Exploring potency and selectivity receptor antagonist profiles using a multilabel classification approach: the human...

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Pharmacology

Background:

• Adenosine receptors (ARs) are key targets in medicinal chemistry, with significant interest in subtype selectivity for therapeutic success.
• Predicting AR subtype selectivity for ligands remains a challenge, with limited ligand-based strategies available.
• Developing selective ligands is crucial for achieving therapeutic efficacy and minimizing off-target effects.

Purpose of the Study:

• To develop and validate a novel multilabel classification approach for predicting adenosine receptor subtype potency and selectivity.
• To combine autocorrelated molecular descriptors of molecular electrostatic potential (autoMEP) with support vector machines (SVMs) for this prediction.
• To create quantitative sieves for simultaneous prediction of potency profiles across human AR subtypes (hA1R, hA2AR, hA2BR, hA3R).

Main Methods:

• Application of a multilabel classification approach using autoMEP descriptors and SVMs.
• Generation of three predictive models based on decreasing potency thresholds.
• Validation of model robustness and reliability using an internal test set.
• Application of the developed strategy to newly synthesized compounds.

Main Results:

• Successful development of three robust multilabel classification models for predicting AR subtype selectivity.
• Accurate prediction of the potency profile and selectivity for a large set of known inverse agonists (>500 compounds).
• Inference of the full adenosine receptor potency spectrum and selectivity profile for 13 novel pyrazolo-triazolo-pyrimidine derivatives.

Conclusions:

• The developed autoMEP-based multilabel classification strategy offers a powerful tool for predicting AR subtype selectivity.
• This approach facilitates the design and prioritization of novel AR-targeting drug candidates.
• The method provides valuable insights into structure-activity relationships for AR ligands.