Identification and Characterization of CD39/Vascular ATP Diphosphohydrolase*

Vascular ATP diphosphohydrolase (ATPDase) is a plasma membrane-bound enzyme that hydrolyses extracellular ATP and ADP to AMP. Analysis of amino acid sequences available from various mammalian and avian ATPDases revealed their close homology with CD39, a putative B-cell activation marker. We, therefore, isolated CD39 cDNA from human endothelial cells and ex- pressed this in COS-7 cells. CD39 was found to have both immunological identity to, and functional characteris- tics of, the vascular ATPDase. We also demonstrated that ATPDase could inhibit platelet aggregation in re- sponse to ADP, collagen, and thrombin, and that this activity in transfected COS-7 cells was lost following exposure to oxidative stress. ATPDase mRNA was present in human placenta, lung, skeletal muscle, kidney, and heart and was not detected in brain. Multiple RNA bands were detected with the CD39 cDNA probe that most probably represent different splicing products. Fi-nally, we identified an unique conserved motif, DLG- GASTQ, that could be crucial for nucleotide binding, activity, and/or structure of ATPDase. Because ATPDase activity is lost with endothelial cell activation, overex-pression of the functional enzyme, or a truncated mu- tant thereof, may prevent platelet activation associated with vascular inflammation. the oxidative ATPDase were performed to determine the di-rect effect of reactive intermediates upon enzymatic function and immunoreactivity of the recombinant CD39/ATPDase. harvested, lysates membrane used for biochemical analysis to determine their effect on stimulated platelet Western blotting.

The quiescent vascular endothelium maintains blood fluidity by inhibiting blood clotting, by the regulation of platelet activation and adhesion, and by promoting fibrinolysis. In large part, the antithrombotic effects are mediated by surface molecules present on resting endothelial cells (EC), 1 such as thrombomodulin and heparan sulfate (1). ECs also have the potential to regulate platelet activation by the synthesis of prostacyclin and nitric oxide and by the surface expression of ATP diphosphohydrolase (ATPDase). This ecto-enzyme (EC 3.6.1.5) has been also described as apyrase, ecto-ATPase, ecto-ADPase, nucleotide phosphohydrolase, or ATP pyrophosphohydrolase (2). ATPDases are generally low abundance proteins that have been difficult to purify, given their sensitivity to detergents and the propensity to co-isolate with other proteins. High levels of ATPDase activity are, however, associated with the vasculature (endothelium, smooth muscle, and cardiac cells), lymphocytes, and platelets (2).
Extracellular tri-and diphosphate nucleosides appear in tissue fluids and plasma as a consequence of lysis of blood cells and tissues and through secretion from platelet-dense granules (3). Vascular ATPDase expressed by quiescent ECs hydrolyzes extracellular ATP and ADP to AMP, which is further converted to adenosine by 5Ј-nucleotidase (3,4). In addition, ADP, which interacts with purinergic P 2t receptors, is a powerful agonist for platelet recruitment, adhesion, and aggregation, whereas adenosine is an antagonist of these processes (5)(6)(7). Thus, the function of ATPDase is critical for the inhibition of platelet aggregation following the hydrolysis of ADP to AMP with the ultimate generation of adenosine (4). We have recently provided evidence for the loss of ATPDase activity following EC activation, such as would occur in vascular inflammatory states or xenograft rejection, where platelet deposition is a consistent component (8,9). Such changes in the level of expression and activity of the ATPDase may, therefore, be of pathogenetic significance and prompted our attempts to identify and study this vascular ecto-enzyme.
The biochemical characterization of ATPDase purified from various tissues and organisms has generally revealed a highly glycosylated protein of molecular mass 70 -100 kDa. This ectoenzyme is Ca 2ϩ -and Mg 2ϩ -dependent, is not sensitive to known inhibitors of the various other ATPases, and hydrolyzes nucleoside tri-and diphosphates but not monophosphates. These enzymatic characteristics have facilitated classification of these proteins into a subgrouping termed E-type ATPases (2).
Recently, ATPDase from human placenta, porcine pancreas, bovine aorta, and chicken gizzard have been purified and partially sequenced (10 -13). Our analysis of these sequence data has shown significant homology to human CD39 ( Fig. 1) (14). CD39 is known to be an acidic glycoprotein with molecular mass 70 -100 kDa that contains two potential transmembrane regions and six potential glycosylation sites. Interestingly, CD39 was originally described as a B-cell activation marker and has been shown to be expressed on the surface of other activated lymphocytes and quiescent vascular endothelium (14). Additionally, CD39 is considered to participate in the enhancement of cell-cell interactions; monoclonal antibodies (mAb) to certain epitopes of CD39 induce homotypic adhesion, probably through involvement of LFA-1 (CD11a/CD18) (15). Potato apyrase (a plant ATPDase) has been recently purified and was independently found to have sequence homology to * This work was supported by Sandoz Pharma, Quebec Heart and Stroke Foundation, and the Natural Sciences and Engineering Research Council of Canada. This is manuscript no. 695 from our laboratory. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) S73813.
¶ Recipient of a studentship award from Fonds pour la Formation de Chercheurs et l'Aide a la Recherche du Quebec and from the Heart and Stroke Foundation of Canada.

MATERIALS AND METHODS
Purification and Sequencing of Mammalian ATPDase-Porcine pancreas and bovine aorta ATPDases were purified according to the methods described previously (11,12). Purified proteins were then sequenced at the Biotechnology Research Institute of Montreal. The sequence of N terminus of the porcine pancreatic ATPDase and five internal sequences from the bovine aortic ATPDase following CNBr cleavage were obtained and were available for our study.
Reverse Transcription-PCR-To amplify a cDNA fragment directly from mRNA, 0.1 g of total RNA was prepared from human umbilical ECs (18) annealed with oligo(dT) 16 primer, and the first strand cDNA synthesis was carried out with Superscript Reverse Transcriptase (Life Technologies, Inc.). The product of this reaction served as a template for PCR amplification.
The following primers were designed based on the CD39 cDNA sequence (14) (GenBank accession no. S73813): forward primer, 5Ј-GAAAGGATCCGAAAACAAAAGCTGCTACT-3Ј; and reverse primer, 5Ј-AGAATCTAGACAGTAAAAGCCAAGGAAGC-3Ј. The DNA fragment of interest was amplified by PCR with the "hot start" technique, according to the Perkin-Elmer protocol. Amplification was carried out for 36 cycles (1 min at 94°C, 2 min at 57°C, and 3 min at 72°C), followed by 10 min at 72°C. The amplified PCR product was examined on 1% agarose gel. Restriction enzyme digestion and sequencing confirmed that the obtained cDNA fragment was CD39. The PCR product was subcloned into the pCRII vector (TA cloning kit; Invitrogen, San Diego, CA); subsequently, plasmid was multiplied in "One Shot" bacteria (Invitrogen), purified, and digested with NotI and SpeI. The insert was purified on glass beads (Bio101, Vista, CA) and then cloned into a mammalian expression vector pCDNA3 (Invitrogen) previously digested with NotI and XbaI. Plasmid DNA was purified with a Qiagen kit (Qiagen, Chatsworth, CA) and checked for the correct orientation of the insert by restriction mapping; this was then followed by confirmation of the correct sequence (Applied Biosystem, model 373A). All restriction enzymes were from New England Biolabs (Beverly, MA).
Transient Transfection-COS-7 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (FCS). Cells were seeded at 5 ϫ 10 5 cells per 30-mm well and transfected 20 -24 h later with Lipofectamine (Life Technologies, Inc.) according to the manufacturer's instructions. Briefly, the cells were exposed to 1 g of DNA (pCDNA3 or pCDNA3-CD39) and 4 l of Lipofectamine in Dulbecco's modified Eagle's medium without FCS for 5 h, followed by the addition of an equal volume of Dulbecco's modified Eagle's medium containing 20% FCS. Twenty-four h after transfection, the culture medium was changed (Dulbecco's modified Eagle's medium/10% FCS), and ϳ70 h posttransfection, COS-7 cells were used for analyses. Control COS-7 cells used for transfection were negative for CD39 as analyzed by Western blotting and FACS with anti-CD39 mAb; ATPDase biochemical activity was negligible.
Cytofluorometric Analysis of the Expression of CD39 -COS-7 cells after transient transfection were harvested by gentle pipetting with Hanks' balanced salt solution (Life Technologies, Inc.) containing 10 mM EDTA. They were washed with buffer (phosphate-buffered saline, 5% FCS, and 0.02% sodium azide) and incubated with either antihuman CD39 mAb, BU61 (IgG 1 ) (Accurate) or an isotype-matched control mAb, anti-CD25/ACT-1 (IgG 1 ) (Dako, Carpinteria, CA) for 30 min on ice. Cells were washed twice and incubated with antimouse IgG conjugated with fluorescein-5-isothiocyanate (Sigma) for 30 min on ice. Finally, cells were washed twice and analyzed by flow cytometry on a FACScan bench top model using Cellquest II software (Becton Dickinson, San Jose, CA). Data were collected from viable cells only, as determined by propidium iodide uptake.
Cell Lysate and Cell Membrane Preparation-Cells were washed three times with Tris-saline buffer, pH 8.0 at 4°C, harvested by scraping in 20 mM Tris, pH 8.0, 50 mM NaCl, 0.1 mM phenylmethylsulfonyl fluoride containing aprotinin (0.02 KIU/ml) and centrifuged 800 ϫ g for 15 min at 4°C. Cells were resuspended in the harvesting buffer and then disrupted in a Potter homogenizer and centrifuged 800 ϫ g for 15 min at 4°C. Supernatants containing cell lysates were used for Western blot analysis. Cell membranes were used for ATPDase activity and platelet aggregation assays. These were prepared by ultracentrifugation of cell lysates at 47,000 rpm for 1 h at 4°C, followed by second centrifugation (57,000 rpm for 50 min at 4°C) of the resuspended pellet. Membranes were resuspended in 7.5% glycerol in 5 mM Tris-HCl, pH 8.0. Protein was measured according to the Bradford method (19).
Western Blotting-SDS-polyacrylamide gel electrophoresis was performed according to Laemmli (20). Proteins (20 g per lane) were separated on a 10% gel under denaturing conditions, transferred to polyvinylidene difluoride membrane (Immobilon P; Millipore, Bedford, MA) by semidry electroblotting and probed with either a rabbit polyclonal antibody to porcine pancreas ATPDase or monoclonal antibody to CD39 (Accurate). Bands were visualized using horseradish peroxidaseconjugated goat antirabbit IgG or goat antimouse IgG (Pierce) and the Enhanced ChemiLuminescence assay (Amersham Life Science, Inc.) according to the manufacturer's instructions.
Polyclonal Antibody to ATPDase-Polyclonal antibody to ATPDase was generated to a peptide sequence corresponding to the N terminus of the purified protein from porcine pancreas (KSDTQETYGALDLGGA) (11). This antibody cross-reacted with ATPDase from bovine aorta, lung, and trachea, as determined by Western blotting.
ATPDase Activity Measured by Hydrolysis of ATP and ADP-COS-7 cell lysates were incubated with either 200 M ADP or 200 M ATP, and Ca 2ϩ -or Mg 2ϩ -dependent release of free phosphate was determined. Malachite green was added to stop the reaction, and absorbance was measured at 610 nm to determine levels of phosphate generation against the standard curve of KH 2 PO 4 (21).
ATPDase Activity Measured by Hydrolysis of [ 14 C]ADP to AMP-Monolayers of transiently transfected COS-7 cells and appropriate control cells were analyzed for the ability to hydrolyze extracellular [ 14 C]ADP (50 Ci/reaction; DuPont NEN) and was carried out on thin layer chromatography (TLC) plates (Whatman Laboratory Division, Clifton, NJ). The solvent system comprised of isobutyl alcohol:1-pentanol:ethylene glycol monoethyl ether:NH 4 OH water at ratios 90:60:180: 90:120. The separated compounds were scanned for radioactivity with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA) and degradation of the [ 14 C]ADP was determined by ImageQuant software according to the manufacturer's instructions.
Platelet Aggregation Assays-Blood was taken after informed consent from apparently healthy human volunteers and anticoagulated with 0.1 volume 3.2% sodium citrate. Platelet rich plasma was prepared by centrifugation of whole blood at 280 ϫ g for 15 min at 22°C. The platelet aggregation assay used a two-sample, four-channel Whole Blood Lumi-Aggregometer, model 560 (Chronolog Corp., Havertown, PA). Platelet-rich plasma was preincubated with COS-7 cell membrane or cell lysate preparations for 10 min at 37°C in a siliconized glass cuvette containing a stirring bar, followed by stimulation with either ADP (5 M), collagen (5 g/ml), or thrombin (0.1 unit/ml) (Chronolog Corp.). Platelet aggregation was recorded for at least 10 min. Data were expressed as the percentage of light transmission with platelet-poor plasma equal to 100%.
Effect of Reactive Oxygen Intermediates on ATPDase Activity-COS-7 cells (control, transfected with pCDNA3 and with pCDNA3-CD39) were exposed for 2 h to either H 2 O 2 (100 M; Sigma) or xanthine oxidase (100 milliunits/ml; Boehringer Mannheim) and xanthine (200 M; Sigma) to directly generate the oxidative stress observed to abrogate EC ATPDase function (8). 2 These experiments were performed to determine the direct effect of reactive oxygen intermediates upon enzymatic function and immunoreactivity of the recombinant CD39/ATPDase. Cells were harvested, and the lysates and membrane preparations were then used for biochemical analysis to determine their effect on stimulated platelet aggregation and were subjected to Western blotting.
Northern Analysis-For tissue mRNA analysis, a Multiple Tissue Northern blot, a charge-modified nylon membrane to which poly(A) ϩ RNA (2 g per lane) from multiple human tissues had been transferred, was used (Clontech Laboratories, San Diego, CA). For the probe, our pCDNA3-CD39 construct was digested with BamHI and XbaI, and CD39 cDNA fragment (1714 base pairs) was purified with the Gene-Clean kit (Bio101) and labeled with [␣-32 P]dATP using Ready-To-Go labeling kit (Pharmacia Biotech Inc.). Prehybridization, hybridization, washes, and stripping of the membrane were carried out with the rapid hybridization protocol from Stratagene (La Jolla, CA). Final washes were at 60°C in 0.2 ϫ sodium saline citrate/0.1% SDS. The blots were exposed to Kodak Biomax Mr film (Eastman Kodak Co.) with intensifying screens at Ϫ80°C for 1 day.

RESULTS
ATPDase Sequence Analysis-We purified ATPDase from pig pancreas and bovine aorta according to the method described earlier and sequenced these proteins (11,12). The obtained sequences originated from the N terminus of the pig pancreas ATPDase and five internal fragments of bovine aorta ATPDase, and all showed remarkable homology to human CD39 (Fig. 1). Fig. 1 also presents previously published partial amino acid sequences of ATPDases purified from human placenta (10) and chicken gizzard (13), which confirm previously unrecognized homology with CD39. The molecular mass of both CD39 and vascular ATPDase is 78 kDa and becomes 54 kDa after removal of Asp-linked oligosaccharides by N-glycosidase (12,15); these data suggest that CD39 and ATPDase may represent the same protein.
Isolation of CD39 cDNA and Its Expression in a Mammalian System-To test the hypothesis that CD39 has ATPDase activity, we isolated CD39 cDNA by reverse transcription-PCR reaction using RNA prepared from human umbilical EC as a template for the first cDNA strand. Primers homologous to CD39 generated PCR product of the expected size, viz. 1714 base pairs. The PCR product was subjected to restriction mapping and then sequenced (data not shown). This confirmed the identity of the PCR fragment with the published sequence of human CD39 (14). The PCR fragment was subcloned into the pCRII vector, excised with NotI and SpeI, and subsequently cloned into a mammalian expression vector, pCDNA3. COS-7 cells were then transiently transfected with pCDNA3-CD39 and used for further analyses. In each experiment, we used nontransfected COS-7 cells and cells transfected with pCDNA3 alone as the controls.
The expression of CD39 on the surface of COS-7 cells was confirmed by flow cytometry and immunocytochemistry. Cells used for flow cytometry analysis were incubated with either mAb to CD39 or to CD25 of the same isotype. Cytofluorometric analysis confirmed the presence of CD39 on the surface of transfected COS-7 cells (Fig. 2). Immunocytostaining confirmed that only cell populations transfected with pCDNA3-CD39, and not the control cells, were stained by the mAb to human CD39 (data not shown).
Immunological Identity between CD39 and ATPDase-To evaluate the identity of CD39 with ATPDase, cell lysates from control and transfected COS-7 cells were analyzed by Western blotting (Fig. 3). Both the mAb to CD39 and a polyclonal antibody to ATPDase recognized a band of the same mobility, with the molecular mass ϳ80 kDa (Fig. 3, A and B, respectively), a finding strongly supporting our hypothesis of identity between CD39 and ATPDase. In addition, the polyclonal antibody recognized a second band around 53 kDa, which could represent a truncated form of the ATPDase nonreactive with the monoclonal antibody to CD39. Consistently, specific CD39 and ATPDase immunoreactivity was detected only in pCDNA3-CD39 transfected cell populations.
ATPDase Activity of CD39 -On the basis of apparent molecular and antigenic similarities between CD39 and ATPDase, we next evaluated selected biochemical functional characteristics of CD39. We incubated COS-7 cell membrane preparations with exogenous ADP or ATP and determined ATPDase activity by measuring divalent cation-dependent phosphate release by standard biochemical techniques. The control COS-7 cells essentially did not hydrolyze ADP or ATP, whereas pCDNA3-CD39 transfected cells exhibited very high levels of ATPDase activity for both substrates (Fig. 4). The ATPDase expressed by CD39 transfected cells did not act on exogenous AMP. In addition, ADP hydrolysis was totally inhibited by 5 mM EDTA, reduced by 60% with 20 mM sodium azide and unaffected by 3 mM ouabain. These inhibitory data are largely consistent with the classification of the vascular ATPDase as an E-type ATPase (2).
We also investigated the activity of expressed enzyme by studying the capacity of intact transfected COS-7 cells to hydrolyze radiolabeled ADP, assayed by thin layer chromatography (TLC). In a representative experiment, transiently transfected cells converted 68% of the 50 Ci of [ 14 C]ADP to [ 14 C]AMP over 20 min, as compared to 11 and 9% for cells transfected with the pCDNA3-only and nontransfected COS-7 cells, respectively, under the identical experimental conditions. Additionally, protein from crude preparations of human umbilical cord vein immunoprecipitated with anti-CD39 also had all of the required characteristics of E-type ATPase. These included specific substrate specificity for ATP and ADP, Ca 2ϩ and Mg 2ϩ dependence, insensitivity to 5 mM tetramizole (an inhibitor of alkaline phosphatase), and inhibition by 20 mM sodium azide by 57 and 47% for ADP and ATP as substrates, respectively.
Effect of CD39/ATPDase on Platelet Aggregation-Because our interest in the vascular ATPDase has concerned the regulation of platelet activation by adenosine nucleotides in the setting of vascular inflammatory disorders, we next investigated the effect of COS-7 cells transfected with pCDNA3-CD39 on platelet aggregation in vitro. Membrane preparations from cells expressing CD39 consistently inhibited platelet aggregation in response to exogenous ADP, whereas membranes from the control cells transfected with pCDNA3-only had no significant effect (Fig. 5A). Comparable specific inhibitory effects were observed when platelets were stimulated by collagen (5 g/ml) or by low levels of thrombin (0.1 unit/ml) in the presence of the transfected COS-7 membranes. Thus, platelet aggregation in response to agonists other than ADP was inhibited only by COS-7 cells expressing CD39 (Fig. 5, B and C). These latter observations also confirm the established dependence of platelet aggregation following collagen and thrombin stimulation upon platelet-released ADP in vitro. Effect of Oxidative Stress on ATPDases Activity-To generate oxidative stress and to evaluate the effect of reactive oxygen intermediates upon the ATPDase function of CD39, COS-7 cells transfected with pCDNA3-CD39 were exposed for 2 h to either H 2 O 2 or xanthine oxidase and xanthine. Cells were harvested, and cell lysates were used for analysis of their effect on platelet aggregation (Fig. 6A). A striking effect of oxidation was observed for the interaction of transfected cells with platelets. After incubation with the xanthine oxidase/xanthine combinations, pCDNA3-CD39 transfected cells almost completely lost their ability to inhibit platelet aggregation, whereas the prior incubations with H 2 O 2 completely abolished the antiaggregatory effect of ATPDase expressed by the transfected COS-7 cells. These results were concordant with the observed loss of ATPDase biochemical activity following oxidative interaction (Fig. 6B) and the reduced immunoreactivity with anti-CD39 monoclonal antibodies (Fig. 6C).
Determination of a Potentially Important Motif in CD39/ ATPDase-Once we had determined identity between CD39 and the ATPDases, the subsequent analysis of available amino acid sequences of the ATPDases purified from different mammalian sources led us to the conclusion that there was only one shared sequence that was common to human placental and the porcine pancreatic ATPDase and also present within human CD39, viz. DLGGASTQ, contained within the N-terminal sequence of the porcine pancreatic ATPDase (Fig. 1).
This sequence was also identified as the putative apyrase conserved region 4 by Handa and Guidotti (16). However, there are some modifications of this sequence for other ATPDases; therefore, this motif may also be presented as (D/E)(L/M/V)GG(A/G)S(T/A/V)Q.
Northern Analysis-To determine tissue-specific CD39/ ATPDase mRNA distribution, we analyzed RNA purified from various human tissues by hybridization with the CD39 cDNA (1714-base pair fragment). We found high levels of CD39 FIG. 5. Anti-platelet aggregatory function of CD39. The activity of newly expressed CD39/ATPDase was determined by analysis of the effect of CD39 transfected and control COS-7 cell membrane preparations (20 g/ml) on platelet aggregation in vitro. Platelets were activated with 5 M ADP (A), 5 g/ml collagen (B), and 0.1 unit/ml thrombin (C). Control nontransfected COS-7 cell membrane preparations had no effect on aggregation (data not shown). PRP, platelet-rich plasma.
FIG. 6. Effect of oxidants on the expressed CD39/ATPDase. COS-7 cells transfected with pCDNA3-CD39 and the appropriate controls were treated for 2 h with either H 2 O 2 (100 M) or xanthine oxidase (100 milliunits/ml) and xanthine (200 M). A, COS-7 cell lysates were then analyzed for their inhibitory effects on platelet aggregation following stimulation with 2.5 M ADP. Xanthine and xanthine oxidase (X/XO) or H 2 O 2 pretreatment abrogated the inhibitory effects of CD39 on platelet aggregation. B, ATPDase biochemical activity was also inhibited following exposure to these oxidative reactions in this representative experiment (data expressed as means; bars, S.D.). C, immunoreactivity of the CD39-transfected COS-7 cell lysates to anti-CD39 was reduced in keeping with loss of ATPDase activities following exposure to X/XO and H 2 O 2 . mRNA in human placenta, lung, skeletal muscle, kidney, and heart but no detectable signal in brain and very little in liver (Fig. 7). We were able to detect as many as five different mobility mRNA transcripts reacting with the same probe, from two up to four of them in the individual tissues. To confirm that these bands were specific for CD39, we hybridized the same blot with 5Ј and 3Ј fragments of CD39 cDNA, 427 and 342 base pairs, respectively. The resulting patterns of hybridization were identical with that one with the full-length probe (data not shown).

DISCUSSION
Our data demonstrate in a persuasive manner that CD39 encodes the vascular ATPDase. This conclusion was originally based on the discovery of sequence homologies between CD39 and human placental ATPDase and bovine aortic ATPDase (Fig. 1). Our hypothesis was further substantiated by other recognized sequence homologies that we noted in mammalian and avian nonvascular ecto-enzymes and the report that the potato soluble apyrase contains certain "apyrase conserved regions" also found in garden pea nucleoside triphosphatase, Saccharomyces cerevisiae golgi guanosine diphosphatase (GDPase), Toxoplasma gondii isoforms of an nucleoside triphosphatase, NTP1 and NTP3, a yeast hypothetical 71.9-kDa protein, a Caenorhabditis elegans 61.3-kDa protein, and human and murine CD39 (16). The following observations also strengthen our conclusion. Both ATPDases and CD39 are known to be membrane glycoproteins with the same molecular weight range. Both proteins have a comparable cellular distribution, are postulated to be involved in cell adhesion, and may undergo up-regulation after viral transformation of certain cells (2,14,15,22).
We were able to generate CD39 cDNA from human umbilical endothelial cells RNA by reverse transcription-PCR. This PCR product, which was of expected size, was then subjected to restriction mapping and sequencing, which confirmed that this product represented true CD39 cDNA (data not shown). The CD39 cDNA was then cloned into the pCDNA3 vector and expressed in COS-7 cells. Using cell membranes or whole-cell lysates, we established that CD39 protein expressed by these transiently transfected cells reacted with both monoclonal antibodies to CD39 and to polyclonal antibodies directed at ATPDase. By FACS analysis using mAb to CD39, we clearly demonstrated that ATPDase was expressed at the surface of COS-7 cells (Fig. 2). Both polyclonal antibody generated by us to the porcine ATPDase N-terminal peptide fragment and cross-reactive with the bovine vascular ATPDase and monoclonal antibody to CD39 detected the appropriate and same mobility band on Western blotting (Fig. 3).
Our functional data show for the first time that the ATPDase activity associated with CD39 expressed by COS-7 cells can hydrolyze the substrate ADP. This specific ADPase enzyme activity can be induced over 100-fold by the COS-7 transfection with CD39. Likewise, we were able to show enzymatic activity of CD39 for the substrate ATP to be induced over 30-fold by COS-7 transfection (394 nmol/mg protein/min in representative experiments). We also have demonstrated that intact transfected COS-7 cells were able to hydrolyze radiolabeled ADP. This latter result obtained by TLC clearly indicated that the active site of ATPDase faces the extracellular milieu. Wang and Guidotti (17) have also established that Epstein-Barr virus-transformed B cells express both CD39 and potentially Ca 2ϩ , Mg 2ϩ apyrase activity. They were able to show by the DEAE-dextran method that COS-7 cells developed 5.4-fold increased ecto-ATPase activity following transfection with CD39 cDNA prepared from B cells, when compared to vector alone (17).
Of potential significance was our observation that preparations of cell membranes from COS-7 cells transfected with pCDNA3-CD39 could inhibit platelet aggregation in response to ADP, collagen, and thrombin (Fig. 5). ADP release from platelet granules is a vital part of the feedback process that amplifies and propagates platelet activation induced by ADP itself or other more potent agonists (3). Hence, ADP may be an important mediator of vascular thrombosis in inflammatory states. It has been suggested that a major role for the ATPDase may be to inhibit ADP or ATP-induced signal transduction in platelets, leukocytes, and vascular endothelium mediated through the purinergic receptors P 2t and P 2y (2,23,24). The hydrolysis of ATP and ADP by ATPDase would remove these purinergic mediators from the extracellular environment and ultimately favor the generation of adenosine (3,4) with the associated anti-inflammatory sequelae related to the interaction with P 1 receptors (2, 4).
CD39 has been shown to play a role in B-cell adhesion, in part related to cellular integrins (14,15). It is further possible that the interaction of adenosine nucleotides with CD39 may also influence cell signaling and integrin affinity for their respective ligands. The identification of CD39 as the vascular ATPDase and the documentation of the significant role in modulating platelet reactivity in vitro will further help to test this hypothesis in several experimental models.
Our observation that exposure of COS-7 CD39 transfectants to reactive oxygen intermediates results in loss of platelet antiaggregatory properties coupled to inhibition of biochemical ATPDase activity is in keeping with the oxidant-dependent loss of ATPDase function noted in association with EC activation in vitro and following vascular injury in vivo (8,9). 2 ATPDase activity has been shown to be lost in vivo with reperfusion injury and this process may be ameliorated by the administration of antioxidants (25). We speculate that this loss, and the resultant decreased capacity to degrade ADP, could play a significant role in the extensive platelet activation and vascular inflammation seen in graft rejection and other forms of vascular injury. Certainly, the intravenous administration of apyrases to experimental animals has been shown to prolong xenograft survival and abrogate the platelet activation and deposition seen in this setting (26).
Because of the variable sensitivity of organs to vascular injury and thrombosis, the tissue-specific distribution for CD39/ FIG. 7. Tissue-specific expression of CD39 mRNA. Identification of mRNA for CD39 in human tissues was undertaken by Northern analysis. Poly(A) ϩ RNA (2 g/ml) from various human tissues was hybridized with CD39 cDNA (A) and glyceraldehyde-3-phosphate dehydrogenase (B) as an internal control (see "Materials and Methods" for additional details).
ATPDase was studied. Northern analysis of RNA extracted from different tissues was, therefore, performed with a full-length CD39 probe (Fig. 7). Among the tissues examined, the strongest signals were observed in certain highly vascularized organs, i.e. placenta, lung, skeletal muscle, and kidney. Heart and liver had lower levels of mRNA transcripts. Two dominant CD39 mRNA transcripts are noted in most tissues akin to the pattern observed when mRNA preparations from cultures of human ECs are studied (data not shown). However, we were also able to detect as many as five different mRNA transcripts reacting with the entire CD39 cDNA probe (Fig. 7) and the 3Ј and 5Ј cDNA regions. These data suggest that these multiple RNA transcripts probably are alternative splicing variants. The pathophysiological significance of this observation is presently undetermined. Interestingly, there was no convincing evidence for the presence of CD39 transcripts in RNA isolated from human brain. This last observation suggests that the specialized vascular tissues of the brain may not express CD39 at levels comparable to the other organs tested and does not explain the data published previously showing the presence of ATPDase activity in nerve tissues and on the external surface of intact synaptosomes (27)(28)(29). Possibly other E-type ATPases or ATPDases unrelated to CD39 are expressed in brain and hepatobiliary tissues, as would be suggested by the finding that ATPase activity may be demonstrated in immunoprecipitated neural cell adhesion molecule from rat brain (reviewed in Ref. 2) and in cell-CAM105 from rat liver (30).
Further investigation and determination of the CD39/ ATPDase ecto-enzymatic active site, putative ATP binding sites (31), and regions sensitive to oxidative reactions by sequential mutagenesis experiments will help elucidate the reason(s) for the potential posttranslational modification or other modulation of ATPDase activity with EC activation. This knowledge should permit us to express CD39/ATPDase in an active form despite EC activation, as we have done for thrombomodulin (32), and to explore the consequences of this intervention in transplantation models associated with vascular inflammation (33).