New QSAR Models to Predict Human Transthyretin Disruption by Per- and Polyfluoroalkyl Substances (PFAS): Development...

Area of Science:

• Environmental Chemistry
• Toxicology
• Computational Chemistry

Background:

• Per- and polyfluoroalkyl substances (PFAS) are widespread environmental contaminants.
• PFAS are suspected of disrupting the thyroid hormone system by binding to human transthyretin (hTTR).
• Experimental data on PFAS-hTTR binding is limited, necessitating predictive models.

Purpose of the Study:

• To develop and validate quantitative structure-activity relationship (QSAR) models for predicting PFAS binding to hTTR.
• To identify structural features of PFAS associated with hTTR disruption.
• To support regulatory risk assessment of PFAS.

Main Methods:

• Development of classification and regression QSAR models using experimental data for 134 PFAS.
• Rigorous validation including bootstrapping, randomization, and external validation to ensure model robustness and predictivity.
• Application of QSARs to the OECD List of PFAS for identifying high-concern compounds.

Main Results:

• Developed QSAR models demonstrated high performance (e.g., classification accuracies of 0.89/0.85, regression R² of 0.81).
• Models possess broader applicability domains than existing QSARs.
• Identified per- and polyfluoroalkyl ether-based, perfluoroalkyl carbonyl, and perfluoroalkane sulfonyl compounds as structural categories of major concern.
• Forty-nine PFAS exhibited stronger binding affinity to hTTR than the natural ligand T4.
• Uncertainty quantification enhanced the reliability of predictions.

Conclusions:

• Novel QSARs provide reliable predictions of PFAS-hTTR binding, addressing data scarcity.
• The models facilitate the identification of hazardous PFAS structures and support risk assessment.
• Implementation in accessible software promotes wider use in research and regulatory contexts.