Controlling the angiogenic switch in developing atherosclerotic plaques: Possible targets for therapeutic intervention

In normal arteries, adventitial blood vessels originating from the vasa vasorum penetrate into the vessel wall at regular intervals and bifurcate around the circumference of the artery. They don't extend beyond the outer media, and diffusion of oxygen and nutrients is limited to approximately 100 μm from the vessel lumen [6]. Angiogenesis is a recognized feature of the atherogenic process in both coronary and carotid disease, intimal neovascularization arising frequently following activation and proliferation of the dense network of vessels in the adventitia adjacent to a plaque [7]. New vasa vasorum (VV) have been identified in developing lesions and these invade the intima at specific sites of medial disruption, whilst the intima and media of complicated plaques in coronary atherosclerotic vessels have been shown to be infiltrated with a tumor-like mass of microvessels which are prone to leak [8]. These angiogenic blood vessels have been identified in complicated regions of human aortic lesions following immuno-staining with antibodies to CD105 (endoglin; which binds only to active endothelial cells) and transforming growth factor-beta 1 (TGFβ-1) [9]. Immature neovessels are most often localized in regions of inflammatory/macrophage cell infiltration (and less common in highly calcified or hyalinized plaques) and at the shoulders of thin-cap atheromas prone to rupture [10]. In a study of coronary artery atherogenesis, lesions were examined from coronary arteries from excised hearts of patients. The highest neovessel content was demonstrated in type VI plaques and this was associated with the highest rate of thrombotic episodes [11], whilst other post-mortem studies showed approximate increases in vessel density of twofold in vulnerable plaques and fourfold in disrupted plaques compared with severely obstructed but stable lesions [12]. Similarly, increased angiogenesis and angiogenic gene expression measured using microarrays, together with significantly thinner plaque fibrous caps was associated with symptomatic carotid atherosclerosis [13].

Some earlier studies failed to identify a direct relationship between plaque angiogenesis and instability and direct evidence for the contribution of VV proliferation and/or intimal neovascularization as a cause of plaque rupture is still limited. Part of the reason for this may have been the lack of suitability and similarity between experimental models of atherosclerosis designed using small mammals, which often produce small plaques with limited neovascularization, possibly due to the much smaller diameter of the vessels and media thickness, compared with human vessels and hence many people do not believe they can be used reliably to predict development of unstable lesions in humans [for a review see, [14]]. Several patient studies also demonstrated that plaque haemorrhage in human coronary atherosclerosis was not always a feature of intimal areas with high microvessel density [15], whilst intimal microvessels were only detected in 64% of restenotic human coronary atherosclerotic specimens [16]. However, although the presence of a vascular cell neointimal region is not a prerequisite for unstable plaque development, overwhelming evidence now points to this process as being a key instigator of plaque fragility leading to rupture. Various studies have demonstrated a clear association between VV activation and neointimal formation. In animal models, VV angiogenesis enhanced neointima formation in both the rabbit collar and rat angioplasty models of carotid artery injury, whilst perivascular adenoviral gene transfer of dominant negative FGFR-1 significantly attenuated VV angiogenesis in response to injury [17]. Further evidence is supplied below:

The irregular nature of blood vessel formation, and its similarity to tumor angiogenesis, suggests that the factors responsible for their growth may be different from those seen during normal wound healing. Newer technological approaches have been able to discriminate between localised micro-environments in individual neointimas using antibodies directed to proteins expressed in actively growing vessels. These studies showed the presence of areas of mature vessel development where haemorrhage was absent as opposed to those containing irregular immature and highly angiogenic vessels where haemorrhage was taking place, and which may help to explain these earlier contadictory findings. For example, Dunmore et al, [18], demonstrated that irregular dysmorphic heamorrhagic vessels without smooth muscle cell coatings, were almost exclusively found in plaques from patients with symptomatic carotid artery disease. Furthermore, these single-layer endothelial tubes co-localized specifically with pockets of VEGF expression providing strong support for the idea that proliferating immature neovessels may be associated with plaque instability and rupture. Similarly, haemorrhagic, leaky blood vessels from unstable carotid plaques proliferate abnormally, and although of relatively large caliber remained immature by virtue of a poor investment with smooth muscle cells and pericytes. These vessels also possessed structural weaknesses with poorly formed EC junctions that may also contribute to instability of the plaque by facilitation of inflammatory cell infiltration and haemorrhagic complications [19], (Figure 1; [20]). In vivo studies showed a strong correlation between areas of increased vascularity, intraplaque haemorrhage and neurological events defined by CT, with subsequent histological staining using anti-CD34 antibodies visualizing EC of the vessels in symptomatic patients following endarterectomy [21].

Intraplaque haemorrhage results in rapid expansion of the plaque necrotic core, due to the fact that red blood cell membranes are a rich source of free cholesterol and phospholipids and the process occurs in association with excessive macrophage infiltration [22]. The size of the necrotic core directly correlates with the risk of plaque rupture. Furthermore, intraplaque haemorrhage and plaque rupture were proportional to neovessel density in coronary atheroma [23]. Adventitial vessels in unstable plaques contain perivascular smooth muscle cells, however, after plaque rupture, the fibrous cap is disrupted with a luminal thrombus and the newer branches of vasa vasorum close to the necrotic core consist almost entirely of a single layer of EC overlying a ruptured, leaky basement membrane, and associated with remnants of red blood cells [24]. Defects are thought to be caused by proteolytic damage from on-going inflammation and release of signalling molecules affecting cell-cell contact.

In the quiescent state, VV of normal arteries provide nutrients to the tunica media whilst the intima is fed by oxygen diffusion from the lumen. The VV remains in a dormant state probably due to the expression and synthesis of anti-angiogenic factors e.g. thrombospondin and endostatin which more than counterbalance the presence of low quantities of pro-angiogenic factors in the micro-environment [39]. However during intimal plaque development, the balance between the angiogenic and anti-angiogenic factors may become altered with increased production of growth factors and cytokines together with a reduction in negative modulators resulting in adventitial vessel angiogenesis at the site proximal to internal vascular damage and plaque growth [40]. Hence the angiogenic switch shifts to on.

Key molecules involved in initiation and/or maintenance of the angiogenic process, and which may prove to be potential therapeutic targets of modulation include the angiopoietin-Tie signaling pathways. Ang-1 and -2 are ligands of the endothelial receptor Tie-2. Ang-1 via Tie-2 binding induces formation of stable blood vessels, whereas Ang-2 destabilizes the interaction between EC and their support cells [41]. In a recent study by Post et al [42], a positive correlation was observed between the ratio of expression of Ang-1/Ang-2, the macrophage count and mean vessel density in complicated fibroatheromatous carotid human plaques. Ang-2 also correlated with the activity of matrix metalloproteinase-2 (MMP-2) expression suggesting a role in development of unstable plaque microvessels. Tie-2 is a tyrosine kinase receptor expressed predominantly on EC, and which has been shown to activate intracellular signal transduction through Akt and ERK1/2 following ligand binding of angiopoietin-1 resulting in angiogenesis both in vitro and in vivo. In relation to tumour neovascularization, expression of Tie-2 was concentrated in vascular hot-spots at the leading edge of invading tissue [43]. Furthermore, Ang-1 is an anti-inflammatory cytokine which can reduce neovessel leakage and vascular permeability partly via inhibition of cell membrane adhesion molecule expression, whilst an increase in Tie-2 together with increased availability of Ang-1 may enhance neointimal vascularization and promote formation of stable vessels less prone to haemorrhage [44]. Placental growth factor (PIGF), a member of the VEGF family, binds to the Flt-1 receptor, is a mediator of inflammation and is highly expressed in atherosclerosis, correlating with both levels of plaque inflammation and microvascular density in symptomatic patients with carotid plaques [45]. Many other pro-angiogenic factors have been demonstrated to be present in neointimal lesions although direct evidence for their involvement in mediating vascularization in vivo is still lacking and hence they have not been described in this review. Worthy of note, however, is the work of Leroyer et al [46] showing that atherosclerotic plaques contained a large number of membrane-shed microparticles originating from the breakdown of apoptotic macrophages. Isolated microparticles were highly angiogenic both in vitro and in vivo and those obtained from symptomatic patients expressed more CD40L and were more potent in mediation of angiogenesis which was blocked in the presence of CD40-neutralising antibodies, suggesting a novel mechanism for stimulation of plaque neovascularization.

Key inhibitors of angiogenesis may be reduced in areas undergoing neovascularization and the balance of pro-angiogenic and anti-angiogenic molecule expression shifts towards blood vessel formation. Thrombospondin-1 (TSP-1) has been implicated in development of vascular diabetic complications and is expressed in developing arterial plaques [47]. Platelet factor IV is expressed by active platelets in plaques [48], whilst increased expression of Collagen XVIII and its proteolytically released endostatin fragment is associated with inhibition of angiogenesis in the vascular wall, and it has been shown that ApoE knockout collagen XVIII minus animals developed significantly more vasa vasorum than wild type [39]. T-cadherin, is involved in maintenance of vascular integrity and growth, is up-regulated in developing atherosclerotic lesions. The effect of T-cadherin on angiogenesis in vivo was evaluated in detail by Rubina et al, [49] using a Matrigel implant model. They demonstrated that T-cadherin over-expression in L929 cells injected in Matrigel inhibited neovascularization of the plug. In vitro, T-cadherin inhibited the directional migration of endothelial cells, capillary growth, and tube formation but had no effect on endothelial cell proliferation, adhesion, or apoptosis.

In vascular plaques, abnormally behaving cells are surrounded by heterogeneous tissue elements, such that the areas of interest/diseased cells may constitute less than 5% of the volume of a tissue biopsy specimen and also may consist of dynamic micro-hotspots which are difficult to analyse in great detail using conventional dissection and non-quantitative global gene or protein microarrays. Hence, techniques such as laser-capture micro-dissection for isolation of specific microvessels in both the VV and intima of evolving lesions could help to identify both the genetic fingerprints associated with blood vessel activation and stimulatory factors associated with the surrounding inflammatory micro-areas in complicated unstable plaque regions. This technique has been used previously successfully by Roy et al, [56] who compared global gene expression between blood vessels isolated by laser-capture from normal skin and identified important de-regulated genes including heparin sulphate 6-O-endosulphatase1 (anti-angiogenic) and a reduction in CD24 (angiogenic) in vessels isolated from within chronic wounded tissue. Our recent un-published work employing Real-time PCr, TaqMan angiogenesis-directed microarrays has demonstrated concurrent over-expression of NOTCH-3/DLL4 and Ang-1/Tie-2 proteins as well as KDR, HBEGF and RAGE in the same microvessels from micro-dissected angiogenic (CD105-positive) vessels from unstable human endarterectomy specimens, whilst these proteins were weakly expressed or not expressed in the same lesion/section in an area containing stable inactive vessels, suggesting a complex but specifically localized activation and signaling cascade is involved in mediating vascularization in growing neointimal lesions. This suggests that actively growing neovessels are re-expressing these embryologically derived genes which have been shown to regulate formation of tip cells during angiogenesis, and therefore may form part of an attempt to regulate formation of stable vessels which could be less prone to leakage [57]. Since inhibition of Notch signalling has been shown to increase the total number of vessels in solid tumour models but reduce the patency and vascular function of the remaining vessels (resulting in "chaotic" vessels), it may have similar effects on neointimal vessels and supplementation could be considered as part of a therapeutic programme of normalization and stabilization of existing vasculature promoting plaque stability. Some of the current research is directed to endothelial progenitor cells. They appear to be mobilized from the bone marrow by angiogenic factors to the adventitial space, and recent studies have shown that progenitor cells in the adventitia can contribute to progression of atherosclerosis, and EPC concentrations were significantly associated with severity and presence of multivessel coronary artery disease [58]. It is possible that these progenitor cells may be involved in development of new vasa as well as neointimal leaky vessels and therefore future therapeutics might also consider methods to restrict their differentiation in angiogenic microenvironments of developing plaques.

Accurate identification of VV susceptible to development of hypoxic microenvirnments is likely to become a possibility in the next few years with the development of contrast-enhanced and targeted ultrasound imaging (Magnoni et al, 2009; Granada et al, 2008) [59, 60]. Adenoviral delivery of genes, directed against angiogenic molecules, to the perivascular surface of these VV in order to inhibit their proliferation would represent a first line of attack, whilst pharmacological treatments to reduce stem cell differentiation and/or reduce local inflammation and cytokine release, appears to be the most likely avenue for future combinational therapies [[61]; Figure 2]. Weak neovascular beds in plaque intima as well as activated adventitial blood vessels are potential targets for molecular imaging and targeted drug therapy, however, the majority of potential imaging and therapeutic agents have been unsuccessful because of their limited capacity to reach and remain stably within the target tissue or cells in vivo. The idea of using peri-vascular combinational gene therapy for inhibition of VV activation was mentioned in an earlier section. Advantages of this outside-in mechanism is that the integrity of the vessel wall would remain intact and high concentrations of gene delivery vectors can be retained in the local environment, producing minimal inflammatory reaction, when compared with current systemic therapies [62].

Molecular imaging for the visualization of vulnerable atherosclerotic plaques may be useful in detecting patients at risk of thrombosis and by recognition of individual cellular and molecular characteristics, provide the potential for offering personalized medical treatment. For example, when large numbers of paramagnetic gadolinium complexes (>50,000) are incorporated onto emulsion particles, the signal enhancement for each binding site is magnified dramatically compared with conventional contrast reagents [[63] and references therein]. Neovessels in coronary plaques of cholesterol-fed apoE mice have been imaged using vascular cell adhesion molecule-1 peptide sequence (VHSPNKK) bound to cross-linked magneto-fluorescent super-paramagnetic iron oxide (CLIO) particles. The particles were shown to selectively bind to aortic plaque vasculature 24 h after injection using MRI and ex-vivo MR and corresponded with histology and fluorescent analysis of tissue samples performed after euthanasia, suggesting potential diagnostic and therapeutic applications [64]. In humans, contrast-enhanced carotid ultrasound imaging was able to identify intra-plaque neovascularization which correlated with later histological findings employing the use of antibodies directed to CD31 and CD34 for identification of the vessels [65]. Similarly, α₅β₃ integrin targeted perfluorocarbon nanoparticles have been successfully used to detect neovasculature in a rabbit model of aortic valve disease suggesting a potential application in humans. In addition, the same technology could be used to home to angiogenic areas or microenvironments, and release targeted knockout genes in order to block VV proliferation, and thereby reduce neointimal expansion or to inhibit neointimal angiogenesis directly or maturate existing fragile rupture-prone vessels. It is important to remember that bioactive molecules of the extracellular matrix including proteoglycans such as hyaluronan whose turnover as regulated by growth factors such as transforming growth factor-β and proteases including matrix metalloproteinases, would also be considered potential therapeutic targets for controlling the angiogenic switch in specific susceptible microenvironments [66]. Anti-angiogenic agents have previously been shown to "prune" tumour vessels and normalize the structure of remaining vessels and therefore might help to stabilize rupture-prone plaques via a similar mechanism [67]. In relation to the process of angiogenesis, fragile endothelial cells of angiogenic neovessels express specific markers such as α₅β₃ integrin, CD105 receptors VEGF-R2 and/or apelin [68], which could be used in this targeting protocol.