Signature analysis of high-throughput transcriptomics screening data for mechanistic inference and chemical grouping

Integrating Transcriptomic and Targeted New Approach Methodologies into a Tiered Framework for Chemical Bioactivity Screening

BACKGROUND

With thousands of chemicals in commerce and the environment, rapid identification of potential hazards is a critical need. Combining broad molecular profiling with targeted in vitro assays, such as high-throughput transcriptomics (HTTr) and receptor screening assays, could improve identification of chemicals that perturb key molecular targets associated with adverse outcomes.

OBJECTIVES

We aimed to link transcriptomic readouts to individual molecular targets and integrate transcriptomic predictions with orthogonal receptor-level assays in a proof-of-concept framework for chemical hazard prioritization.

METHODS

Transcriptomic profiles generated via TempO-Seq in U-2 OS and HepaRG cell lines were used to develop signatures composed of genes uniquely responsive to reference chemicals for distinct molecular targets. These signatures were applied to 75 reference and 1,126 nonreference chemicals screened via HTTr in both cell lines. Selective bioactivity toward each signature was determined by comparing potency estimates against the bulk of transcriptomic bioactivity for each chemical. Chemicals predicted by transcriptomics were confirmed for target bioactivity and selectivity using available orthogonal assay data from the US Environmental Protection Agency ToxCast program. A subset of 37 selectively acting chemicals from HTTr that did not have sufficient orthogonal data were prospectively tested using one of five receptor-level assays.

RESULTS

Of the 1,126 nonreference chemicals screened, 201 demonstrated selective bioactivity in at least one transcriptomic signature and 57 were confirmed as selective nuclear receptor agonists. Chemicals bioactive for each signature were significantly associated with orthogonal assay bioactivity, and signature-based points-of-departure were equally or more sensitive than biological pathway altering concentrations in 95.4% of signature-prioritized chemicals. Prospective profiling found that 18 of 37 (49%) chemicals without prior orthogonal assay data were bioactive against the predicted receptor.

DISCUSSION

Our work demonstrates that integrating transcriptomics with targeted orthogonal assays in a tiered framework can support Next Generation Risk Assessment by informing putative molecular targets and prioritizing chemicals for further testing. https://doi.org/10.1289/EHP16024.

Identification of mutagenicity, MOA, and dose response analysis

Bates et al. (2023) developed a cancer risk assessment framework to evaluate dietary carcinogens. The framework 1) evaluates gene mutation as an early key event of cancer development; 2) considers the dose metric appropriate based on mode of action understanding; and 3) integrates the appropriate dose metric category with relevant exposure data to evaluate dose response options and cancer level of concern for the specified exposure scenario. Here, we test the framework with three demonstrated rodent carcinogens with varying human cancer assessments and underlying cancer biology: acrylamide, aflatoxin B1, and β-myrcene. While traditional cancer assessment approaches might characterize these chemicals as potential human carcinogens based primarily on rodent tumorigenicity data, the framework evaluates the cancer MOA in the context of exposure patterns to provide more information on conditions that may increase risk. We found that mutation is an early key event for aflatoxin B1 carcinogenicity, and linear low-dose extrapolation is an appropriate approach. In contrast, MOA data support a dose threshold-based approach for acrylamide and β-myrcene, and their respective dietary consumption patterns suggest a low concern for cancer. The framework provides a more nuanced approach to cancer risk assessment and provides for a more informed risk management decision.

Decidual Disrupting Effects of Low-Dose Benzophenone-Type UV Filters in Human Endometrial Stromal Cells via ER/PR/FOXO1 Signaling

Exposure to endocrine disrupting chemicals (EDCs), particularly benzophenone (BP)-type UV filters, has been epidemiologically linked to endometrium-related reproductive risks in women. However, their effects on hormone-driven endometrial events and key receptor signaling at the human cellular level remain unexplored. Herein, using human primary endometrial stromal cells (HESCs), we investigated the disrupting effects of five BP congeners and deciphered the underlying mechanism on decidualization, a functional change of the endometrium preparing for pregnancy. BP-8, its two metabolites, BP-3, and BP-1 at 10 nM significantly disrupted progesterone-dependent decidualization in HESCs, marked by 1.5-1.8-fold and 2.2-2.6-fold upregulation of IGFBP-1 and LEFTY, respectively. Decidual transcriptional activators, WNT-FOXO1, were significantly induced by BPs, which are implicated in G2 phase cell arrest (from 3.26% to 8.93%) and apoptosis (from 12.29% to 25.61%). Mechanistically, the inhibition of estrogen receptor α (ERα) effectively alleviated these decidual disrupting effects. BPs increased the transcription of ERα and progesterone receptor (PR) signaling and enhanced nuclear translocation and interaction between ERα and PR during decidualization. The ERα-mediated enhancement of PR signaling activity by BPs was further validated in progesterone response element-luciferase transfected cells. Collectively, our findings elucidate the molecular pathway through which BPs disrupt endometrial decidualization via ERα/PR/FOXO1, providing critical mechanistic insights for the reproductive risk assessment of BPs and structurally related EDCs.

Cell Painting for cytotoxicity and mode-of-action analysis in primary human hepatocytes

High-throughput, human-relevant approaches for predicting chemical toxicity are urgently needed for better decision-making in human health. Here, we apply image-based profiling (the Cell Painting assay) and two cytotoxicity assays (metabolic and membrane damage readouts) to primary human hepatocytes after exposure to eight concentrations of 1085 compounds that include pharmaceuticals, pesticides, and industrial chemicals with known liver toxicity-related outcomes. Three computational methods (CellProfiler, a Cell Painting-specific convolutional neural network, and a pretrained vision transformer) were compared to extract morphology features from single cells or entire images. We used these morphology features to predict activity in the measured cytotoxicity assays, as well as in 412 curated ToxCast assays that span cytotoxicity, cell-based, and cell-free categories. We found that the morphological profiles detect compound bioactivity at lower concentrations than standard cytotoxicity assays. In supervised analyses, they predict cytotoxicity and targeted cell-based assay readouts, but not cell-free assay readouts. We also found that the various feature extraction methods performed relatively similarly and that filtering out non-bioactive or cytotoxic concentrations did not boost supervised assay prediction performance for any assay endpoint category, although it did have a large influence on unsupervised cluster analysis. We envision that image-based profiling could serve as a key component of modern safety assessment.

Concentrations, composition profiles, and in vitro-in silico-based mixture risk assessment of bisphenol A and its analogs in plant-based foods

The substitution of bisphenol A (BPA) with structurally similar analogs has raised concerns due to their comparable estrogenic activities. Considering the high consumption of plant-based foods, assessing the risks posed by bisphenols (BPs) in such dietary sources is essential. However, limited exposure and animal toxicological data on BP analogs hinder comprehensive risk assessments. This study investigated 16 BPs in 23 plant-based foods from Taiwan and estimated their dietary exposure across age groups. High-throughput toxicokinetic modeling was used to convert in vitro ToxCast estrogen receptor (ER) bioactive concentrations into human-equivalent points of departure (PODs), which were compared to PODs derived from animal studies and applied to assess mixture risks through the margin of exposure based on the common ER pathway. In total, 7 BPs were detected, and most samples (85.9 %) contained detectable concentrations. Total concentrations of the 7 BPs (∑7BP) ranged from 0.06 ± 0.11 ng/g to 26.60 ± 72.18 ng/g, with BPA being the most predominant (63 % of the mean ∑7BP concentrations), followed by bisphenol S (19 %) and 4,4-bisphenol F (13 %). In vitro-in silico-derived PODs were comparable to or even more protective than in vivo animal-derived PODs. For most population groups, combined exposure to multiple BPs from plant-based foods is generally not a risk concern for ER pathway perturbation, although potential concerns in worst-case scenarios cannot be excluded. This study advances the understanding of the dietary risks associated with BP mixtures and illustrates the potential of in vitro-in silico approaches for assessing human health risks from environmental contaminants.

High-throughput gene expression analysis with TempO-LINC sensitively resolves complex brain, lung and kidney heterogeneity at single-cell resolution

We report the development and performance of a novel genomics platform, TempO-LINC, for conducting high-throughput transcriptomic analysis on single cells and nuclei. TempO-LINC works by adding cell-identifying molecular barcodes onto highly selective and high-sensitivity gene expression probes within fixed cells, without having to first generate cDNA. Using an instrument-free combinatorial indexing approach, all probes within the same fixed cell receive an identical barcode, enabling the reconstruction of single-cell gene expression profiles across as few as several hundred cells and up to 100,000 + cells per sample. The TempO-LINC approach is easily scalable based on the number of barcodes and rounds of barcoding performed; however, for the experiments reported in this study, the assay utilized over 5.3 million unique barcodes. TempO-LINC offers a robust protocol for fixing and banking cells and displays high-sensitivity gene detection from multiple diverse sample types. We show that TempO-LINC has a multiplet rate of less than 1.1% and a cell capture rate of ~ 50%. Although the assay can accurately profile the whole transcriptome (19,683 human, 21,400 mouse and 21,119 rat genes), it can be targeted to measure only actionable/informative genes and molecular pathways of interest - thereby reducing sequencing requirements. In this study, we applied TempO-LINC to profile the transcriptomes of more than 90,000 cells across multiple species and sample types, including nuclei from mouse lung, kidney and brain tissues. The data demonstrated the ability to identify and annotate more than 50 unique cell populations and positively correlate expression of cell type-specific molecular markers within them. TempO-LINC is a robust new single-cell technology that is ideal for large-scale applications/studies with high data quality.

A transcriptomic biomarker predictive of cell proliferation for use in adverse outcome pathway-informed testing and assessment

High-throughput transcriptomics (HTTr) is increasingly being used to identify molecular targets of chemicals that can be linked to adverse outcomes. Cell proliferation (CP) is an important key event in chemical carcinogenesis. Here, we describe the construction and characterization of a gene expression biomarker that is predictive of the CP status in human and rodent tissues. The biomarker was constructed from 30 genes known to be increased in expression in prostate cancers relative to surrounding tissues and in cycling human MCF-7 cells after estrogen receptor (ER) agonist exposure. Using a large compendium of gene expression profiles to test utility, the biomarker could identify increases in CP in (i) 308 out of 367 tumor vs. normal surrounding tissue comparisons from 6 human organs, (ii) MCF-7 cells after activation of ER, (iii) after partial hepatectomy in mice and rats, and (iv) the livers of mice and rats after exposure to nongenotoxic hepatocarcinogens. The biomarker identified suppression of CP (i) under conditions of p53 activation by DNA damaging agents in human cells, (ii) in human A549 lung cells exposed to therapeutic anticancer kinase inhibitors (dasatinib, nilotnib), and (iii) in the mouse liver when comparing high levels of CP at birth to the low background levels in the adult. The responses using the biomarker were similar to those observed using conventional markers of CP including PCNA, Ki67, and BrdU labeling. The CP biomarker will be a useful tool for interpretation of HTTr data streams to identify CP status after exposure to chemicals in human cells or in rodent tissues.