Humans are exposed to chemicals that leach from plastics, yet many remain data-poor and lack toxicological evaluation. High-throughput transcriptomics (HTTr) provides a scalable way to screen chemicals and generate mechanistic insights relevant to human health risk assessment. We applied HTTr in MCF-7 breast cancer cells to assess chemicals across several concentrations (0.001–50 µM) used in plastics and dyes. Transcriptomic points of departure (tPODs) were derived from general gene perturbations, including pathway-level and estrogen receptor α (ERα)–specific changes. We also used transcriptomic biomarkers to evaluate ERα activity and cellular stress responses. Most plastic chemicals showed similar toxicological potency, with tPODs falling within one order of magnitude. Bisphenol K was the most potent, activating ERα at the lowest concentrations, followed by plastic additive 08 and bisphenol A (BPA). Despite similar overall potency, transcriptomic biomarkers revealed distinct mechanisms. Chemicals structurally similar to BPA activated ERα, whereas others with different functional groups inhibited the ERα biomarker. Pathway and upstream regulator analyses further indicated that BPA-like chemicals consistently perturbed ERα-related pathways, while other chemicals enriched fewer gene sets and often produced opposite directional responses. At the highest concentrations tested, several chemicals also activated stress-response biomarkers and suppressed proliferation. Overall, these results suggest that while many plastic chemicals exhibit comparable in vitro potency, they diverge in ERα regulation and downstream biological pathways. This study demonstrates the reproducibility and value of HTTr for chemical screening, supports grouping ERα-active plastic chemicals for read-across, and underscores the need for additional evaluation of plastic chemicals.

Accurate mutation detection and quantification are crucial for understanding mutagenesis and its potential health implications. Traditional in vivo mutagenicity assays, such as the transgenic rodent gene mutation assay, are limited by their focus on single reporter genes and inability to efficiently generate mutation spectra. Error-corrected sequencing (ECS) technologies like Duplex Sequencing (DS) offer significant advantages, including extremely low error rates and the ability to measure mutation frequencies (MFs) across various tissues and model organisms. Before ECS approaches can be adopted for regulatory purposes, their performance characteristics, particularly the type 1 error rate, must be rigorously established. We evaluated the type 1 error rate of DS through empirical analysis of vehicle control data and complementary simulation studies. Using 138 control mouse liver samples from 28 studies analyzed with the TwinStrand Mouse Mutagenesis Panel, we performed variance component analysis and found that experiment-level variability exceeds within-experiment sample variability. To evaluate the impact of between-study heterogeneity, we simulated overdispersed binomial data informed by the observed variance components. Removing the most variable studies reduced overdispersion and improved control of the type 1 error rate. Our findings demonstrate that DS maintains appropriate type 1 error rates (~0.05) when study heterogeneity is limited and at least four samples per group are used. Under greater overdispersion, sample sizes of five or six per group may be needed to achieve comparable control of the type 1 error rate. These results underscore the importance of combining empirical and simulation-based approaches to evaluate and optimize the statistical performance of emerging genomic technologies.

Human health risk assessment of engineered nanomaterials (ENM; materials with any dimension or structure between 1 and 100 nm) is challenged by the large number of compounds requiring assessment and the intricate association between physicochemical properties and toxicity. Previously, a number of high-content and high-throughput Novel Approach Methodologies (NAMs) were applied to investigate the impact of dissolution on toxicity induced by metal oxide (MOs) nanoparticles (NPs) in lung epithelial cells. This study evaluated the applicability of the data, including information from cell viability, transcriptomics, and genotoxicity endpoints, to conduct potency grouping and hazard identification for 7 individual MOs and their forms-NPs, dissolved equivalents, and bulk microparticle (MPs) analogues (18 total compounds). Benchmark concentration modeling was performed across a range of benchmark responses (BMRs), followed by hierarchical clustering to facilitate potency grouping and hazard identification. Correlation was assessed between endpoints used for potency grouping, and between endpoints and primary particle size, specific surface area, and solubility in cell culture medium across all BMRs. Instantaneously dissolving zinc oxide (ZnO) NPs presented similar potency to dissolved zinc and ZnO MPs, while aluminum oxide, iron oxide, and titanium dioxide NPs showed low potency. Thus, these particles were considered low priority for further testing. Copper oxide, nickel oxide and manganese dioxide NPs exhibited distinct potency and hazards compared to their dissolved or MPs forms, implying these particles warrant further assessment. Material solubility and the form of MOs were associated with endpoint potency. These results demonstrate the applicability of in vitro NAMs-based potency screening for first-tier assessment of MONP-induced acute toxicity.

High-throughput transcriptomics (HTTr) is increasingly used to derive transcriptomic points of departure (tPODs) for chemical screening and prioritization, yet the robustness of these estimates across studies with differing experimental designs remains unclear. Here, we compared tPODs for bisphenol A (BPA) across multiple HTTr studies conducted in MCF-7 breast cancer cells, including datasets from our laboratory and others. Although these studies employed broadly similar approaches, they differed in key methodological features, including estrogen-depletion protocols, exposure duration, and maximum test concentration. Using a standardized downstream bioinformatic workflow, we evaluated the consistency of BPA transcriptomic potency estimates and assessed factors contributing to variability across studies. Overall, five of seven studies yielded BPA potency estimates within a similar concentration range, supporting the utility of HTTr for comparative potency assessment despite some inter-study variability. Notably, studies conducted under estrogen-depleted conditions yielded higher potency estimates relative to those performed under standard culture conditions. Similarly, longer exposure durations were associated with higher potency estimates. These findings indicate that, while tPODs are generally reproducible across HTTr studies, experimental conditions, particularly estrogen depletion and exposure duration, can influence potency estimates in MCF-7 cells. However, these factors were not systematically varied or independently controlled across datasets, and therefore their individual contributions cannot be definitively disentangled in the present analysis. This work highlights the importance of standardizing hormone conditions and exposure durations when applying HTTr to screen estrogenic chemicals. Collectively, these results support the use of HTTr for chemical prioritization while underscoring the need for harmonized experimental design in endocrine-relevant in vitro models.

PURPOSE:The DNA damage response (DDR) and repair pathways are well-characterized mechanisms that maintain genomic integrity following genotoxic stress. However, the dose at which these pathways are transcriptionally activated for ionizing radiation and how they vary across individuals remains an important area for investigation. The TGx-DDI transcriptomic biomarker panel, a 64-gene signature originally developed in TK6 lymphoblastic cells, was designed to distinguish DNA damage-inducing (DDI) from non-DDI agents through key transcriptional responses in the DDR pathway. While validated for chemical exposures, its application to ionizing radiation remains limited. Here, we evaluate whether TGx-DDI can detect coordinated DDR-associated transcriptional activation following X-ray exposure at the individual donor level across a range of radiation doses and dose-rates. MATERIALS AND METHODS:We conducted a meta-analysis of previously generated transcriptomic data from primary human lymphocytes exposed to X-rays at nine doses (0.05-6 Gy) under two dose-rates (0.05 Gy/min, LDR; 1 Gy/min, HDR). Transcriptomic profiling was performed 24 hours post-irradiation using the targeted TempO-Seq™ platform. RESULTS:TGx-DDI classification outcomes of the dataset showed clear dose- and dose-rate dependencies. Under both dose rate conditions, DDI classifications were observed at most doses across donors, although non-DDI- calls occurred at lower doses in a subset of individuals. More TGx-DDI responsive donors exhibited detectable DDI responses at very low doses (as low as 0.05 Gy), whereas less TGx-DDI responsive donors showed responses only at higher doses (≥1 Gy) or were non-responsive. CONCLUSION:Overall, TGx-DDI demonstrated proof of concept by identifying radiation-associated transcriptional activation of DDR at moderate to high doses, while revealing differences in DDR-associated transcriptional responses across donors at low doses. This study represents the first application of TGx-DDI at the individual level, providing preliminary evidence that transcriptomic biomarkers may be used for individualized radiation response profiling in future studies.

Chronic exposure to polycyclic aromatic hydrocarbons (PAHs) is associated with increased risk of cancer through a mutagenic mode of action. We investigated how exposure duration influences mutagenicity in bone marrow (BM) and liver of MutaMouse males orally exposed to increasing doses of benzo[b]fluoranthene (BbF), a priority PAH, for 28, 90, or 180 days. We applied Duplex Sequencing (DS) across 20 endogenous loci to assess dose- and time-dependent effects on mutation frequency, clonal expansion of mutant cells, mutation distribution, and spectra. Mutation frequencies increased with BbF dose and exposure duration in both tissues. After 90 and 180 days of exposure to 25 mg/kg/day BbF, mutation frequency was 1.8x and 2.4x higher in BM, and 3.8x and 8.6x higher in liver, than 28-day exposures. Clonally derived mutations increased significantly with exposure duration in BM but not liver. Intergenic loci incurred more mutations than genic regions in both tissues across all exposure durations. Mutation spectra were dominated by C:G>A:T transversions and was enriched in SBS4, the mutation signature associated with tobacco-induced human lung cancer. Benchmark dose (BMD50) modeling showed increasing BbF potency with longer exposures. BMD50 decreased from 8.2 (28d) to 3.7 (90d) and 3.0 (180d) mg/kg/day in BM; in liver, BMD50 shifted from 14.3 (28d) to 4.3 (90d) and 3.9 (180d) mg/kg/day. The observed tissue-specific responses may be attributed to differences in proliferation rate, metabolic activity, and cellular lifespan. Our findings provide insights into the mutagenic impacts of prolonged PAH exposure and highlight tissue-specific differences in mutation susceptibility.

Ionizing radiation elicits complex cellular responses that are influenced not only by total dose but also by the rate at which the dose is delivered. Understanding how dose rate modulates molecular outcomes is important for accurate risk assessment. In this study, we apply an integrative multi-omics approach combining transcriptomic and proteomic profiling while adjusting for covariates to investigate how differential dose rates of ionizing radiation alter gene and protein expression in human lymphocytes. Particular emphasis is placed on identifying dose-rate-specific alterations in key molecular pathways. Peripheral blood from 14 healthy donors (8 males, 6 females) was irradiatedex vivowith x-rays at 0.05 Gy min-1(DR1) and 1.0 Gy min-1(DR2) across a dose range from 0 to 6 Gy. Gene expression was assessed using TempO-Seq™, and relative protein abundance was determined by mass spectrometry. Differential expression analysis was conducted using edgeR and limma, adjusting for sex, age, and leukocyte counts (false discovery rate < 0.05). Multi-omics integration was performed using regularised canonical correlation analysis (rCCA) implemented in mixOmics, followed by Reactome pathway enrichment analysis. We identified 2477 and 2612 differentially expressed genes at DR1 and DR2, respectively, and 368 and 386 differentially expressed proteins. To assess dose discrimination, we examined sample separation in the space defined by the average canonical variates from transcriptomic and proteomic datasets using rCCA. Covariate adjustment improved dose discrimination, particularly above 0.5 Gy. Using a correlation cut-off threshold of 0.5 in rCCA, 212 (DR1) and 276 (DR2) highly correlated gene-protein pairs were identified. DR2 exposure was associated with stronger enrichment of stress-related pathways, including unfolded protein response, senescence and oncogenic kinase signalling. In contrast, DR1 induced enrichment of pathways associated with immune engagement, including antigen presentation. At both dose rates, transcriptomic changes highlighted upstream regulatory processes (chromatin modelling) and proteomic changes captured downstream functional pathways such as immune activity and apoptosis. The multi-omics approach with covariate adjustment revealed key radiation-responsive pathways and dose-rate-dependent molecular differences, highlighting the value of integrating transcriptomic and proteomic data to better understand radiation effects.

Duplex Sequencing (DS) is an ultra-accurate, error-corrected next generation sequencing (ecNGS) technology for mutation analysis. A working group (WG) within Health and Environmental Sciences Institute's Genetic Toxicology Technical Committee is investigating the suitability of ecNGS for regulatory mutagenicity testing, using DS as a model. Initial steps to promote acceptance require demonstrating technical reproducibility across DS-experienced and inexperienced laboratories and establishing the method's sensitivity relative to conventional tests. Thus, the WG conducted a 'reconstruction experiment' to evaluate the transferability, reproducibility, and sensitivity of DS. TwinStrand Biosciences first applied DS to establish mutation frequency (MF) in DNA samples extracted from the livers of an untreated Sprague Dawley rat, or rats treated with either 100 mg/kg/day benzo[a]pyrene (B[a]P) for ten days or 40 mg/kg/day N-ethyl-N-nitrosourea (ENU) for three days. Using the measured MF in these original samples, mixtures were then constructed using the B[a]P- and ENU-treated samples to create "MF standards" with target MFs 1.2-, 1.5-, and 2-fold greater than the untreated control. Aliquots of these standards were distributed to seven laboratories in North America and Europe. DS libraries were prepared by each laboratory and TwinStrand. All eight laboratories met library preparation and assay performance metrics to yield high quality sequencing data with MF in the expected 'MF standard' range. The measured MF and mutation spectra were nearly identical across the laboratories and a 2-fold increase in MF could readily be identified in all labs relative to the untreated controls. The results confirm the high reproducibility and sensitivity of DS for mutagenicity assessment.

Most per- and poly-fluoroalkyl substances (PFAS) lack toxicity data, and the hazards associated with different PFAS chemical structures have not been systematically assessed using in vivo models. To address this gap, we compared the toxicity of nine PFAS in embryo-larval zebrafish, an emerging alternative to conventional in vivo models. Exposures were conducted from 0 to 5 days post-fertilization with semi-static renewal. We then evaluated three apical toxicity endpoints (developmental toxicity (mortality/malformation), swimming behaviour, and metabolic activity) alongside gene expression changes using high-throughput transcriptomics. These data were used to derive apical and transcriptomic points of departure (aPODs and tPODs, respectively). Transcriptomic benchmark concentration modeling in BMDExpress v3.2 was performed to derive tPODs using multiple approaches. Overall, PFAS potency increased with longer fluorinated carbon chain lengths and was greater for PFAS containing sulfonic groups. tPODs were generally the most sensitive endpoints, typically falling within a 10-fold range below aPODs. Our results support previous findings that tPODs provide suitably conservative PODs for chemical toxicity assessment. Our results contribute new data on PFAS early-life stage toxicity and demonstrate an economical and ethically viable high-throughput platform for systematic evaluation of chemical hazards and potencies for risk assessment applications.

BACKGROUND:The current radiation protection framework extrapolates health risks from high-dose exposures based on a linear, no-threshold model. However, empirical data on molecular effects below 0.1 Gy are lacking, creating uncertainties in risk assessments. To address this, we used benchmark dose (BMD) modeling, commonly applied in chemical hazard assessment, to analyze gene and protein expression changes in human white blood cells, providing insights into dose-response relationships following low-dose radiation (LDR) exposure. METHODS:Blood samples were collected from 14 participants (6 females, 8 males). Lymphocytes were isolated, cultured, and exposed to X-irradiation at nine doses (0-6 Gy) at 0.05 Gy/min. Transcriptomic and proteomic changes were assessed 24 h post-exposure. BMD modeling was applied to each endpoint, and the data were grouped into distinct dose-response patterns. Pathway analysis identified cellular functions associated with these patterns, offering insight into the biological effects of LDR. RESULTS:BMD modeling identified 1,204 genes and 168 proteins with dose-response relationships, with median BMD lower confidence limits (BMDLs) of 1.38 Gy and 0.21 Gy, respectively. Transcriptional and proteomic responses exhibited complex patterns, including exponential, biphasic, and hypersensitivity responses, with peak activity between 0.05-0.25 Gy, followed by a decline or plateau. Pathway analysis revealed changes in genes and proteins related to DNA damage, cell cycle, cellular stress, metabolism, immune function, and cancer, with DNA damage response genes showing BMDLs below 0.1 Gy. CONCLUSIONS:This study shows that molecular dose-response relationships can be complex and non-linear, emphasizing the need for further research to better understand the effects of LDR.

High-throughput gene expression studies commonly employ pathway analyses to infer biological meaning from lists of differentially expressed genes (DEGs). In toxicology and pharmacology studies, treatment groups are analysed against vehicle controls to identify DEGs and altered pathways. Previously, we empirically quantified false-positive rates of DEGs in gene expression data from pools of vehicle-treated zebrafish embryos to determine appropriate study designs (sample and pool size). Here, the same data were subject to Over-Representation Analysis (ORA) and Gene Set Enrichment Analysis (GSEA) to identify false-positive enriched pathways. As expected, the number of false-positive ORA results was lowest where pool and sample sizes were largest (conditions which also generated the fewest significant DEGs). In contrast, the frequency of GSEA false-positives generated through the fast GSEA (fgsea) algorithm increased with pool and sample size and was highest for simulations that generated 0 DEGs, with ribosomal gene sets significantly enriched with the highest frequency. We describe 2 distinct mechanisms by which GSEA generated these false-positive results, both of which are most likely to generate significant gene sets under conditions where expression differences are particularly low. Finally, GSEA analyses were repeated using 1 alternative GSEA algorithm (CERNO) and 11 different ranking statistics. In almost every analysis, the number of significant results was highest where pool size was highest, with ribosome as the more frequently enriched gene set, suggesting our observations to be generalizable to different implementations of GSEA. These results from zebrafish embryos suggest caution in interpreting any GSEA results in contrasts where there are no DEGs.

Some everyday consumer products contain endocrine disruptors like bisphenol A (BPA) and its replacements. To date, most in vitro chemical screening to evaluate these compounds has been accomplished using immortalized cell lines, which differ significantly from human tissues. Our goal was to test BPA and select alternatives previously screened in breast cancer cells for toxicological potency and mechanism of action in human mammary epithelial cells (HMECs). HMECs from three human donors were exposed to BPA and four alternative chemicals (in concentration response format from 0.001 to 50 µM) for 48 h and global transcriptomic changes were quantified. Transcriptomic biomarker analysis was employed to explore chemically induced estrogen receptor alpha (ERα) activation and alterations in stress response pathways. Benchmark concentration (BMC) analysis was applied to gene expression data to derive transcriptomic points of departure (tPODs) to compare chemicals for potency. Pathway and upstream regulator analysis was applied among the genes fitting BMCs. All chemicals had tPODs within a single order of magnitude. Bisphenol AF (BPAF) was the most potent, followed by tetramethyl bisphenol F (TMBPF), bisphenol C (BPC), 4,4'-bisphenol S (BPS), and BPA. None of the chemicals activated the ERα biomarker. Some stress response biomarkers were activated at high exposure concentrations. Genes fitting BMCs clustered chemicals into two groups, with one group (BPAF and TMBPF) primarily inhibiting expression patterns and the other (BPC, BPS, and BPA) mostly activating. These data suggest that the BPA alternatives tested have similar toxicological potencies in HMECs and oppositely enrich various gene sets.

Per- and polyfluoroalkyl substances (PFAS) are persistent and widespread contaminants. Epidemiological effects of PFAS include increased serum cholesterol, decreased immune response to vaccination and disease, and increased incidence of cancer; however, PFAS modes of action remain unclear. Herein, we analyzed gene expression data from human liver spheroids that were exposed to several concentrations of 24 different PFAS. Benchmark concentration (BMC) response modeling was used to identify the 250 lowest gene BMCs for each PFAS. Hierarchical clustering analysis revealed 4 functionally diverse gene sets. Each gene set was affected by a distinct group of PFAS, whereas individual PFAS were usually part of more than 1 PFAS group. The biological roles of these gene sets relate to: (1) cholesterol biogenesis and cholesterol clearance (downregulated by 7 fluorocarbon or longer PFAS), putatively through discordance of cholesterol sensing by SCAP and LXR due to membrane integration of PFAS; (2) lipolysis (upregulated by 8 carbon or shorter PFAS); (3) innate immunity (downregulated by most PFAS); and (4) adaptive immunity (downregulated by sulfonate-type PFAS). The distinctions between the 4 PFAS groups suggest that PFAS can act through at least 4 independent mechanisms. The molecular characteristics of each PFAS group may be useful for understanding the molecular interactions leading to their effect on gene expression. Inclusion of some PFAS congeners in more than one PFAS group suggests that individual PFAS can act through multiple unrelated molecular interactions. This transcriptomic analysis offers a major advancement to the understanding of the molecular mechanisms underlying the effects of PFAS exposure and provides guidance for future work that may strengthen links between PFAS exposure and their proposed effects on human health.

Understanding the mechanisms by which environmental chemicals cause toxicity is necessary for effective human health risk assessment. High-throughput transcriptomics (HTTr) can be used to inform risk assessment on toxicological mechanisms, hazards, and potencies. We applied HTTr to elucidate the molecular mechanisms by which per- and polyfluoroalkyl substances (PFAS) cause liver perturbations. We contrasted transcriptomic profiles of PFOA, PFBS, PFOS, and PFDS against transcriptomic profiles from established liver-toxic and non-toxic reference compounds, alongside peroxisome proliferator-activated receptors (PPARs) agonists. Our analysis was conducted on metabolically competent 3-D human liver spheroids produced from primary cells from 10 donors. Pathway analysis showed that PFOS and PFDS perturb many of the same pathways as the known liver-toxic compounds in the spheroids, and that the cholesterol biosynthesis pathways are significantly affected by exposure to these compounds. PFOA alters lipid metabolism-related pathways but its expression profile does not closely match reference compounds. PFBS upregulates many degradation-related pathways and targets many of the same pathways as the PPAR agonists and acetaminophen. Our transcriptional analysis does not support the claim that these PFAS are DNA-damaging in this model. A multidimensional scaling (MDS) analysis revealed that PFOS, PFOA, and PFDS cluster together in the same multidimensional space as liver-damaging compounds, whereas PFBS clusters more closely with the non-liver-damaging compounds. Benchmark concentration-response modeling predicts that all the PFAS are bioactive in the liver. Overall, our results show that these PFAS produce unique transcriptional changes but also alter pathways associated with established liver-toxic chemicals in this liver spheroid model.