Integrating deep learning and molecular dynamics simulations for FXR antagonist discovery

Area of Science:

• Biochemistry
• Pharmacology
• Computational Chemistry

Background:

• Farnesoid X receptor (FXR) is crucial for bile acid, lipid, and glucose balance.
• FXR antagonists offer therapeutic potential for cholestasis, metabolic disorders, and cancers.
• Lack of approved FXR antagonists necessitates novel drug discovery approaches.

Purpose of the Study:

• To develop deep learning models for predicting FXR antagonistic activity and toxicity.
• To screen human metabolites for potential FXR antagonists.
• To identify novel FXR antagonist candidates for metabolic disease treatment.

Main Methods:

• Developed deep learning models for predicting FXR antagonistic activity (ANTCL) and toxicity (TOXCL).
• Screened 217,345 compounds from the HMDB database.
• Utilized molecular dynamics simulations and binding free energy calculations.

Main Results:

• Identified eleven human metabolite candidates with significant FXR binding potential.
• Five complexes showed enhanced stability compared to the reference Gly-MCA.
• HMDB0253354 (Fulvestrant) and HMDB0242367 (ZM 189154) exhibited favorable binding free energies.
• Hydrophobic interactions involving MET328, PHE329, and ALA291 were key to complex stability.

Conclusions:

• Deep learning is effective for discovering FXR antagonists.
• HMDB0253354 and HMDB0242367 are promising candidates for metabolic disease therapies.
• Further investigation of these compounds could lead to new treatments.