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

Area of Science:

• Toxicology
• Computational Biology
• Genomics

Background:

• High-throughput transcriptomics (HTTr) profiles gene expression to assess chemical biological activity in vitro.
• Previous studies screened 44 chemicals; this work expands screening to 1,751 additional chemicals from the EPA's ToxCast collection.

Purpose of the Study:

• To evaluate the utility of concentration-response modeling of signature scores for identifying molecular targets and clustering chemicals by bioactivity.
• To predict mechanisms of action (MeOAs) and group chemicals with similar activity profiles using HTTr data.

Main Methods:

• Screened 1,751 chemicals using HTTr in MCF7 cells across 8 concentrations with a 6-hour exposure.
• Performed clustering and enrichment analyses based on signature catalog annotations and ToxPrint chemotypes.
• Utilized Uniform Manifold Approximation and Projection (UMAP) for embedding and novel modulator identification.

Main Results:

• Identified known and predicted MeOAs, including estrogen receptor (ER), glucocorticoid receptor (GR), and NRF2/KEAP/ARE pathway modulation.
• Demonstrated HTTr's ability to stratify chemicals by ER agonist potency and distinguish agonists from antagonists.
• UMAP identified novel ER modulators and explored bioactivity of structurally related chemicals when combined with ToxPrint chemotype enrichment.

Conclusions:

• HTTr in MCF7 cells is effective for predicting chemical mechanisms of action and grouping compounds by bioactivity.
• This approach can inform chemical risk assessment by establishing in vitro points of departure and predicting MeOAs.
• The study highlights the value of transcriptomics for understanding chemical interactions and predicting toxicological profiles.