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

Pluripotent stem cell assays: Modalities and applications for predictive developmental toxicity

This manuscript provides a review focused on embryonic stem cell-based models and their place within the landscape of alternative developmental toxicity assays. Against the background of the principles of developmental toxicology, the wide diversity of alternative methods using pluripotent stem cells developed in this area over the past half century is reviewed. In order to provide an overview of available models, a systematic scoping review was conducted following a published protocol with inclusion criteria, which were applied to select the assays. Critical aspects including biological domain, readout endpoint, availability of standardized protocols, chemical domain, reproducibility and predictive power of each assay are described in detail, in order to review the applicability and limitations of the platform in general and progress moving forward to implementation. The horizon of innovative routes of promoting regulatory implementation of alternative methods is scanned, and recommendations for further work are given.

A modular mechanistic framework for estimating exposure to SVOCs: Next steps for modeling emission and partitioning of plasticizers and PFAS

Estimates of human exposure to semi-volatile organic compounds (SVOCs) such as phthalates, phthalate alternatives, and some per- and polyfluoroalkyl substances (PFAS) are required for the risk-based evaluation of chemicals. Recently, a modular mechanistic modeling framework to rapidly predict SVOC emission and partitioning in indoor environments has been presented, in which several mechanistically consistent source emission categories (SECs) were identified. However, not all SECs have well-developed emission models. In addition, data on model parameters are missing even for frequently studied SVOCs. These knowledge gaps impede the comprehensive prediction of the fate of SVOCs indoors. In this paper, sets of high-priority phthalates, phthalate alternatives, and PFAS were identified based on chemical occurrence indoors and additional selection criteria. These high-priority chemicals served as the basis for exploring model parameter availability for existing indoor SVOC emission and partitioning models. The results reveal that additional experimental and modeling work is needed to fully understand the behavior of SVOCs indoors and to predict exposures with greater confidence and lower uncertainty. Modeling approaches to fill some of the identified gaps are proposed. The prioritized sets of chemicals and proposed new modeling approaches will help guide future research. The inclusion of polar phases in the framework will further expand its applicability and scope. IMPACT STATEMENT: This paper compiles data on high-priority chemicals commonly found indoors and information on the availability of applicable models and model parameters to predict emission, partitioning, and subsequent exposure to these chemicals. Modeling approaches for a selection of the missing SECs (source emission categories) are proposed, to illustrate the path forward. The comprehensive data set helps inform researchers, exposure assessors, and policy makers to better understand the state of the science regarding modeling of indoor exposure to semi-volatile organic compounds (SVOCs) and per- and polyfluoroalkyl substances (PFAS).

IVIVE: Facilitating the Use of In Vitro Toxicity Data in Risk Assessment and Decision Making

During the past few decades, the science of toxicology has been undergoing a transformation from observational to predictive science. New approach methodologies (NAMs), including in vitro assays, in silico models, read-across, and in vitro to in vivo extrapolation (IVIVE), are being developed to reduce, refine, or replace whole animal testing, encouraging the judicious use of time and resources. Some of these methods have advanced past the exploratory research stage and are beginning to gain acceptance for the risk assessment of chemicals. A review of the recent literature reveals a burst of IVIVE publications over the past decade. In this review, we propose operational definitions for IVIVE, present literature examples for several common toxicity endpoints, and highlight their implications in decision-making processes across various federal agencies, as well as international organizations, including those in the European Union (EU). The current challenges and future needs are also summarized for IVIVE. In addition to refining and reducing the number of animals in traditional toxicity testing protocols and being used for prioritizing chemical testing, the goal to use IVIVE to facilitate the replacement of animal models can be achieved through their continued evolution and development, including a strategic plan to qualify IVIVE methods for regulatory acceptance.

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.

Latent Variables Capture Pathway-Level Points of Departure in High-Throughput Toxicogenomic Data

Estimation of points of departure (PoDs) from high-throughput transcriptomic data (HTTr) represents a key step in the development of next-generation risk assessment (NGRA). Current approaches mainly rely on single key gene targets, which are constrained by the information currently available in the knowledge base and make interpretation challenging as scientists need to interpret PoDs for thousands of genes or hundreds of pathways. In this work, we aimed to address these issues by developing a computational workflow to investigate the pathway concentration-response relationships in a way that is not fully constrained by known biology and also facilitates interpretation. We employed the Pathway-Level Information ExtractoR (PLIER) to identify latent variables (LVs) describing biological activity and then investigated in vitro LVs' concentration-response relationships using the ToxCast pipeline. We applied this methodology to a published transcriptomic concentration-response data set for 44 chemicals in MCF-7 cells and showed that our workflow can capture known biological activity and discriminate between estrogenic and antiestrogenic compounds as well as activity not aligning with the existing knowledge base, which may be relevant in a risk assessment scenario. Moreover, we were able to identify the known estrogen activity in compounds that are not well-established ER agonists/antagonists supporting the use of the workflow in read-across. Next, we transferred its application to chemical compounds tested in HepG2, HepaRG, and MCF-7 cells and showed that PoD estimates are in strong agreement with those estimated using a recently developed Bayesian approach (cor = 0.89) and in weak agreement with those estimated using a well-established approach such as BMDExpress2 (cor = 0.57). These results demonstrate the effectiveness of using PLIER in a concentration-response scenario to investigate pathway activity in a way that is not fully constrained by the knowledge base and to ease the biological interpretation and support the development of an NGRA framework with the ability to improve current risk assessment strategies for chemicals using new approach methodologies.

A Framework that Considers the Impacts of Time, Cost, and Uncertainty in the Determination of the Cost Effectiveness of Toxicity-Testing Methodologies

Regulatory agencies are required to evaluate the impacts of thousands of chemicals. Toxicological tests currently used in such evaluations are time-consuming and resource intensive; however, advances in toxicology and related fields are providing new testing methodologies that reduce the cost and time required for testing. The selection of a preferred methodology is challenging because the new methodologies vary in duration and cost, and the data they generate vary in the level of uncertainty. This article presents a framework for performing cost-effectiveness analyses (CEAs) of toxicity tests that account for cost, duration, and uncertainty. This is achieved by using an output metric-the cost per correct regulatory decision-that reflects the three elements. The framework is demonstrated in two example CEAs, one for a simple decision of risk acceptability and a second, more complex decision, involving the selection of regulatory actions. Each example CEA evaluates five hypothetical toxicity-testing methodologies which differ with respect to cost, time, and uncertainty. The results of the examples indicate that either a fivefold reduction in cost or duration can be a larger driver of the selection of an optimal toxicity-testing methodology than a fivefold reduction in uncertainty. Uncertainty becomes of similar importance to cost and duration when decisionmakers are required to make more complex decisions that require the determination of small differences in risk predictions. The framework presented in this article may provide a useful basis for the identification of cost-effective methods for toxicity testing of large numbers of chemicals.

Beyond AOPs: A Mechanistic Evaluation of NAMs in DART Testing

New Approach Methodologies (NAMs) promise to offer a unique opportunity to enable human-relevant safety decisions to be made without the need for animal testing in the context of exposure-driven Next Generation Risk Assessment (NGRA). Protecting human health against the potential effects a chemical may have on embryo-foetal development and/or aspects of reproductive biology using NGRA is particularly challenging. These are not single endpoint or health effects and risk assessments have traditionally relied on data from Developmental and Reproductive Toxicity (DART) tests in animals. There are numerous Adverse Outcome Pathways (AOPs) that can lead to DART, which means defining and developing strict testing strategies for every AOP, to predict apical outcomes, is neither a tenable goal nor a necessity to ensure NAM-based safety assessments are fit-for-purpose. Instead, a pragmatic approach is needed that uses the available knowledge and data to ensure NAM-based exposure-led safety assessments are sufficiently protective. To this end, the mechanistic and biological coverage of existing NAMs for DART were assessed and gaps to be addressed were identified, allowing the development of an approach that relies on generating data relevant to the overall mechanisms involved in human reproduction and embryo-foetal development. Using the knowledge of cellular processes and signalling pathways underlying the key stages in reproduction and development, we have developed a broad outline of endpoints informative of DART. When the existing NAMs were compared against this outline to determine whether they provide comprehensive coverage when integrated in a framework, we found them to generally cover the reproductive and developmental processes underlying the traditionally evaluated apical endpoint studies. The application of this safety assessment framework is illustrated using an exposure-led case study.

Predicting molecular initiating events using chemical target annotations and gene expression

BACKGROUND

The advent of high-throughput transcriptomic screening technologies has resulted in a wealth of publicly available gene expression data associated with chemical treatments. From a regulatory perspective, data sets that cover a large chemical space and contain reference chemicals offer utility for the prediction of molecular initiating events associated with chemical exposure. Here, we integrate data from a large compendium of transcriptomic responses to chemical exposure with a comprehensive database of chemical-protein associations to train binary classifiers that predict mechanism(s) of action from transcriptomic responses. First, we linked reference chemicals present in the LINCS L1000 gene expression data collection to chemical identifiers in RefChemDB, a database of chemical-protein interactions. Next, we trained binary classifiers on MCF7 human breast cancer cell line derived gene expression profiles and chemical-protein labels using six classification algorithms to identify optimal analysis parameters. To validate classifier accuracy, we used holdout data sets, training-excluded reference chemicals, and empirical significance testing of null models derived from permuted chemical-protein associations. To identify classifiers that have variable predicting performance across training data derived from different cellular contexts, we trained a separate set of binary classifiers on the PC3 human prostate cancer cell line.

RESULTS

We trained classifiers using expression data associated with chemical treatments linked to 51 molecular initiating events. This analysis identified and validated 9 high-performing classifiers with empirical p-values lower than 0.05 and internal accuracies ranging from 0.73 to 0.94 and holdout accuracies of 0.68 to 0.92. High-ranking predictions for training-excluded reference chemicals demonstrating that predictive accuracy extends beyond the set of chemicals used in classifier training. To explore differences in classifier performance as a function of training data cellular context, MCF7-trained classifier accuracies were compared to classifiers trained on the PC3 gene expression data for the same molecular initiating events.

CONCLUSIONS

This methodology can offer insight in prioritizing candidate perturbagens of interest for targeted screens. This approach can also help guide the selection of relevant cellular contexts for screening classes of candidate perturbagens using cell line specific model performance.

A gene expression biomarker for predictive toxicology to identify chemical modulators of NF-κB

The nuclear factor-kappa B (NF-κB) is a transcription factor with important roles in inflammation, immune response, and oncogenesis. Dysregulation of NF-κB signaling is associated with inflammation and certain cancers. We developed a gene expression biomarker predictive of NF-κB modulation and used the biomarker to screen a large compendia of gene expression data. The biomarker consists of 108 genes responsive to tumor necrosis factor α in the absence but not the presence of IκB, an inhibitor of NF-κB. Using a set of 450 profiles from cells treated with immunomodulatory factors with known NF-κB activity, the balanced accuracy for prediction of NF-κB activation was > 90%. The biomarker was used to screen a microarray compendium consisting of 12,061 microarray comparisons from human cells exposed to 2,672 individual chemicals to identify chemicals that could cause toxic effects through NF-κB. There were 215 and 49 chemicals that were identified as putative or known NF-κB activators or suppressors, respectively. NF-κB activators were also identified using two high-throughput screening assays; 165 out of the ~3,800 chemicals (ToxCast assay) and 55 out of ~7,500 unique compounds (Tox21 assay) were identified as potential activators. A set of 32 chemicals not previously associated with NF-κB activation and which partially overlapped between the different screens were selected for validation in wild-type and NFKB1-null HeLa cells. Using RT-qPCR and targeted RNA-Seq, 31 of the 32 chemicals were confirmed to be NF-κB activators. These results comprehensively identify a set of chemicals that could cause toxic effects through NF-κB.

Implementing organ-on-chip in a next-generation risk assessment of chemicals: a review

Organ-on-chip (OoC) technology is full of engineering and biological challenges, but it has the potential to revolutionize the Next-Generation Risk Assessment of novel ingredients for consumer products and chemicals. A successful incorporation of OoC technology into the Next-Generation Risk Assessment toolbox depends on the robustness of the microfluidic devices and the organ tissue models used. Recent advances in standardized device manufacturing, organ tissue cultivation and growth protocols offer the ability to bridge the gaps towards the implementation of organ-on-chip technology. Next-Generation Risk Assessment is an exposure-led and hypothesis-driven tiered approach to risk assessment using detailed human exposure information and the application of appropriate new (non-animal) toxicological testing approaches. Organ-on-chip presents a promising in vitro approach by combining human cell culturing with dynamic microfluidics to improve physiological emulation. Here, we critically review commercial organ-on-chip devices, as well as recent tissue culture model studies of the skin, intestinal barrier and liver as the main metabolic organ to be used on-chip for Next-Generation Risk Assessment. Finally, microfluidically linked tissue combinations such as skin-liver and intestine-liver in organ-on-chip devices are reviewed as they form a relevant aspect for advancing toxicokinetic and toxicodynamic studies. We point to recent achievements and challenges to overcome, to advance non-animal, human-relevant safety studies.

The use of Bayesian methodology in the development and validation of a tiered assessment approach towards prediction of rat acute oral toxicity

There exists consensus that the traditional means by which safety of chemicals is assessed-namely through reliance upon apical outcomes obtained following in vivo testing-is increasingly unfit for purpose. Whilst efforts in development of suitable alternatives continue, few have achieved levels of robustness required for regulatory acceptance. An array of "new approach methodologies" (NAM) for determining toxic effect, spanning in vitro and in silico spheres, have by now emerged. It has been suggested, intuitively, that combining data obtained from across these sources might serve to enhance overall confidence in derived judgment. This concept may be formalised in the "tiered assessment" approach, whereby evidence gathered through a sequential NAM testing strategy is exploited so to infer the properties of a compound of interest. Our intention has been to provide an illustration of how such a scheme might be developed and applied within a practical setting-adopting for this purpose the endpoint of rat acute oral lethality. Bayesian statistical inference is drawn upon to enable quantification of degree of confidence that a substance might ultimately belong to one of five LD50-associated toxicity categories. Informing this is evidence acquired both from existing in silico and in vitro resources, alongside a purposely-constructed random forest model and structural alert set. Results indicate that the combination of in silico methodologies provides moderately conservative estimations of hazard, conducive for application in safety assessment, and for which levels of certainty are defined. Accordingly, scope for potential extension of approach to further toxicological endpoints is demonstrated.

Innovation in regulatory approaches for endocrine disrupting chemicals: The journey to risk assessment modernization in Canada

Globally, regulatory authorities grapple with the challenge of assessing the hazards and risks to human and ecosystem health that may result from exposure to chemicals that disrupt the normal functioning of endocrine systems. Rapidly increasing number of chemicals in commerce, coupled with the reliance on traditional, costly animal experiments for hazard characterization - often with limited sensitivity to many important mechanisms of endocrine disruption -, presents ongoing challenges for chemical regulation. The consequence is a limited number of chemicals for which there is sufficient data to assess if there is endocrine toxicity and hence few chemicals with thorough hazard characterization. To address this challenge, regulatory assessment of endocrine disrupting chemicals (EDCs) is benefiting from a revolution in toxicology that focuses on New Approach Methodologies (NAMs) to more rapidly identify, prioritize, and assess the potential risks from exposure to chemicals using novel, more efficient, and more mechanistically driven methodologies and tools. Incorporated into Integrated Approaches to Testing and Assessment (IATA) and guided by conceptual frameworks such as Adverse Outcome Pathways (AOPs), emerging approaches focus initially on molecular interactions between the test chemical and potentially vulnerable biological systems instead of the need for animal toxicity data. These new toxicity testing methods can be complemented with in silico and computational toxicology approaches, including those that predict chemical kinetics. Coupled with exposure data, these will inform risk-based decision-making approaches. Canada is part of a global network collaborating on building confidence in the use of NAMs for regulatory assessment of EDCs. Herein, we review the current approaches to EDC regulation globally (mainly from the perspective of human health), and provide a perspective on how the advances for regulatory testing and assessment can be applied and discuss the promises and challenges faced in adopting these novel approaches to minimize risks due to EDC exposure in Canada, and our world.

Use of in vitro metabolism and biokinetics assays to refine predicted in vivo and in vitro internal exposure to the cosmetic ingredient, phenoxyethanol, for use in risk assessment

A novel approach was developed to help characterize the biokinetics of the cosmetic ingredient, phenoxyethanol, to help assess the safety of the parent and its major stable metabolite. In the first step of this non-animal tiered approach, primary human hepatocytes were used to confirm or refute in silico predicted metabolites, and elucidate the intrinsic clearance of phenoxyethanol. A key result was the identification of the major metabolite, phenoxyacetic acid (PAA), the exposure to which in the kidney was subsequently predicted to far exceed that of phenoxyethanol in blood or other tissues. Therefore, a novel aspect of this approach was to measure in the subsequent step the formation of PAA in the cells dosed with phenoxyethanol that were used to provide points of departure (PoDs) and express the intracellular exposure as the Cmax and AUC24. This enabled the calculation of the intracellular concentrations of parent and metabolite at the PoD in the cells used to derive this value. These concentrations can be compared with in vivo tissue levels to conclude on the safety margin. The lessons from this case study will help to inform the design of other non-animal safety assessments.

Characterization of the mechanistic linkages between iodothyronine deiodinase inhibition and impaired thyroid-mediated growth and development in Xenopus laevis using iopanoic acid

Iodothyronine deiodinases (DIO) are key enzymes that influence tissue-specific thyroid hormone levels during thyroid-mediated amphibian metamorphosis. Within the larger context of evaluating chemicals for thyroid system disrupting potential, chemical activity toward DIOs is being evaluated using high-throughput in vitro screening assays as part of U.S. EPA's ToxCast program. However, existing data gaps preclude any inferences between in vitro chemical inhibition of DIOs and in vivo outcomes relevant to ecological risk assessment. This study aimed to generate targeted data in a laboratory model species (Xenopus laevis) using a model DIO inhibitor, iopanoic acid (IOP), to characterize linkages between in vitro potency, in vivo biochemical responses, and adverse organismal outcomes. In vitro potency of IOP toward DIOs was evaluated using previously developed in vitro screening assays, which showed concentration-dependent inhibition of human DIO1 (IC50: 97 µM) and DIO2 (IC50: 231 µM) but did not inhibit human or X. laevis DIO3 under the assay conditions. In vivo exposure of larval X. laevis to 0, 2.6, 5.3 and 10.5 µM IOP caused thyroid-related biochemical profiles in the thyroid gland and plasma consistent with hyperthyroxinemia but resulted in delayed metamorphosis and significantly reduced growth in the highest two exposure concentrations. Independent evaluations of dio gene expression ontogeny, together with existing literature, supported interpretation of IOP-mediated effects resulting in a proposed adverse outcome pathway for DIO2 inhibition leading to altered amphibian metamorphosis. This study highlights the types of mechanistic data needed to move toward predicting in vivo outcomes of regulatory concern from in vitro bioactivity data.

Optimization of Human Neural Progenitor Cells for an Imaging-Based High-Throughput Phenotypic Profiling Assay for Developmental Neurotoxicity Screening

Studies in in vivo rodent models have been the accepted approach by regulatory agencies to evaluate potential developmental neurotoxicity (DNT) of chemicals for decades. These studies, however, are inefficient and cannot meet the demand for the thousands of chemicals that need to be assessed for DNT hazard. As such, several in vitro new approach methods (NAMs) have been developed to circumvent limitations of these traditional studies. The DNT NAMs, some of which utilize human-derived cell models, are intended to be employed in a testing battery approach, each focused on a specific neurodevelopmental process. The need for multiple assays, however, to evaluate each process can prolong testing and prioritization of chemicals for more in depth assessments. Therefore, a multi-endpoint higher-throughput approach to assess DNT hazard potential would be of value. Accordingly, we have adapted a high-throughput phenotypic profiling (HTPP) approach for use with human-derived neural progenitor (hNP1) cells. HTPP is a fluorescence-based assay that quantitatively measures alterations in cellular morphology. This approach, however, required optimization of several laboratory procedures prior to chemical screening. First, we had to determine an appropriate cell plating density in 384-well plates. We then had to identify the minimum laminin concentration required for optimal cell growth and attachment. And finally, we had to evaluate whether addition of antibiotics to the culture medium would alter cellular morphology. We selected 6,000 cells/well as an appropriate plating density, 20 µg/ml laminin for optimal cell growth and attachment, and antibiotic addition in the culture medium. After optimizing hNP1 cell culture conditions for HTPP, it was then necessary to select appropriate in-plate assay controls from a reference chemical set. These reference chemicals were previously demonstrated to elicit unique phenotypic profiles in various other cell types. Aphidicolin, bafilomycin A1, berberine chloride, and cucurbitacin I induced robust phenotypic profiles as compared to dimethyl sulfoxide vehicle control in the hNP1 cells, and thus can be employed as in-plate assay controls for subsequent chemical screens. We have optimized HTPP for hNP1 cells, and consequently this approach can now be assessed as a potential NAM for DNT hazard evaluation and results compared to previously developed DNT assays.

The AOP-DB RDF: Applying FAIR Principles to the Semantic Integration of AOP Data Using the Research Description Framework

Computational toxicology is central to the current transformation occurring in toxicology and chemical risk assessment. There is a need for more efficient use of existing data to characterize human toxicological response data for environmental chemicals in the US and Europe. The Adverse Outcome Pathway (AOP) framework helps to organize existing mechanistic information and contributes to what is currently being described as New Approach Methodologies (NAMs). AOP knowledge and data are currently submitted directly by users and stored in the AOP-Wiki (https://aopwiki.org/). Automatic and systematic parsing of AOP-Wiki data is challenging, so we have created the EPA Adverse Outcome Pathway Database. The AOP-DB, developed by the US EPA to assist in the biological and mechanistic characterization of AOP data, provides a broad, systems-level overview of the biological context of AOPs. Here we describe the recent semantic mapping efforts for the AOP-DB, and how this process facilitates the integration of AOP-DB data with other toxicologically relevant datasets through a use case example.

A biologically based model to quantitatively assess the role of the nuclear receptors liver X (LXR), and pregnane X (PXR) on chemically induced hepatic steatosis

Hepatic steatosis is characterized by the intracellular increase of free fatty acids (FFAs) in the form of triglycerides in hepatocytes. This hepatic adverse outcome can be caused by many factors, including exposure to drugs or environmental toxicants. Mechanistically, accumulation of lipids in the liver can take place via several mechanisms such as de novo synthesis and/or uptake of FFAs from serum via high fat content diets. De novo synthesis of FFAs within the liver is mediated by the liver X receptor (LXR), and their uptake into the liver is mediated through the pregnane X receptor (PXR). We investigated the impact of chemical exposure on FFAs hepatic content via activation of LXR and PXR by integrating chemical-specific physiologically based pharmacokinetic (PBPK) models with a quantitative toxicology systems (QTS) model of hepatic lipid homeostasis. Three known agonists of LXR and/or PXR were modeled: T0901317 (antagonist for both receptors), GW3965 (LXR only), and Rifampicin (PXR only). Model predictions showed that T0901317 caused the most FFAs build-up in the liver, followed by Rifampicin and then GW3965. These modeling results highlight the importance of PXR activation for serum FFAs uptake into the liver while suggesting that increased hepatic FAAs de novo synthesis alone may not be enough to cause appreciable accumulation of lipids in the liver under normal environmental exposure levels. Moreover, the overall PBPK-hepatic lipids quantitative model can be used to screen chemicals for their potential to cause in vivo hepatic lipid content buildup in view of their in vitro potential to activate the nuclear receptors and their exposure levels.

Exploration of the DARTable Genome- a Resource Enabling Data-Driven NAMs for Developmental and Reproductive Toxicity Prediction

The identification of developmental and reproductive toxicity (DART) is a critical component of toxicological evaluations of chemical safety. Adverse Outcome Pathways (AOPs) provide a framework to describe biological processes leading to a toxic effect and can provide insights in understanding the mechanisms underlying toxicological endpoints and aid the development of new approach methods (NAMs). Integrated approaches to testing and assessment (IATA) can be developed based on AOP knowledge and can serve as pragmatic approaches to chemical hazard characterization using NAMs. However, DART effects remain difficult to predict given the diversity of biological mechanisms operating during ontogenesis and consequently, the considerable number of potential molecular initiating events (MIEs) that might trigger a DART Adverse Outcome (DART AO). Consequently, two challenges that need to be overcome to create an AOP-based DART IATA are having sufficient knowledge of relevant biology and using this knowledge to determine the appropriate selection of cell systems that provide sufficient coverage of that biology. The wealth of modern biological and bioinformatics data can be used to provide this knowledge. Here we demonstrate the utility of bioinformatics analyses to address these questions. We integrated known DART MIEs with gene-developmental phenotype information to curate the hypothetical human DARTable genome (HDG, ∼5 k genes) which represents the comprehensive set of biomarkers for DART. Using network analysis of the human interactome, we show that HDG genes have distinct connectivity compared to other genes. HDG genes have higher node degree with lower neighborhood connectivity, betweenness centralities and average shortest path length. Therefore, HDG is highly connected to itself and to the wider network and not only to their local community. Also, by comparison with the Druggable Genome we show how the HDG can be prioritized to identify potential MIEs based on potential to interact with small molecules. We demonstrate how the HDG in combination with gene expression data can be used to select a panel of relevant cell lines (RD-1, OVCAR-3) for inclusion in an IATA and conclude that bioinformatic analyses can provide the necessary insights and serve as a resource for the development of a screening panel for a DART IATA.

Biological system considerations for application of toxicogenomics in next-generation risk assessment and predictive toxicology

There is increasing interest in using modern 'omics technologies, such as whole transcriptome sequencing, to inform decisions about human health safety and chemical toxicity hazard. High throughput methodologies using in vitro assays offer a path forward in reducing or eliminating animal testing. However, many aspects of these technologies need assessment before they will gain the trust of regulators and the public as viable alternative test methods for human health and safety. We used a high throughput whole transcriptome sequence assay (TempO-Seq) to assess the use of three widely used cancer cell lines (HepG2, MCF7, and Ishikawa cells) as in vitro systems for determination of cellular modes of action for two well studied compounds with canonical liver responses: ketoconazole and phenobarbital. We evaluated transcriptomic data to infer points of departure for use in risk analyses of compounds. Both compounds displayed shortcomings in evidence for canonical liver-related responses in any cell line, despite a strong dose response in all three. This raises questions about the competence of simple, mono-cultured cancer cell lines as appropriate surrogates for some adverse effects or toxic endpoints. Points of departure derived from benchmark doses were highly consistent across all three cell lines however, indicating the use of transcriptomic BMD analyses for such purposes would be a reliable and consistent approach.

Use of Physiologically-Based Kinetics Modelling to Reliably Predict Internal Concentrations of the UV Filter, Homosalate, After Repeated Oral and Topical Application

Ethical and legal considerations have led to increased use of non-animal methods to evaluate the safety of chemicals for human use. We describe the development and qualification of a physiologically-based kinetics (PBK) model for the cosmetic UV filter ingredient, homosalate, to support its safety without the need of generating further animal data. The intravenous (IV) rat PBK model, using PK-Sim®, was developed and validated using legacy in vivo data generated prior to the 2013 EU animal-testing ban. Input data included literature or predicted physicochemical and pharmacokinetic properties. The refined IV rat PBK model was subject to sensitivity analysis to identify homosalate-specific sensitive parameters impacting the prediction of Cmax (more sensitive than AUC(0-∞)). These were then considered, together with population modeling, to calculate the confidence interval (CI) 95% Cmax and AUC(0-∞). Final model parameters were established by visual inspection of the simulations and biological plausibility. The IV rat model was extrapolated to oral administration, and used to estimate internal exposures to doses tested in an oral repeated dose toxicity study. Next, a human PBK dermal model was developed using measured human in vitro ADME data and a module to represent the dermal route. Model performance was confirmed by comparing predicted and measured values from a US-FDA clinical trial (Identifier: NCT03582215, https://clinicaltrials.gov/). Final exposure estimations were obtained in a virtual population and considering the in vitro and input parameter uncertainty. This model was then used to estimate the Cmax and AUC(0-24 h) of homosalate according to consumer use in a sunscreen. The developed rat and human PBK models had a good biological basis and reproduced in vivo legacy rat and human clinical kinetics data. They also complied with the most recent WHO and OECD recommendations for assessing the confidence level. In conclusion, we have developed a PBK model which predicted reasonably well the internal exposure of homosalate according to different exposure scenarios with a medium to high level of confidence. In the absence of in vivo data, such human PBK models will be the heart of future completely non-animal risk assessments; therefore, valid approaches will be key in gaining their regulatory acceptance. Clinical Trial Registration: https://clinicaltrials.gov/, identifier, NCT03582215.

A 10-step framework for use of read-across (RAX) in next generation risk assessment (NGRA) for cosmetics safety assessment

This paper presents a 10-step read-across (RAX) framework for use in cases where a threshold of toxicological concern (TTC) approach to cosmetics safety assessment is not possible. RAX builds on established approaches that have existed for more than two decades using chemical properties and in silico toxicology predictions, by further substantiating hypotheses on toxicological similarity of substances, and integrating new approach methodologies (NAM) in the biological and kinetic domains. NAM include new types of data on biological observations from, for example, in vitro assays, toxicogenomics, metabolomics, receptor binding screens and uses physiologically-based kinetic (PBK) modelling to inform about systemic exposure. NAM data can help to substantiate a mode/mechanism of action (MoA), and if similar chemicals can be shown to work by a similar MoA, a next generation risk assessment (NGRA) may be performed with acceptable confidence for a data-poor target substance with no or inadequate safety data, based on RAX approaches using data-rich analogue(s), and taking account of potency or kinetic/dynamic differences.

Quantitative non-targeted analysis: Bridging the gap between contaminant discovery and risk characterization

Chemical risk assessments follow a long-standing paradigm that integrates hazard, dose-response, and exposure information to facilitate quantitative risk characterization. Targeted analytical measurement data directly support risk assessment activities, as well as downstream risk management and compliance monitoring efforts. Yet, targeted methods have struggled to keep pace with the demands for data regarding the vast, and growing, number of known chemicals. Many contemporary monitoring studies therefore utilize non-targeted analysis (NTA) methods to screen for known chemicals with limited risk information. Qualitative NTA data has enabled identification of previously unknown compounds and characterization of data-poor compounds in support of hazard identification and exposure assessment efforts. In spite of this, NTA data have seen limited use in risk-based decision making due to uncertainties surrounding their quantitative interpretation. Significant efforts have been made in recent years to bridge this quantitative gap. Based on these advancements, quantitative NTA data, when coupled with other high-throughput data streams and predictive models, are poised to directly support 21st-century risk-based decisions. This article highlights components of the chemical risk assessment process that are influenced by NTA data, surveys the existing literature for approaches to derive quantitative estimates of chemicals from NTA measurements, and presents a conceptual framework for incorporating NTA data into contemporary risk assessment frameworks.

Implementing in vitro bioactivity data to modernize priority setting of chemical inventories

Internationally, there are thousands of existing and newly introduced chemicals in commerce, highlighting the ongoing importance of innovative approaches to identify emerging chemicals of concern. For many chemicals, there is a paucity of hazard and exposure data. Thus, there is a crucial need for efficient and robust approaches to address data gaps and support risk-based prioritization. Several studies have demonstrated the utility of in vitro bioactivity data from the ToxCast program in deriving points of departure (PODs). ToxCast contains data for nearly 1,400 endpoints per chemical, and the bioactivity concentrations, indicative of potential adverse outcomes, can be converted to human-equivalent PODs using high-throughput toxicokinetics (HTTK) modeling. However, data gaps need to be addressed for broader application: the limited chemical space of HTTK and quantitative high-throughput screening data. Here we explore the applicability of in silico models to address these data needs. Specifically, we used ADMET predictor for HTTK predictions and a generalized read-across approach to predict ToxCast bioactivity potency. We applied these models to profile 5,801 chemicals on Canada’s Domestic Substances List (DSL). To evaluate the approach’s performance, bioactivity PODs were compared with in vivo results from the EPA Toxicity Values database for 1,042 DSL chemicals. Comparisons demonstrated that the bioac­tivity PODs, based on ToxCast data or read-across, were conservative for 95% of the chemicals. Comparing bioactivity PODs to human exposure estimates supports the identification of chemicals of potential interest for further work. The bioac­tivity workflow shows promise as a powerful screening tool to support effective triaging of chemical inventories.

OUP accepted manuscript

In vivo developmental neurotoxicity (DNT) testing is resource intensive and lacks information on cellular processes affected by chemicals. To address this, DNT new approach methodologies (NAMs) are being evaluated, including: the microelectrode array neuronal network formation assay; and high-content imaging to evaluate proliferation, apoptosis, neurite outgrowth, and synaptogenesis. This work addresses 3 hypotheses: (1) a broad screening battery provides a sensitive marker of DNT bioactivity; (2) selective bioactivity (occurring at noncytotoxic concentrations) may indicate functional processes disrupted; and, (3) a subset of endpoints may optimally classify chemicals with in vivo evidence for DNT. The dataset was comprised of 92 chemicals screened in all 57 assay endpoints sourced from publicly available data, including a set of DNT NAM evaluation chemicals with putative positives (53) and negatives (13). The DNT NAM battery provides a sensitive marker of DNT bioactivity, particularly in cytotoxicity and network connectivity parameters. Hierarchical clustering suggested potency (including cytotoxicity) was important for classifying positive chemicals with high sensitivity (93%) but failed to distinguish patterns of disrupted functional processes. In contrast, clustering of selective values revealed informative patterns of differential activity but demonstrated lower sensitivity (74%). The false negatives were associated with several limitations, such as the maximal concentration tested or gaps in the biology captured by the current battery. This work demonstrates that this multi-dimensional assay suite provides a sensitive biomarker for DNT bioactivity, with selective activity providing possible insight into specific functional processes affected by chemical exposure and a basis for further research.

The effect of noise on the predictive limit of QSAR models

A key challenge in the field of Quantitative Structure Activity Relationships (QSAR) is how to effectively treat experimental error in the training and evaluation of computational models. It is often assumed in the field of QSAR that models cannot produce predictions which are more accurate than their training data. Additionally, it is implicitly assumed, by necessity, that data points in test sets or validation sets do not contain error, and that each data point is a population mean. This work proposes the hypothesis that QSAR models can make predictions which are more accurate than their training data and that the error-free test set assumption leads to a significant misevaluation of model performance. This work used 8 datasets with six different common QSAR endpoints, because different endpoints should have different amounts of experimental error associated with varying complexity of the measurements. Up to 15 levels of simulated Gaussian distributed random error was added to the datasets, and models were built on the error laden datasets using five different algorithms. The models were trained on the error laden data, evaluated on error-laden test sets, and evaluated on error-free test sets. The results show that for each level of added error, the RMSE for evaluation on the error free test sets was always better. The results support the hypothesis that, at least under the conditions of Gaussian distributed random error, QSAR models can make predictions which are more accurate than their training data, and that the evaluation of models on error laden test and validation sets may give a flawed measure of model performance. These results have implications for how QSAR models are evaluated, especially for disciplines where experimental error is very large, such as in computational toxicology.

Establishment of a 13 genes-based molecular prediction score model to discriminate the neurotoxic potential of food relevant-chemicals

Although many neurotoxicity prediction studies of food additives have been developed, they are applicable in a qualitative way. We aimed to develop a novel prediction score that is described quantitatively and precisely. We examined cell viability, reactive oxygen species activity, intracellular calcium and RNA transcription level of potential prediction related genes to develop a high-throughput neurotoxicity test method in vitro to screen the neurotoxicity of hazardous factors in food using AI-based machine learning. We trained artificial intelligence models (random forest and neural network) to predict neurotoxicity precisely, establishing a universal classification assessment score (CA-Score) that relies on the expression status of only 13 of prediction related genes. The CA-Score system is almost universally applicable to food risk factors (p<0.05) in a manner independent of platform (microarray or RNA sequencing) by being compared with cut-off value 23.487 to judge whether it's neurotoxic or not. We finally validated our prediction with the external validation of CA-Score on neural precursor cells derived from embryonic stem cells. Therefore, we draw a conclusion that the AI-based machine learning including neural network and random forest is likely to provide a useful tool for large-scale screening of neurotoxicity in food risk factors.

Evaluating adaptive stress response gene signatures using transcriptomics

Stress response pathways (SRPs) mitigate the cellular effects of chemicals, but excessive perturbation can lead to adverse outcomes. Here, we investigated a computational approach to evaluate SRP activity from transcriptomic data using gene set enrichment analysis (GSEA). We extracted published gene signatures for DNA damage response (DDR), unfolded protein response (UPR), heat shock response (HSR), response to hypoxia (HPX), metal-associated response (MTL), and oxidative stress response (OSR) from the Molecular Signatures Database (MSigDB). Next, we used a gene-frequency approach to build consensus SRP signatures of varying lengths from 50 to 477 genes. We then prepared a reference dataset from perturbagens associated with SRPs from the literature with their transcriptomic profiles retrieved from public repositories. Lastly, we used receiver-operator characteristic analysis to evaluate the GSEA scores from matching transcriptomic reference profiles to SRP signatures. Our consensus signatures performed better than or as well as published signatures for 4 out of the 6 SRPs, with the best consensus signature area under the curve (% performance relative to median of published signatures) of 1.00 for DDR (109%), 0.86 for UPR (169%), 0.99 for HTS (103%), 1.00 for HPX (104%), 0.74 for MTL (150%) and 0.83 for OSR (148%). The best matches between transcriptomic profiles and SRP signatures correctly classified perturbagens in 78% and 88% of the cases by first and second rank, respectively. We believe this approach can characterize SRP activity for new chemicals using transcriptomics with further evaluation.

Reproducibility and robustness of high-throughput S1500+ transcriptomics on primary rat hepatocytes for chemical-induced hepatotoxicity assessment

•

TempO-Seq assays of rat hepatocytes in collagen sandwich are highly reproducible.

•

Gene expression analysis shows S1500+ is representative of the whole transcriptome.

•

Connectivity mapping shows consistency between TempO-Seq and Affymetrix data.

•

Gene set enrichment shows consistency between S1500+ and the whole transcriptome.

•

Gene set enrichment using hallmark gene sets informs hepatotoxicity.

Cell-based in vitro models coupled with high-throughput transcriptomics (HTTr) are increasingly utilized as alternative methods to animal-based toxicity testing. Here, using a panel of 14 chemicals with different risks of human drug-induced liver injury (DILI) and two dosing concentrations, we evaluated an HTTr platform comprised of collagen sandwich primary rat hepatocyte culture and the TempO-Seq surrogate S1500+ (ST) assay. First, the HTTr platform was found to exhibit high reproducibility between technical and biological replicates (r greater than 0.85). Connectivity mapping analysis further demonstrated a high level of inter-platform reproducibility between TempO-Seq data and Affymetrix GeneChip data from the Open TG-GATES project. Second, the TempO-Seq ST assay was shown to be a robust surrogate to the whole transcriptome (WT) assay in capturing chemical-induced changes in gene expression, as evident from correlation analysis, PCA and unsupervised hierarchical clustering. Gene set enrichment analysis (GSEA) using the Hallmark gene set collection also demonstrated consistency in enrichment scores between ST and WT assays. Lastly, unsupervised hierarchical clustering of hallmark enrichment scores broadly divided the samples into hepatotoxic, intermediate, and non-hepatotoxic groups. Xenobiotic metabolism, bile acid metabolism, apoptosis, p53 pathway, and coagulation were found to be the key hallmarks driving the clustering. Taken together, our results established the reproducibility and performance of collagen sandwich culture in combination with TempO-Seq S1500+ assay, and demonstrated the utility of GSEA using the hallmark gene set collection to identify potential hepatotoxicants for further validation.

Sourcing data on chemical properties and hazard data from the US-EPA CompTox Chemicals Dashboard: A practical guide for human risk assessment

For the past six decades, human health risk assessment of chemicals has relied on in vivo data from human epidemiological and experimental animal toxicological studies to inform the derivation of non-cancer toxicity values. The ongoing evolution of this risk assessment paradigm in an environmental landscape of data-poor chemicals has highlighted the need to develop and implement non-testing methods, so-called New Approach Methodologies (NAMs). NAMs include a growing number of in silico and in vitro data streams designed to inform hazard properties of chemicals, including kinetics and dynamics at different levels of biological organization, environmental fate and transport, and exposure. NAMs provide a fit-for-purpose science-basis for human hazard and risk characterization of chemicals ranging from data-gap filling applications to broad evidence-based decision-making. Systematic assembly and delivery of empirical and predicted data for chemicals are paramount to advancing chemical evaluation, and software tools serve an essential role in delivering these data to the scientific community. The CompTox Chemicals Dashboard (from here on referred to as the "Dashboard") is one such tool and is a publicly available web-based application developed by the US Environmental Protection Agency to provide access to chemistry, toxicity and exposure information for ~900,000 chemicals. The Dashboard is increasingly becoming a valuable resource for assessors tasked with the evaluation of potential human health risks associated with chemical exposures. In this context, the significant amount of information present in the Dashboard facilitates: 1) assembly of information on physicochemical properties and environmental fate and transport and exposure parameters and metrics; 2) identification of cancer and non-cancer health effects from extant human and experimental animal studies in the public domain and/or information not available in the public domain (i.e., "grey literature"); 3) systematic literature searching and review for developing cancer and non-cancer hazard evidence bases; and 4) access to mechanistic information that can aid or augment the analysis of traditional toxicology evidence bases, or potentially, serve as the primary basis for informing hazard identification and dose-response when traditional bioassay data are lacking. Finally, in silico predictive tools developed to conduct structure-activity or read-across analyses are also available within the Dashboard. This practical tutorial is intended to address key questions from the human health risk assessment community dealing with chemicals in both food and in the environment. Perspectives for future development or refinement of the Dashboard highlight foreseen activities to further support the research and risk assessment community in cancer and non-cancer chemical evaluations.

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

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

Paving the way for application of next generation risk assessment to safety decision-making for cosmetic ingredients

Next generation risk assessment (NGRA) is an exposure-led, hypothesis-driven approach that has the potential to support animal-free safety decision-making. However, significant effort is needed to develop and test the in vitro and in silico (computational) approaches that underpin NGRA to enable confident application in a regulatory context. A workshop was held in Montreal in 2019 to discuss where effort needs to be focussed and to agree on the steps needed to ensure safety decisions made on cosmetic ingredients are robust and protective. Workshop participants explored whether NGRA for cosmetic ingredients can be protective of human health, and reviewed examples of NGRA for cosmetic ingredients. From the limited examples available, it is clear that NGRA is still in its infancy, and further case studies are needed to determine whether safety decisions are sufficiently protective and not overly conservative. Seven areas were identified to help progress application of NGRA, including further investments in case studies that elaborate on scenarios frequently encountered by industry and regulators, including those where a ‘high risk’ conclusion would be expected. These will provide confidence that the tools and approaches can reliably discern differing levels of risk. Furthermore, frameworks to guide performance and reporting should be developed.

Screening for androgen agonists using autonomously bioluminescent HEK293 reporter cells

Due to the public health concerns of endocrine-disrupting chemicals, there is an increasing demand to develop improved high-throughput detection assays for enhanced exposure control and risk assessment. A substrate-free, autobioluminescent HEK293_ARE/Gal4-Lux assay was developed to screen compounds for their ability to induce androgen receptor (AR)-mediated transcriptional activation. The assay was validated against a group of 40 recommended chemicals and achieved an overall 87.5% accuracy in qualitatively classifying positive and negative AR agonists. The HEK293_ARE/Gal4-Lux assay was demonstrated as a suitable tool for Tier 1 AR agonist screening. By eliminating exogenous substrate, this assay provided a significant advantage over traditional reporter assays by enabling higher-throughput screening with reduced testing costs while maintaining detection accuracy.

A human optimized version of the bacterial luciferase gene cassette was developed such that bioluminescence is controlled by exposure to androgen-disruptor chemicals. This cassette, along with the androgen receptor gene, was co-transfected into an HEK293 human cell host that naturally lacks hormone receptors. The resulting reporter cell line was used to screen compounds for androgenic activity in a low cost, high throughput format.

Repeat-dose toxicity prediction with Generalized Read-Across (GenRA) using targeted transcriptomic data: A proof-of-concept case study

Read-across is a data gap filling technique utilized to predict the toxicity of a target chemical using data from similar analogues. Recent efforts such as Generalized Read-Across (GenRA) facilitate automated read-across predictions for untested chemicals. GenRA makes predictions of toxicity outcomes based on "neighboring" chemicals characterized by chemical and bioactivity fingerprints. Here we investigated the impact of biological similarities on neighborhood formation and read-across performance in predicting hazard (based on repeat-dose testing outcomes from US EPA ToxRefDB v2.0). We used targeted transcriptomic data on 93 genes for 1060 chemicals in HepaRG^™ cells that measure nuclear receptor activation, xenobiotic metabolism, cellular stress, cell cycle progression, and apoptosis. Transcriptomic similarity between chemicals was calculated using binary hit-calls from concentration-response data for each gene. We evaluated GenRA performance in predicting ToxRefDB v2.0 hazard outcomes using the area under the Receiver Operating Characteristic (ROC) curve (AUC) for the baseline approach (chemical fingerprints) versus transcriptomic fingerprints and a combination of both (hybrid). For all endpoints, there were significant but only modest improvements in ROC AUC scores of 0.01 (2.1%) and 0.04 (7.3%) with transcriptomic and hybrid descriptors, respectively. However, for liver-specific toxicity endpoints, ROC AUC scores improved by 10% and 17% for transcriptomic and hybrid descriptors, respectively. Our findings suggest that using hybrid descriptors formed by combining chemical and targeted transcriptomic information can improve in vivo toxicity predictions in the right context.

Estimating Hepatotoxic Doses Using High-Content Imaging in Primary Hepatocytes

Using in vitro data to estimate point of departure (POD) values is an essential component of new approach methodologies (NAM)-based chemical risk assessments. In this case study, we evaluated a NAM for hepatotoxicity based on rat primary hepatocytes, high-content imaging (HCI), and toxicokinetic modeling. First, we treated rat primary hepatocytes with 10 concentrations (0.2 to 100 µM) of 51 chemicals that produced hepatotoxicity in repeat-dose subchronic and chronic exposures. Second, we used HCI to measure endoplasmic reticulum stress, mitochondrial function, lysosomal mass, steatosis, apoptosis, DNA texture, nuclear size, and cell number at 24, 48, and 72 h and calculated concentrations at 50% maximal activity (AC50). Third, we estimated administered equivalent doses (AEDs) from AC50 values using toxicokinetic modeling. AEDs using physiologically-based toxicokinetic models were 4.1-fold (SD 6.3) and 8.1-fold (SD 15.5) lower than subchronic and chronic lowest observed adverse effect levels (LOAELs), respectively. In contrast, AEDs from ToxCast and Tox21 assays were 89.8-fold (SD 149.5) and 168-fold (SD 323.7) lower than subchronic and chronic LOAELs. Individual HCI end-points also estimated AEDs for specific hepatic lesions that were lower than in vivo PODs. Lastly, AEDs were similar for different in vitro exposure durations, but steady-state toxicokinetic models produced 7.6-fold lower estimates than dynamic physiologically-based ones. Our findings suggest that NAMs from diverse cell types provide conservative estimates of PODs. In contrast, NAMs based on the same species and cell type as the adverse outcome may produce estimates closer to the traditional in vivo PODs.

Application of an in Vitro Assay to Identify Chemicals That Increase Estradiol and Progesterone Synthesis and Are Potential Breast Cancer Risk Factors

BACKGROUND

Established breast cancer risk factors, such as hormone replacement therapy and reproductive history, are thought to act by increasing estrogen and progesterone (P4) activity.

OBJECTIVE

We aimed to use in vitro screening data to identify chemicals that increase the synthesis of estradiol (E2) or P4 and evaluate potential risks.

METHOD

Using data from a high-throughput (HT) in vitro steroidogenesis assay developed for the U.S. Environmental Protection Agency (EPA) ToxCast program, we identified chemicals that increased estradiol (E2-up) or progesterone (P4-up) in human H295R adrenocortical carcinoma cells. We prioritized chemicals by their activity. We compiled in vivo studies and assessments about carcinogenicity and reproductive/developmental (repro/dev) toxicity. We identified exposure sources and predicted intakes from the U.S. EPA's ExpoCast.

RESULTS

We found 296 chemicals increased E2 (182) or P4 (185), with 71 chemicals increasing both. In vivo data often showed effects consistent with this mechanism. Of the E2- and P4-up chemicals, about 30% were likely repro/dev toxicants or carcinogens, whereas only 5-13% were classified as unlikely. However, most of the chemicals had insufficient in vivo data to evaluate their effects. Of 45 chemicals associated with mammary gland effects, and also tested in the H294R assay, 29 increased E2 or P4, including the well-known mammary carcinogen 7,12-dimethylbenz(a)anthracene. E2- and P4-up chemicals include pesticides, consumer product ingredients, food additives, and drinking water contaminants.

DISCUSSION

The U.S. EPA's in vitro screening data identified several hundred chemicals that should be considered as potential risk factors for breast cancer because they increased E2 or P4 synthesis. In vitro data is a helpful addition to current toxicity assessments, which are not sensitive to mammary gland effects. Relevant effects on the mammary gland are often not noticed or are dismissed, including for 2,4-dichlorophenol and cyfluthrin. Fifty-three active E2-up and 59 active P4-up chemicals that are in consumer products, food, pesticides, or drugs have not been evaluated for carcinogenic potential and are priorities for study and exposure reduction. https://doi.org/10.1289/EHP8608.

A Gene Expression Biomarker Predicts Heat Shock Factor 1 Activation in a Gene Expression Compendium

The United States Environmental Protection Agency (US EPA) recently developed a tiered testing strategy to use advances in high-throughput transcriptomics (HTTr) testing to identify molecular targets of thousands of environmental chemicals that can be linked to adverse outcomes. Here, we describe a method that uses a gene expression biomarker to predict chemical activation of heat shock factor 1 (HSF1), a transcription factor critical for proteome maintenance. The HSF1 biomarker was built from transcript profiles derived from A375 cells exposed to a HSF1-activating heat shock protein (HSP) 90 inhibitor in the presence or absence of HSF1 expression. The resultant 44 identified genes included those that (1) are dependent on HSF1 for regulation, (2) have direct interactions with HSF1 assessed by ChIP-Seq, and (3) are in the molecular chaperone family. To test for accuracy, the biomarker was compared in a pairwise manner to gene lists derived from treatments with known HSF1 activity (HSP and proteasomal inhibitors) using the correlation-based Running Fisher test; the balanced accuracy for prediction was 96%. A microarray compendium consisting of 12,092 microarray comparisons from human cells exposed to 2670 individual chemicals was screened using our approach; 112 and 19 chemicals were identified as putative HSF1 activators or suppressors, respectively, and most appear to be novel modulators. A large percentage of the chemical treatments that induced HSF1 also induced oxidant-activated NRF2 (∼46%). For five compounds or mixtures, we found that NRF2 activation occurred at lower concentrations or at earlier times than HSF1 activation, supporting the concept of a tiered cellular protection system dependent on the level of chemical-induced stress. The approach described here could be used to identify environmentally relevant chemical HSF1 activators in HTTr data sets.

Mechanistic Computational Model for Extrapolating In vitro Thyroid Peroxidase (TPO) Inhibition Data to Predict Serum Thyroid Hormone Levels in Rats

High throughput (HTP) in vitro assays are developed to screen chemicals for their potential to inhibit thyroid hormones (THs) synthesis. Some of these experiments, such as the thyroid peroxidase (TPO) inhibition assay, are based on thyroid microsomal extracts. However, the regulation of thyroid disruption chemicals (TDCs) is based on THs in vivo serum levels. This necessitates the estimation of TDCs in vivo tissue levels in the thyroid where THs synthesis inhibition by TPO takes place. The in vivo tissue levels of chemicals are controlled by pharmacokinetic determinants such as absorption, distribution, metabolism and excretion (ADME), and can be described quantitatively in physiologically based pharmacokinetic (PBPK) models. An integrative computational model including chemical specific PBPK and TH kinetics models provides a mechanistic quantitative approach to translate thyroidal HTP in vitro assays to in vivo measures of circulating THs serum levels. This computational framework is developed to quantitatively establish the linkage between applied dose, chemical thyroid tissue levels, thyroid TPO inhibition potential, and in vivo TH serum levels. Once this link is established quantitively, the overall model is used to calibrate the TH kinetics parameters using experimental data for THs levels in thyroid tissue and serum for the two drugs Propylthiouracil (PTU) and Methimazole (MMI). The calibrated quantitative framework is then evaluated against literature data for the environmental chemical ethylenethiourea (ETU). The linkage of PBPK and TH kinetics models illustrates a computational framework that can be extrapolated to humans to screen chemicals based on their exposure levels and potential to disrupt serum THs levels in vivo.

Determination of "fitness-for-purpose" of quantitative structure-activity relationship (QSAR) models to predict (eco-)toxicological endpoints for regulatory use

In silico models are used to predict toxicity and molecular properties in chemical safety assessment, gaining widespread regulatory use under a number of legislations globally. This study has rationalised previously published criteria to evaluate quantitative structure-activity relationships (QSARs) in terms of their uncertainty, variability and potential areas of bias, into ten assessment components, or higher level groupings. The components have been mapped onto specific regulatory uses (i.e. data gap filling for risk assessment, classification and labelling, and screening and prioritisation) identifying different levels of uncertainty that may be acceptable for each. Twelve published QSARs were evaluated using the components, such that their potential use could be identified. High uncertainty was commonly observed with the presentation of data, mechanistic interpretability, incorporation of toxicokinetics and the relevance of the data for regulatory purposes. The assessment components help to guide strategies that can be implemented to improve acceptability of QSARs through the reduction of uncertainties. It is anticipated that model developers could apply the assessment components from the model design phase (e.g. through problem formulation) through to their documentation and use. The application of the components provides the possibility to assess QSARs in a meaningful manner and demonstrate their fitness-for-purpose against pre-defined criteria.

Curated Data In - Trustworthy In Silico Models Out: The Impact of Data Quality on the Reliability of Artificial Intelligence Models as Alternatives to Animal Testing

New Approach Methodologies (NAMs) that employ artificial intelligence (AI) for predicting adverse effects of chemicals have generated optimistic expectations as alternatives to animal testing. However, the major underappreciated challenge in developing robust and predictive AI models is the impact of the quality of the input data on the model accuracy. Indeed, poor data reproducibility and quality have been frequently cited as factors contributing to the crisis in biomedical research, as well as similar shortcomings in the fields of toxicology and chemistry. In this article, we review the most recent efforts to improve confidence in the robustness of toxicological data and investigate the impact that data curation has on the confidence in model predictions. We also present two case studies demonstrating the effect of data curation on the performance of AI models for predicting skin sensitisation and skin irritation. We show that, whereas models generated with uncurated data had a 7-24% higher correct classification rate (CCR), the perceived performance was, in fact, inflated owing to the high number of duplicates in the training set. We assert that data curation is a critical step in building computational models, to help ensure that reliable predictions of chemical toxicity are achieved through use of the models.

High-Throughput Transcriptomics Platform for Screening Environmental Chemicals

New approach methodologies (NAMs) that efficiently provide information about chemical hazard without using whole animals are needed to accelerate the pace of chemical risk assessments. Technological advancements in gene expression assays have made in vitro high-throughput transcriptomics (HTTr) a feasible option for NAMs-based hazard characterization of environmental chemicals. In this study, we evaluated the Templated Oligo with Sequencing Readout (TempO-Seq) assay for HTTr concentration-response screening of a small set of chemicals in the human-derived MCF7 cell model. Our experimental design included a variety of reference samples and reference chemical treatments in order to objectively evaluate TempO-Seq assay performance. To facilitate analysis of these data, we developed a robust and scalable bioinformatics pipeline using open-source tools. We also developed a novel gene expression signature-based concentration-response modeling approach and compared the results to a previously implemented workflow for concentration-response analysis of transcriptomics data using BMDExpress. Analysis of reference samples and reference chemical treatments demonstrated highly reproducible differential gene expression signatures. In addition, we found that aggregating signals from individual genes into gene signatures prior to concentration-response modeling yielded in vitro transcriptional biological pathway altering concentrations (BPACs) that were closely aligned with previous ToxCast high-throughput screening assays. Often these identified signatures were associated with the known molecular target of the chemicals in our test set as the most sensitive components of the overall transcriptional response. This work has resulted in a novel and scalable in vitro HTTr workflow that is suitable for high-throughput hazard evaluation of environmental chemicals.

Single-cell profiling for advancing birth defects research and prevention

Cellular analysis of developmental processes and toxicities has traditionally entailed bulk methods (e.g., transcriptomics) that lack single cell resolution or tissue localization methods (e.g., immunostaining) that allow only a few genes to be monitored in each experiment. Recent technological advances have enabled interrogation of genomic function at the single-cell level, providing new opportunities to unravel developmental pathways and processes with unprecedented resolution. Here, we review emerging technologies of single-cell RNA-sequencing (scRNA-seq) to globally characterize the gene expression sets of different cell types and how different cell types emerge from earlier cell states in development. Cell atlases of experimental embryology and human embryogenesis at single-cell resolution will provide an encyclopedia of genes that define key stages from gastrulation to organogenesis. This technology, combined with computational models to discover key organizational principles, was recognized by Science magazine as the "Breakthrough of the year" for 2018 due to transformative potential on the way we study how human cells mature over a lifetime, how tissues regenerate, and how cells change in diseases (e.g., patient-derived organoids to screen disease-specific targets and design precision therapy). Profiling transcriptomes at the single-cell level can fulfill the need for greater detail in the molecular progression of all cell lineages, from pluripotency to adulthood and how cell-cell signaling pathways control progression at every step. Translational opportunities emerge for elucidating pathogenesis of genetic birth defects with cellular precision and improvements for predictive toxicology of chemical teratogenesis.

FutureTox IV Workshop Summary: Predictive Toxicology for Healthy Children

FutureTox IV, a Society of Toxicology Contemporary Concepts in Toxicology workshop, was held in November 2018. Building upon FutureTox I, II, and III, this conference focused on the latest science and technology for in vitro profiling and in silico modeling as it relates to predictive developmental and reproductive toxicity (DART). Publicly available high-throughput screening data sets are now available for broad in vitro profiling of bioactivities across large inventories of chemicals. Coupling this vast amount of mechanistic data with a deeper understanding of molecular embryology and post-natal development lays the groundwork for using new approach methodologies (NAMs) to evaluate chemical toxicity, drug efficacy, and safety assessment for embryo-fetal development. NAM is a term recently adopted in reference to any technology, methodology, approach, or combination thereof that can be used to provide information on chemical hazard and risk assessment to avoid the use of intact animals (U.S. Environmental Protection Agency [EPA], Strategic plan to promote the development and implementation of alternative test methods within the tsca program, 2018, https://www.epa.gov/sites/production/files/2018-06/documents/epa_alt_strat_plan_6-20-18_clean_final.pdf). There are challenges to implementing NAMs to evaluate chemicals for developmental toxicity compared with adult toxicity. This forum article reviews the 2018 workshop activities, highlighting challenges and opportunities for applying NAMs for adverse pregnancy outcomes (eg, preterm labor, malformations, low birth weight) as well as disorders manifesting postnatally (eg, neurodevelopmental impairment, breast cancer, cardiovascular disease, fertility). DART is an important concern for different regulatory statutes and test guidelines. Leveraging advancements in such approaches and the accompanying efficiencies to detecting potential hazards to human development are the unifying concepts toward implementing NAMs in DART testing. Although use of NAMs for higher level regulatory decision making is still on the horizon, the conference highlighted novel testing platforms and computational models that cover multiple levels of biological organization, with the unique temporal dynamics of embryonic development, and novel approaches for estimating toxicokinetic parameters essential in supporting in vitro to in vivo extrapolation.

Stem Cells for Next Level Toxicity Testing in the 21st Century

The call for a paradigm change in toxicology from the United States National Research Council in 2007 initiates awareness for the invention and use of human-relevant alternative methods for toxicological hazard assessment. Simple 2D in vitro systems may serve as first screening tools, however, recent developments infer the need for more complex, multicellular organotypic models, which are superior in mimicking the complexity of human organs. In this review article most critical organs for toxicity assessment, i.e., skin, brain, thyroid system, lung, heart, liver, kidney, and intestine are discussed with regards to their functions in health and disease. Embracing the manifold modes-of-action how xenobiotic compounds can interfere with physiological organ functions and cause toxicity, the need for translation of such multifaceted organ features into the dish seems obvious. Currently used in vitro methods for toxicological applications and ongoing developments not yet arrived in toxicity testing are discussed, especially highlighting the potential of models based on embryonic stem cells and induced pluripotent stem cells of human origin. Finally, the application of innovative technologies like organs-on-a-chip and genome editing point toward a toxicological paradigm change moves into action.

Non-Animal Strategies for Toxicity Assessment of Nanoscale Materials: Role of Adverse Outcome Pathways in the Selection of Endpoints

Faster, cheaper, sensitive, and mechanisms-based animal alternatives are needed to address the safety assessment needs of the growing number of nanomaterials (NM) and their sophisticated property variants. Specifically, strategies that help identify and prioritize alternative schemes involving individual test models, toxicity endpoints, and assays for the assessment of adverse outcomes, as well as strategies that enable validation and refinement of these schemes for the regulatory acceptance are needed. In this review, two strategies 1) the current nanotoxicology literature review and 2) the adverse outcome pathways (AOPs) framework, a systematic process that allows the assembly of available mechanistic information concerning a toxicological response in a simple modular format, are presented. The review highlights 1) the most frequently assessed and reported ad hoc in vivo and in vitro toxicity measurements in the literature, 2) various AOPs of relevance to inhalation toxicity of NM that are presently under development, and 3) their applicability in identifying key events of toxicity for targeted in vitro assay development. Finally, using an existing AOP for lung fibrosis, the specific combinations of cell types, exposure and test systems, and assays that are experimentally supported and thus, can be used for assessing NM-induced lung fibrosis, are proposed.

Predictive Capability of QSAR Models Based on the CompTox Zebrafish Embryo Assays: An Imbalanced Classification Problem

The CompTox Chemistry Dashboard (ToxCast) contains one of the largest public databases on Zebrafish (Danio rerio) developmental toxicity. The data consists of 19 toxicological endpoints on unique 1018 compounds measured in relatively low concentration ranges. The endpoints are related to developmental effects occurring in dechorionated zebrafish embryos for 120 hours post fertilization and monitored via gross malformations and mortality. We report the predictive capability of 209 quantitative structure-activity relationship (QSAR) models developed by machine learning methods using penalization techniques and diverse model quality metrics to cope with the imbalanced endpoints. All these QSAR models were generated to test how the imbalanced classification (toxic or non-toxic) endpoints could be predicted regardless which of three algorithms is used: logistic regression, multi-layer perceptron, or random forests. Additionally, QSAR toxicity models are developed starting from sets of classical molecular descriptors, structural fingerprints and their combinations. Only 8 out of 209 models passed the 0.20 Matthew's correlation coefficient value defined a priori as a threshold for acceptable model quality on the test sets. The best models were obtained for endpoints mortality (MORT), ActivityScore and JAW (deformation). The low predictability of the QSAR model developed from the zebrafish embryotoxicity data in the database is mainly due to a higher sensitivity of 19 measurements of endpoints carried out on dechorionated embryos at low concentrations.

The Challenges of 21st Century Neurotoxicology: The Case of Neurotoxicology Applied to Nanomaterials

After a short background discussing engineered nanomaterials (ENMs) and their physicochemical properties and applications, the present perspective paper highlights the main specific points that need to be considered when examining the question of neurotoxicity of nanomaterials. It underlines the necessity to integrate parameters, specific tools, and tests from multiple sources that make neurotoxicology when applied to nanomaterials particularly complex. Bringing together the knowledge of multiple disciplines e.g., nanotoxicology to neurotoxicology, is necessary to build integrated neurotoxicology for the third decade of the 21st Century. This article focuses on the greatest challenges and opportunities offered by this specific field. It highlights the scientific, methodological, political, regulatory, and educational issues. Scientific and methodological challenges include the determination of ENMs physicochemical parameters, the lack of information about protein corona modes of action, target organs, and cells and dose– response functions of ENMs. The need of standardization of data collection and harmonization of dedicated neurotoxicological protocols are also addressed. This article highlights how to address those challenges through innovative methods and tools, and our work also ventures to sketch the first list of substances that should be urgently prioritized for human modern neurotoxicology. Finally, political support with dedicated funding at the national and international levels must also be used to engage the communities concerned to set up dedicated educational program on this novel field.

A Gene Expression Biomarker Identifies Chemical Modulators of Estrogen Receptor α in an MCF-7 Microarray Compendium

Identification of chemicals that affect hormone-regulated systems will help to predict endocrine disruption. In our previous study, a 46 gene biomarker was found to be an accurate predictor of estrogen receptor (ER) α modulation in chemically treated MCF-7 cells. Here, potential ERα modulators were identified using the biomarker by screening a microarray compendium consisting of ∼1600 gene expression comparisons representing exposure to ∼1200 chemicals. A total of ∼170 chemicals were identified as potential ERα modulators. In the Connectivity Map 2.0 collection, 75 and 39 chemicals were predicted to activate or suppress ERα, and they included 12 and six known ERα agonists and antagonists/selective ERα modulators, respectively. Nineteen and eight of the total number were also identified as active in an ERα transactivation assay carried out in an MCF-7-derived cell line used to screen the Tox21 10K chemical library in agonist or antagonist modes, respectively. Chemicals predicted to modulate ERα in MCF-7 cells were examined further using global and targeted gene expression in wild-type and ERα-null cells, transactivation assays, and cell-free ERα coregulator interaction assays. Environmental chemicals classified as weak and very weak agonists were confirmed to activate ERα including apigenin, kaempferol, and oxybenzone. Novel activators included digoxin, nabumetone, ivermectin, and six progestins. Novel suppressors included emetine, mifepristone, niclosamide, and proscillaridin. Our strategy will be useful to identify environmentally relevant ERα modulators in future high-throughput transcriptomic screens.

Image-based profiling for drug discovery: due for a machine-learning upgrade?

Image-based profiling is a maturing strategy by which the rich information present in biological images is reduced to a multidimensional profile, a collection of extracted image-based features. These profiles can be mined for relevant patterns, revealing unexpected biological activity that is useful for many steps in the drug discovery process. Such applications include identifying disease-associated screenable phenotypes, understanding disease mechanisms and predicting a drug's activity, toxicity or mechanism of action. Several of these applications have been recently validated and have moved into production mode within academia and the pharmaceutical industry. Some of these have yielded disappointing results in practice but are now of renewed interest due to improved machine-learning strategies that better leverage image-based information. Although challenges remain, novel computational technologies such as deep learning and single-cell methods that better capture the biological information in images hold promise for accelerating drug discovery.