Building predictive in vitro pulmonary toxicity assays using high-throughput imaging and artificial intelligence

Area of Science:

• Toxicology and Pharmacology
• Computational Biology and Bioinformatics
• Cell Biology

Background:

• Human lungs are vulnerable to xenobiotic toxicity, but understanding direct cellular effects of pulmonotoxic chemicals is challenging.
• A universal in vitro assay for predicting chemical-induced lung toxicity is currently lacking.
• Existing assays, like cell-viability tests, have limitations in accurately predicting in vivo pulmonotoxicity.

Purpose of the Study:

• To develop and validate a novel in vitro assay for predicting pulmonotoxicity using high-throughput imaging and artificial intelligence.
• To identify human lung cell lines and phenotypic features most predictive of in vivo pulmonotoxicity.
• To establish a more accurate and efficient method for assessing chemical safety for human lungs.

Main Methods:

• Development of the High-throughput In vitro Phenotypic Profiling for Toxicity Prediction (HIPPTox) assay.
• Utilizing high-throughput imaging and AI to analyze quantitative phenotypic features in human lung cell lines.
• Selection of the BEAS-2B cell line and two specific phenotypic features for the final assay model.

Main Results:

• The HIPPTox assay, using BEAS-2B cells, accurately classified 33 reference chemicals (88.8% balanced accuracy), outperforming standard cell-viability assays (77.1%).
• HIPPTox demonstrated high sensitivity (84.6%) and specificity (93.0%) in predicting human pulmonotoxicity.
• Experimental validation confirmed that predicted pulmonotoxic chemicals induce DNA damage and activate the DNA-damage response (DDR) pathway.

Conclusions:

• HIPPTox provides a robust and accurate in vitro method for predicting chemical-induced lung toxicity.
• The assay effectively identifies common modes of action for pulmonotoxic chemicals, such as DNA damage.
• HIPPTox has the potential to be adapted for other organ-specific toxicity assessments, accelerating safety testing.