A structure-based model for predicting serum albumin binding.

Area of Science:

• Pharmacokinetics and Drug Metabolism
• Structural Biology
• Computational Chemistry

Background:

• Compound distribution and metabolism are influenced by binding to human serum albumin (HSA).
• Existing experimental and QSAR methods have limitations in accurately predicting binding affinity for novel compounds.
• Albumin's flexibility and promiscuity present challenges for binding prediction.

Purpose of the Study:

• To develop a structure-based model for predicting compound binding affinity to human serum albumin.
• To improve the accuracy of HSA binding predictions for novel chemical entities.
• To complement existing experimental and computational approaches for assessing drug-protein interactions.

Main Methods:

• Developed a structure-based computational model for HSA binding.
• Incorporated albumin's inherent flexibility and promiscuity into the prediction approach.
• Utilized a weighted combination of predicted logP and docking scores to differentiate binders from nonbinders.

Main Results:

• The developed model accurately distinguished between compounds that bind and do not bind to HSA.
• A weighted combination of predicted logP and docking score proved most effective for classification.
• The model demonstrated successful prediction of HSA binding in a large test set of therapeutics with experimental data.

Conclusions:

• A novel structure-based model effectively predicts human serum albumin binding affinity.
• The model's accuracy is enhanced by considering albumin's flexibility and using a combination of logP and docking scores.
• This computational tool offers a valuable method for assessing drug candidates' potential for HSA binding.