Autotaxin inhibition: development and application of computational tools to identify site-selective lead compounds

Autotaxin (ATX) catalyzes the conversion of lysophosphatidyl choline (LPC) to lysophosphatidic acid (LPA). Both ATX and LPA have been linked to pathophysiologies ranging from cancer to neuropathic pain. Inhibition of LPA production by ATX is therefore of therapeutic interest. Here we report the application of previously-developed, subsite-targeted pharmacophore models in a screening workflow that involves either docking or binary QSAR as secondary filters to identify ATX inhibitors from previously unreported structural types, four of which have sub-micromolar inhibition constants. Cell-based assays demonstrate that ATX inhibition and cytotoxicity structure-activity-relationships (SAR) exhibit selectivity cliffs, characterized by structurally similar compounds exhibiting similar biological activities with respect to ATX inhibition but very different biological activities with respect to cytotoxicity. Thus, general cytotoxicity should not be used as an early filter to eliminate candidate ATX inhibitor scaffolds from further SAR studies. Assays using two substrates of vastly different sizes demonstrate that the tools developed to identify compounds binding outside the central core of the active site did identify compounds acting at an allosteric site. In contrast, tools developed to identify active-site directed compounds did not identify active-site directed compounds. The stronger volume overlap imposed when selecting screening candidates expected to bind outside the active site is likely responsible for the stronger match between intended and actual target site.