Toward Reducing hERG Affinities for DAT Inhibitors with a Combined Machine Learning and Molecular Modeling Approach.

Area of Science:

• Neuroscience
• Pharmacology
• Computational Chemistry

Background:

• Psychostimulant drugs like cocaine block dopamine transporter (DAT), leading to abuse.
• Atypical DAT inhibitors show promise for treating psychostimulant use disorders by blocking drug-induced behaviors.
• Off-target binding, particularly to the hERG channel, can cause dangerous side effects like ventricular tachycardia.

Purpose of the Study:

• To establish a counter screening platform for DAT and hERG binding.
• To leverage machine learning, experimental validation, and molecular simulations for drug development.
• To identify structural elements influencing DAT and hERG binding for rational drug optimization.

Main Methods:

• Machine learning-based quantitative structure-activity relationship (QSAR) modeling.
• Experimental validation including chemical synthesis and pharmacological evaluation.
• Molecular modeling and simulations to identify key structural determinants of binding.

Main Results:

• Robust QSAR models for DAT inhibitors were established and validated.
• QSAR models demonstrated predictive power even with data subsets from specific experimental approaches.
• Molecular simulations elucidated structural factors behind differential DAT and hERG binding affinities.

Conclusions:

• The developed platform enables the design of safer atypical DAT inhibitors with reduced hERG channel interaction.
• This approach facilitates the rational optimization of lead compounds for treating psychostimulant use disorders.
• The findings support targeting specific protein functional states for improved drug development.