Autotaxin is an exoenzyme possessing 5'-nucleotide phosphodiesterase/ATP pyrophosphatase and ATPase activities.

Autotaxin (ATX) is an extracellular enzyme and an autocrine motility factor that stimulates pertussis toxin-sensitive chemotaxis in human melanoma cells at picomolar to nanomolar concentrations. This 125-kDa glycoprotein contains a peptide sequence identified as the catalytic site in type I alkaline phosphodiesterases (PDEs), and it possesses 5′-nucleotide PDE (EC 3.1.4.1) activity (Stracke, M. L., Krutzsch, H. C., Unsworth, E. J., Årestad, A., Cioce, V., Schiffmann, E., and Liotta, L. (1992) J. Biol. Chem. 267, 2524-2529; Murata, J., Lee, H. Y., Clair, T., Krutsch, H. C., Årestad, A. A., Sobel, M. E., Liotta, L. A., and Stracke, M. L. (1994) J. Biol. Chem. 269, 30479-30484). ATX binds ATP and is phosphorylated only on threonine. Thr210 at the PDE active site of ATX is required for phosphorylation, 5′-nucleotide PDE, and motility-stimulating activities (Lee, H. Y., Clair, T., Mulvaney, P. T., Woodhouse, E. C., Aznavoorian, S., Liotta, L. A., and Stracke, M. L. (1996) J. Biol. Chem. 271, 24408-24412). In this article we report that the phosphorylation of ATX is a transient event, being stable at 0°C but unstable at 37°C, and that ATX has adenosine-5′-triphosphatase (ATPase; EC 3.6.1.3) and ATP pyrophosphatase (EC 3.6.1.8) activities. Thus ATX catalyzes the hydrolysis of the phosphodiester bond on either side of the β-phosphate of ATP. ATX also catalyzes the hydrolysis of GTP to GDP and GMP, of either AMP or PPi to Pi, and the hydrolysis of NAD to AMP, and each of these substrates can serve as a phosphate donor in the phosphorylation of ATX. ATX possesses no detectable protein kinase activity toward histone, myelin basic protein, or casein. These results lead to the proposal that ATX is capable of at least two alternative reaction mechanisms, threonine (T-type) ATPase and 5′-nucleotide PDE/ATP pyrophosphatase, with a common site (Thr210) for the formation of covalently bound reaction intermediates threonine phosphate and threonine adenylate, respectively.

Early studies on digestive enzymes responsible for RNA degradation identified a class of enzymes characterized by their reaction product, a 5Ј-monophosphate nucleotide, and their activity toward p-nitrophenyl-thymidine monophosphate (⌽-TMP) (8). This type I PDE activity has also been detected in a variety of mammalian tissues, their plasma membranes, and cell surfaces (9 -11). The unifying features of these activities, in addition to the reaction product, are the broad specificity for substrates and competitive inhibitors, the alkaline pH optimum, and the ability to hydrolyze the phosphodiester bond between the ␣and ␤-phosphates in nucleoside polyphosphates. ATX possesses type I PDE activity and also induces a known biological response, the potent stimulation of cellular locomotion; thus it is possible to investigate the role of this enzyme reaction center in extracellular signal transduction.
The reaction mechanism for type I PDE has been described as involving formation of nucleotidylated threonine as a covalently bound reaction intermediate (4), and PC-1 can be autophosphorylated on this threonine at the PDE catalytic center using [␥-32 P]ATP (12). Previous studies from this laboratory on ATX with point mutations at the PDE active site showed that the corresponding threonine in ATX (Thr 210 ) is required for its chemotactic, 5Ј-nucleotide PDE and threonine phosphorylation activities, and that phosphorylation-deficient, 5Ј-nucleotide PDE-competent ATX (K209L) is fully active in the stimulation of cellular motility (3). These findings suggested that the dephosphorylated state of ATX is a biologically active form and prompted us to investigate the relationship between the phosphorylation state and the catalytic properties of ATX. These earlier studies had also shown that phospho-ATX contains the ␥and not the ␣-phosphate from ATP but addressed neither the stability of this construct nor the fate of the ␤-phosphate. In addition, unanswered questions remained concerning the nucleotide reaction products, the ability of ATX to use substrates other than ATP, and the possibility that the phosphorylation of ATX was due to the presence of a co-purifying protein kinase. We have resolved these issues by characterizing the enzymatic activities of ATX using homogeneously pure recombinant ATX (rATX) derived from the human teratocarcinoma cell line N-tera2D1 (13) and partially purified ATX (A2058 ATX) from A2058 human melanoma cells.

EXPERIMENTAL PROCEDURES
Materials-Histone IIA, myelin basic protein, casein, 2-mercaptoethanol, magnesium chloride, sodium chloride, dibasic potassium * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. phosphate, p-nitrophenyl thymidine-5Ј-monophosphate and other nucleotides, and Tris-HCl were from Sigma. HEPES buffer was from Life Technologies, Inc. Electrophoresis buffer was from Bio-Rad. Ethylene glycol was from Fisher. Radioactive materials were from ICN (Costa Mesa, CA) (ATP, 8-azido-ATP, and GTP), DuPont NEN (ATP, P i , PP i , and NAD), Amersham Corp. (AMP), and American Radiolabeled Chemicals (St. Louis, MO) (ATP).
Purification of ATX-The purification of ATX from A2058 cells was performed as described previously (1) through the weak anion exchange step. rATX was purified to homogeneity as follows. ATX cDNA, which included the full-length open reading frame, was subcloned into the plasmid vector pMJ601 (14) and then transfected into vaccinia virus (15). BS-C-1 cells were infected with recombinant virus, and then the culture lysate was collected and filtered with an Easy Flow filter, molecular mass cutoff, 300 kDa (Sartorius), to remove virus particles. The lysate was concentrated on an Amicon ultrafiltration device, a Diaflo YM30 membrane, and then was sequentially fractionated through agarose-bound concanavalin A (Vector Laboratories, Inc., Burlingame, CA) as described (1) and either anion exchange on ZORBAX BioSeries-WAX (MAC-MOD, Chadds Ford, PA) as described (3) or ATPagarose. For the ATP-agarose step active fractions from concanavalin A chromatography were concentrated and dialyzed into T/EG buffer (50 mM Tris-HCl and 20% ethylene glycol), and a 0.75-ml sample of this concentrate was applied to a 2-ml bed of ATP-agarose resin (C-8 linked through a nine-carbon spacer) in a 10-ml Econocolumn (Bio-Rad), which had been equilibrated with 10 volume of T/EG buffer. An additional 6 ml of T/EG buffer was added, and the column was stoppered and gently rocked at 4°C. After 2 h the column was drained, and the resin was washed with an additional 12 ml of T/EG buffer. At this point rATX was eluted from the ATP-agarose resin by the addition of 6 ml of T/EG buffer containing 1 M NaCl, stoppering the column, and rocking it gently at room temperature for an additional 2 h. The column was drained at room temperature and washed with 12 ml of T/EG buffer containing 1 M NaCl. Fractions were tested for motility in chemotaxis assays and for purity by silver stain of an SDS-PAGE gel. The pooled fractions were dialyzed in T/EG buffer and stored at 5°C.
Analytical Gel Electrophoresis-Protein samples were analyzed by SDS/PAGE in a Tris glycine buffer system using precast 8 -16% gradient minigels (Novex, San Diego, CA). Gels were stained using a Daiichi Silver Stain II kit (Integrated Separation Systems, Natick, MA) and dried using an air drying rack system kit (Novex), each according to the manufacturer's instructions.
Type I (5Ј-Nucleotide) PDE Assay-The 5Ј-nucleotide PDE activity of ATX was measured using a modification of the method of Razzell and Khorana (8). Samples (20 l) were incubated in 1.8-ml microtubes in a final volume of 100 l containing p-nitrophenyl thymidine-5Ј-monophosphate at the indicated concentrations and either 50 mM Tris-HCl, pH 8.9, or 50 mM HEPES, pH 7.3. After 20 min at 37°C, reactions were terminated by the addition of 900 l 0.1 N NaOH. The reaction product was quantified by reading the absorbance at 410 nm (A 410 ϫ 64 ϭ nmol of p-nitrophenol  32 PP i (1 Ci/mmol), or [ 32 P](adenylate) NAD (800 Ci/mmol)), each at a concentration of 10 M, were incubated with and without ATX, at the indicated times and temperatures, in 1.8-ml microtubes containing 100 M MgCl 2 and 50 mM HEPES, pH 7.3. For analysis of protein, 20-l reactions were terminated by addition of 10 l of 2 ϫ gel sample buffer (Novex), and products were resolved by SDS-PAGE, silver staining, and autoradiography (XAR film, Eastman Kodak Co.). Where indicated, using the autoradiogram and silver stain to localize the band, radioactivity in individual protein bands in dried gels was quantified directly using a BioScan SpotCount apparatus (BioScan Inc., Washington, DC; inventor's prototype, kindly provided by Richard Braverman, NCI). Nucleotide and phosphate products (2-l aliquots) from ATX-catalyzed reactions (10 l) were resolved by ascending TLC on polyethyleneiminecoated sheets (J. T. Baker, Phillipsburg, NJ) in 0.85 M dibasic potassium phosphate, pH 3.4. Radioactive spots were localized and quantified as described above for 32 P-labeled protein bands in gels. The reaction product concentration in a radioactive spot was calculated as a fraction of the total radioactivity in all spots in the chromatogram. ATX-dependent hydrolysis was calculated as the difference in product concentration in corresponding areas of chromatograms from reactions performed in the presence and absence of ATX.
Photoaffinity Labeling-The binding of ATP to ATX was detected by photoaffinity labeling (16). Samples (10 l) were incubated in 1.8-ml microtubes in a final volume of 20 l containing 100 M MgCl 2 , 50 mM HEPES, pH 7.3, and 10 M [␣-32 P]8-azido-ATP (10 Ci/mmol, ICN). After 90 min at 0°C this ligand was photoactivated by irradiation at 254 nm using a hand-held UV lamp (UVG-54; UVP, Inc., San Gabriel, CA) placed directly over the uncapped tubes for 30 s. Immediately following photolysis, reactions were terminated, and protein products were analyzed as described above for 32 P-labeled protein bands. No radioactive bands were detected in samples that had not been irradiated.

ATX Has 5Ј-Nucleotide PDE Activity at Physiological pH-
Type I PDE enzymes have characteristic alkaline pH optima, whereas the cellular motility-stimulating activity of ATX, which depends on the presence of an intact PDE active site, is expressed at neutral pH. We therefore sought to determine whether ATX displays 5Ј-nucleotide PDE activity under conditions of physiological pH and low substrate concentrations. The 5Ј-nucleotide PDE activity of rATX was assayed at pH 7.3 and 8.9 (Fig. 1). This activity, measured at high concentrations (5 mM) of substrate (⌽-TMP), is 3-fold greater at alkaline than at physiological pH. Measurement of the PDE activity of ATX at substrate concentrations below 1 mM, however, reveals that the reaction velocity measured at pH 7.3 is not significantly different from that detected at pH 8.9. Thus, at physiological temperature and pH and at low substrate concentrations, ATX has readily detectable 5Ј-nucleotide PDE activity.
Autotaxin Catalyzes a Phosphorylation-Dephosphorylation Cycle-The observation (3) that phosphorylation-deficient (K209L) ATX stimulates cellular motility suggested that dephospho-ATX is biologically active and led us to investigate the stability of phospho-ATX. The incorporation of the ␥-phosphate of ATP into A2058 ATX and the dephosphorylation of phospho-ATX at physiological temperature are depicted in (Fig. 2). Samples of A2058 ATX were incubated at 0°C with 10 M [␥-32 P]ATP for the indicated times, and reaction products were analyzed by SDS-PAGE and autoradiography (Fig. 2, A and C).
To detect the dephosphorylation of phospho-ATX, the phosphorylated product of the on reaction was dialyzed at 4°C, divided into aliquots, and further incubated either at 0 or 37°C for the indicated times (Fig. 2, B and C). Silver staining (data not shown) showed equal protein loading, and only one band of labeled protein from each reaction was detected by autoradiography. Phospho-ATX accumulates for about 90 min and re-  Fig. 3B, last lane) indicates that the dephosphorylation of phospho-ATX is an enzymatic activity, requiring the native conformation of ATX. The ATX protein itself is stable under dephosphorylation conditions (silver stain; Fig. 3B, third lane). These results with homogeneously pure rATX demonstrate that the on-off cycle of phosphorylation is an intrinsic property of ATX and is not due to a co-purifying protein kinase and/or phosphoprotein phosphatase.
In addition to this activity toward ATP, ATX can be phosphorylated using [␥-32 P]GTP, [ 32 P]AMP, [ 32 P](adenylate) NAD or 32 PP i (data not shown). In an attempt to detect protein kinase activity in ATX, we have used histone, casein, and myelin basic protein as possible phosphoacceptors. Incubation of each of these proteins with [␥-32 P]ATP resulted in the incorporation of label into histone or myelin basic protein but not into casein in the absence of any exogenous catalyst, whereas the inclusion of A2058 ATX in these reactions resulted in the phosphorylation of ATX as well but had no effect on the endogenous incorporation of label into these proteins (data not shown). These experiments fail to demonstrate protein kinase activity in ATX but leave open the possibility that such activity exists given appropriate cofactors and/or specific substrates.
ATX Has ATPase and ATP Pyrophosphatase Activities-The production of nucleotides and phosphates from ATP by ATX was detected using ATP that had been labeled with 32 P in either the ␣or the ␥-phosphate position. Incubation of rATX with [␣-32 P]ATP results in the production of [ 32 P]ADP and [ 32 P]AMP (Fig. 4, A and B). Identical incubations were performed with [␥-32 P]ATP and resulted in the production of 32 P i and 32 PP i (Fig. 4, C and D). ADP and P i are produced at more than twice the rate as AMP and PP i , respectively. Under the assay conditions used, each of these reaction products accumulates at a constant rate for 60 min (data not shown); the chromatograms shown are from 40-min incubations. These results demonstrate that ATX is able to hydrolyze the phosphodiester bonds in ATP on either side of the ␤-phosphate, which is then contained in either of the reaction products, ADP or PP i , resulting from either ATPase (EC 3.6.1.3) or ATP pyrophosphatase (EC 3.6.1.8) activities, respectively.
The substrates other than ATP that serve as phosphate donors in the phosphorylation of ATX were tested for their susceptibility to hydrolysis by ATX. Incubation with rATX results in the production of [ 32  There Is Competition between Substrates for the 5Ј-Nucleotide PDE, ATP Binding, and Phosphorylation Activities of ATX-ATP binds noncovalently to ATX (3), and ATX uses nucleotides as substrates, but the nature and number of nucleotide binding sites in ATX are not known. The data on enzyme catalysis by ATX presented here can be explained by the existence of a single nucleotide binding site, and this interpretation is supported by experiments showing competition between substrates for the various activities of ATX, as depicted (Fig. 5). Either ⌽-TMP or ATP can compete with [␣-32 P]8-azido-ATP in the ATP binding assay (Fig. 5A) or with [␥-32 P]ATP in the phosphorylation of ATX (Fig. 5B). ⌽-TMP and ATP are comparable in their ability to inhibit ATP binding, whereas ⌽-TMP competes markedly less well than ATP in the phosphorylation assay.
A comparison of the ability of ATP and various ATP derivatives to inhibit the 5Ј-nucleotide PDE reaction is shown in Fig.  5C. The nucleotide analogs, which lack a hydrolyzable phosphate at the ␤-␥ position (AMP-CP and AMP-PCP), are relatively less efficient as inhibitors than those that contain this phosphate (AMP-CPP and ATP), showing that the ability of ATP to inhibit the reaction depends at least partially on the presence of a hydrolyzable phosphate in the ␥ position. DISCUSSION In this study we have shown that homogeneously pure rATX catalyzes 5Ј-nucleotide PDE activity (Fig. 1) under physiological conditions and is indistinguishable from A2058 ATX (purified from a human melanoma cell line), based on the kinetics of threonine phosphorylation and dephosphorylation (Figs. 2 and  3). In addition we have shown that, with ATP as a substrate, ATX has ATPase (producing ADP and P i ) and ATP pyrophosphatase (producing AMP and PP i ) activities (Fig. 4).
Since both the 5Ј-nucleotide PDE and ATP pyrophosphatase activities of ATX hydrolyze the ␣-␤ phosphodiester bond in their respective nucleotide substrates, it is probable that these two activities result from the same reaction mechanism. ATX is labeled by either [ 32 P](adenylate) NAD or [ 32 P]AMP (this report) but not by [␣-32 P] ATP (3; data not shown). It is possible that ATX preferentially hydrolyzes, and incorporates phosphate from, the highest energy phosphoester bond in the substrate, which, in the case of ATP, is the ␤-␥ phosphodiester bond. Such a preference would also explain the observations that ATP and ⌽-TMP are comparable in their ability to compete for the noncovalent binding of [␣-32 P]8-N 3 -ATP to ATX (Fig. 5A), but that ATP is effective at much lower concentrations than ⌽-TMP in inhibiting phosphorylation of ATX by [␥-32 P]ATP (Fig. 5B). Consistent with this possibility is the observation (Fig. 5C) that ATP derivatives that lack a hydrolyzable bond at the ␤-␥ position (AMP-CP and AMP-PCP) are less effective as inhibitors of the ATX-catalyzed 5Ј-nucleotide PDE reaction than derivatives that contain a hydrolyzable bond at this position (AMP-CPP and ATP). The interesting suggestion (17) that there may be competition between the phosphorylation and phosphodiesterase activities of PC-1 may be relevant to these unresolved questions regarding ATX. The simplest interpretation of the competition between substrates for ATP binding, 5Ј-nucleotide PDE, and phosphorylation (Fig.  5) is that a single nucleotide binding site is used by ATX for each of these enzymatic functions, but definitive resolution of this question awaits more extensive enzyme inhibition and nucleotide binding studies.
GTP, NAD, AMP, and PP i are susceptible to hydrolysis by ATX and serve as phosphate donors in its phosphorylation. The hydrolysis of PP i to P i occurs in a number of intracellular energy-conserving reactions (18), but the relationship between these reactions and the inorganic pyrophosphatase activity of ATX is not clear. The predominant products of ATP hydrolysis by ATX in vitro are ADP and P i , but the substrates and products of in vivo catalysis by ATX in the stimulation of tumor cell motility are not known. With the ability to hydrolyze nucleoside polyphosphates at a variety of positions, ATX may catalyze nucleotidase cascades (19,20). ATX hydrolyzes substrates other than ATP, and the facility with which these substrates phosphorylate ATX suggests that in catalyzing each of the various hydrolytic reactions, ATX uses a covalently bound, phosphate-containing reaction intermediate. The data presented in this article strongly suggest that this is indeed the case for the ATPase reaction catalyzed by ATX. Fig. 6  proposed model for the formation of covalently bound reaction intermediates in the catalytic action of ATX toward ATP. ATX is proposed to be capable of at least two alternative mechanisms, ATPase and 5Ј-nucleotide PDE/ATP pyrophosphatase, each of which uses Thr 210 as the site for the formation of the covalently bound reaction intermediate. The phosphothreonine intermediate in the ATPase reaction mechanism (Fig. 6, reaction 1) contains only the ␥-phosphate from ATP and is stable at 0°C and unstable at 37°C (Figs. 2 and 3). The depicted formation of the adenylyl threonine intermediate (Fig. 6, reaction 2) is based on the reported mechanism for 5Ј-nucleotide PDE (4). According to this proposal the phosphorylation-dephosphorylation cycle of ATX is a integral part of the ATPase reaction mechanism, and ATX is atypical among known ATPases (21,22) in that it uses a phosphorylated threonine as a covalently bound reaction intermediate. Unequivocal demonstration of the identity of the phosphorylation-dephosphorylation cycle of ATX with its ATPase activity awaits analysis in progress designed to show that a single point mutation simultaneously abolishes both of these activities. This mutational analysis is also being used to investigate the possibility that the same relationship holds between the other phosphorylation substrates and their hydrolysis by ATX.
Among the proteins with sequence homology to ATX the most well characterized is the ectoprotein PC-1. ATX and PC-1 each contain two tandem somatomedin B regions, the loop region of an EF-hand, and a type I PDE catalytic site and possess 5Ј-nucleotide PDE activity (2,23). Studies on the effect of pH on the PDE activity of PC-1 (12), assayed at a substrate (⌽-TMP) concentration of 0.5 mM, show optimum activity at alkaline pH, a characteristic that is typical of type I PDE enzymes. The 5Ј-nucleotide PDE activity of ATX at submillimolar substrate concentrations (Fig. 1) does not show this preference for alkaline pH. These data suggest that ATX and PC-1 may differ in this respect, and that catalysis of the 5Јnucleotide PDE reaction by ATX is physiologically relevant.
[␣-32 P]ATP has been reported to label purified PC-1 (threonine at the PDE active site) (12) as well as immunoprecipitated or cell surface PC-1 (24). Attempts to label ATX with [␣-32 P]ATP have been unsuccessful (3; data not shown). It is possible that adenylyl ATX, formed during incubation of ATP with ATX, exists only as a short-lived 5Ј-nucleotide PDE/ATP pyrophosphatase reaction intermediate and that its extremely transient nature precludes detection under the conditions and quantities of ATX used. Such a characteristic would also explain the efficiency of this ATX-catalyzed reaction at physiological pH, a property previously unreported among the type I PDE enzymes. The dephosphorylation of phospho-ATX also differs from that of PC-1 in that it occurs after dialysis to remove exogenous nucleotides, which are reported to be stimulatory and necessary for the dephosphorylation of phospho-PC-1 (17).
This and other distinctions in the enzymatic characteristics between PC-1 and ATX may arise, at least in part, from a difference in the sequence of the nucleotide binding site. PC-1 (5) contains the glycine-rich GXGXXG sequence found in nucleotide-binding proteins (25) along with the downstream lysine invariably found in protein kinases (26), and this region may serve as an ATP binding site. Although ATX (2) has extensive homology to PC-1, it does not contain this sequence, nor does ATX contain a perfect match to any of the other P-loop type sequences found in adenine and guanine nucleotide-binding proteins (27). Although the nature of the ATP binding site(s) in PDE enzymes is not yet defined, ATP clearly binds to ATX (3; this report), and both PC-1 (28) and ATX (3; this report) have been purified to homogeneity using ATP-agarose chromatography. The failure to detect protein kinase activity in ATX is not unexpected considering the lack of sequence similarity to known protein kinases.
The discovery of the heterotrimeric G-proteins and the nature of their interaction with adenylyl cyclase (29) revealed a mechanism by which intracellular enzyme catalysis participates in signal transduction, but such a role for extracellular enzyme activity has not been established. Human angiogenin has been reported to have both RNase and angiogenic activities (30); the thymidine phosphorylase activity of platelet-derived endothelial cell growth factor may be responsible for its chemotactic activity (31); antibodies directed against alkaline phosphatase activity have been shown to inhibit cell migration during development of the axolotl pronephric duct (32); and the pertussis toxin-sensitive stimulation of tumor cell motility by ATX requires an intact 5Ј-nucleotide PDE reaction site (3). Together these observations suggest that extracellular enzyme catalysis may also have a role in transmembrane signaling.
Extracellular nucleosides and nucleotides participate in a variety of biological processes, including signal transduction through purinoreceptors (33,34) and nucleoside phosphate and phosphoprotein metabolism by ectoenzymes (35)(36)(37)(38)(39)(40). Adenosine has been shown to promote angiogenesis in the chick egg system (41) and chemotaxis in endothelial (42) and immune (43) cells and to have complex effects on chemotaxis in neutrophils (44). Extracellular nucleotidases such as ATX may serve to deplete ATP and/or ADP as a cytoprotective mechanism (45) or to terminate P 2 purinoreceptor-mediated signals (34). Also, enzyme catalysis by ATX may provide AMP and/or adenosine to initiate P 1 purinoreceptor-mediated signals (33), or it may participate in salvage pathways by facilitating the capture and FIG. 6. Proposal for shared-site catalysis. Two different reaction intermediates are proposed to form at a common site in ATX. Cleavage of the ␤-␥ phosphodiester bond in ATP by ATX involves covalently bound phosphate (from the ␥ position) at Thr 210 (reaction 1) and requires the presence of an adjacent lysine (Lys 209 ). The subsequent hydrolysis of this phosphate from ATX completes the reaction (ATPase). Cleavage of the ␣-␤ phosphodiester bond in ATP by ATX involves covalently bound adenylate (including phosphate from the ␣-position) at Thr 210 (reaction 2). The subsequent hydrolysis of this adenylate from ATX completes the reaction (ATP pyrophosphatase). reuptake of nucleosides (46). Cell adhesion molecule 105 has been identified as an ecto-ATPase with implications for cell-cell interaction (47), and a rat liver ecto-ATPase has been identified as a canalicular bile acid transport protein (48). Since phosphorylation-deficient ATX (K209L) is biologically active (3), the ATPase activity of ATX may be dispensable for the stimulation of cellular motility. On the other hand, the stability of phospho-ATX in vivo is unknown, and since the dephospho form of ATX is apparently an active state, the possibility of a regulatory role for the phosphorylation of ATX is not excluded.
Continuing investigations on autotaxin are designed to test the hypothesis that the phosphorylated forms of ATX are enzyme-bound reaction intermediates in the hydrolysis of phosphoester bonds and to study the relationship between the 5Јnucleotide PDE/ATP pyrophosphatase activity of ATX and its stimulation of cellular motility, as well as the influence of the phosphorylation state and ATPase activities on these properties of ATX.