Having established the structural requirements for

Having established the structural requirements for potency in both the biochemical and biomarker (CRA) assays, we evaluated key compounds in the metabolic stability assay. Disappointingly, we found that all compounds had high intrinsic clearances in mouse microsomes. To ascertain whether the in vitro clearance correlated to in vivo blood clearance, we evaluated a number of compounds in intravenous PK studies in CD-1 mice. shows that the majority of compounds had high blood clearances when tested in vivo in CD-1 mice Compound , however, had a moderate blood clearance (41mL/min/kg). Scaling of the in vitro intrinsic clearance of , using the well-stirred model and incorporating free fraction, predicted an in vivo hepatic clearance of 25mL/min/kg indicating that the moderate in vivo clearance of was driven by the high protein binding in the mouse (). The same correlation was observed in the rat. Oral bioavailability of was moderate at 41% in Balb-c / mice, and the compound showed a proportional increase in exposure when orally dosed at 10, 30 and 100mg/kg (). We conducted PK studies in SD rats and found that had a reasonable PK profile in this species (Cl=19ml/min/kg, half life=5.4h, =28%). The compound was well tolerated in Balb-c / mice following five days of dosing at 100mg/kg bid. We demonstrated that reduced LPA levels in vivo in MDA-MB-231- tumour bearing Balb-c / mice with an IC of 0.013μM for LPA C18:1 (). The compound was effective in reducing LPA C18:1 levels in vitro in human plasma with an IC of 0.038μM (LPA levels were measured using a UPLC/MS/MS assay, see ). Compound was also shown to inhibit the migration of 4T1 GLPG0634 with an EC of 0.025μM in a LPC/ATX mediated transwell migration assay with recovery of inhibition in the presence of LPA (). The crystal structure of mouse ATX alone and in complex with LPAs with different acyl-chain lengths and saturations has recently been reported. We solved the X-ray crystal structure of bound to mouse ATX (PDB accession code ) and this revealed that the compound interacts with the protein in an unexpected binding mode (). The compound binds to the lipophilic acyl chain pocket of ATX and also to the LPA exit channel. The inhibitor binds approximately 5Å from the catalytic centre in contrast to some ATX inhibitors reported in the literature which bind to the catalytic zinc ions and/or threonine 209 residue., This mode of binding which avoids interaction with the catalytic zinc ions may offer selectivity advantages over other inhibitors. In summary we have identified and optimised a novel series of ATX inhibitors based on the imidazopyridine scaffold. This work led to the discovery of a leading compound with excellent enzyme potency and drug-like properties. The compound potently inhibited the production of LPA in human plasma and has suitable PK properties for in vivo efficacy studies. We are currently evaluating the lead compound (CRT0273750) in ATX/LPA-dependent models of cancer.
Introduction Autotaxin (ATX or ENPP2) is a secreted 100kDa glycoprotein (Stracke et al., 1992). It is a unique member of the ectonucleotide pyrophosphatase/phosphodiesterase (ENPP) family (Stefan et al., 2005), and the only family member with lysoPLD activity, hydrolysing lysophosphatidylcholine (LPC) into lysophosphatidic acid (LPA) (Tokumura et al., 2002, Umezu-Goto et al., 2002) (Fig. 1). LPA is a signalling lipid that activates many cellular pathways by binding to six specific G-protein coupled receptors (LPA1–6) (Noguchi et al., 2009). ATX driven LPA signalling is implicated in vascular and neural development, tumour progression, as well as metastasis, inflammation, fibrotic disease and neuropathic pain (Moolenaar and Hla, 2012, Moolenaar and Perrakis, 2011). Others and we have previously crystallized and determined the 3-dimensional structure of ATX, which shows a compact and robust architecture for a multi-domain protein (Hausmann et al., 2011, Nishimasu et al., 2011). It consists of four domains, two repetitive N-terminal somatomedin-B-like domains, a catalytic phosphodiesterase domain (PDE) and an inert nuclease domain (NUC). The crystal structures of ATX showed that the architecture of the active site around the Threonine 209 (Thr209) nucleophile, including the two zinc ions (Zn1 and Zn2) and their coordination shell, are similar to the closest bacterial homologue of ATX, the nucleotide pyrophosphatase from Xanthomonas axonopodis (XacNPP PDB: 2GSO) (Zalatan et al., 2006). Both enzymes belong to the superfamily of alkaline phosphatases (AP) (Gijsbers et al., 2001, Zalatan et al., 2006), which encompasses a diverse set of metalloenzymes (Galperin et al., 1998), including for example the name-giving member of this superfamily the Escherichia coli alkaline phosphatase (ECAP) (Kim and Wyckoff, 1991), but also the alkaline phosphatase from Sphingomonas (SPAP) (Bihani et al., 2011). ECAP and SPAP preferentially hydrolyse phosphomonoesters (Bihani et al., 2011, Coleman, 1992), while XacNPP preferentially hydrolyses phosphodiesters (Gijsbers et al., 2001, Zalatan et al., 2006), like ATX (Stefan et al., 2005). Phosphate monoester and diester hydrolysis are considered to follow a two-step in-line displacement mechanism, in a dissociative fashion for phosphomonoesterases and in an associative fashion for phosphodiesterases (Cassano et al., 2002, Cleland and Hengge, 1995). During the dissociative reaction mechanism the cleavage of the phosphorus leaving group bond precedes the formation of the new phosphorus-nucleophile bond. In contrast, in the associative mechanism no significant bond cleavage to the leaving group occurs, while the new bond to the nucleophile is established (Lassila et al., 2011, Lopez-Canut et al., 2010). Despite the clear similarity of the active sites between various superfamily members, obvious differences are also detectable. For instance, the active site of ECAP is of tri-metallic nature, with an additionally coordinated Mg2+ ion. It has been suggested, that this Mg2+-ion site, has been evolutionary replaced by Lys171 in SPAP (Bihani et al., 2011).