Geometry-encoded molecular dynamics enables deep learning insights into P450 regiospecificity control

Area of Science:

• Biochemistry
• Molecular Biology
• Pharmacology

Background:

• Cytochrome P450 1A2 (CYP1A2) is a key enzyme in drug metabolism.
• CYP1A2 exhibits pluripotency, generating diverse metabolites from single substrates.
• Substrate orientation influences CYP1A2's regioselectivity and metabolic outcomes.

Purpose of the Study:

• To investigate the mechanisms underlying CYP1A2's substrate-dependent regioselectivity.
• To elucidate how substrate binding and orientation dictate metabolite formation.
• To understand the role of active site subsites in controlling caffeine oxidation.

Main Methods:

• Computational modeling, including molecular dynamics simulations.
• Geometric encoding of molecular trajectories.
• Dimensional reduction and differential machine learning techniques.
• Analysis of caffeine oxidation pathways.

Main Results:

• A two-subsite model was proposed, controlling sequential caffeine binding and orientation.
• Substrate exchange between subsites is regulated by a phenylalanine gate.
• Substrate face flipping is hindered within the active site, influencing orientation.
• CYP1A2 regioselectivity arises from local factors, subsite interactions, and pre-orientation.

Conclusions:

• CYP1A2's pluripotency is governed by dynamic interactions within its active site.
• Substrate pre-orientation and subsite-specific controls are critical for regioselectivity.
• Understanding these mechanisms aids in predicting drug metabolism and metabolite profiles.