Inhibition and substrate recognition--a computational approach applied to HIV protease

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Developing effective inhibitors for HIV protease is crucial for antiviral therapy.
• Accurately predicting inhibitor strength computationally can accelerate drug development.
• Understanding specific ligand-protein interactions is key to designing potent drugs.

Purpose of the Study:

• To develop and validate a computational method for predicting inhibitor strength against HIV protease.
• To identify a minimal set of critical amino acid residues involved in inhibitor binding.
• To gain insights into the molecular interactions driving HIV protease inhibition.

Main Methods:

• A computational approach using interaction energies between inhibitors and a consensus HIV protease structure.
• Application of a genetic algorithm to identify key amino acid residues and optimize predictive models.
• Docking of inhibitors into the consensus structure derived from X-ray crystallographic data.
• Validation of the predictive model using external datasets.

Main Results:

• The developed method achieved a q2 value of 0.63 for internal validation.
• External validation demonstrated prediction accuracy between 0.9 and 1.5 log10 units.
• The models identified crucial subsets of residues (9-20) out of 198 total residues involved in inhibition.
• These key residues are distributed across the enzyme's subsites.

Conclusions:

• The computational approach effectively predicts HIV protease inhibitor strength.
• The identified minimal sets of residues provide valuable insights into essential drug-target interactions.
• This method facilitates quantitative prediction and can guide the design of novel HIV protease inhibitors.