Integrin αvβ3 as a target in the prevention of neointimal hyperplasia

Structure, function, and distribution of αvβ3 integrin 

Integrins are a family of transmembrane glycoproteins that mediate cell–cell and cell–matrix interaction.5 All known members of this superfamily are noncovalently associated heterodimers composed of an α and a β subunit. At present, at least 8 β and 18 α subunits have been characterized, and these subunits associate to generate at least 24 different integrins.5 For instance, subunit β3 associates with subunits αIIb and αv to generate integrins αIIbβ3 and αvβ3. Integrins are type I membrane proteins with a large extracellular, a transmembrane and a short cytoplasmic domains. The interaction between integrins and their ligands, besides mediating cell adhesion, is important in a number of cellular processes.6

One of the most prevalent integrins, αvβ3 is expressed on almost all the cells originating from the mesenchyme and on a variety of cell types in the blood vessel, including endothelial cells, SMCs, fibroblasts, macrophage, and platelets. It is known to mediate many biologic events such as migration of vascular SMCs, adhesion of osteoclasts to the bone matrix, and angiogenesis. It is the most promiscuous integrin, for it binds to many different ligands and a number of extracellular matrix proteins, including vitronectin, fibronectin, osteopontin, fibrinogen, and von Willebrand factor, by the interaction with the Arg-Gly-Asp (RGD) motif.5, 7 Conversely, a related integrin, αIIbβ3 is exclusively expressed on platelets and is largely responsible for the final cohesive phase of platelet activation in vivo, such as platelet aggregation supported by the binding of adhesive protein.8 Of interest is that the αVbβ3 integrin recognize the same RGD motif and binds to the same extracellular matrix proteins.9, 10

Osteopontin one of the ligands for αvβ3, contains the canonical integrin recognition sequence, RGD, and binds to αvβ3 integrin through the sequence.11 In vitro studies have demonstrated that osteopontin promotes the migration of cultured rat arterial SMCs12 and human coronary artery SMCs.13 Previous data showed that osteopontin was coordinately expressed with β3 integrins in the vessel wall and that a blockade of αvβ3 resulted in a reduction of neointimal formation in animal models after vascular injury.14 These data suggest that αvβ3 binding to osteopontin is important in mediating SMC migration from the media to the neointima in vivo.

The integrin-mediated adhesion of cells to extracellular matrix leads bidirectional intracellular signaling events that regulate cell migration, survival, and proliferation. In outside-in signaling, ligand binding activates intracellular signaling pathways. In inside-out signaling, signals received by other receptors activate intracellular signaling pathways that impinge on integrin cytoplasmic domains and change the extracellular domain conformation for binding to ligands.5

Recent studies have shown that αvβ3 expression on SMCs is subject to regulation and is increased by treatment with thrombin,15 transforming growth factor-β (TGF-β), and platelet-derived growth factor-BB (PDGF-BB).16 In endothelial cells, vascular endothelial growth factor (VEGF) can induce activation of αvβ3 and nuclear factor-κB3 (NF-κB3), which leads to suppression of p53 and p21WAF1/CIP1.17, 18 Moreover αvβ3, along with membrane type 1 (MT1) matrix metalloproteinase-1 (MMP-1), is associated with MMP-2 at the cell surface.19, 20

The MMPs belong to a family of zinc-dependent endopeptidases that degrade many components of the extracellular matrix. Most MMPs are secreted in a latent form (pro-MMP), and a specific multistep activation process is required to convert pro-MMP to proteolytic active forms. Localization of functionally active MMP on the cell surface is an essential and tightly regulated element during a variety of normal and disease processes, such as tumor cell invasion.21 For instance, MMP-2 is activated at the cell surface of invasive cells by a multimeric receptor/activation complex consisting of the tissue inhibitor of metalloproteinase 2 (TIMP2), and MT1-MMP.22

In line with the theory of cellular invasion requiring a coordinated expression of proteolytic enzymes and adhesion molecules, Hofmann et al23 suggested that functional cooperation of MT1-MMP and αvβ3 is critical for spatial and temporal control of extracellular matrix proteolysis in human melanoma cells. They indicated that joint MT1-MMP and αvβ3 might enforce the most efficient docking and activation of MMP-2, and in turn, facilitate cellular locomotion. Furthermore, Brooks et al19 demonstrated that the functionally active form of MMP-2 on the cell surface seems to predominantly involve αvβ3 in angiogenesis and concomitant melanoma growth.

αvβ3 integrin and animal models of arterial injury 

Our group first reported the potential therapeutic benefit of αvβ3 blockade in the prevention of intimal hyperplasia and restenosis.24 We demonstrated that PDGF, a potent chemotactic agent present in the arterial wall after injury, regulates the surface distribution of αvβ3 on the SMC surface in cell culture. Using indirect immunofluorescence, focal adhesions containing αvβ3 were localized to the leading edge of migrating cells when stimulated with PDGF. In contrast, αvβ3 was evenly distributed on the surface of SMCs grown in the absence of PDGF. These results suggest that a redistribution of αvβ3 in focal adhesion is necessary for SMC motility.

In an in vitro assay, we determined that PDGF-induced human SMC migration is mediated by αvβ3 by using a blocking antibody to αvβ3 (LM609). This PDGF-mediated migration was also attenuated with a αvβ3-blocking RGD peptide (GpenGRGDSPCA), demonstrating that the RGD sequence is the binding site in the extracellular matrix proteins.

We also tested the effects of the local administration of this RGD peptide in a rabbit model of carotid balloon angioplasty injury. This RGD antagonist was delivered to the adventitia of the injured artery and inhibited the new intimal (neointimal) lesion formation by 70%. Neointimal hyperplasia seen in an animal model should be distinguished from intimal hyperplasia seen in humans because there are no inherent SMCs in the noninjured intimal layer in most, normocholesterolemic animals. The same peptide locally applied to the carotid artery through an adventitial Pluronic gel (Sigma-Aldrich, Inc., St. Louis, Mo) in rats led to a 92% reduction in neointimal hyperplasia after a similar balloon angioplasty injury.25

In humans, αvβ3 is present both in normal artery and in sites of SMC accumulation and angiogenesis in atherosclerotic plaques.26 It is generally detectable in normal artery only along the luminal surface, with minimal expression in the media.26, 27 Several studies in animal models have shown that arterial injury is a stimulus for expression of αvβ3 by endothelial cells and medial SMCs.24, 27 For instance, Srivatsa et al28 showed in the pig coronary stent model that an upregulation of αvβ3 takes place at sites of cell accumulation within the neointima and adventitia at 7 days after arterial injury, followed by persistent high levels of αvβ3 expression within the media and neointima up to 21 days, decreasing towards baseline by 28 days. Indeed, Table I29, 30, 31, 32, 33, 34, 35 summarizes a number of reports demonstrating the efficacy of αvβ3 and αIIbβ3 antagonists in the reduction of neointimal hyperplasia in a variety of species.

Table I. Scientific investigations on the effectiveness of the β3 integrin antagonists in the inhibition of intimal lesion formation in various animal models of arterial injury

	Species (ref)	Antagonist	β3 integrin	Artery	Injury type	IH lesion reduction?
	Rat25, 29	ReoPr	αvβ3, αIIbβ3	Carotid	Angioplasty	Yes
	Rat30	Gpen	αvβ3	Carotid	Angioplasty	Yes
	Hamster31	Gpen	αvβ3	Carotid	Angioplasty	Yes
	Hamster32	FK633	αIIbβ3	Carotid	Angioplasty	Yes
	Rabbit24	Gpen	αvβ3	Carotid	Angioplasty	Yes
	Rabbit33	Vitaxi	αvβ3	Carotid	Angioplasty	Yes
	Rabbit34	AZ-1	αIIbβ3	Femoral	Angioplasty	No
	Pig28	XJ735	αvβ3	Coronary	Stenting	Yes
	Monkey35	ReoPro	αvβ3, αIIbβ3	Iliac	Angioplasty	No
	Monkey35	ReoPro	αvβ3, αIIbβ3	Subclavian	Stenting	No

IH, Intimal hyperplastic.

αvβ3 integrin and the clinical trials of restenosis 

The results from the animal studies were consistent with findings from the early clinical trials examining the effect of various antagonists to platelet integrin αIIbβ3 and SMC integrin αvβ3 on the issue of long-term benefit of reduced target lesion revascularization (Table II).36, 37, 38, 39, 40 In the Evaluation of Platelet IIb/IIIa Inhibition for Prevention of Ischemic Complications (EPIC) trial, ReoPro (abciximab, an monoclonal antibody fragment directed against the β3 integrin; Centocor, Inc, Horsham, Pa) was effective in limiting the need for late revascularization after angioplasty for at least 3 years after treatment.41 A subsequent study confirmed that ReoPro treatment reduced ischemic complications and late mortality, particularly in the diabetic population.37

Table II. Clinical trials on the effectiveness of the β3 integrin antagonists in the inhibition of restenosis in the coronary arteries

	Study (ref)	Antagonist	β3 integrin	Injury type	TLR
	EPIC36	ReoPro	αvβ3, αIIbβ3	Angioplasty, atherectomy	Reduced
	Lincoff et al37	ReoPro	αvβ3, αIIbβ3	Stenting	Reduced
	IMPACT II38	Integrilin	αIIbβ3	Angioplasty	No difference
	ERASER39	ReoPro	αvβ3, αIIbβ3	Stenting	No difference
	CAPTURE40	ReoPro	αvβ3, αIIbβ3	Angioplasty	No difference

TLR, Targeted lesion revascularization; EPIC, Evaluation of Platelet IIb/IIIa Inhibition for Prevention of Ischemic Complications; IMPACT, Integrilin to Minimize Platelet Aggregation and Coronary Thrombosis; ERASER, Evaluation of ReoPro And Stenting to Eliminate Restenosis; CAPTURE, C7E3 Fab Antiplatelet Therapy in Unstable Refractory Angina.

In the Integrilin to Minimize Platelet Aggregation and Coronary Thrombosis (IMPACT) II trial, however, Integrilin (an agent with anti-αIIbβ3 activity but without specific αvβ3 inhibitory activity; Millennium Pharmaceuticals, Cambridge, Mass) was ineffective in reducing coronary revascularizations in the same clinical setting as the EPIC trial.38 As in the animal studies, these clinical results suggest that αIIbβ3 integrin inhibition had no place in the treatment of coronary restenosis. However, a more detailed clinical study, the Evaluation of ReoPro And Stenting to Eliminate Restenosis (ERASER) trial, revealed that ReoPro given at the time of or a short duration after coronary angioplasty and stenting had little or no effect on the size of the intimal hyperplastic lesion as measured by intravascular ultrasound.39

Still, these clinical trials did not adequately address the role of αvβ3 in restenosis because short-term infusions of Integrilin and ReoPro would not be expected to block αvβ3 during crucial periods of vascular repair. Bleeding complications limited the long-term administration of these antagonists during percutaneous intervention to humans. Indeed, there is no certainty the local concentrations of these antagonists in the vessel wall are sufficient to inhibit αvβ3 integrin clinically.

αvβ3 integrin and the clinical significance based on animal models 

Because the exact role of αvβ3 in intimal hyperplastic lesion formation and restenosis remains unknown, it became critical to re-examine the precise mechanism of action of αvβ3 in cell culture and in animal models. Indeed, Azrin et al34 tested in a hypercholesterolemic rabbit model of balloon angioplasty with pre-existing atherosclerotic lesions an antibody (AZ-1) that binds to the rabbit platelet αIIbβ3 and inhibits platelet function in vivo. No significant differences were found in intimal hyperplastic lesion formation between the AZ-1 antibody-treated and control groups 4 weeks after angioplasty. In this case, the αIIbβ3 antagonist failed to inhibit restenosis in the setting of pre-existing intimal lesion, similar to the human clinical situation where SMC migration is not a requisite for intimal lesion generation.

Although the case could be made only against the αIIbβ3 antagonists in the treatment of intimal hyperplasia, Deitch et al35 reported that ReoPro failed to reduce intimal hyperplastic lesion formation in atherosclerotic nonhuman primates after angioplasty and stenting in separate arteries, suggesting that blockade of SMC migration is not critical in the setting of pre-existing intimal SMCs.

Since the publication of the reports on the ineffectiveness of αvβ3 blockade on the intimal hyperplasia development in complex animal models, Smyth et al42 used a combination of guidewire-induced endothelial denudation and arterial ligation to demonstrate that β3-integrin deficiency (β3–/–) did not have a role in intimal lesion formation. Our group, however, later classified the injury methodology by creating three distinct injury patterns that differed in the extent of medial injury induced in these β3–/– mice: (1) guidewire probe–induced transmural injury with medial disruption, (2) nonmedial disruptive ligation injury, and (3) eccentric medial disruptive injury, followed by arterial ligation.43 We believed that injury induced by guidewire probe generated more transmural mechanical damage to the media over a longer segment of the vessel compared with the ligation injury, which generates a more modest, focal lesion with stagnant flow and thrombosis.

As before, we showed that β3-integrin deficiency did not protect against neointimal lesion formation after a significant medial disruption seen with guidewire probe injury. In contrast, in the setting of arterial ligation injury, β3-integrin deficiency protected against neointimal lesion formation at 1, 2, and 3 weeks, and 3 months after injury. When the combination of medial disruption and arterial ligation was used in β3–/– mice, eccentric neointimal lesion formation occurred only at the site of disruption. The lack of neointimal lesion formation on the opposite, nondisrupted section is consistent with the dependence of neointimal formation on the mechanical disruption of the internal elastic lamina and media as described by others.44

One satisfactory explanation for only discrepancies from the arterial injury patterns in β3–/– mice is that different models or methodologies accentuate the various, distinct functions of β3 integrins. For instance, Carmeliet et al45 compared intima formation induced by mechanical injury in mice deficient in plasminogen activator inhibitor 1 (PAI-1) and in wild-type mice and demonstrated that PAI-1 blocks intimal thickening by inhibiting the migration of SMCs. In contrast, Peng et al46 demonstrated that when ligation-induced intima formation was examined in PAI-1+/+ and PAI-1–/– mice, PAI-1 promoted neointimal thickening. One can conclude that these injury models emphasize a contrasting cascade of events despite the apparently simple injuries.

Moreover, Tanaka et al47 showed in a carotid ligation injury model that there is minimal bone marrow–derived cell contribution to the neointimal lesion development. Therefore, in the carotid ligation model of intimal hyperplasia, the seminal event appears to be a directional cellular migration from the media to the neointima with the aid of αvβ3, with little or no contribution from the bone marrow (Fig, B). Hence, although β3-integrin blockade effectively reduces neointimal hyperplasia in animal models, this blockade may not be effective for prevention of neointimal lesion formation in the less defined, more disruptive injury induced by percutaneous transluminal coronary angioplasty in human coronary arteries (Fig, A).

View Large Image Download to PowerPoint

Fig. Leading animal models of arterial injury. A, The balloon angioplasty model results in a disruptive injury to the intima and media leading to neointimal hyperplastic lesion formation with a significant smooth muscle cell (SMC) contribution from the bone marrow. B, The carotid ligation (flow cessation) model results in little or no disruptive injury to the media. The seminal event appears to be a directional SMC migration from the media to the neointima with little or no contribution from the bone marrow. C, Like the carotid ligation model, the critical event appears to be SMC migration with minimal or no medial disruption and bone marrow contribution.

Future directions 

Complications with hemodialysis access constitute a major cause of morbidity for patients with end-stage kidney disease. In the United States alone, expanded polytetrafluoroethylene (ePTFE) grafts are used for permanent vascular access in approximately 70% of the 250,000 patients receiving hemodialysis.48 Currently, the primary patency rates of these ePTFE grafts at 1 and 2 years are 50% and 25%, respectively, while hospitalizations related to hemodialysis access cost well over $1 billion dollars per annum.49

The failure of hemodialysis access grafts is predominantly due to a neointimal hyperplastic response in the region of the venous anastomosis resulting in reduction of shunt flow and ineffective hemodialysis. The access then needs to be revised by an open surgical revision or a percutaneous angioplasty or stenting, or both. By then, placement of another hemodialysis access at a difference site is not far off, quickly exhausting the available sites predominantly in the upper extremities.

Castier et al50 recently created an arteriovenous fistula model in mice that demonstrated a rapid neointimal hyperplasia development at the anastomosis, the site most relevant to the clinical problem of venous neointimal hyperplasia and acute thrombosis. This arteriovenous fistula model results in traumatic injury to the blood vessels involved and turbulent blood flow near the anastomosis, along with a compliance mismatch between artery and vein, which are believed to be factors that produce the rapid neointimal lesion formation. They further demonstrated that like the arterial ligation injury model, the neointimal SMCs of the arteriovenous fistula anastomosis do not originate from bone marrow stem cells. Hence, they have demonstrated that this animal model and the clinical situation of arteriovenous access surgery for hemodialysis are uniquely suited to target the seminal event in the formation of the neointimal lesion formation, the SMC migration (Fig, C).

Unlike intimal hyperplasia seen with preocclusive atherosclerotic arteries after angioplasty and stenting, neointimal hyperplasia is seen with an anastomosis involving a synthetic graft (eg, ePTFE, Dacron) and a relatively disease-free segment of vein or artery. There is thus no preprocedural, stenotic intimal plaque with abundant resident SMCs, and therefore, adhesion and directional migration (relocation) of SMCs into the provisional matrix on the luminal surface are indeed the seminal events, not unlike the invasive tumor cells (metastasis).

In this setting of end-stage kidney disease and arteriovenous access, targeting αvβ3 integrin could have a significant impact on the prevention of neointimal hyperplasia. Indeed, animal studies examining the role of the αvβ3 antagonists on the long-term patency of the arteriovenous accesses need to be performed in the setting of uremia, and then perhaps properly designed clinical trials in this patient population with kidney disease might ultimately provide a clinical problem to this old solution.