(Received for publication, September 5, 1996)
From the Laboratory of Pathology, Division of Clinical Sciences, NCI, National Institutes of Health, Bethesda, Maryland 20892
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
-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.
Autotaxin (ATX)1 is a 125-kDa
glycoprotein secreted by the human melanoma cell line A2058. ATX
stimulates both random and directed motility in its producer cells
(1), and its recent cloning and sequencing (2) has revealed homology
with the active site of bovine intestinal 5-nucleotide PDE (EC
3.1.4.1) (4) and extensive homology with the ectoprotein PC-1 (5), the
brain-type PDE I-nucleotide pyrophosphatase gene 2 (6), and the rat
neural differentiation antigen gp130RB13-6 (7). ATX
contains two tandem somatomedin B regions, the loop region of an
EF-hand and a type I PDE catalytic site, and possesses 5
-nucleotide
PDE activity (2) .
Early studies on digestive enzymes responsible for RNA degradation
identified a class of enzymes characterized by their reaction product,
a 5-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, 10, 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 [-32P]ATP (12).
Previous studies from this laboratory on ATX with point mutations at
the PDE active site showed that the corresponding threonine in ATX
(Thr210) is required for its chemotactic, 5
-nucleotide PDE
and threonine phosphorylation activities, and that
phosphorylation-deficient, 5
-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.
Histone IIA, myelin basic protein, casein,
2-mercaptoethanol, magnesium chloride, sodium chloride, dibasic
potassium phosphate, p-nitrophenyl
thymidine-5-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, Pi,
PPi, and NAD), Amersham Corp. (AMP), and American
Radiolabeled Chemicals (St. Louis, MO) (ATP).
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 ATP-agarose. 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 ElectrophoresisProtein 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 (5The 5-nucleotide PDE
activity of ATX was measured using a modification of the method of
Razzell and Khorana (8). Samples (20 µl) were incubated in 1.8-ml
microtubes in a final volume of 100 µl containing
p-nitrophenyl thymidine-5
-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 µl 0.1 N NaOH. The reaction product
was quantified by reading the absorbance at 410 nm
(A410 × 64 = nmol of
p-nitrophenol).
Various 32P-labeled substrates
([-32P]ATP or [
-32P]ATP (25 Ci/mmol),
[
-32P]GTP or [
-32P]GTP (25 Ci/mmol),
[32P]AMP (3000 Ci/mmol), 32PPi (1 Ci/mmol), or [32P](adenylate) NAD (800 Ci/mmol)), each at
a concentration of 10 µM, were incubated with and without
ATX, at the indicated times and temperatures, in 1.8-ml microtubes
containing 100 µM MgCl2 and 50 mM
HEPES, pH 7.3. For analysis of protein, 20-µl reactions were
terminated by addition of 10 µl of 2 × 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-µl aliquots) from ATX-catalyzed reactions (10 µl) were resolved by ascending TLC on
polyethyleneimine-coated 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
32P-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.
The binding of ATP to ATX was
detected by photoaffinity labeling (16). Samples (10 µl) were
incubated in 1.8-ml microtubes in a final volume of 20 µl containing
100 µM MgCl2, 50 mM HEPES, pH
7.3, and 10 µM [-32P]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
32P-labeled protein bands. No radioactive bands were
detected in samples that had not been irradiated.
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 5-nucleotide PDE
activity under conditions of physiological pH and low substrate
concentrations. The 5
-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 5
-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 µM
[
-32P]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 remains stable at 0 °C; in contrast, phospho-ATX is unstable at
37 °C, displaying a half-life of about 10 min.
rATX displayed similar properties of phosphorylation (Fig.
3A) and dephosphorylation (Fig.
3B). The stability of phospho-ATX at 37 °C in the
presence of SDS (autoradiogram; 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 [-32P]GTP, [32P]AMP,
[32P](adenylate) NAD or 32PPi
(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
[
-32P]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.
The
production of nucleotides and phosphates from ATP by ATX was detected
using ATP that had been labeled with 32P in either the -
or the
-phosphate position. Incubation of rATX with
[
-32P]ATP results in the production of
[32P]ADP and [32P]AMP (Fig.
4, A and B). Identical incubations
were performed with [
-32P]ATP and resulted in the
production of 32Pi and
32PPi (Fig. 4, C and D).
ADP and Pi are produced at more than twice the rate as AMP
and PPi, 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
PPi, 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
[32P]GDP and [32P]GMP from
[-32P]GTP, 32Pi from
[32P]AMP or 32PPi, and
[32P]AMP from [32P](adenylate) NAD (data
not shown).
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
[
-32P]8-azido-ATP in the ATP binding assay (Fig.
5A) or with [
-32P]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 5-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.
In this study we have shown that homogeneously pure rATX catalyzes
5-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 Pi)
and ATP pyrophosphatase (producing AMP and PPi) activities (Fig. 4).
Since both the 5-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
[32P](adenylate) NAD or [32P]AMP (this
report) but not by [
-32P] 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
[
-32P]8-N3-ATP to ATX (Fig.
5A), but that ATP is effective at much lower concentrations
than
-TMP in inhibiting phosphorylation of ATX by
[
-32P]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 5
-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,
5
-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 PPi are susceptible to hydrolysis by ATX
and serve as phosphate donors in its phosphorylation. The hydrolysis of
PPi to Pi 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 Pi, 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 depicts
a 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 5-nucleotide PDE/ATP pyrophosphatase, each of which uses
Thr210 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 5
-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 5-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 5
-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 5
-nucleotide PDE
reaction by ATX is physiologically relevant. [
-32P]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 [
-32P]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
5
-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 5-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 P2 purinoreceptor-mediated signals (34). Also, enzyme catalysis by ATX may provide AMP and/or adenosine to initiate P1 purinoreceptor-mediated signals (33), or it may participate in salvage pathways by facilitating the capture and 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 5-nucleotide 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.
We thank Dr. Elliott Schiffmann for invaluable discussions throughout the course of this work.