Combined Modeling Approaches for Assessing Sodium-Iodide Symporter Inhibition.

Area of Science:

• Toxicology
• Computational Chemistry
• Neuroscience

Background:

• The sodium-iodide symporter (NIS) is vital for thyroid hormone synthesis, crucial for brain development.
• NIS inhibition by environmental chemicals can lead to neurodevelopmental disorders like autism and lowered IQ.
• Accurate prediction of developmental neurotoxicity (DNT) is essential for risk assessment.

Purpose of the Study:

• To develop and validate an in silico framework for predicting NIS inhibitors.
• To identify environmental chemicals with DNT potential by modeling NIS inhibition.
• To support next-generation risk assessment strategies.

Main Methods:

• Applied docking-based virtual screening of NIS inhibitors.
• Trained machine learning models (RF, XGB, SVM) using ECFP4 and CDDDs.
• Validated models using 9-fold cross-validation and an internal test set.

Main Results:

• Combined ML and docking predictions improved discrimination (ROC AUC of 0.77).
• Optimal thresholds yielded MCC of 0.32 and balanced accuracy of 0.78.
• A robust framework was developed using 1412 diverse compounds.

Conclusions:

• The study presents a novel, robust computational framework for predicting NIS inhibition.
• This approach enhances the identification of DNT-causing chemicals.
• The developed method represents a new approach for toxicological risk assessment.