Background
Thrombocytopenia is a major adverse effect of high dose systemic administration of adenoviral (Ad) gene therapy vectors. While a previous report did not find platelet activation by Ad [
1], recent studies have shown that Ad may activate platelets [
2] and binds
in vivo to murine thrombocytes resulting in hepatic sequestration [
3]. Ad-induced thrombocytopenia has been shown to be dose-dependent, saturable and reversible [
4], compatible with a ligand-receptor mechanism. Recently, binding of Ad to platelet was indirectly suggested following interference of platelet adhesion to fibronectin after incubation with Ad [
2]. In this study we developed a direct flow cytometry assay to quantitatively analyze Ad attachment to human platelets
in vitro and to characterize their interaction.
Many microorganisms in addition to Ad have evolved to facilitate cell entry via RGD recognition of cell surface integrins. For example, integrins mediate RGD-dependent attachment of picornaviruses [
5,
6] and bacteria [
7,
8]. In contrast, Group C Ad primarily attaches to the cell surface via the fiber protein knob binding to CAR [
9] (coxsackie and Ad receptor). Next, Ad internalizes primarily utilizing αVβ3 integrin [
10], and to a lesser extent αVβ5 integrin [
11], via interaction of the RGD-containing Ad penton base protein. In addition to αVβ3 and αVβ5, other integrin receptors for Ad may include αVβ1, and α5β1 [
12]. Because Ad uses both CAR and αV integrins, we used our flow cytometry assay to evaluate CAR expression in platelets and integrin-mediated Ad binding to platelets.
Discussion
Platelets bind physiological ligands in an RGD-dependent manner, e.g. FBG, von Willerband factor (VWF), fibronectin and vitronectin. Non-physiological platelet integrin ligands include disintegrins (cyclic RGD-based polypeptides in snake venoms) and a number of microorganisms. One of the most extensively characterized pathogens with respect to nucleated cellular integrin interaction is Ad. Cell entry by Ad viruses initially involves attachment of the Ad fiber knob to the primary Ad receptor, CAR [
6], followed simultaneously or subsequently by binding of any of the five RGD protrusions on the Ad penton base protein to cellular αV integrins heterodimerized to specific β chains [
10,
11]. A critical requirement for Ad infection is the interaction of membrane αV integrin with the RGD-displaying Ad penton base. This interaction has been previously demonstrated via inhibition of Ad cell entry by RGD peptides and antibodies to αV integrins [
10,
16]. Integrin receptors are heterodimers comprised of α and β subunits whose specific sequence and activation-dependent conformation determine their ligand affinity. The ligand motif for a number of integrins is based on an arginine-glycine-aspartate (RGD) sequence and variations on the RGD theme determine specific ligand-integrin recognition. For example, fibrinogen (FBG) binding to αIIbβ3 depends on prior inside-out signaling, resulting in platelet priming and conformational αIIbβ3 transition into a high-affinity state [
17].
In the current study, we developed a direct flow cytometry approach to characterize Ad binding to human platelets, focusing on platelet integrin-mediated binding. Optimization of the methodology could show a number of pertinent findings. First, Ad binding to human platelets can be manipulated
in vitro by combining a divalent ion and thrombin activation (Fig.
2). Second, Ad binding activates platelets
in vitro (Fig.
2). Third, an optimal MOI in the order of 10 (Fig.
3) was observed for Ad attachment to the platelet surface (Fig.
4). This optimal MOI is compatible with the ratio of 40 between the spherical surface areas of platelets and Ad, given respective diameters of ~3 m and ~150 nm. Fourth, Ad attachment to human platelets is at least partially mediated by platelet integrins, as evident by blocking assays using anti-αVβ3 monoclonal antibody and RGD peptidomimetics (Fig.
5).
Fifth, although CAR was previously reported to be expressed in human platelets both at the level of RNA and using flow cytomtery [
2], our studies show CAR deficiency in normal human platelets (Fig.
6).
Because CAR mediates homotypic cell adhesion, it is generally present in specialized intracellular junctions, including the cardiac intercalated disk and the adherens junction of polarized epithelial cells [
12]. Although CAR is abundantly expressed in epithelial cells during embryogenesis, its expression in adult mice is restricted to fewer cell types, contrasting with the homogeneous expression pattern of αV-integrins [
18]. Thus, in bone marrow hematopoeitic lineages CAR expression is minute [
19,
20]. Othman et al employed the RmcB anti-CAR antibody and did not report a CAR-negative cell line to demonstrate the specificity of the anti-CAR antibody [
2]. In contrast, we confirmed specificity of the rabbit H-300 polyclonal anti-CAR antibody in both CAR-positive and negative cell lines prior to testing CAR expression in platelets. While variations in the specificity of the anti-CAR antibodies employed may account for the discrepancy between our results and Othman et al [
2], further studies are required to conclusively define CAR expression in human platelets. However, our blocking assays, along with the recent observation that Ad serotype 11 can efficiently bind to mouse platelets fiber-independently [
21], further highlight the role of platelet integrins as mediators of Ad binding. Other integrins expressed by platelets include a5b1 and a1b1 [
22,
23]. While these are not well established as Ad receptors, the recent finding of Ad interference with platelet adhesion to fibronectin [
2] may suggest that Ad may also bind to the fibronectin receptor a5b1.
Previously, αIIbβ3 (gpIIb/IIIa), the primary platelet FBG receptor was reported to mediate platelet attachment of the intracellular bacterial microorganisms, chlamydia and borrelia. These studies employed blocking assays using abciximab (an anti- αIIbβ3 antibody) and RGD peptides [
24‐
26], at 4–400 fold higher blocking concentrations than employed in this study. However, while αVβ3 may partially mediate attachment of Ad to platelets (Fig.
5), αIIbβ3 does not appear to play a significant role in Ad binding to platelets, as evident by lack of blockade by a monoclonal antibody against αIIbβ3 (Fig.
5), and by avid adherence of Ad to αIIbβ3-deficient platelets from a patient with Glanzmann thrombasthenia (Fig.
7). Of note, unlike borrelia binding to platelets that requires prior platelet activation [
24,
25], Ad could also efficiently bind to naïve platelets, although platelet activation along with MnCl enhanced Ad binding (Fig.
2).
Glanzmann thrombasthenia (GT) is a rare, inherited disorder of platelet function characterized by mucocutaneous hemorrhage caused by mutations in the αIIbβ3 gene. The major laboratory finding in GT is a profound defect in platelet aggregation caused by a failure of αIIbβ3 to bind FBG. In this study we observed minute expression of αIIbβ3 on platelets from a patient with GT. A typical mutation in this patient's kindred was previously found to completely abolish αIIβ expression while a very low β3 (IIIα) expression level was still detected [
15]. Thus, unlike other αIIbβ3 mutations in other GT kindreds where both αIIβ and β3 were completely absent, platelets from this GT subpopulation maintain the potential to express some αVβ3 [
15]. In this context, while we failed to document substantial αVβ3 expression on normal platelets, Coller et al had measured ~100 αVβ3 receptors per platelet, i.e. 0.25% of the number of αIIbβ3 receptors per platelet [
15].
Our findings on platelet-Ad interaction
in vitro may have implications on the biodistribution of Ad
in vivo. Previously, partial platelet depletion did not alter Ad biodistribution and it was postulated that Ad attachment to platelets may occur only in a small fraction of platelets [
24]. However, we speculate that blockade of platelet integrins
in vivo will alter Ad biodistribution. Recently, attachment of Ad particles to platelets resulted in platelet-leukocyte aggregates [
3], VWF and p-selectin-mediated thrombocytopenia [
2] via clearance by the reticuloendothelial system and the complement pathway [
27]. While this scenario may complicate Ad-based gene delivery, it may reflect an evolutionary-conserved defense mechanism allowing to efficiently clear circulating RGD-displaying microorganisms such as Ad. In support of this hypothesis, RGD display on the Ad fiber knob, in addition to the natural RGD ligand on the Ad penton base, has been reported to result in paradoxically diminished systemic tissue distribution [
16]. Thus, fundamental to future rationalized systemic Ad-based gene delivery endeavors in humans is the molecular characterization of Ad-platelet interaction
in vivo. Additionally, Ad biodistribution and toxicity may differ in GT patients from healthy subjects.
Taken together, we report a direct flow cytometry assay to characterize Ad binding to platelets. This approach may eventually be employed to determine the exact integrin profile accounting for Ad attachment to human platelets and therefore may have implications on systemic Ad biodistribution.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
NS designed study, performed experiments and wrote manuscript, GE performed experiments, DKG intellectual contribution and reagents, LK performed experiments, SUS intellectual contribution and provided cell lines, YSH designed experiments and wrote manuscript.