You are here:
Development and use of a high-throughput phenotypic profiling assay for bioactivity screening of environmental chemicals
Citation:
Nyffeler, J., C. Willis, K. Friedman, AND J. Harrill. Development and use of a high-throughput phenotypic profiling assay for bioactivity screening of environmental chemicals. Presented at Society for Laboratory Automation and Screening (SLAS) annual conference, Washington, DC, February 02 - 06, 2019. https://doi.org/10.23645/epacomptox.7753688
Impact/Purpose:
Here we describe how we adapted an existing assay that evaluates changes in organelle morphology (Bray et al. (2016)) to use for screening of environmental chemicals. We investigate how in vitro thresholds for bioactivity change across time and biological space and how these in vitro potency estimates relate to potency estimates from in vivo toxicology studies.
Description:
Imaging-based phenotypic profiling is a screening method that combines automated microscopy and image analysis to measure a large variety of features at the single cell level. Here, we describe how we adapted an existing assay that evaluates changes in organelle morphology (Bray et al. (2016)) to use for screening of environmental chemicals. We investigate how in vitro thresholds for bioactivity change across time and biological space and how these in vitro potency estimates relate to potency estimates from in vivo toxicology studies. Six cell lines were plated in 384-well plates and after 24 h treated with 7 concentrations (1/2 log10 spacing, n=3) of test chemicals. After 3 – 48 h of exposure, cells were labeled with MitoTracker (mitochondria), fixed, permeabilized and labeled with Hoechst-33342 (nuclei), SYTO14 (nucleoli) and fluorescent conjugates of concanavalin A (ER), phalloidin (actin cytoskeleton), and wheat germ agglutinin (Golgi/plasma membrane). Confocal images were acquired using an Opera Phenix HCS system and analyzed using Harmony software, yielding ~1300 features per cell. Cell-level data were normalized to DMSO controls using median absolute deviation. Benchmark dose (BMD) modeling was performed on well-level median values with BMDExpress. The threshold for in vitro bioactivity was defined as the lower end of the BMD distribution. In vitro potency estimates were converted to oral equivalent doses using the httk R package and compared to in vivo oral toxicity data available in the ToxValDB database. Assay performance was measured using 14 phenotypic reference chemicals. Distinct patterns of affected features were observed across the chemical set and, were mostly consistent with observations from the literature. For these chemicals, bioactivity thresholds were determined after 3, 6, 12, 24 and 48 h of exposure in U-2 OS cells. For all chemicals, potency estimates varied less than 1 order of magnitude between 6 – 48 h. Testing the same 14 reference chemicals in 6 human cell lines at 48 h of exposure revealed that the potency estimates vary less than 2 orders of magnitudes across cell types. In general, phenotypic profiling BMDs were at least as sensitive as cytotoxicity BMDs, while in some cases phenotypic profiling BMDs were >10x lower than cytotoxicity BMDs. Finally, 79 environmental chemicals were screened in U-2 OS cells and 47 (59%) were found to be active. For active chemicals, 40/47 (85%) had an in vitro potency estimate comparable to or below a corresponding potency estimate from in vivo toxicity studies. These findings indicate that the phenotypic profiling assay could be used for bioactivity screening of environmental chemicals. This abstract does not necessarily reflect USEPA policy.
URLs/Downloads:
DOI: Development and use of a high-throughput phenotypic profiling assay for bioactivity screening of environmental chemicals
SLAS2019_POSTER_V5_NYFFELER_M.PDF (PDF, NA pp, 3773.914 KB, about PDF)