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.

Analysis of variability in the rabbit skin irritation assay.

The in vivo rabbit test is the benchmark against which new approach methodologies for skin irritation are usually compared. No alternative method offers a complete replacement of animal use for this endpoint for all regulatory applications. Variability in the animal reference data may be a limiting factor in identifying a replacement. We established a curated data set of 2624 test records, representing 990 substances, each tested at least twice, to characterize the reproducibility of the in vivo assay. Methodological deviations from guidelines were noted, and multiple data sets with differing tolerances for deviations were created. Conditional probabilities were used to evaluate the reproducibility of the in vivo method in identification of U.S. Environmental Protection Agency or Globally Harmonized System hazard categories. Chemicals classified as moderate irritants at least once were classified as mild or non-irritants at least 40% of the time when tested repeatedly. Variability was greatest between mild and moderate irritants, which both had less than a 50% likelihood of being replicated. Increased reproducibility was observed when a binary categorization between corrosives/moderate irritants and mild/non-irritants was used. This analysis indicates that variability present in the rabbit skin irritation test should be considered when evaluating nonanimal alternative methods as potential replacements.

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.

Thresholds Derived From Common Measures in Rat Studies Are Predictive of Liver Tumorigenic Chemicals.

We hypothesized that typical tissue and clinical chemistry (ClinChem) end points measured in rat toxicity studies exhibit chemical-independent biological thresholds beyond which cancer occurs. Using the rat in vivo TG-GATES study, 75 chemicals were examined across chemical-dose-time comparisons that could be linked to liver tumor outcomes. Thresholds for liver weight to body weight (LW/BW) and 21 serum ClinChem end points were defined as the maximum and minimum values for those exposures that did not lead to liver tumors in rats. Upper thresholds were identified for LW/BW (117%), aspartate aminotransferase (195%), alanine aminotransferase (141%), alkaline phosphatase (152%), and total bilirubin (115%), and lower thresholds were identified for phospholipids (82%), relative albumin (93%), total cholesterol (82%), and total protein (94%). Thresholds derived from the TG-GATES data set were consistent across other acute and subchronic rat studies. A training set of ClinChem and LW/BW thresholds derived from a 38 chemical training set from TG-GATES was predictive of liver tumor outcomes for a test set of 37 independent TG-GATES chemicals (91%). The thresholds were most predictive when applied to 7d treatments (98%). These findings provide support that biological thresholds for common end points in rodent studies can be used to predict chemical tumorigenic potential.

Reduction of charge impurities in a silicon metal-oxide-semiconductor quantum dot qubit device patterned with nano-imprint lithography.

The silicon metal-oxide-semiconductor quantum dot architecture is a leading approach for the physical implementation of semiconductor quantum computing. One major challenge for scalable quantum dots is the presence of charge impurities. Electron-beam lithography (EBL), almost universally used to fabricate quantum dot devices, is known to create such defects at the Si/SiO2 interface. To eliminate the need for EBL, we have transferred the metal gate pattern of a quantum dot onto the silicon substrate using nano-imprint lithography. Critical features with 50 nm scale and separation can be dependably reproduced. By characterizing the bias-dependent charge transport through a quantum point contact barrier, the prevalence of impurities is found to be largely diminished in nano-imprinted devices when compared to similar electron-beam-written counterparts. High-quality charge transport and charge sensing of several quantum dots are obtained. Additionally, gate noise is measured with an average of 1.5 µeV Hz-1/2 equivalent to previous measurements made on devices fabricated with EBL, which suggests that the leading source of impurities produced by EBL are deep, fixed charges. This work offers a path toward reliable quantum dot operation in MOS by improving fabrication techniques to reduce charge impurities.

Adverse outcome pathway-driven identification of rat liver tumorigens in short-term assays.

Chemicals induce liver cancer in rodents through well characterized adverse outcome pathways (AOPs). We hypothesized that measurement of molecular initiating events (MIEs) and downstream key events (KEs) in liver cancer AOPs in short-term assays will allow early identification of chemicals and their associated doses that are likely to be tumorigenic in the liver in two-year bioassays. We tested this hypothesis using the rat in vivo TG-GATES study data to measure MIEs (genotoxicity, cytotoxicity, AhR, CAR, ER, PPARα) and associated KEs (oxidative stress, cell proliferation, liver to body weights) across 77 chemicals that could be linked to doses with previously established effects on rat liver tumor induction. Gene expression biomarkers for MIEs generally considered to be rodent specific and human irrelevant (CAR, PPARα) and for MIEs that would be considered of greater risk at human relevant exposures (ER, AhR) were built using microarray comparisons from the livers of rats treated with prototypical activators of the receptors. The genotoxicity biomarker, also a potentially human relevant MIE, was comprised of 7 p53-responsive genes known to be induced upon DNA damage. The ability of the biomarkers to accurately predict MIE activation ranged from 91% to 98%. The Toxicological Priority Index (ToxPi) was used to rank chemicals based on their ability to activate MIEs/KEs. Chemicals administered at tumorigenic doses clearly gave the highest ranked scores. Our AOP-directed approach could be used in short term assays to identify chemicals and their doses that would be predicted to cause liver tumors in rats.

High-power broadband Yb-free clad-pumped EDFA for L-band DWDM applications.

We demonstrated high-power broadband Yb-free clad-pumped erbium-doped fiber amplifier (EDFA) using commercial available low-cost 976 nm multimode diodes. An output power +33 dBm with less than ±1 dB natural gain flatness over a gain bandwidth of 33 nm (1570.3-1603.3 nm) was obtained using a new double-clad erbium-doped fiber which has a large core diameter of 17 µm and low NA of 0.11. The saturated output power, net gain, noise figure, and optical power conversion efficiency of the clad-pumped EDFA were characterized, and system impact of LP11 mode and four-wave-mixing nonlinearity were also evaluated.

Cellular toxicity associated with exposure to perfluorinated carboxylates (PFCAs) and their metabolic precursors.

The biotransformation of fluorotelomer based compounds yields saturated and unsaturated fluorotelomer aldehydes (FTALs and FTUALs, respectively) and carboxylic acids (FTCAs and FTUCAs, respectively) as intermediate metabolites that subsequently transform to perfluorinated carboxylic acids (PFCAs). Previous studies have demonstrated that the FTCAs and FTUCAs are 1 to 5 orders of magnitude more toxic than PFCAs after exposure to aquatic organisms. Additionally, FTUALs have demonstrated reactivity with proteins, which may be associated with toxicity through the inhibition of protein function. The purpose of this study was to carry out a comprehensive assessment of the relative toxicity between PFCAs and their intermediate precursor metabolites: the FTALs, FTUALs, FTCAs, and FTUCAs. Analytes were separately incubated with human liver epithelial (THLE-2) cells to assess how varying the functional group and the fluorinated chain length affects cell viability. For each analyte, dose-response EC50 values were calculated. The EC50 values for FTUCAs and FTCAs were similar, with values ranging from 22 ± 9 and 24 ± 9 µM for the 10:2 congeners to 1004 ± 20 and 1004 ± 24 µM for the 4:2 congeners, respectively. The EC50 values for the PFCAs ranged from 65 ± 41 (PFDA) to 1361 ± 146 (PFBA) µM. The range of toxicity between PFCAs and their acid precursors were similar. However, the comparative toxicity between the 6:2 and 8:2 congeners and their corresponding PFCA had toxicity thresholds that varied depending on the functional headgroup, where FTUALs ≥ FTALs > FTUCAs ≥ FTCAs > PFCAs. For all PFCAs and acid precursors, toxicity depended on the length of the fluorinated chain, where the longer chain lengths yielded greater bioaccumulation and enhanced toxicity, results which agreed with those previously reported. By contrast, FTALs and FTUALs were the most toxic of all the analytes examined, where toxicity was enhanced at shorter chain lengths, with EC50 values of 7 ± 1 µM (6:2 FTUAL) and 8.6 ± 0.8 µM (6:2 FTAL). DNA adducts were not detectable for the aldehyde precursors, using a quantitative long-range PCR method. Our data provide the first evidence that aldehyde intermediates have demonstrated toxicity in cellular systems that is more significant than PFCAs and their corresponding acid intermediates.

Evidence to challenge the universality of the Horiuti-Polanyi mechanism for hydrogenation in heterogeneous catalysis: origin and trend of the preference of a non-Horiuti-Polanyi mechanism.

The Horiuti-Polanyi mechanism has been considered to be universal for explaining the mechanisms of hydrogenation reactions in heterogeneous catalysis for several decades. In this work, we examine this mechanism for the hydrogenation of acrolein, the simplest α,ß-unsaturated aldehyde, in gold-based systems as well as some other metals using extensive first-principles calculations. It is found that a non-Horiuti-Polanyi mechanism is favored in some cases. Furthermore, the physical origin and trend of this mechanism are revealed and discussed regarding the geometrical and electronic effects, which will have a significant influence on current understandings on heterogeneous catalytic hydrogenation reactions and the future catalyst design for these reactions.

Translational infidelity-induced protein stress results from a deficiency in Trm9-catalyzed tRNA modifications.

Correct codon-anticodon pairing promotes translational fidelity, with these interactions greatly facilitated by modified nucleosides found in tRNA. We hypothesized that wobble uridine modifications catalyzed by tRNA methyltransferase 9 (Trm9) are essential for translational fidelity. In support, we have used phenotypic, reporter and protein-based assays to demonstrate increased translational infidelity in trm9Δ Saccharomyces cerevisiae cells. Codon reengineering studies suggest that Trm9-catalyzed tRNA modifications promote fidelity during the translation of specific genes, those rich in arginine and glutamic acid codons from mixed boxes. Using quantitative tRNA modification analysis, we determined that trm9Δ cells are only deficient in 2 of 23 tRNA modifications, with those 2, 5-methoxycarbonylmethyluridine (mcm ( 5) U) and 5-methoxycarbonylmethyl-2-thiouridine (mcm ( 5) s ( 2) U), classified as key determinants of translational fidelity. We also show that in the absence of mcm ( 5) U and mcm ( 5) s ( 2) U, the resulting translational infidelity promotes protein errors and activation of unfolded protein and heat shock responses. These data support a model in which Trm9-catalyzed tRNA modifications promote fidelity during the translation of specific transcripts, with decreased wobble base modification leading to translational infidelity, protein errors and activation of protein stress response pathways.

Cross-species Functionome analysis identifies proteins associated with DNA repair, translation and aerobic respiration as conserved modulators of UV-toxicity.

Cellular responses to DNA damage can prevent mutations and death. In this study, we have used high throughput screens and developed a comparative genomic approach, termed Functionome mapping, to discover conserved responses to UVC-damage. Functionome mapping uses gene ontology (GO) information to link proteins with similar biological functions from different organisms, and we have used it to compare 303, 311 and 288 UVC-toxicity modulating proteins from Escherichia coli, Schizosaccharomyces pombe and Saccharomyces cerevisiae, respectively. We have demonstrated that all three organisms use DNA repair, translation and aerobic respiration associated processes to modulate the toxicity of UVC, with these last two categories highlighting the importance of ribosomal proteins and electron transport machinery. Our study has demonstrated that comparative genomic approaches can be used to identify conserved responses to damage, and suggest roles for translational machinery and components of energy metabolism in optimizing the DNA damage response.

Data curation to support toxicity assessments using the Integrated Chemical Environment.

Humans are exposed to large numbers of chemicals during their daily activities. To assess and understand potential health impacts of chemical exposure, investigators and regulators need access to reliable toxicity data. In particular, reliable toxicity data for a wide range of chemistries are needed to support development of new approach methodologies (NAMs) such as computational models, which offer increased throughput relative to traditional approaches and reduce or replace animal use. NAMs development and evaluation require chemically diverse data sets that are typically constructed by incorporating results from multiple studies into a single, integrated view; however, integrating data is not always a straightforward task. Primary study sources often vary in the way data are organized and reported. Metadata and information needed to support interoperability and provide context are often lacking, which necessitates literature research on the assay prior to attempting data integration. The Integrated Chemical Environment (ICE) was developed to support the development, evaluation, and application of NAMs. ICE provides curated toxicity data and computational tools to integrate and explore available information, thus facilitating knowledge discovery and interoperability. This paper describes the data curation workflow for integrating data into ICE. Data destined for ICE undergo rigorous harmonization, standardization, and formatting processes using both automated and manual expert-driven approaches. These processes improve the utility of the data for diverse analyses and facilitate application within ICE or a user's external workflow while preserving data integrity and context. ICE data curation provides the structure, reliability, and accessibility needed for data to support chemical assessments.

Genomic comparisons between hepatocarcinogenic and non-hepatocarcinogenic organophosphate insecticides in the mouse liver.

Short-term biomarkers of toxicity have an increasingly important role in the screening and prioritization of new chemicals. In this study, we examined early indicators of liver toxicity for three reference organophosphate (OP) chemicals, which are among the most widely used insecticides in the world. The OP methidathion was previously shown to increase the incidence of liver toxicity, including hepatocellular tumors, in male mice. To provide insights into the adverse outcome pathway (AOP) that underlies these tumors, effects of methidathion in the male mouse liver were examined after 7 and 28 day exposures and compared to those of two other OPs that either do not increase (fenthion) or possibly suppress liver cancer (parathion) in mice. None of the chemicals caused increases in liver weight/body weight or histopathological changes in the liver. Parathion decreased liver cell proliferation after 7 and 28 days while the other chemicals had no effects. There was no evidence for hepatotoxicity in any of the treatment groups. Full-genome microarray analysis of the livers from the 7 and 28 day treatments demonstrated that methidathion and fenthion regulated a large number of overlapping genes, while parathion regulated a unique set of genes. Examination of cytochrome P450 enzyme activities and use of predictive gene expression biomarkers found no consistent evidence for activation of AhR, CAR, PXR, or PPARα. Parathion suppressed the male-specific gene expression pattern through STAT5b, similar to genetic and dietary conditions that decrease liver tumor incidence in mice. Overall, these findings indicate that methidathion causes liver cancer by a mechanism that does not involve common mechanisms of liver cancer induction.

Retrospective analysis of dermal absorption triple pack data.

Dermal toxicity is driven by the ability of a substance to penetrate the skin. The "triple pack" approach, which combines in vivo rat, in vitro rat, and in vitro human data, is used to calculate an estimated human dermal absorption factor (DAF). To assess the feasibility of deriving a DAF using only in vitro data, we retrospectively evaluated agrochemical formulations to compare the DAF derived from each individual method to the DAF generated from the triple pack approach. For most of the formulations evaluated, the in vitro rat method generated a similar or higher DAF value than the in vivo method. Absorption through in vitro human skin was similar to or less than that observed in rat skin for all formulations. For most of the formulations, the human in vitro method provided a similar or higher estimate of dermal absorption than the triple pack approach. For human health risk assessment, in vitro assays using human skin would be preferable, as they would be directly relevant to the species of interest and avoid overestimation of dermal absorption using rat models. However, rat in vitro studies would still have utility in the absence of human in vitro data. In vitro rat data provide estimates of dermal absorption that are at least as protective as in vivo rat data and thus could also be considered adequate for use in estab­lishing DAFs. The comparisons presented support potentially using in vitro data alone for DAF derivation for human health risk assessment of pesticides.

Early-life mitochondrial DNA damage results in lifelong deficits in energy production mediated by redox signaling in Caenorhabditis elegans.

The consequences of damage to the mitochondrial genome (mtDNA) are poorly understood, although mtDNA is more susceptible to damage resulting from some genotoxicants than nuclear DNA (nucDNA), and many environmental toxicants target the mitochondria. Reports from the toxicological literature suggest that exposure to early-life mitochondrial damage could lead to deleterious consequences later in life (the "Developmental Origins of Health and Disease" paradigm), but reports from other fields often report beneficial ("mitohormetic") responses to such damage. Here, we tested the effects of low (causing no change in lifespan) levels of ultraviolet C (UVC)-induced, irreparable mtDNA damage during early development in Caenorhabditis elegans. This exposure led to life-long reductions in mtDNA copy number and steady-state ATP levels, accompanied by increased oxygen consumption and altered metabolite profiles, suggesting inefficient mitochondrial function. Exposed nematodes were also developmentally delayed, reached smaller adult size, and were rendered more susceptible to subsequent exposure to chemical mitotoxicants. Metabolomic and genetic analysis of key signaling and metabolic pathways supported redox and mitochondrial stress-response signaling during early development as a mechanism for establishing these persistent alterations. Our results highlight the importance of early-life exposures to environmental pollutants, especially in the context of exposure to chemicals that target mitochondria.

In Silico Approaches In Carcinogenicity Hazard Assessment: Current Status and Future Needs.

Historically, identifying carcinogens has relied primarily on tumor studies in rodents, which require enormous resources in both money and time. In silico models have been developed for predicting rodent carcinogens but have not yet found general regulatory acceptance, in part due to the lack of a generally accepted protocol for performing such an assessment as well as limitations in predictive performance and scope. There remains a need for additional, improved in silico carcinogenicity models, especially ones that are more human-relevant, for use in research and regulatory decision-making. As part of an international effort to develop in silico toxicological protocols, a consortium of toxicologists, computational scientists, and regulatory scientists across several industries and governmental agencies evaluated the extent to which in silico models exist for each of the recently defined 10 key characteristics (KCs) of carcinogens. This position paper summarizes the current status of in silico tools for the assessment of each KC and identifies the data gaps that need to be addressed before a comprehensive in silico carcinogenicity protocol can be developed for regulatory use.

Systems based mapping demonstrates that recovery from alkylation damage requires DNA repair, RNA processing, and translation associated networks.

The identification of cellular responses to damage can promote mechanistic insight into stress signalling. We have screened a library of 3968 Escherichia coli gene-deletion mutants to identify 99 gene products that modulate the toxicity of the alkylating agent methyl methanesulfonate (MMS). We have developed an ontology mapping approach to identify functional categories over-represented with MMS-toxicity modulating proteins and demonstrate that, in addition to DNA re-synthesis (replication, recombination, and repair), proteins involved in mRNA processing and translation influence viability after MMS damage. We have also mapped our MMS-toxicity modulating proteins onto an E. coli protein interactome and identified a sub-network consisting of 32 proteins functioning in DNA repair, mRNA processing, and translation. Clustering coefficient analysis identified seven highly connected MMS-toxicity modulating proteins associated with translation and mRNA processing, with the high connectivity suggestive of a coordinated response. Corresponding results from reporter assays support the idea that the SOS response is influenced by activities associated with the mRNA-translation interface.

A genome-wide screen in Saccharomyces cerevisiae reveals pathways affected by arsenic toxicity.

We have used Saccharomyces cerevisiae to identify toxicologically important proteins and pathways involved in arsenic-induced toxicity and carcinogenicity in humans. We performed a systemic screen of the complete set of 4733 haploid S. cerevisiae single-gene-deletion mutants to identify those that have decreased or increased growth, relative to wild type, after exposure to sodium arsenite (NaAsO(2)). IC(50) values for all mutants were determined to further validate our results. Ultimately we identified 248 mutants sensitive to arsenite and 5 mutants resistant to arsenite exposure. We analyzed the proteins corresponding to arsenite-sensitive mutants and determined that they belonged to functional categories that include protein binding, phosphate metabolism, vacuolar/lysosomal transport, protein targeting, sorting, and translocation, cell growth/morphogenesis, cell polarity and filament formation. Furthermore, these data were mapped onto a protein interactome to identify arsenite-toxicity-modulating networks. These networks are associated with the cytoskeleton, ubiquitination, histone acetylation and the MAPK signaling pathway. Our studies have potential implications for understanding toxicity and carcinogenesis in arsenic-induced human conditions, such as cancer and aging.

A Set of Six Gene Expression Biomarkers Identify Rat Liver Tumorigens in Short-term Assays.

Chemical-induced liver cancer occurs in rodents through well-characterized adverse outcome pathways. We hypothesized that measurement of the 6 most common molecular initiating events (MIEs) in liver cancer adverse outcome pathways in short-term assays using only gene expression will allow early identification of chemicals and their associated doses that are likely to be tumorigenic in the liver in 2-year bioassays. We tested this hypothesis using transcript data from a rat liver microarray compendium consisting of 2013 comparisons of 146 chemicals administered at doses with previously established effects on rat liver tumor induction. Five MIEs were measured using previously characterized gene expression biomarkers composed of gene sets predictive for genotoxicity and activation of 1 or more xenobiotic receptors (aryl hydrocarbon receptor, constitutive activated receptor, estrogen receptor, and peroxisome proliferator-activated receptor α). Because chronic injury can be important in tumorigenesis, we also developed a biomarker for cytotoxicity that had a 96% balanced accuracy. Characterization of the genes in each biomarker set using the unsupervised TXG-MAP network model demonstrated that the genes were associated with distinct functional coexpression modules. Using the Toxicological Priority Index to rank chemicals based on their ability to activate the MIEs showed that chemicals administered at tumorigenic doses clearly gave the highest ranked scores. Balanced accuracies using thresholds derived from either TG-GATES or DrugMatrix data sets to predict tumorigenicity in independent sets of chemicals were up to 93%. These results show that a MIE-directed approach using only gene expression biomarkers could be used in short-term assays to identify chemicals and their doses that cause tumors.

Mining a human transcriptome database for chemical modulators of NRF2.

Nuclear factor erythroid-2 related factor 2 (NRF2) encoded by the NFE2L2 gene is a transcription factor critical for protecting cells from chemically-induced oxidative stress. We developed computational procedures to identify chemical modulators of NRF2 in a large database of human microarray data. A gene expression biomarker was built from statistically-filtered gene lists derived from microarray experiments in primary human hepatocytes and cancer cell lines exposed to NRF2-activating chemicals (oltipraz, sulforaphane, CDDO-Im) or in which the NRF2 suppressor Keap1 was knocked down by siRNA. Directionally consistent biomarker genes were further filtered for those dependent on NRF2 using a microarray dataset from cells after NFE2L2 siRNA knockdown. The resulting 143-gene biomarker was evaluated as a predictive tool using the correlation-based Running Fisher algorithm. Using 59 gene expression comparisons from chemically-treated cells with known NRF2 activating potential, the biomarker gave a balanced accuracy of 93%. The biomarker was comprised of many well-known NRF2 target genes (AKR1B10, AKR1C1, NQO1, TXNRD1, SRXN1, GCLC, GCLM), 69% of which were found to be bound directly by NRF2 using ChIP-Seq. NRF2 activity was assessed across ~9840 microarray comparisons from ~1460 studies examining the effects of ~2260 chemicals in human cell lines. A total of 260 and 43 chemicals were found to activate or suppress NRF2, respectively, most of which have not been previously reported to modulate NRF2 activity. Using a NRF2-responsive reporter gene in HepG2 cells, we confirmed the activity of a set of chemicals predicted using the biomarker. The biomarker will be useful for future gene expression screening studies of environmentally-relevant chemicals.