Structure-Selectivity Relationship Prediction of Tau Imaging Tracers Using Machine Learning-Assisted QSAR Models and...

Area of Science:

• Neuroscience
• Computational Chemistry
• Pharmacology

Background:

• Tau fibrils are key pathological hallmarks in tauopathies.
• Current imaging agents lack well-understood molecular binding mechanisms.
• Highly selective agents are needed for accurate tau imaging.

Purpose of the Study:

• Investigate the structural basis of selective tau aggregate binding.
• Identify essential molecular moieties for selective binding.
• Facilitate rational design of novel tau imaging agents.

Main Methods:

• Machine learning-based quantitative structure-activity relationship (QSAR) classification.
• Molecular docking and molecular dynamics (MD) simulations.
• MM/PBSA binding free-energy calculations.

Main Results:

• A Random Forest QSAR model achieved high accuracy (96.6% selective, 97.6% non-selective).
• Ligand 63 demonstrated superior binding energy to tau and PHF6.
• Key selective moieties include N-heterocycles, nitrogen atom positions, and tertiary amines.

Conclusions:

• Computational approaches elucidate the structure-selectivity relationship for tau binding.
• Identified molecular features are crucial for designing selective tau imaging agents.
• This knowledge will guide the development of improved positron emission tomography tracers.