In Silico Guidance for In Vitro Androgen and Glucocorticoid Receptor ToxCast Assays.

Molecular initiating events (MIEs) are key events in adverse outcome pathways that link molecular chemistry to target biology. As they are based on chemistry, these interactions are excellent targets for computational chemistry approaches to in silico modeling. In this work, we aim to link ligand chemical structures to MIEs for androgen receptor (AR) and glucocorticoid receptor (GR) binding using ToxCast data. This has been done using an automated computational algorithm to perform maximal common substructure searches on chemical binders for each target from the ToxCast dataset. The models developed show a high level of accuracy, correctly assigning 87.20% of AR binders and 96.81% of GR binders in a 25% test set using holdout cross-validation. The 2D structural alerts developed can be used as in silico models to predict these MIEs and as guidance for in vitro ToxCast assays to confirm hits. These models can target such experimental work, reducing the number of assays to be performed to gain required toxicological insight. Development of these models has also allowed some structural alerts to be identified as predictors for agonist or antagonist behavior at the receptor target. This work represents a first step in using computational methods to guide and target experimental approaches.

Evaluating potential refinements to existing Threshold of Toxicological Concern (TTC) values for environmentally-relevant compounds.

The Toxic Substances Control Act (TSCA) mandates the US EPA perform risk-based prioritisation of chemicals in commerce and then, for high-priority substances, develop risk evaluations that integrate toxicity data with exposure information. One approach being considered for data poor chemicals is the Threshold of Toxicological Concern (TTC). Here, TTC values derived using oral (sub)chronic No Observable (Adverse) Effect Level (NO(A)EL) data from the EPA's Toxicity Values database (ToxValDB) were compared with published TTC values from Munro et al. (1996). A total of 4554 chemicals with structures present in ToxValDB were assigned into their respective TTC categories using the Toxtree software tool, of which toxicity data was available for 1304 substances. The TTC values derived from ToxValDB were similar, but not identical to the Munro TTC values: Cramer I ((ToxValDB) 37.3 c. f. (Munro) 30 µg/kg-day), Cramer II (34.6 c. f. 9.1 µg/kg-day) and Cramer III (3.9 c. f. 1.5 µg/kg-day). Cramer III 5th percentile values were found to be statistically different. Chemical features of the two Cramer III datasets were evaluated to account for the differences. TTC values derived from this expanded dataset substantiated the original TTC values, reaffirming the utility of TTC as a promising tool in a risk-based prioritisation approach.

Development of an in Silico Profiler for Mitochondrial Toxicity.

This study outlines the analysis of mitochondrial toxicity for a variety of pharmaceutical drugs extracted from Zhang et al. ((2009) Toxicol. In Vitro, 23, 134-140). These chemicals were grouped into categories based upon structural similarity. Subsequently, mechanistic analysis was undertaken for each category to identify the molecular initiating event driving mitochondrial toxicity. The mechanistic information elucidated during the analysis enabled mechanism-based structural alerts to be developed and combined together to form an in silico profiler. This profiler is envisaged to be used to develop chemical categories based upon similar mechanisms as part of the adverse outcome pathway paradigm. Additionally, the profiler could be utilized in screening large data sets in order to identify chemicals with the potential to induce mitochondrial toxicity.

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.

Optimizing androgen receptor prioritization using high-throughput assay-based activity models.

Introduction: Computational models using data from high-throughput screening assays have promise for prioritizing and screening chemicals for testing under the U.S. Environmental Protection Agency's Endocrine Disruptor Screening Program (EDSP). The purpose of this work was to demonstrate a data processing method for the determination of optimal minimal assay batteries from a larger comprehensive model, to provide a uniform method of evaluating the performance of future minimal assay batteries compared with the androgen receptor (AR) pathway model, and to incorporate chemical cluster analysis into this evaluation. Although several of the assays in the AR pathway model are no longer available through the original vendor, this approach could be used for future evaluations of minimal assay models for prioritization and screening. Methods: We compared two previously published models and found that an expanded 14-assay model had higher sensitivity for antagonists, whereas the original 11-assay model had slightly higher sensitivity for agonists. We then investigated subsets of assays in the original AR pathway model to optimize overall testing strategies that minimize cost while maintaining sensitivity across a broad chemical space. Results and Discussion: Evaluation of the critical assays across subset models derived from the 14-assay model identified three critical assays for predicting antagonism and two critical assays for predicting agonism. A minimum of nine assays is required for predicting agonism and antagonism with high sensitivity (95%). However, testing workflows guided by chemical structure-based clusters can reduce the average number of assays needed per chemical by basing the assays selected for testing on the likelihood of a chemical being an AR agonist, according to its structure. Our results show that a multi-stage testing workflow can provide 95% sensitivity while requiring only 48% of the resources required for running all assays from the original full models. The resources can be reduced further by incorporating in silico activity predictions. Conclusion: This work illustrates a data-driven approach that incorporates chemical clustering and simultaneous consideration of antagonism and agonism mechanisms to more efficiently screen chemicals. This case study provides a proof of concept for prioritization and screening strategies that can be utilized in future analyses to minimize the overall number of assays needed for predicting AR activity, which will maximize the number of chemicals that can be tested and allow data-driven prioritization of chemicals for further screening under the EDSP.

Smoking cessation, harm reduction, and biomarkers protocols in the PhenX Toolkit: Tools for standardized data collection.

The use of standard protocols in studies supports consistent data collection, improves data quality, and facilitates cross-study analyses. Funded by the National Institutes of Health, the PhenX (consensus measures for Phenotypes and eXposures) Toolkit is a catalog of recommended measurement protocols that address a wide range of research topics and are suitable for inclusion in a variety of study designs. In 2020, a PhenX Working Group of smoking cessation experts followed a well-established consensus process to identify and recommend measurement protocols suitable for inclusion in smoking cessation and smoking harm reduction studies. The broader scientific community was invited to review and provide feedback on the preliminary recommendation of the Working Group. Fourteen selected protocols for measuring smoking cessation, harm reduction, and biomarkers research associated with smoking cessation were released in the PhenX Toolkit ( https://www.phenxtoolkit.org) in February 2021. These protocols complement existing PhenX Toolkit content related to tobacco regulatory research, substance use and addiction research, and other measures of smoking-related health outcomes. Adopting well-established protocols enables consistent data collection and facilitates comparing and combining data across studies, potentially increasing the scientific impact of individual studies.

Derivation of New Threshold of Toxicological Concern Values for Exposure via Inhalation for Environmentally-Relevant Chemicals.

The requirements of amended Toxic Substances Control Act (TSCA) stipulates that the US Environmental Protection Agency (US EPA) evaluate existing chemicals and make risk based assessments. There are ~33,000 substances that are active in commerce on the TSCA public non-confidential inventory, many of which lack available toxicity and exposure information to inform risk-based decision making. One approach to facilitate the assessment of these substances being considered is the Threshold of Toxicological Concern (TTC). TTC values are intended to identify safe levels of exposure for data poor substances. TTC values derived based on non-cancer data notably by Munro et al. (1996) are well-established and are in routine use for food additive applications however far less attention has been focused on developing TTC values where inhalation is the route of exposure. Here, an effort was made to derive new inhalation TTC values using the EPA's Toxicity Values database, ToxValDB. A total of 4,703 substances captured in ToxValDB were assigned into their respective TTC categories using the Kroes module within the Toxtree software tool and custom profilers developed in Nelms et al. (2019) and Patlewicz et al. (2018). For the substances assigned into the 3 Cramer classes, the 5th percentiles were calculated from the empirical cumulative distributions of No observed (adverse) effect level (concentration) values. The 5th percentiles were converted to their respective TTC values and compared with published values reported by Escher et al. (2010) and Carthew et al. (2009). The TTC values derived from ToxValDB were orders of magnitude more conservative, further, Cramer classification was not found to be effective at discriminating potencies. Instead, use of aquatic toxicity modes of action such as Verhaar et al. (1992) were found to be effective at separating substances in terms of their potencies and new TTC thresholds were derived.

An evaluation of the performance of selected (Q)SARs/expert systems for predicting acute oral toxicity.

Multiple US agencies use acute oral toxicity data in a variety of regulatory contexts. One of the ad-hoc groups that the US Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM) established to implement the ICCVAM Strategic Roadmap was the Acute Toxicity Workgroup (ATWG) to support the development, acceptance, and actualisation of new approach methodologies (NAMs). One of the ATWG charges was to evaluate in vitro and in silico methods for predicting rat acute systemic toxicity. Collaboratively, the NTP Interagency Center for the Evaluation of Alternative Toxicological Methods (NICEATM) and the US Environmental Protection Agency (US EPA) collected a large body of rat oral acute toxicity data (~16,713 studies for 11,992 substances) to serve as a reference set to evaluate the performance and coverage of new and existing models as well as build understanding of the inherent variability of the animal data. Here, we focus on evaluating in silico models for predicting the Lethal Dose (LD50) as implemented within two expert systems, TIMES and TEST. The performance and coverage were evaluated against the reference dataset. The performance of both models were similar, but TEST was able to make predictions for more chemicals than TIMES. The subset of the data with multiple (>3) LD50 values was used to evaluate the variability in data and served as a benchmark to compare model performance. Enrichment analysis was conducted using ToxPrint chemical fingerprints to identify the types of chemicals where predictions lay outside the upper 95% confidence interval. Overall, TEST and TIMES models performed similarly but had different chemical features associated with low accuracy predictions, reaffirming that these models are complementary and both worth evaluation when seeking to predict rat LD50 values.

Identification of novel activators of the metal responsive transcription factor (MTF-1) using a gene expression biomarker in a microarray compendium.

Environmental exposure to metals is known to cause a number of human toxicities including cancer. Metal-responsive transcription factor 1 (MTF-1) is an important component of metal regulation systems in mammalian cells. Here, we describe a novel method to identify chemicals that activate MTF-1 based on microarray profiling data. MTF-1 biomarker genes were identified that exhibited consistent, robust expression across 10 microarray comparisons examining the effects of metals (zinc, nickel, lead, arsenic, mercury, and silver) on gene expression in human cells. A subset of the resulting 81 biomarker genes was shown to be altered by knockdown of the MTF1 gene including metallothionein family members and a zinc transporter. The ability to correctly identify treatment conditions that activate MTF-1 was determined by comparing the biomarker to microarray comparisons from cells exposed to reference metal activators of MTF-1 using the rank-based Running Fisher algorithm. The balanced accuracy for prediction was 93%. The biomarker was then used to identify organic chemicals that activate MTF-1 from a compendium of 11 725 human gene expression comparisons representing 2582 chemicals. There were 700 chemicals identified that included those known to interact with cellular metals, such as clioquinol and disulfiram, as well as a set of novel chemicals. All nine of the novel chemicals selected for validation were confirmed to activate MTF-1 biomarker genes in MCF-7 cells and to lesser extents in MTF1-null cells by qPCR and targeted RNA-Seq. Overall, our work demonstrates that the biomarker for MTF-1 coupled with the Running Fisher test is a reliable strategy to identify novel chemical modulators of metal homeostasis using gene expression profiling.

Comparing and contrasting the coverage of publicly available structural alerts for protein binding.

The molecular initiating event for many mechanisms of toxicological action comprise the reactive, covalent binding between an exogenous electrophile and an endogenous nucleophile. The target sites for electrophiles are typically peptides, proteins, enzymes or DNA. Of these, the formation of covalent adducts with proteins and DNA are perhaps the most established as they are most closely associated with skin sensitisation and genotoxicity endpoints. As such, being able to identify electrophilic features within a chemical structure provides a starting point to characterise its reactivity profile. There are a number of software tools that have been developed to help identify structural features indicative of electrophilic reactive potential to address various purposes, including: 1) to facilitate category formation for read-across of toxicity effects such as skin sensitisation potential, as well as 2) to profile substances to identify potential confounding factors to rationalise their activity in high-throughput screening (HTS) assays. Here, three such schemes that have been published in the literature as collections of SMARTS patterns and their associated chemical-biological reaction domains have been compared. The goals are 1) to better understand their scope and coverage, and 2) to assess their performance relative to a published skin sensitisation dataset where manual annotations to assign likely mechanistic domains based on expert judgement were already available. The 3 schemes were then applied to the Tox21 library and the consensus outcome was reported to highlight the proportion of chemicals likely to exhibit a reactivity response, specific to a mechanistic reaction domain, but non-specific with respect to target-tissue based activity. ToxPrint fingerprints were computed and activity enrichments computed to compare the structural features identified for the skin sensitisation dataset and Tox21 chemicals for each 'consensus' reaction domain. Enriched ToxPrints were also used to identify ToxCast assays potentially informative for reactivity.

Differentiating Pathway-Specific From Nonspecific Effects in High-Throughput Toxicity Data: A Foundation for Prioritizing Adverse Outcome Pathway Development.

The U.S. Environmental Protection Agency's ToxCast program has screened thousands of chemicals for biological activity, primarily using high-throughput in vitro bioassays. Adverse outcome pathways (AOPs) offer a means to link pathway-specific biological activities with potential apical effects relevant to risk assessors. Thus, efforts are underway to develop AOPs relevant to pathway-specific perturbations detected in ToxCast assays. Previous work identified a "cytotoxic burst" (CTB) phenomenon wherein large numbers of the ToxCast assays begin to respond at or near test chemical concentrations that elicit cytotoxicity, and a statistical approach to defining the bounds of the CTB was developed. To focus AOP development on the molecular targets corresponding to ToxCast assays indicating pathway-specific effects, we conducted a meta-analysis to identify which assays most frequently respond at concentrations below the CTB. A preliminary list of potentially important, target-specific assays was determined by ranking assays by the fraction of chemical hits below the CTB compared with the number of chemicals tested. Additional priority assays were identified using a diagnostic-odds-ratio approach which gives greater ranking to assays with high specificity but low responsivity. Combined, the two prioritization methods identified several novel targets (e.g., peripheral benzodiazepine and progesterone receptors) to prioritize for AOP development, and affirmed the importance of a number of existing AOPs aligned with ToxCast targets (e.g., thyroperoxidase, estrogen receptor, aromatase). The prioritization approaches did not appear to be influenced by inter-assay differences in chemical bioavailability. Furthermore, the outcomes were robust based on a variety of different parameters used to define the CTB.

A Mechanistic Framework for Integrating Chemical Structure and High-Throughput Screening Results to Improve Toxicity Predictions.

Adverse Outcome Pathways (AOPs) establish a connection between a molecular initiating event (MIE) and an adverse outcome. Detailed understanding of the MIE provides the ideal data for determining chemical properties required to elicit the MIE. This study utilized high-throughput screening data from the ToxCast program, coupled with chemical structural information, to generate chemical clusters using three similarity methods pertaining to nine MIEs within an AOP network for hepatic steatosis. Three case studies demonstrate the utility of the mechanistic information held by the MIE for integrating biological and chemical data. Evaluation of the chemical clusters activating the glucocorticoid receptor identified activity differences in chemicals within a cluster. Comparison of the estrogen receptor results with previous work showed that bioactivity data and structural alerts can be combined to improve predictions in a customizable way where bioactivity data are limited. The aryl hydrocarbon receptor (AHR) highlighted that while structural data can be used to offset limited data for new screening efforts, not all ToxCast targets have sufficient data to define robust chemical clusters. In this context, an alternative to additional receptor assays is proposed where assays for proximal key events downstream of AHR activation could be used to enhance confidence in active calls. These case studies illustrate how the AOP framework can support an iterative process whereby in vitro toxicity testing and chemical structure can be combined to improve toxicity predictions. In vitro assays can inform the development of structural alerts linking chemical structure to toxicity. Consequently, structurally related chemical groups can facilitate identification of assays that would be informative for a specific MIE. Together, these activities form a virtuous cycle where the mechanistic basis for the in vitro results and the breadth of the structural alerts continually improve over time to better predict activity of chemicals for which limited toxicity data exist.

Editor's Highlight: Negative Predictors of Carcinogenicity for Environmental Chemicals.

Recent international efforts have led to proposals for modified carcinogenicity testing paradigms based on data from shorter-term studies. The main goal of the current study was to evaluate the negative predictive value (NPV) of short-term toxicity indicators on carcinogenicity study outcomes and cancer classifications for chemicals previously reviewed by the U.S. Environmental Protection Agency (EPA). Pathology data were analyzed from over 900 acceptable 2-sex guideline subchronic (3-month) and carcinogenicity studies in the U.S. EPA Toxicity Reference Database. Chemical cancer classifications were obtained from annual reports of the U.S. EPA Office of Pesticide Programs. Histopathologic risk signals and evidence of hormonal perturbation in subchronic rat studies provided 56% NPV for any tumor outcome in the rat or mouse and 75% NPV for cancer classifications not requiring quantitative risk assessment (qRA). In comparison, lack of activity in a battery of 35 in vitro cytotoxicity assays from the U.S. EPA ToxCast library provided 49% NPV for any tumor outcome and 80% NPV for cancer classifications not requiring qRA. These findings support the idea that the absence of short-term bioactivity may provide useful information for prioritizing chemicals based on potential carcinogenic risk. Additional data streams are needed to further refine these models.

Accelerating Adverse Outcome Pathway Development Using Publicly Available Data Sources.

The adverse outcome pathway (AOP) concept links molecular perturbations with organism and population-level outcomes to support high-throughput toxicity (HTT) testing. International efforts are underway to define AOPs and store the information supporting these AOPs in a central knowledge base; however, this process is currently labor-intensive and time-consuming. Publicly available data sources provide a wealth of information that could be used to define computationally predicted AOPs (cpAOPs), which could serve as a basis for creating expert-derived AOPs in a much more efficient way. Computational tools for mining large datasets provide the means for extracting and organizing the information captured in these public data sources. Using cpAOPs as a starting point for expert-derived AOPs should accelerate AOP development. Coupling this with tools to coordinate and facilitate the expert development efforts will increase the number and quality of AOPs produced, which should play a key role in advancing the adoption of HTT testing, thereby reducing the use of animals in toxicity testing and greatly increasing the number of chemicals that can be tested.

Methods for assigning confidence to toxicity data with multiple values--Identifying experimental outliers.

The assessment of data quality is a crucial element in many disciplines such as predictive toxicology and risk assessment. Currently, the reliability of toxicity data is assessed on the basis of testing information alone (adherence to Good Laboratory Practice (GLP), detailed testing protocols, etc.). Common practice is to take one toxicity data point per compound - usually the one with the apparently highest reliability. All other toxicity data points (for the same experiment and compound) from other sources are neglected. To show the benefits of incorporating the "less reliable" data, a simple, independent, statistical approach to assess data quality and reliability on a mathematical basis was developed. A large data set of toxicity values to Aliivibrio fischeri was assessed. The data set contained 1813 data points for 1227 different compounds, including 203 identified as non-polar narcotic. Log KOW values were calculated and non-polar narcosis quantitative structure-activity relationship (QSAR) models were built. A statistical approach to data quality assessment, which is based on data outlier omission and confidence scoring, improved the linear QSARs. The results indicate that a beneficial method for using large data sets containing multiple data values per compound and highly variable study data has been developed. Furthermore this statistical approach can help to develop novel QSARs and support risk assessment by obtaining more reliable values for biological endpoints.