Archived Webinar

Phenotypic profiling for high-throughput chemical screening at the U.S. EPA

Presenter: Johanna Nyffeler, ORISE Fellow with US EPA

Registration is now open:

Presenter: Johanna Nyffeler, ORISE Fellow with US EPA

Thursday, September 24, 2020
10:00AM - 11:00AM ET

The Center for Computational Toxicology & Exposure (CCTE) at U.S. EPA developed a blueprint for utilization of new approach methodologies (NAMs) in computational toxicology. The tiered approach relies on high-throughput profiling assays in the first step in characterizing the biological activity of chemicals. Chemicals are run in concentration-response in the assays for two purposes:

  1. To derive a potency estimate for chemical bioactivity. This in vitro potency estimate can then be used to prioritize chemicals and/or compare the bioactivity to exposure estimates.
  2. To gain information about putative mechanisms-of-actions. Depending on the mechanisms-of-action, different second and third-tier assays may then be performed to confirm hypothesized bioactivity.

CCTE currently has two such profiling assays: High-throughput transcriptomics and imaging-based high-throughput phenotypic profiling (HTPP). In the webinar, I will outline our efforts to achieve these two purposes for HTPP assay.

For HTPP, we are using an assay called ‘Cell Painting’, that labels multiple cellular organelles (nucleus, nucleoli, endoplasmic reticulum, golgi, actin skeleton, plasma membrane, mitochondria) to measure changes in cell morphology in response to chemical treatment. We have operationalized the assay for 384 well plates to evaluate the phenotypic effects of > 1200 chemicals at 8 concentrations in human U-2 OS sarcoma cells. This analysis generates 1300 features per cell. We have derived potency estimates (i.e. phenotype altering concentrations, PACs) for all active chemicals.

For 303 chemicals, in vitro-to-in vivo extrapolation was performed to compare the potency estimates to available in vivo data. For 78% of the chemicals, the HTPP potency was within two orders of magnitude from the in vivo point-of-departure. Moreover, for 72% of chemicals, HTPP was comparable or more conservative than the in vivo point-of-departure.

For each active chemical, a phenotypic profile was derived from the 1300 measured features. Profiles were then compared using Pearson correlation. Among the tested chemicals were 179 chemicals with target annotations in the RefChemDB database. For the retinoic acid, and glucocorticoid receptor pathways, model chemicals activating the same target also display high profile similarity. Moreover, by comparing all tested chemicals to the annotated chemicals, we identified five test chemicals with high profile similarity to retinoids. Of those, four were not previously identified as modulators of the retinoic acid pathway.

To summarize, our results to date indicate that the HTPP assay can be used to derive potency estimates as well as some mechanistic information that can both be used for prioritization of chemicals.

This abstract does necessarily reflect U.S. EPA policy. Mention of trade names is not an endorsement or recommendation for use.