Autotaxin (ATX; also known as ENPP2) is one of seven mammalian ectonucleotide pyrophosphatases/phosphodiesterases (ENPPs; also known as NPPs), which hydrolyse pyrophosphate or phosphodiester bonds in a range of extracellular mol- ecules 1 . ATX is unique among the ENPPs, in that it functions as a lysophospho- lipase D (lysoPLD), generating the signal- ling phospholipid lysophosphatidic acid (LPA; monoacylglycerol-3-phosphate) from lysophosphatidylcholine (LPC), an abun- dant plasma phospholipid 2,3 . The closest relatives of ATX, ENPP1 and ENPP3, con- vert ATP into pyrophosphate, a regulator of mineralization and calcification processes in bone and vascular smooth muscle 1,4 . ENPP6 and ENPP7 are choline-specific ectophospholipase Cs, probably serving catabolic rather than signalling roles, and the activities of ENPP4 and ENPP5 have yet to be defined 1 . Originally discovered as an enigmatic ‘autocrine motility factor’ in melanoma cells 5 , ATX has emerged as the key LPA-producing enzyme in plasma and tissues. LPA is a potent extracellular sig- nalling molecule with multiple physio- logical actions, particularly as an inducer of cell migration, proliferation and survival. LPA signals through six distinct guanine-nucleotide-binding protein (G protein)-coupled receptors (termed LPA 1–6 ) found on the surface of many dif- ferent cell types 6–8 .These LPA receptors connect to multiple downstream effector pathways, including those driven by RAS and by RHO and RAC GTPases, the out- comes of which depend on the LPA recep- tor expression profile in a given cell type 6,8,9 (FIG. 1). In addition to its growth factor-like activities, LPA regulates other diverse pro- cesses, including neurite remodelling 10–12 , cytokine production 13 and ion channel activity 14,15 . Studies in mice have revealed that the ATX–LPA signalling axis has a key role in a remarkably wide range of physiological and pathological processes, ranging from vascu- lar and neural development to lymphocyte homing, fibrosis and cancer (BOX 1). But, despite much progress made in understand- ing the action of LPA, the mechanisms by which ATX acts have remained elusive, owing primarily to the lack of structural information about this enzyme. In particu- lar, it has been unclear what determines the unique lysoPLD activity of ATX and how this activity is regulated. Moreover, although it is expected that the levels of bioactive LPA are under tight spatial and temporal control, it is unknown how newly formed LPA is released from ATX and spe- cifically targeted to its receptors. Structural studies have now begun to shed light on these issues but, naturally, also raise new questions and challenges 16,17 . Here, we dis- cuss new insights and inferences from the structure of ATX. We propose that ATX can interact with target cells via specific cell-surface molecules, including integ- rins and heparan sulphate proteoglycans (HSPs), as well as by direct membrane association, to facilitate LPA release near to its cognate receptors. In this way, ATX not only drives the formation of LPA but also ensures specificity in LPA signalling. ATX, a rigid multidomain structure In common with its closest relatives (ENPP1 and ENPP3), ATX has two amino-terminal somatomedin B-like (SMB) domains, which could mediate protein–protein interactions, a central phosphodiesterase (PDE) domain and a carboxy-terminal nuclease-like (NUC) domain that is catalytically inactive (FIG. 2a,b). The crystal structure of ATX shows that the central catalytic PDE domain interacts extensively with the SMB domains on one side and with the NUC domain on the other, an interaction that is reinforced by an N-linked glycan positioned between the two domains and an inter-domain disulphide bridge 16,17 . Furthermore, a large ‘lasso’ loop domain, starting at the end of the PDE domain, wraps tightly around the NUC domain and enters the NUC fold from the opposite side. All of these features serve to maintain the structural rigidity of the ATX catalytic domain (but still allow internal motions within this domain). Thus, it seems that evolution has selected for an extended range of interactions that bind the NUC domain tightly to the PDE domain. OPINION Insights into autotaxin: how to produce and present a lipid mediator Wouter H. Moolenaar and Anastassis Perrakis Abstract | Autotaxin (ATX) is a secreted phosphodiesterase that produces the lipid mediator lysophosphatidic acid (LPA). LPA acts through specific guanine-nucleotide- binding protein (G protein)-coupled receptors to stimulate migration, proliferation, survival and other functions in many cell types. ATX is important in vivo for processes as diverse as vasculogenesis, lymphocyte trafficking and tumour progression. However, the inner workings of ATX have long been elusive, in terms of both its substrate specificity and how localized LPA signalling is achieved. Structural studies have shown how ATX recognizes its substrates and may interact with the cell surface to promote specificity in LPA signalling. We propose that ATX can interact with target cells via specific cell-surface molecules ... to facilitate LPA release near to its cognate receptors. PERSPECTIVES 674 | OCTOBER 2011 | VOLUME 12 www.nature.com/reviews/molcellbio © 2011 Macmillan Publishers Limited. All rights reserved