Antibodies to CD36 (GPIV) Inhibit Platelet Adhesion to Subendothelial Surfaces Under Flow Conditions
The membrane glycoprotein CD36 (glycoprotein [GP] IV) has previously been shown to accelerate the initial interaction of platelets with purified type I collagen in both static and flow systems. In the present study, the role of CD36 on platelet interaction with physiologically relevant collagenous surfaces was addressed. Using arterial subendothelium (SE) and endothelial cell extracellular matrix (ECM), studies were performed under flow conditions with annular and parallel-plate perfusion chambers, respectively, at a shear rate of 800 s−1 for 2, 5, and 10 minutes. Perfusates consisted of citrated normal blood samples incubated with Fab fragments of a monospecific polyclonal anti-CD36 antibody or with each of three new anti-CD36 monoclonal antibodies (MoAbs) that inhibit platelet adhesion to purified type I collagen in a static system (131.4, 131.5, and 131.7). Perfusions over SE were also carried out using citrated blood samples from a Naka-negative donor, whose platelets lack CD36. Morphometric evaluation of the perfused samples showed that polyclonal anti-CD36 Fab and the three monoclonal anti-CD36 antibodies inhibited platelet adhesion to the two substrates by 40% after 2 minutes of perfusion and by 30% after 5 minutes (P<.005 on SE and P<.01 on ECM), but at 10 minutes, significant inhibition was seen only on SE with polyclonal anti-CD36 Fab. Similar inhibitions were seen with Naka-negative platelets on SE. These studies demonstrate that CD36 plays a role in the early stages of platelet adhesion to physiologically relevant subendothelial surfaces.
- Received November 14, 1995.
- Revision received February 27, 1996.
After disruption of the endothelial cell monolayer of the vessel wall, subendothelial components are exposed to circulating platelets. After their initial contact, platelets become activated, undergo morphological changes, and secrete their granule contents. During these events, surface receptors for adhesive proteins of plasma and SE are expressed on the membrane in a functional form.1 Collagen is the most thrombogenic protein of the vascular SE, and numerous putative collagen receptors have been suggested, but only GPIa/IIa(α2β1),2 3 CD36 (alternatively known as GPIV or GPIIIb), and GPVI4 5 have so far been found to meet most of the criteria proposed to validate a possible collagen receptor.
Static microtiter assays using washed normal platelets blocked with polyclonal anti-CD36 Fab fragments6 or washed platelets of the Naka-negative phenotype, which constitutively lack CD36,7 have shown that CD36 has its greatest effect within the first 10 minutes of platelet interaction with purified type I collagen. This effect is most readily seen in the absence of Mg2+, which promotes the quantitatively greater integrin-mediated adhesion involving α2β1, but both Mg2+-dependent and Mg2+-independent pathways would be in progress under physiological conditions. These results were confirmed under flow conditions in a parallel-plate perfusion system over purified type I collagen at a shear rate of 800 s−1, using both reconstituted citrated whole blood from platelets exposed to polyclonal anti-CD36 Fab fragments and citrated whole blood from a Naka-negative donor.8 With both preparations, adhesion was only ≈50% of control values after 2 minutes of perfusion, the earliest time point that can be reasonably measured in this apparatus, but reached 70% to 90% of control values at 10 minutes.
Although individuals of the Naka-negative phenotype do not appear to suffer any bleeding problems, this may be due in part to the fact that the incidence of the Naka-negative phenotype has been sought in predominantly young, healthy blood donor populations. However, examination of 806 patients with bleeding problems suggested that in some cases, a partial deficiency of CD36 may exacerbate the effect of autoantibodies against GPIIb/IIIa, GPIa/IIa, and CD36, which did not themselves inhibit platelet function.9
CD36 is a member of a newly recognized gene family that may be involved in adhesive processes and protein trafficking. Apart from mediating the early stages of platelet-collagen interaction, other adhesive functions of CD36 include its role as a thrombospondin receptor and as a receptor on endothelial and other cells for sequestrin present on the surface of Plasmodium falciparum–infected erythrocytes.10 However, its role in collagen adhesion appears to involve an epitope distinct from that mediating other adhesive functions11 12 or to involve antigenic changes resulting from phosphorylation/dephosphorylation of a specific serine residue.13
The aim of the present study was to evaluate the role of CD36 on platelet attachment to physiologically relevant collagenous surfaces rather than the purified type I collagen previously used. Biochemical and immunocytochemical methods have demonstrated the presence in SE of collagen types I, III, IV, and V, whereas the ECM generated by endothelial cells in culture lacks collagen type I but expresses collagen types III, IV, and V. The reactivity of the different types of collagen with respect to platelets is still subject to debate, and the literature contains conflicting reports about the influence of different types of collagen on platelet reactivity. This confusion may be due in part to different methods of collagen purification and the loss of the nonhelical telopeptides following pepsin extractions used in the isolation of certain collagens. In the present study, normal citrated human whole blood samples were perfused over either deendothelialized rabbit aortic segments or human endothelial cell ECM in the presence of polyclonal and monoclonal anti-CD36 antibodies by using annular and parallel-plate perfusion chambers, respectively. Adhesion of Naka-negative platelets, which lack CD36, to vascular SE was also evaluated.
Whole blood samples anticoagulated with citrate-phosphate-dextrose (final citrate concentration of 19 mmol/L) were obtained from healthy volunteer blood donors by venipuncture. Whole blood was similarly collected on two occasions from a previously described and well-characterized Naka-negative donor with normal platelet count and hematocrit (ARC-36).7 14 None of the donors had taken drugs affecting platelet function in the 2 weeks before the studies.
Human Endothelial Cell Cultures
Endothelial cells were isolated from human umbilical veins with collagenase (2% in PBS, 15 minutes, 37°C) (Boehringer Mannheim) as previously described15 with minor modifications16 and were maintained and subcultured at 37°C in a 5% CO2 humidified incubator in Medium 199 (Biochrom KG Laboratories) supplemented with 1 mmol/L glutamine, 100 U/mL penicillin, 50 μg/mL streptomycin (Flow Laboratories), and 20% pooled human serum. The culture medium was changed every 48 hours. The cells were identified as endothelial cells both morphologically and by the presence of von Willebrand factor, as detected by immunofluorescence. After the second passage, the cells were subcultured to confluence on 1% gelatin-coated glass coverslips.
The monospecific polyclonal rabbit anti-CD36 fragments were prepared as previously described,7 8 as were the monoclonal antibodies 131.4, 131.5, and 131.7 raised against purified native CD36 and directed against the collagen-binding domain of CD36.17
Perfusions were carried out over human umbilical vein endothelial cell ECM and rabbit arterial SE under flow conditions, using annular18 and parallel-plate19 perfusion chambers, respectively. Perfusates consisted of (1) control samples of citrated whole blood incubated with an isotype-matched (IgG2) preimmune antibody, (2) samples incubated with Fab fragments of a polyclonal anti-CD36 antibody, (3) samples previously treated with three new monoclonal anti-CD36 antibodies (131.4, 131.5, and 131.7), and (4) blood samples from a Naka-negative donor, whose platelets lack CD36. The amount of antibody added to blood aliquots was calculated to achieve final concentrations of 5 μg/mL, and preincubation was carried out for 30 minutes before perfusions were initiated. Perfusions were carried out at a shear rate of 800 s−1 for 2, 5, and 10 minutes at 37°C. All experiments were carried out in a paired fashion; that is, perfusion was carried out with blood from each donor in the presence of both the appropriate antibody and nonimmune IgG.
Perfusion Over SE
Deendothelialized rabbit aorta segments were prepared by chymotryptic digestion20 and used in perfusion experiments carried out in annular chambers as previously described.21 After perfusion, segments were fixed, dehydrated with alcohols, embedded in JB-4 resin, thin sectioned for light microscopy, and stained with toluidine blue. Morphometric analysis of individual sections established the total surface coverage of contacted, adherent, and aggregated platelets as a percentage of the total length of section analyzed.
Perfusion Over ECM
ECM produced by the confluent endothelial cell cultures was exposed by incubation of the coverslips with 2% EGTA for 60 minutes.16 For each perfusion, two coverslips covered with ECM were inserted into the chamber, and blood samples that had been previously incubated with anti-CD36 antibodies or with an irrelevant isotype-matched IgG were then circulated for 2, 5, and 10 minutes at 37°C. After perfusion, coverslips were removed from the chamber, rinsed with 0.15 mol/L PBS, pH 7.4, and then fixed in 0.5% glutaraldehyde in 0.15 mol/L PBS, pH 7.4, overnight at 4°C. Fixed coverslips were stained in a solution of toluidine blue (0.02% in distilled water) for 5 minutes at 22°C, mounted in aqueous medium and used for en face evaluations.
The extent of platelet interaction with the ECM was evaluated morphometrically en face by light microscopy at ×400 magnification in 10 microscope fields per coverslip. Images from the perfused ECM-coated coverslips were obtained with a video camera (Panasonic) adapted to a light microscope (Polyvar, Reichter-Jung). The video camera provides images to the automated image analysis system (BIOCOM, Les Ulis Cedex), which is connected to a computer (Deskpro 386, Compaq) with a graphic card adapter. Images are digitized, and areas of the image with gray levels in a selected range can be measured. This capability facilitates the selection of groups of interacting platelets and the quantification of the area covered by platelets. Results obtained are expressed as %SC.
Adhesion to SE
Platelet interaction with SE increased with increasing time of perfusion, but surface coverage in the presence of anti-CD36 Fab was always less than in its absence in the four donors examined (Fig 1⇓, upper panel). The quantitative data showed that surface coverage (mean±SEM, n=4 at each time point) in control samples after 2, 5, and 10 minutes of perfusion was 13.7±2.0%, 30.8±0.6%, and 42.5±2.5%, respectively. In contrast, in the Fab-blocked samples, surface coverage at these times was 7.2±1.7%, 17.6±1.6%, and 30.6±3.1%. These values are significantly less than controls at a level of P<.05. When expressed as percentage of control, the relative %SC of antibody-blocked samples after 2, 5, and 10 minutes of perfusion was 53%, 57%, and 72% of control values in the absence of anti-CD36 Fab fragments.
In further experiments, adhesion to SE was measured in the presence of the anti-CD36 MoAbs 131.4, 131.5, and 131.7 (Fig 2⇓, upper panel). Adhesion (%SC) to SE in control experiments (n=4) was 18.0±0.3% at 2 minutes, 31.2±0.3% at 5 minutes, and 41.7±4.3% at 10 minutes, using blood samples that had been preincubated with nonimmune IgG2. Surface coverage was decreased at all three time points in the presence of the 131 series MoAbs (5 μg/mL final concentration in each case). In the presence of 131.4, surface coverage was 55% of control values at 2 minutes of perfusion, 60% at 5 minutes, and 80% at 10 minutes. With 131.5, inhibition was 70%, 76%, and 99% of controls at 2, 5, and 10 minutes, respectively (not shown). With 131.7, the surface coverage was 60%, 67%, and 87% at 2, 5, and 10 minutes, respectively. For all three MoAbs, the reductions in surface coverage on SE are significant (P<.005, n=4) for 2- and 5-minute perfusions but are not significant at 10 minutes.
Perfusion Over ECM
In a second series of experiments, blood samples (n=4) were perfused in the parallel-plate perfusion chamber over coverslips coated with ECM (Fig 1⇑, lower panel). In this case, the attachment of control platelets, expressed as percent of the total surface examined, was 23.5±1.3% at 2 minutes (mean±SEM), 28.9±1.8% at 5 minutes, and 44.3±3.4% at 10 minutes, whereas with blood samples preincubated with anti-CD36 fragments, attachment was 17.6±1.5%, 22.4±1.7%, and 37.6±3.9% at 2, 5, and 10 minutes, respectively. These values are significantly less than control values at a level of P<.05 at both 2 and 5 minutes of perfusion but are not significantly different at 10 minutes. As a percentage of controls, values at 2, 5, and 10 minutes were 75%, 77%, and 85%, respectively.
Similar results were obtained using the anti-CD36 MoAbs to inhibit platelet adhesion to ECM in the parallel-plate chamber (Fig 2⇑, lower panel). The surface coverage obtained in control experiments, in the presence of nonimmune IgG2, was 24.4±0.8%, 33.9±1.5%, and 55.5±1.56% at 2, 5, and 10 minutes of perfusion, respectively. In the presence of MoAbs 131.4 and 131.7, adhesion was about 65% and 70% of control values at 2 minutes of perfusion (P<.01). After 5 minutes of perfusion, %SC in the presence of 131.4 and 131.7 was 63% (P<.01) and 87% (P<.05), respectively, but no significant inhibitory effect was seen after 10 minutes of perfusion with either antibody. The MoAb 131.5 did not induce statistically significant inhibition of platelet adhesion of ECM at any time point.
Fig 3⇓ shows micrographs of platelet adhesion on ECM after perfusion at 800 s−1 for 5 minutes, in the presence of IgG2 (A) or when blood samples were previously incubated with the anti-CD36 MoAb 131.4 (B). Visual inspection shows reduced coverage and a reduced degree of activation, as exhibited by reduced pseudopod extension in the presence of the antibody.
Adhesion of Naka-Negative Platelets to SE
Naka-negative platelets showed a decreased interaction with SE, which was most marked after 2 minutes of perfusion but was also detectable after 5 and 10 minutes of perfusion (Fig 4⇓). Adhesion of Naka-negative platelets on SE, using the annular perfusion chamber, was also evaluated at 800 s−1, and perfusions were carried out at the three time points (2, 5, and 10 minutes). Evaluation of the surface covered by platelets was performed on cross sections of the vascular segments previously fixed and embedded. Quantitatively, control samples gave surface coverage of 14.7%, 32.3%, and 45.2% at 2, 5, and 10 minutes of perfusion, while surface coverage of Naka-negative platelets was 5.9%, 15.2%, and 27.3% at 2, 5, and 10 minutes (average of data from two different experiments performed with blood from the same donor); that is, surface coverage by Naka-negative platelets on SE after 2, 5, and 10 minutes of perfusion was 40%, 50%, and 60%, respectively, of control values.
The onset of thrombogenesis involves the initial attachment of platelets to collagenous subendothelial surfaces, extension of pseudopodia and spreading on the surface, and finally the formation of aggregates and thrombi. The two subendothelial surfaces thought to be most involved in these events are the ECM, which is exposed following the loss of endothelial cells, and the subendothelial surface resulting from more extensive damage to the vessel wall.
As outlined earlier, we have previously used purified type I collagen as a well-defined substrate for evaluating the role of CD36 in platelet adhesion, including the inhibitory effects under static and flow conditions of monospecific polyclonal anti-CD36 Fab fragments, anti-CD36 MoAbs, and platelets from Naka-negative donors lacking CD36. In the present case, we have examined these effects in the interaction of platelets with the more complex but more physiologically relevant substrates of the ECM laid down by human endothelial cells and the SE remaining after removal of endothelial cells from rabbit aortic segments.
Both of these substrates have been used extensively for in vitro studies of platelet adhesion and thrombogenesis. The subendothelial surface of enzymatically digested rabbit aortic segments consists of a three-dimensional meshwork of collagen fibrils in an unreactive elastin matrix and appears to be highly reactive to platelets. While ECM produced by cultured human endothelial cells supports adhesion and has been compared to human blood vessel SE, it is less reactive to platelets in aggregate formation than are human renal or rabbit arterial segments.21 These two substrates, SE and ECM, differ in collagen composition insofar as types I, III, IV, and V have been identified in the SE, whereas ECM lacks type I collagen and types III, IV, and V have been identified.
In the present studies we have shown that platelet interaction with either substrate is reduced in the presence of anti-CD36 antibodies. Visual observation of the en face preparations of adhesion to ECM indicates that the degree of pseudopod formation and other morphological parameters of platelet activation are more advanced in the absence of anti-CD36 Fab fragments than in their presence.
Despite the molecular and structural complexity of ECM and SE compared with purified type I collagen, the time-dependent course of platelet adhesion and the inhibition seen with polyclonal and monoclonal anti-CD36 antibodies are similar with each of the three substrates. Adhesion to each substrate, as measured by surface coverage, is about 20% after 2 minutes of perfusion, about 30% after 5 minutes, and about 50% after 10 minutes. Broadly, the anti-CD36 MoAbs and polyclonal Fab fragments reduced adhesion to about 60% of control values after 2 minutes of perfusion, to about 70% after 5 minutes of perfusion, and to about 80% after 10 minutes of perfusion.
These values are similar to the values found for the adhesion of Naka-negative platelets to type I collagen in our previous studies and to SE in the current work. However, certain differences in antibody effects may be noted: for example, all three MoAbs inhibit adhesion to type I collagen in the static system, while adhesion to type I collagen or ECM in the parallel-plate perfusion chamber is inhibited by MoAbs 131.4 and 131.7 but not by MoAb 131.5; on the other hand, adhesion to SE is significantly inhibited by all three antibodies. Moreover, the CD36-dependent phase of adhesion to SE and ECM appears to be complete after 5 minutes of perfusion, since none of the antibodies caused significant inhibition at the 10-minute time point except the polyclonal anti-CD36, and that was effective only with adhesion to SE. In contrast, all the antibodies except 131.5 could inhibit adhesion to type I collagen even after 10 minutes of perfusion. These results suggest that CD36-dependent adhesion but not overall adhesion may be occurring more rapidly on SE and ECM as substrates compared with purified type I collagen.
SE and ECM also contain other adhesive substrates, but the anti-CD36 MoAbs do not inhibit platelet adhesion to fibrinogen, fibronectin, or vitronectin under Mg2+-independent or Mg2+-dependent conditions.17 Furthermore, thrombospondin, another proposed ligand for CD36, does not support adhesion in citrated blood22 as used here. Thus, the inhibitory effects observed here are largely, if not solely, due to changes in platelet-collagen interaction.
Although these perfusion studies and complementary studies from other laboratories12 13 23 24 support a role for CD36 in platelet-collagen interaction, this role has been questioned because of inability to detect reduced adhesion of platelets to different collagen types in two studies involving perfusion of Naka-negative blood.25 26 These apparent differences may be explainable on a methodological basis, considering differences in the nature of the substrates, anticoagulants, shear rates, and perfusion times. In addition to type I collagen, both perfusion studies used collagen types III and IV prepared by pepsin digestion, which would lack the nonhelical telopeptides necessary to support Mg2+-independent adhesion,27 and some, but not all, perfusions used high shear rates (1600 and 2600 s−1), which are in the range expected for von Willebrand factor–dependent adhesion. These high shear rates could also have caused the peaking of CD36-dependent effects within the stated perfusion times of 5 minutes, since maximum CD36-dependent effects are seen at the initial 2-minute time point at shear rates of 800 s−1. Finally, one of the studies26 used low-molecular-weight heparin as anticoagulant, which these investigators have shown can lead to platelet activation due to tissue factor–dependent thrombin generation28 and which maintains physiological levels of Mg2+ that could obscure the quantitatively smaller CD36-mediated events that may occur at earlier time points.
While indirect, our studies show that the group of collagens present in ECM (types III, IV, and V) are able to support CD36-mediated platelet adhesion, as does type I collagen, whether purified or present in SE. Taken together or separately, our studies demonstrate that CD36 plays a role in the earliest stages of platelet interaction with collagen, although its contribution to the pathophysiology of thrombogenesis remains to be determined.
Selected Abbreviations and Acronyms
|%SC||=||percentage of total surface covered by platelets|
Dr Diaz-Ricart is supported by the Spanish Ministry of Education and Science (MEC-Fulbright 1994). Additional support for these studies was provided by an award (PM95/0103) from the Dirección General de Investigación Científica y Técnica (DGICYT) and CIRIT GRO93-9121 to Dr Ordinas and by US Public Health Service grant HL40858 to Dr Jamieson.
Nieuwenhuis HK, Sakariassen KS, Houdijk WPM, Nievelstein PFEM, Sixma JJ. Deficiency of platelet membrane glycoprotein Ia associated with a decreased platelet adhesion to subendothelium: a defect in platelet spreading. Blood. 1986;68:692-695.
Moroi M, Jung SM, Okuma M, Shinmyozu K. A patient with platelets deficient in glycoprotein VI that lack both collagen-induced aggregation and adhesion. J Clin Invest. 1989;84:1440-1445.
Tandon NN, Kralisz U, Jamieson GA. Identification of GPIV (CD36) as a primary receptor for platelet-collagen adhesion. J Biol Chem. 1989;264:7576-7583.
Tandon NN, Ockenhouse CF, Greco NJ, Jamieson GA. Adhesive functions of platelets lacking GPIV (CD36). Blood. 1991;78:2809-2813.
Diaz-Ricart M, Tandon NN, Carretero M, Ordinas A, Bastida E, Jamieson GA. Platelets lacking functional CD36 (glycoprotein IV) show reduced adhesion to collagen in flowing whole blood. Blood. 1993;82:491-496.
Beer JH, Rabaglio M, Berchtold P, von Felten A, Clemetson KJ, Tsakiris DA, Kehrel B, Bradenberger S. Autoantibodies against the platelet glycoproteins (GP) IIb/IIIa, Ia/IIa, and IV and partial deficiency in GPIV in a patient with a bleeding disorder and a defective platelet collagen interaction. Blood. 1993;82:820-829.
Greenwalt DE, Lipsky RH, Ockenhouse CF, Ikeda H, Tandon NN, Jamieson GA. Membrane glycoprotein CD36: a review of its roles in adherence, signal transduction and transfusion medicine. Blood. 1992;80:1105-1115.
Tandon NN, Diaz-Ricart M, Lotfalian B, Ockenhouse CF, Jamieson GA. Different domains of CD36 mediate collagen adhesion and malarial cytoadherence. Thromb Haemost. 1995;73:1742. Abstract.
Mercier N, Catimel B, Reck MP, Pellecchia D, McGregor JL. Identification of a functional site on CD36 involved in the interaction between platelets and collagen. Platelets. 1995;6:139-145.
Asch AS, Liu I, Briccetti FM, Barnwell JW, Kwakye-Berko F, Dokun A, Goldberger J, Pernambuco M. Analysis of CD36 binding domains: ligand specificity controlled by dephosphorylation of an ectodomain. Science. 1993;262:1436-1440.
Yamamoto N, Ikeda H, Tandon NN, Herman J, Tomiyama Y, Mitani T, Sekiguchi S, Lipsky R, Kralisz U, Jamieson GA. A platelet membrane glycoprotein (GP) deficiency in healthy blood donors: Naka− platelets lack detectable GPIV (CD36). Blood. 1990;76:1698-1703.
Jaffe E, Minick C, Adelman B, Becker C, Nachman R. Synthesis of basement membrane collagen by cultured human endothelial cells. J Exp Med. 1976;144:209-225.
Aznar-Salatti J, Bastida E, Haas TA, Escolar G, Ordinas A, de Groot PG, Buchanan MR. Platelet adhesion to exposed endothelial cell extracellular matrices is influenced by the method of preparation. Arterioscler Thromb. 1991;11:436-442.
Agbanyo FR, Sixma JJ, de Groot PG, Languino LR, Plow EF. Thrombospondin-platelet interactions: role of divalent cations, wall shear rate and platelet membrane glycoproteins. J Clin Invest. 1993;92:288-296.
Li W-X, Howard RJ, Leung LL. Identification of SVTCG in thrombospondin as the conformation-dependent, high affinity binding site for its receptor, CD36. J Biol Chem. 1993;268:16179-16184.
Leung LLK, Li WX, McGregor JL, Albrecht G, Howard RJ. CD36 peptides enhance or inhibit CD36-thrombospondin binding: a two-step process of ligand-receptor interaction. J Biol Chem. 1992;267:18244-18250.
McKeown L, Vail M, Williams S, Kramer W, Hansmann K, Gralnick H. Platelet adhesion to collagen in individuals lacking glycoprotein IV. Blood. 1994;83:2866-2871.
Saelman EUM, Kehrel B, Hese KM, de Groot PG, Sixma JJ, Nieuwenhuis HK. Platelet adhesion to collagen and endothelial cell matrix under flow conditions is not dependent on platelet glycoprotein IV. Blood. 1994;83:3240-3244.
Zijenah LS, Morton LF, Barnes MJ. Platelet adhesion to collagen: factors affecting Mg2+-dependent and bivalent-cation–independent adhesion. Biochem J. 1990;268:481-486.
Lozano M, Bos A, de Groot PG, van Willigen G, Meuleman DG, Ordinas A, Sixma JJ. Suitability of low-molecular-weight heparin(oid)s and a pentasaccharide for an in vitro human thrombosis model. Arterioscler Thromb.. 1994;14:1215-1222.