The war against pain : the design, synthesis and testing of potential COX-2 selective inhibitors.
Much research has focused on the inhibition of the cyclooxygenase (COX) enzyme as this protein is responsible for the first step in the pain pathway in the conversion of arachidonic acid into prostaglandins and thromboxanes. The binding of curcumin and celecoxib, a known cyclooxygenase-2 (COX-2) selective inhibitor, was investigated computationally in order to identify important ligand-protein interactions which would need to be mimicked by a novel COX-2 selective compound. Initial investigations into the binding of curcumin identified the lesser diketone tautomer as having potential COX-2 selective activity. Two novel COX-2 selective compounds were designed using moieties common amongst known COX-2 selective compounds and moieties found in curcumin. Initial docking and binding scores showed that these compounds interacted in a similar manner with the protein as did celecoxib. Modifications to these initial compounds yielded two classes of compounds which explored the impact of the substitutions on the docking and binding scores, the poses and the ligand-protein interactions. All modifications made resulted in enhanced binding towards COX-2, and in a number of cases a reduction in the binding scores for COX-1. Thirty of the 166 compounds designed were selected for synthesis and biological screening as these compounds exemplified the range of changes observed in the full complement of compounds. Retrosynthesis yielded two potential synthetic pathways, and while the first path proved unsuccessful, the second route, which makes use of convenient reaction conditions, afforded the compounds in modest to good yields. Complete NMR spectroscopic analysis was carried out on all compounds, with Diffusion Ordered Spectroscopy used to determine the diffusion coefficients and hydrodynamic radii of two compounds and illustrated the dependence of these measurements on the properties of the medium. NMR Analysis of Molecular Flexibility In Solution (NAMFIS) analysis of one of the final compounds identified six conformers as existing in solution, based on the comparison of experimentally derived Nuclear Overhauser Enhancement (NOE) data with the results from a conformational analysis carried out in silico. Four of the six poses are responsible for >95% of the solution population, with one pose comprising almost 50%. All but one of the poses show Root Mean Squared Deviation (RMSD) values of less than 2 Å when compared to the predicted pose, indicating that any of these poses could bind into the protein. Initial inhibition screening results of the unsubstituted parent benzenesulfonate compound appeared to show three-fold selectivity of COX-2 over COX-1 at 100 nM. Testing of the substituted compounds revealed that these compounds are not COX-2 selective as desired, rather a number show promise as COX-1 selective compounds, with inhibition scores of over 40%, and several other compounds show potential as non-selective COX inhibitors. There is no obvious correlation between the inhibition results and either the Glide XP docking scores or the Prime binding scores, and as such, additional computational analysis as well as experimental testing is required to identify a correlation between the theoretical results and the experimental data, and illustrates that computational results cannot be the sole criterion on which selectivity is judged.