You are here:
High-throughput Transcriptomics Concentration-Response Screen of 1593 Chemicals in MCF7 Cells
Citation:
Judson, R., L. Everett, D. Haggard, J. Bundy, B. Chambers, L. Taylor, B. Vallanat, I. Shah, AND J. Harrill. High-throughput Transcriptomics Concentration-Response Screen of 1593 Chemicals in MCF7 Cells. Society of Toxicology 2021 Virtual Annual Meeting, Virtual, NC, March 12 - 26, 2021. https://doi.org/10.23645/epacomptox.17130014
Impact/Purpose:
Poster presented to the Society of Toxicology 2021 Virtual Annual Meeting March 2021. Concentration-response transcriptomics data can now be generated on hundreds to thousands of compounds. We have twin goals of using this data for hazard identification (what pathways / targets will a chemical activate) and estimation of points of departure (POD). Data can be analyzed at the gene or gene-set / signature level. Here we present preliminary work on using signature modeling methods to analyze data from a 1593-chemical screen run by EPA.
Description:
There is an increasing push to make use of New Approach Methods (NAMs) to help in the assessment of chemical risk. NAMs are non-animal approaches that include a variety of in vitro and computational methods such as high-throughput screening, transcriptomics and QSAR modeling. Transcriptomics, under the term toxicogenomics, has been used for many years to broadly probe the cellular response to chemicals, but only in the last few years has the cost of this technology come down enough to allow testing of large numbers of compounds in concentration-response format. Here we report results of a high-throughput transcriptomics (HTTr) screen of 1593 chemicals (including drugs, food and cosmetics ingredients, industrial chemicals, pesticides) in a breast cancer cell line (MCF7) in 8 concentrations spanning 3 orders of magnitude using the TempO-seq whole transcriptome assay. Raw count data was processed to produce probe-wise log2 fold changes (l2fc) values using the R package DESeq2 and then concentration-response modeling was performed at the level of “signatures”. Signatures are directed or undirected gene sets that capture the coordinated response of pathway perturbations. For this step, we used a custom R package called httrpathway that allows the use of a variety of signature scoring methods and concentration-response models. To evaluate the accuracy of this approach, the set of screened chemicals was annotated by molecular target, where known, to designate a set of reference chemicals. We observe strong signals in the reference chemicals for several nuclear receptor (NR) pathways (estrogen, retinoic acid, glucocorticoid) as well as non-NR pathways (calcium modulating ligand (CAMLG) and ATPase). In addition to pathway-level information, we calculate potencies as benchmark dose (BMDs) for each signature, and then summarize these to a chemical-level potency. For estrogen receptor targeting chemicals, we compared HTTr potencies with those from high-throughput screening assays and see a high correlation (R2=0.8). This data set and similar ones from other cell types (U2OS and HepaRG) are allowing us to explore pathway-level activity and potency for a large number of chemicals of environmental interest. This abstract does not necessarily reflect US EPA policy.
URLs/Downloads:
DOI: High-throughput Transcriptomics Concentration-Response Screen of 1593 Chemicals in MCF7 Cells
JUDSON SOT 2021 HTTR POSTER V03.PDF (PDF, NA pp, 529.836 KB, about PDF)