In this paper we describe the
In this paper, we describe the formation of a structure-based pharmacophore which lead to the discovery of several hydrophobic, yet non-lipid inhibitors of ATX. These compounds docked within the same volume occupied by the initial non-lipid inhibitors of ATX used to build the pharmacophore. Violations to Lipinski’s Rule of Five were calculated for each cox inhibitor to filter out compounds that are not drug-like. In order to sample the entirety of the chemical space found by the pharmacophore, compounds were grouped together into clusters based on similarity. Representatives from each cluster were tested for ATX inhibition using two assays, one using a FRET-based substrate, FS-3, and the other using a nucleotide substrate, p-nitrophenyl 5′-thymidine monophosphate. Of the seventy-two compounds tested, four inhibited FS-3 hydrolysis of ATX by 50% or greater at a concentration of 10μM. Sixty-six analogs of one lead were synthesized to explore the structure–activity relationship of this novel scaffold. Thirty-six compounds inhibited ATX-catalyzed FS-3 hydrolysis by 50% or greater at a concentration of 10μM, with one compound having a sub-micromolar potency.
Discussion ATX has lysophospholipase D activity2, 3 leading to specific GPCR activation by its product LPA, The ATX-LPA axis has received interest as a drug target42, 43 due to its role in human disease. Initial headway toward ATX inhibition began with non-specific metal chelators, which inhibited ATX activity when applied in millimolar concentrations and which have numerous off target effects. Once LPA and S1P, both bioactive lipid products of ATX hydrolysis, were identified as feedback inhibitors of ATX, research shifted into inhibitory properties of LPA/S1P analogs.46, 47, 48, 49 Although, some of these compounds had promising in vitro potencies, lipids are unlikely to be orally bioavailable drug candidates according to Lipinski’s rules. Efforts to generate better drug candidate ATX inhibitors resulted in identification of non-lipid inhibitors.35, 49, 50, 41 The first small molecule, non-lipid inhibitor was reported in 2008. Other diverse small molecules were discovered in efforts to improve potency.23, 35, 49, 41, 51, 52, 53 Because protein structure and function are closely related, it is important to understand how molecules bind to ATX in an effort to better design/identify potent inhibitors. The recently published crystal structures of ATX, enzyme inhibition kinetics, and molecular docking studies show distinct binding regions for small molecule ATX inhibitors.26, 27, 28, 29, 30 Small molecule, non-lipid inhibitors are divided between their interactions with either the polar active site30, 41, 51, 52 and the distant hydrophobic domain.28, 29, 30, 41, 54 In the present study, a structure-based pharmacophore was generated for the hydrophobic region of ATX in order to discover novel inhibitor scaffolds which are likely to show selectivity for ATX over other NPP enzymes (NPP6 and NPP7) that lack the hydrophobic pocket. These molecules would also be smaller and more rigid than the relatively large and flexible LPA which would likely prevent their interaction with known LPA GPCR. After searching a large database of nearly 400,000 compounds, seventy-two candidate inhibitors were tested with four showing inhibition of ATX (Table 2). The most potent and synthetically tractable inhibitor, GRI 392104 (IC50 4μM) was used as a scaffold to explore a structure–activity relationship. Sixty-six analogs (Table 1) were synthesized and analyzed for ATX inhibition. This study used two synthetic substrates to differentiate binding modes without the need to solve crystallographic structures as both Albers et al. and Kawaguchi et al. did, which follows a precedent set by Hoeglund et al., Saunders et al., and Fells et al.28, 41, 52, 54, 55 FS-3 (Echelon) is a large substrate with increased fluorescence after hydrolysis by ATX. FS-3 should occupy the entire binding pocket occupied by LPC, the endogenous substrate, stretching from the polar active site down into the hydrophobic pocket. Another substrate, 4-nitrophenyl thymidine-5′-monophosphate (pNP-TMP, Sigma) is synthetic nucleotide which can be hydrolyzed to release 4-nitrophenolate to absorb light at 405nm. This smaller substrate, pNP-TMP, should only reside in the polar active site and not extend into the hydrophobic region. Inhibitors that block FS-3 hydrolysis but not the phosphodiesterase activity of ATX toward pNP-TMP suggests interactions distant from the polar active site deeper in the hydrophobic region of the catalytic domain. However, some analogs may still have positive or negative allosteric effects on pNP-TMP hydrolysis, resulting in either an increase or decrease in ATX activity on that substrate (for instance, 32 inhibits FS-3 hydrolysis 52% but increases ATX-catalyzed hydrolysis of pNP-TMP while 22 inhibits ATX activity for both FS-3 and pNP-TMP to varying degrees).