You are here:
Gene Signature Concentration-Response Modeling of High-Throughput Transcriptomics (HTTr) Data for Mechanistic Prediction and Potency Estimation
Citation:
Harrill, J., L. Everett, D. Haggard, J. Bundy, L. Taylor, B. Chambers, C. Willis, I. Shah, AND R. Judson. Gene Signature Concentration-Response Modeling of High-Throughput Transcriptomics (HTTr) Data for Mechanistic Prediction and Potency Estimation. Society of Toxicology 61st Annual Meeting and ToxExpo 2022, San Diego, CA, March 27 - 31, 2022. https://doi.org/10.23645/epacomptox.19400294
Impact/Purpose:
Poster presented to the Society of Toxicology 61st Annual Meeting and ToxExpo March 2022. The US EPA Center for Computational Toxicology and Exposure has research programs focused on developing the tools, approaches and data needed to accelerate the pace of chemical risk assessment and foster incorporation of non-traditional toxicity testing data into regulatory decision-making processes. High-throughput transcriptomics (HTTr) with TempO-Seq is a promising technology for comprehensive bioactivity screening of chemicals. We have developed robust laboratory and bioinformatics workflows for generating HTTr data on thousands of chemicals in concentration-response. Information gleaned from these assays includes molecular point of departures representing the threshold for perturbation of cellular biology and transcriptional profiles that can be used to identify chemicals that act through similar mechanisms of action. This information would be of interest to scientists using or contemplating the use of new approach methodologies (NAMs) for next generation risk assessment (NGRA).
Description:
US EPA has developed a tiered hazard evaluation strategy using new approach methods (NAMs) as part of an effort to incorporate non-traditional toxicity testing into regulatory decision-making processes. The first tier of this strategy specifies the use of high-throughput profiling (HTP) to rapidly evaluate the biological activity of chemicals in human-derived cell models. High-throughput transcriptomics (HTTr) using the whole transcriptome TempO-Seq assay has been identified as an HTP assay that can be used to: 1) identify molecular points of departure (mPODs) for perturbation of cellular biology by chemicals and 2) generate gene expression profiles for mechanism of action prediction. Previously, we developed laboratory workflows for conducting large scale HTTr screens in 384-well format and data analysis pipelines for mPOD determination based on concentration-response (CR) modeling of transcriptomic signature enrichment scores. Here, we applied those methods to screen 1784 chemicals in MCF7 adenocarcinoma cells (6 hr exposure, 8 concentrations up to 100 uM) using a collection of 11,037 signatures gathered from public sources (Connectivity Map (CMAP), Molecular Signatures (MSigDB), BioPlanet, DisGeNET and others). The R package tcplfit2 was used to conduct the CR modeling. Overall, 99% of chemicals had at least one active signature as defined using tcplfit2 outputs with “hitcall” >= 0.9, “top_over_cutoff” >= 1 and a benchmark concentration (BMC) less than the highest concentration tested. There were 148 chemicals with mPOD (defined as the 5th percentile BMC of active signatures) less than 1 uM. The distribution of signature BMCs for a subset of these chemicals displayed a peak of biological activity at low test concentrations, followed by a second, larger peak of biological activity at higher concentrations. Chemicals behaving as such tended to be those with known specificity for a molecular target that is expressed in MCF7 cells and included estrogen receptor agonists, selective estrogen receptor modulators (SERMs), glucocorticoid receptor agonists and retinoic acid receptor agonists. Hierarchical clustering based on signed area under the curve (sAUC) clearly separated these chemical classes from one another and also identified a set of polycyclic aromatic hydrocarbons (PAHs) as having similar biological activities. This abstract does not represent USEPA policy.
URLs/Downloads:
DOI: Gene Signature Concentration-Response Modeling of High-Throughput Transcriptomics (HTTr) Data for Mechanistic Prediction and Potency Estimation
SOT2022_JHARRILL V5.PDF (PDF, NA pp, 1263.887 KB, about PDF)