Abstract

Purpose: To design, synthesize, and test new lead molecules that inhibit both COX-1 and COX-2 forms of the cyclooxygenase (COX) enzyme, with higher affinity for COX-2 than for COX-1.

Methodology: We used the ADMET Design Suite™ (ADMET Predictor™, MedChem Studio™, and MedChem Designer™) and GastroPlus™ software programs [1] with public-domain data [2-19]. Starting with ~550 COX-1 and COX-2 inhibitors, we generated structure-activity relationships (SAR) and identified activity cliffs using MedChem Studio. Using the ADMET Modeler™ portion of ADMET Predictor, we trained artificial neural network ensemble (ANNE) classification and regression models to predict inhibition and selectivity. Novel scaffolds were then generated using de novo design tools in MedChem Studio. Nine virtual libraries based on these scaffolds were created. in silico screening was performed based on predicted activities and ADMET Risk™ scores. GastroPlus™ simulations were run to determine likely dosing regimens that would achieve target plasma concentrations. Eight compounds were selected to be synthesized and tested for COX-1 and COX2 inhibition, and for selected physicochemical and biopharmaceutical properties. Only four were successfully synthesized. Synthesis was performed by the Southern Research Institute (Birmingham, AL) and testing was performed by Eurofins CEREP (Seattle, WA) and Absorption Systems, Ltd. (Exton, PA).

By Robert D. Clark, Michael Lawless, Michael B. Bolger, Walter S. Woltosz