of thousands of compounds for SARS-CoV-2 M^pro that preserve the core scaffold. We utilized and implemented several computational tools such as structural alert and toxicity analysis, high throughput virtual screening, ML-based 3D quantitative structure-activity relationships, multi-parameter optimization, and graph neural networks on generated candidates to predict biological activity and binding affinity in advance. As a result of these combined computational endeavors, eight promising candidates were singled out and put through experimental testing using Native Mass Spectrometry and FRET-based functional assays. Two of the tested compounds with quinazoline-2-thiol and acetylpiperidine core moieties showed IC[Formula: see text] values in the low micromolar range: [Formula: see text] [Formula: see text]M and 3.41±0.0015 [Formula: see text]M, respectively. Molecular dynamics simulations further highlight that binding of these compounds results in allosteric modulations within the chain B and the interface domains of the M^pro. Our integrated approach provides a platform for data driven lead optimization with rapid characterization and experimental validation in a closed loop that could be applied to other potential protein targets.

A tiered testing strategy based on in vitro phenotypic and transcriptomic data for selecting representative petroleum UVCBs for toxicity evaluation in vivo

Hazard evaluation of substances of "unknown or variable composition, complex reaction products and biological materials" (UVCBs) remains a major challenge in regulatory science because their chemical composition is difficult to ascertain. Petroleum substances are representative UVCBs and human cell-based data have been previously used to substantiate their groupings for regulatory submissions. We hypothesized that a combination of phenotypic and transcriptomic data could be integrated to make decisions as to selection of group-representative worst-case petroleum UVCBs for subsequent toxicity evaluation in vivo. We used data obtained from 141 substances from 16 manufacturing categories previously tested in 6 human cell types (induced pluripotent stem cell [iPSC]-derived hepatocytes, cardiomyocytes, neurons, and endothelial cells, and MCF7 and A375 cell lines). Benchmark doses for gene-substance combinations were calculated, and both transcriptomic and phenotype-derived points of departure (PODs) were obtained. Correlation analysis and machine learning were used to assess associations between phenotypic and transcriptional PODs and to determine the most informative cell types and assays, thus representing a cost-effective integrated testing strategy. We found that 2 cell types-iPSC-derived-hepatocytes and -cardiomyocytes-contributed the most informative and protective PODs and may be used to inform selection of representative petroleum UVCBs for further toxicity evaluation in vivo. Overall, although the use of new approach methodologies to prioritize UVCBs has not been widely adopted, our study proposes a tiered testing strategy based on iPSC-derived hepatocytes and cardiomyocytes to inform selection of representative worst-case petroleum UVCBs from each manufacturing category for further toxicity evaluation in vivo.