Evaluation of molecular model-based discovery of ecto-5'-nucleotidase inhibitors on the basis of X-ray structures

The enzyme ecto-5'-nucleotidase (e5NT, CD73), a metallophosphoesterase, is a critical component of adenosine metabolism and signaling and implicated in different disease states. Therefore, attempts have been made to discover inhibitors of e5NT. For example, a virtual screening study using a molecular model of the enzyme has led to the identification of a new series of sulfonamide-containing e5NT inhibitors. The recent availability of several X-ray structures of human e5NT in complex with inhibitors has made it possible to re-evaluate this model building and virtual screening effort. We have assessed the quality of the model in detail and analyzed the question why it was possible to identify a new series of inhibitors on the basis of model-based docking calculations. The model utilized for virtual screening was found to be topologically correct and approach experimental accuracy in the active site region. Two key features within the active site were identified as major determinants for the successfully identification of inhibitors. Taken together, the results rationalize the computer-aided discovery of sulfonamide inhibitors of e5NT and provide further support for the use of carefully built protein models for virtual screening.