Thromboregulation by Endothelial Cells
Significance for Occlusive Vascular Diseases
Abstract—During their 7- to 9-day lifespan in the circulation, platelets perform an ill-defined baseline function that maintains the integrity of the vasculature. In thrombocytopenic states, there is an increase in vascular permeability and fragility, which is presumably due to absence of this platelet function. In sharp contrast, biochemical or physical injury in the coronary, carotid, or peripheral arteries induces platelet activation and platelet recruitment, which can progress to thrombotic vascular occlusion. Because there is 1 death every 33 seconds from vascular occlusion in the United States, this problem has critical public health implications. In this review, we describe the characterization of a novel potential antithrombotic agent with a unique mode of action—biochemical “deletion” of ADP from an activated platelet releasate, which thereby inhibits platelet recruitment and further activation.
- Received October 11, 2000.
- Accepted November 20, 2000.
Several years ago, in the course of studying interactions between cells of the vascular wall and those in the circulation, we demonstrated that platelets in proximity to endothelial cells do not respond to any added agonist in vitro. Experiments initiated in the late 1980s cumulatively led to the conclusion that endothelial cell CD39, an ecto-ADPase, was mainly responsible for this phenomenon. CD39 rapidly and preferentially metabolizes ADP released from activated platelets. ADP is the final common pathway for platelet recruitment and thrombus formation, and platelet aggregation and recruitment are abolished by CD39. Our present working hypothesis is that CD39 represents a novel antithrombotic agent for treating high-risk patients who have activated platelets in their circulation, the identifying characteristic of coronary artery occlusion and thrombotic stroke.
We generated a recombinant soluble form of human CD39 (solCD39), a glycosylated protein of 66 kDa whose enzymatic and biological properties are identical to the full-length form of this enzyme. In our in vitro experiments, solCD39 blocks ADP-induced human platelet aggregation and inhibits collagen-induced and thrombin receptor agonist peptide (TRAP)-induced platelet reactivity. We also studied solCD39 in vivo in a murine model of stroke, which was shown to be driven by excessive platelet recruitment. In studies with CD39 wild-type (CD39+/+) mice, solCD39 completely abolished ADP-induced platelet aggregation and strongly inhibited collagen- and arachidonate-induced platelet reactivity ex vivo. When solCD39 was administered before transient intraluminal middle cerebral artery occlusion, it reduced ipsilateral fibrin deposition, decreased 111In-platelet deposition, and increased postischemic blood flow 2-fold at 24 hours. These results were superior to those we obtained with aspirin pretreatment.
CD39 null (CD39-/-) mice, which we generated by the deletion of exons 4 to 6 (apyrase-conserved regions [ACRs] 2 to 4), have a normal phenotype and normal hematologic profiles and bleeding times, but they exhibit a decrease in postischemic perfusion and an increase in cerebral infarct volume when they are compared with genotypic CD39+/+ control mice in our stroke model. “Reconstitution” of CD39 null mice with solCD39 reversed these pathological changes. Thus, the CD39-/- mice were actually rescued from cerebral injury by solCD39, thereby fulfilling Koch’s postulates.1 These experiments have led us to hypothesize that solCD39 has potential as a novel therapeutic agent for thrombotic stroke.
Interactions Between Platelets and Vascular Endothelium
Thrombosis is a multicellular process.2 To gain further insight into the pathogenesis of vascular occlusion, cell-cell interactions must be studied in detail. Cell-cell interactions and interactions between blood cells and the vessel wall are critical components of the hemostatic and thrombotic process. Several of such reactions occur via “transcellular metabolism,” a locution we developed to indicate reciprocal or collaborative metabolic steps initiated by signaling molecules from different cell types.3 These steps are particularly relevant with regard to vascular endothelial cells and platelets. We have demonstrated that endothelial cells in proximity to platelets in motion will downregulate their activity via at least 3 different metabolic pathways. The first involves a cell-associated aspirin-insensitive nucleotidase, ecto-ADPase/CD39.4 The second system consists of a short-lived fluid-phase inhibitory signaling pathway involving eicosanoids such as prostacyclin (PGI2) and prostaglandin D2.5 The third is the NO system, which is also a fluid-phase autacoid that inhibits platelet reactivity by elevating levels of cGMP.6 We termed these substances “thromboregulators.” When PGI2 and NO activities are blocked by aspirin and hemoglobin, respectively, platelet inhibition is solely due to the metabolism of ADP in the platelet releasate after agonist-induced platelet reactivity. Under these conditions, metabolism of ADP by CD39 results in the loss of platelet responsiveness, release, recruitment, and aggregation.4 It is important to emphasize that platelet inhibition by CD39 is unique because there is no other consequent blockade of platelet function per se; the deletion of ADP from the activated platelet releasate is solely responsible for the abolition of platelet recruitment. This is the modus operandi of CD39 as an inhibitor of platelet aggregation and recruitment.
Characterization of the Human Endothelial Cell Ecto-ADPase as CD39
Our initial studies of the platelet inhibitory activity of endothelial cells during the 1970s and 1980s led to the conclusion that this property was due to endothelial cell eicosanoid and/or NO production. This early hypothesis was tested in 1990 by incubating aspirin-treated human umbilical vein endothelial cells (HUVECs) with radiolabeled ADP. Under these experimental conditions, PGI2 could not form, and any NO that was generated either rapidly decayed or was blocked by our addition of purified oxyhemoglobin to the system. The added ADP and any metabolites formed were separated and identified by radioactive thin-layer chromatography (Figure 1⇓). Under these experimental conditions, AMP accumulated transiently and was further metabolized to adenosine, followed by deamination to inosine and hypoxanthine. Importantly, these experiments also demonstrated that cell-free supernatants from HUVECs incubated with [14C]ADP could not induce aggregation in platelet-rich plasma because the ADP had been metabolized by a component of the endothelial cell surface, presumably the ecto-nucleotidase system.4
We now know that the molecule primarily responsible for platelet inhibition as described above is indeed an ADPase. Specifically, it is a membrane-associated ecto-nucleotidase of the E type.7 8 This enzyme has several interesting characteristics, which include the following: it is dependent on calcium (although magnesium can partially substitute) but not affected by specific inhibitors of P-, F-, and V-type ATPases, and it metabolizes ATP and ADP but not AMP. These biochemical activities identify the HUVEC enzyme as an apyrase (ATP diphosphohydrolase [ATPDase]), EC184.108.40.206,7 now classified as E-NTPDase-1.8
In 1996, Handa and Guidotti9 purified a soluble apyrase from potato tubers and cloned its cDNA. Sequence analysis revealed 25% amino acid identity and 48% amino acid homology with human CD39.9 Maliszewski et al10 had cloned CD39 as a cell-surface glycoprotein expressed on activated B cells. It was also present on NK cells and subsets of T cells as well as on some preparations of HUVECs.11 Interestingly, nucleotidases with homology to CD39 and potato apyrase are expressed throughout the animal and vegetable kingdoms, in species as varied as the garden pea, Caenorhabditis elegans, and Toxoplasma.9 At least 4 regions within these molecules demonstrate extraordinary homology and are designated ACRs.9
Several observations have indicated that the HUVEC ADPase is identical to CD39.12 More than 95% of the ADPase activity from a preparation purified from HUVEC membranes can be immunoprecipitated with several anti-human CD39 antibodies. Confocal microscopy and indirect immunofluorescence studies localized CD39 to the cell surface of HUVECs. When COS cells are transfected with a vector containing cDNA from either human or murine CD39, the expression of CD39 and ecto-ADPase activity are demonstrable immunologically and biochemically, respectively, on the COS cell surface. Polymerase chain reaction (PCR) analyses using either authentic human CD39 cDNA or cDNA synthesized from HUVEC mRNA will result in products of identical size for each of 4 different CD39-specific primer pairs. When these PCR products were sequenced, their complete identity was confirmed. The PCR products encompassed 75% of the coding region of CD39,12 including the original 4 ACR apyrase domains (Figure 2⇓). Furthermore, Northern analyses demonstrated that HUVEC and MP-1 cells (from which CD39 was originally cloned) contain the same sized messages for CD39.12 Other laboratories have reported ATPDases from different cell sources.13 14
When COS cells were transfected with human or murine CD39 cDNA, they blocked ADP-induced platelet aggregation.12 We found that the transfectants metabolized ADP to AMP within 3 minutes. This time frame correlates with the events leading to the formation of a hemostatic platelet plug or thrombus. This event parallels the time course for platelet inhibition by cells expressing CD39 and is also commensurate with the respective ADPase activities of these cells (Figure 3⇓). The data emphasize the importance of CD39 as a thromboregulator and represent the first direct demonstration of a physiological function for CD39 as an ADPase: blockade of platelet responsiveness to the prothrombotic agonist ADP via its metabolism to AMP. This phenomenon might represent the evolution of an endothelial mechanism targeted toward metabolism of prothrombotic platelet-derived nucleotides, thereby controlling excessive platelet accumulation as well as maintaining blood fluidity.
Recombinant Soluble CD39/Ecto-ADPase
The biochemical and biological properties of CD39 suggest an entirely new strategy for therapeutic intervention in platelet-driven occlusive vascular diseases. Although treatment of patients with aspirin inhibits the prothrombotic action of thromboxane, it simultaneously prevents the formation of the antithrombotic eicosanoid, PGI2. This represents a limitation in the therapeutic use of aspirin. Moreover, aspirin is a nondiscriminating acetylating agent, as we had demonstrated in 1970,15 and its use is accompanied by undesirable side effects. CD39 is aspirin independent and inhibits platelet reactivity even when eicosanoid formation and NO production are blocked. Importantly, the action of CD39 is not on the platelet itself; rather, it metabolizes ADP appearing in the activated platelet releasate. This action serves to abolish platelet recruitment.
Because we found that ADPase/CD39 is an effective physiological and constitutively expressed endothelial cell inhibitor of platelet reactivity, we postulated that a soluble form of the human enzyme would represent a promising new systemic antithrombotic modality for thrombosis-prone patients with a low threshold for platelet activation to be evaluated in vivo and ex vivo. This could be accomplished in the setting of acute, subacute, or chronic clinical situations. Thus, a recombinant soluble form of human CD39 based on the structure of full-length CD39 was designed (Figure 2⇑).16 CD39 contains 2 transmembrane regions near the amino and carboxyl termini, respectively. These domains serve to anchor the native protein in the cell membrane. Modeling studies, antibody epitope analyses, and sequence homology have demonstrated that the portion of the molecule between the transmembrane regions is external to the cell.10 The extracellular region contains the ACR characteristics of compounds in the apyrase family, in concordance with the hypothesis that the external portion of CD39 is critical for its ecto-ADPase activity. The ability of CD39-expressing cells to metabolize extracellular nucleotides supports the extracellular localization of the enzymatic portion of the molecule. The fact that intracellular nucleotide concentrations are maintained in the millimolar range further suggests that the active site of CD39 is not exposed to the cytoplasm.
For the generation of solCD39, the extracellular domain, encoding 439 amino acids, was isolated with the use of oligonucleotide cassettes and PCR, and was placed in a mammalian expression vector.16 Addition of the interleukin-2 leader sequence ensured that the recombinant molecule was secreted. After transfection with this solCD39-encoding plasmid, COS cells generated levels of ATPase and ADPase activity in their conditioned medium that linearly increased for a 5-day period. SolCD39 was isolated from conditioned medium derived from transiently transfected COS cells via immunoaffinity chromatography with the use of an anti-CD39 monoclonal antibody. This procedure yielded a single ≈66-kDa protein with both ATPase and ADPase activities, suggesting that the molecule had been properly glycosylated in this cell system. Incubation of the purified protein with N-glycanase to remove N-linked oligosaccharides yielded a band with the predicted molecular weight of 52 kDa.16
Purified solCD39 blocks ADP-induced platelet aggregation in vitro and also inhibits collagen-induced platelet reactivity.16 Aggregation induced by TRAP is also strongly blocked by CD39 (Figure 4⇓). This has led us to postulate that collagen and TRAP depend more on released ADP for recruitment and aggregation than was previously appreciated and underscores the possibilities of solCD39 as an antithrombotic agent.
Site-Directed Mutagenesis Studies of Amino Acids in the ACR Regions
To identify amino acids critically important for the enzymatic and biological activity of CD39, we performed site-directed mutagenesis studies within the highly conserved ACR of solCD39. Mutations of Glu174 to Ala (E174A) and Ser218 to Ala (S218A) resulted in the loss of solCD39 enzymatic activity. Moreover, the Tyr127 to Ala (Y127A) mutant lost 50% to 60% of ADPase and ATPase activity. The ADPase activity of wild-type solCD39 and of each mutant was generally greater with calcium than with magnesium, but for ATPase activity, no such preference was observed. Y127A demonstrated the highest calcium/magnesium ADPase activity ratio, 2.8-fold higher than that of the wild type, although its enzyme activity was greatly reduced. Enzymatic activity always correlated with biological activity in our platelet aggregation system. Thus, E174A, completely devoid of enzymatic activity, failed to inhibit platelet responsiveness. S218A, with 91% loss of ADPase activity, could still reverse platelet aggregation, albeit much less effectively than wild-type solCD39. These data indicate that Glu174 and Ser218 are essential for both the enzymatic and biological activity of solCD39, whereas Tyr127 appears to play an important role as well.17
Summary and Future Possibilities
A CHO cell–based solCD39 expression system, capable of growing in a serum-free medium, has been developed to increase protein production and to facilitate protein purification. The conditioned medium from these CHO cells contains 20-fold more ATPase and ADPase activity than that obtained from COS cells.16 After the administration of this solCD39 preparation to mice, enzymatic activity was measurable for an extended period of time. The elimination phase half-life was ≈2 days.16 Experimental results with a novel soluble form of recombinant human ecto-ADPase, solCD39, indicate a potential for a new class of antithrombotic agents acting by metabolism of the ADP in an activated platelet releasate. Thus, solCD39 blocks and reverses platelet activation, preventing the recruitment of additional platelets into a growing thrombus. In this manner, the extent of occlusion as well as damage to the vascular wall associated with cardiac and cerebral vascular events, such as stroke, myocardial infarction, angioplasty, and stenting, can largely be attenuated (Figure 5⇓). Furthermore, because of its independent mode of action on the platelet releasate, solCD39 could be administered in combination with currently used therapeutic modalities, such as heparin, aspirin, and glycoprotein IIb/IIIa antagonists.
This study was supported in part by Merit Review grants from the Department of Veterans Affairs and by National Institutes of Health grants HL-47073, HL-46403, HL-59488, NS-41462, and NS-41460.
Koch R. Über bakteriologische Forschung: Verhandlungen des X. Internationalen Medizinischen Congresses Berlin, 4-9. August 1890. Berlin, Germany: 1891;1:35–47.
Marcus AJ. Thrombosis and inflammation as multicellular processes: pathophysiological significance of transcellular metabolism. Blood. 1990;76:1903–1907.
Marcus AJ, Safier LB, Hajjar KA, Ullman HL, Islam N, Broekman MJ, Eiroa AM. Inhibition of platelet function by an aspirin-insensitive endothelial cell ADPase: thromboregulation by endothelial cells. J Clin Invest. 1991;88:1690–1696.
Marcus AJ, Weksler BB, Jaffe EA, Broekman MJ. Synthesis of prostacyclin from platelet-derived endoperoxides by cultured human endothelial cells. J Clin Invest. 1980;66:979–986.
Broekman MJ, Eiroa AM, Marcus AJ. Inhibition of human platelet reactivity by endothelium-derived relaxing factor from human umbilical vein endothelial cells in suspension: blockade of aggregation and secretion by an aspirin-insensitive mechanism. Blood. 1991;78:1033–1040.
Zimmermann H, Beaudoin AR, Bollen M, Goding JW, Guidotti G, Kirley TL, Robson SC, Sano K. Proposed nomenclature for two novel nucleotide hydrolyzing enzyme families expressed on the cell surface. In: Vanduffel L, Lemmens R, eds. Ecto-ATPases and Related Ectonucleotidases: Proceedings of the Second International Workshop on Ecto-ATPases and Related Ectonucleotidases. Maastricht, the Netherlands: Shaker Publishing; 2000:1–8.
Maliszewski CR, Delespesse GJ, Schoenborn MA, Armitage RJ, Fanslow WC, Nakajima T, Baker E, Sutherland GR, Poindexter K, Birks C, et al. The CD39 lymphoid cell activation antigen: molecular cloning and structural characterization. J Immunol. 1994;153:3574–3583.
Kansas GS, Wood GS, Tedder TF. Expression, distribution, and biochemistry of human CD39: role in activation-associated homotypic adhesion of lymphocytes. J Immunol. 1991;146:2235–2244.
Kaczmarek E, Koziak K, Sévigny J, Siegel JB, Anrather J, Beaudoin AR, Bach FH, Robson SC. Identification and characterization of CD39 vascular ATP diphosphohydrolase. J Biol Chem. 1996;271:33116–33122.
Al-Mondhiry H, Marcus AJ, Spaet TH. On the mechanism of platelet function inhibition by acetylsalicylic acid. Proc Soc Exp Biol Med. 1970;133:632–636.
Drosopoulos JHF, Broekman MJ, Islam N, Maliszewski CR, Gayle RB III, Marcus AJ. Site-directed mutagenesis of human endothelial cell ecto-ADPase/soluble CD39: requirement of glutamate174 and serine218 for enzyme activity and inhibition of platelet recruitment. Biochemistry. 2000;39:6936–6943.