Selective inhibition of the COX-2 enzyme is the central tenet underpinning the success of new arthritis medications and likely plays a key role in increased protection against various forms of colorectal cancer. Computational simulations of COX-2 inhibitors will provide a powerful tool to advance the understanding of potency and selectivity at the molecular level. An efficient simulation protocol for selective COX-2 inhibitors will be developed using extended linear response (ELR) theory. Scoring functions will be constructed using physical descriptors derived from Monte Carlo simulations of a three dimensional structural model of the enzyme and inhibitor in solution. The scoring functions will be trained using the wealth of available experimental binding data for celecoxib analogs and will be used to predict binding affinities for newly conceived inhibitors. The effectiveness of the ELR approach will also be compared to other medicinally important systems currently under investigation (e.g., HIV-1 reverse transcriptase). Ultimately, a balance of methodological rigor and computational efficiency is sought to accurately and rapidly predict binding affinities for hundreds of potentially more effective COX-2 inhibitors at a fraction of the cost of combinatorial synthesis and enzyme assaying.
