Plaque-Associated Vasa Vasorum in Aged Apolipoprotein E–Deficient Mice Exhibit Proatherogenic Functional Features In Vivo

Introduction

Human atherosclerotic plaques are characterized by local expansion of the vasa vasorum and development of intraplaque microvasculature. The extent of vascularization has been shown to correlate with progression stage, inflammatory content of the plaque, and plaque instability.^1,2 Ever since Köster suggested that advanced atherosclerotic lesions are associated with an increased density in vasa vasorum,³ a growing number of studies have shown that both intraplaque microvasculature and vasa vasorum expansion enhance human atherosclerotic plaque progression.^1,4,5 Intraplaque microvessels are thought to be derived mainly from the vasa vasorum by angiogenic sprouting, followed by ingrowth into the lesion.^6–8 The angiogenic response itself is considered to arise from the gradual development of lesional hypoxia attributable to (1) increased physical distance to the lumen, exceeding the diffusion distance of oxygen, and (2) increased local metabolic activity of the plaque’s inflammatory content.^1,9 Additionally, the endothelium of the newly formed microvessels is prone to continuous exposure to inflammatory mediators, which could lead to impaired functionality and increased permeability, caused by deteriorated interendothelial junctions and basement membrane detachment.¹⁰ Dysfunctional microvessels would favor influx of circulating erythrocytes, leukocytes, and blood platelets into the plaque,^6,7,11 thereby promoting plaque expansion, destabilization, and thrombogenicity.

Despite the clinical relevance of plaque neovascularization, experimental animal studies on functionality of plaque microvessels have been hampered by their low incidence, or even lack thereof, in mouse models of atherosclerosis. Proangiogenic interventions in murine models have shown to increase plaque development and disease burden,^12–16 whereas antiangiogenic therapy decreases atherosclerosis development,^17–20 implicating a potent role for angiogenesis in experimental atherosclerosis models. However, the interventions also influenced the inflammatory response, thereby affecting lesion development independent of plaque neovascularization. In general, animal studies have failed to show a direct causal relation between plaque vascularization, on the one hand, and plaque progression and destabilization, on the other.

Lack of adequate detection tools also hampered functional studies of plaque microvessels. Several methods like magnetic resonance imaging^21,22 or micro- and nano-computed tomography²³ have been applied to investigate plaque vascularization in murine models of atherosclerosis. These methods lack the spatial and temporal resolution to detect functional and structural properties of individual microvessels, or to distinguish between the artery and microvessels. Also, these techniques lack the ability to detect multiple imaging probes simultaneously. Multiphoton laser scanning microscopy (MPLSM) can overcome these shortcomings and has previously been used for in situ imaging of healthy and atherosclerotic arteries at subcellular resolution,^24,25 as well as for quantifying angiogenic activity in models of tumor vascularization and myocardial infarction.^26,27 In addition, MPLSM has been used to image large blood vessels in vivo, triggered on heart rate and respiration to reduce motion artifacts.^28,29 More recently, MPLSM was used to quantify blood flow and hematocrit in blood vessels,³⁰ substantiating MPLSM as a valuable tool for functional studies of the (micro)vasculature.

Here, we studied plaque microvessels to elucidate their structural and functional characteristics. Using MPLSM imaging in a novel in vivo imaging strategy, plaque microvessels could be visualized in aged apolipoprotein E–deficient (ApoE^−/−) mice. More importantly, angiogenic activity could be visualized and functional characteristics, including blood flow, adhesion and extravasation of leukocytes, and microvascular permeability, could be quantified and were found to differ significantly from control microvascular beds.

Methods

Animals

For ex vivo and in vivo experiments, aged ApoE^−/− mice (40–50 weeks of age; n=51) with advanced atherosclerotic plaques were used. Young ApoE^−/− mice (12–16 weeks of age; n=8), without or with only initial atherosclerotic plaques, and age-matched wild-type C57Bl/6 mice (n=20) served as controls. Mice were supplied by the institutional animal facility and were housed in conventional cages and received a normal chow diet. Experiments were approved by the local animal ethics committee and performed according to institutional guidelines.

Imaging

MPLSM imaging was performed on a Leica SP5 imaging platform (Leica Microsystems) that integrates multiphoton microscopy with fast resonant scanning, and uses a Compact Ultrafast Ti:Sapphire Laser (Chameleon, Coherent). Anti-CD31 antibody (BD Pharmingen) was used to detect vascular endothelium, and cyclic Asn–Gly–Arg was used to target CD13, a marker of angiogenic vascular endothelium. Biotinylated anti–Ephrin-B2 and anti–EphB4 (R&D Systems) antibodies were used to assess arterial or venous lineage of microvessels, respectively. Biotinylated antibodies were conjugated with streptavidin-quantum dot 525 (green) or streptavidin-quantum dot 585 (red) as indicated. To examine perfusion of microvessels and microvascular permeability, 70- and 150-kDa fluorescein isothiocyanate-labeled dextrans (Sigma) were used. For studying blood flow and leukocyte trafficking, leukocytes were labeled using Rhodamin 6G or Cell Tracker Red (CMTPX, Invitrogen).

Statistics

Data are presented as mean±SEM, unless stated otherwise. Statistics were performed using Graphpad Prism 5.0 or PASW Statistics 18. The nonparametric Mann-Whitney U test was applied to adhesion, extravasation, and blood flow velocity data. Microvascular permeability data were analyzed using curve fitting regression analysis. Fisher exact test was used for examining differences between groups for presence of plaque-associated microvasculature. Results were considered statistically different when P<0.05. Expanded methods are available in the online-only Data Supplement.

Results

Vasa Vasorum Were Visualized and Morphologically Characterized In Vivo

The vasa vasorum were studied in detail using in vivo and in situ MPLSM after labeling with anti-CD31, allowing characterization of microvessel morphology throughout the adventitia. Characteristics like microvessel diameter were similar in the adventitia of the carotid artery and abdominal aorta (Table 1). Based on the size and orientation, 2 types of plaque-specific microvessels could be discriminated. The larger vasa vasorum microvessels (median diameter, 18.5 μm) generally ran in parallel to the artery, whereas smaller microvessels (median diameter, 5.9 μm) mostly covered the artery radially. The adventitia surrounding the atherosclerotic plaque showed several microvessels (Figure 1A and 1B, Movies I and II in the online-only Data Supplement), which was confirmed by immunohistochemistry for CD31 and von Willebrand factor (Figure I in the online-only Data Supplement). Microvessels in peri-adventitial fat were excluded from analysis based on collagen imaging of the adventitia using second harmonic generation and imaging of autofluorescence of the peri-adventitial fat. Microvessels were mostly situated within 40 to 80 μm from the outer elastic lamina, but occasionally—at sites of large advanced plaques—microvessels would be situated within 15 to 20 μm from the outer elastic lamina. Vasa vasorum were exclusively located at atherosclerotic foci around the carotid bifurcation and were neither observed further upstream at the nondiseased common carotid artery nor in younger ApoE^−/− mice or age-matched C57Bl/6 controls (Table 2). Individual endothelial cells (Figure 1C, arrowheads) of the vasa vasorum and their nuclei (Figure 1C, arrows) could be delineated.

• Download figure
• Download PowerPoint

In vivo measurements displayed a predominantly arterial phenotype of the vasa vasorum (Figure 1D), as observed by Ephrin-B2 and Eph-B4 staining. Moreover, on local or bilateral ligation of the jugular vein for up to 30 minutes, functional parameters (eg, microvessel diameter) remained unaltered (Figure II in the online-only Data Supplement), arguing a marginal role of the venous system in controlling the vasa vasorum.

Plaque-Associated Vasa Vasorum Showed Angiogenic Activity In Vivo

The particular localization of the microvasculature at sites of plaque formation and absence of microvasculature in young atherosclerotic mice suggests angiogenesis underlies the development of these microvessels. Therefore, a cyclic Asn–Gly–Arg probe targeting the angiogenic marker CD13 and specific for angiogenic endothelium on in vivo injection²⁶ was used to investigate whether ongoing angiogenesis was present. The vasa vasorum showed CD31⁺/CD13^– microvessels accompanied by occasional CD31^low/CD13⁺ sprouts (Figure 1E). At the site of sprouting, part of the endothelium expressed both CD31 and CD13 (Figure 1E, inset).

Plaque-Associated Vasa Vasorum Were Actively Perfused and Hyperpermeable

Because the normal physiology and perfusion of vasa vasorum microvessels were unknown, this was assessed by injection of 70-kDa and 150-kDa fluorescein isothiocyanate-dextrans, which indicated active perfusion (Figure 1F, green) of the vasa vasorum (red: CD31) in vivo. Also, sprouting of new microvessels was observed (Figure 1F, arrowhead), with new sprouts displaying low CD31 expression. Additionally, a rise in fluorescence intensity (ie, dextran concentration) in the adventitia was observed (Figure 1F, asterisk).

Concurrently, dextran injection allowed measurement of microvascular permeability by calculating the time-dependent rise in extraluminal fluorescence intensity compared with intraluminal fluorescence intensity. Although intraluminal dextran levels remained constant for >60 minutes, the extraluminal content of 70-kDa dextran increased with time to an extraluminal/intraluminal ratio of ≈0.46 after 35 minutes in plaque-associated vasa vasorum (Figure 2A-1 and 2B-1). Because vasa vasorum, in accordance with literature,^19,31 was not observed in nondiseased arteries, we used the similar sized microvasculature of the nearby sternohyoid muscle as a control. In the sternohyoid muscle, only a limited increase in extravascular dextran concentration was observed (Figure 2A-2 and 2B-2). On quantification, extravascular dextran concentration (Figure 2C, black line, P<0.05) was significantly increased only in vasa vasorum, but not in control microvessels (Figure 2C, dotted line), indicative of enhanced microvascular permeability in the vasa vasorum. Additionally, extravasation of 70-kDa dextran occurred earlier in vasa vasorum compared with control microvessels (Figure 2C). Because of myocyte autofluorescence, baseline extra/intraluminal fluorescence ratios varied slightly; however, all experiments showed an identical pattern of significantly increased microvascular permeability of the vasa vasorum. Of note, microvascular extra-vasation of 150-kDa dextran did not significantly increase over time, neither in the vasa vasorum or in control microvessels (not shown).

• Download figure
• Download PowerPoint

A size-dependent increase in permeability could be caused by several mechanisms. It is known that reduction or absence of endothelial glycocalyx in microvessels can, independently from the interendothelial junctions, lead to increased permeability and leukocyte adhesion.³² In addition, disruption of the endothelial barrier may cause hyperpermeability. To determine what caused the increased permeability, we examined glycocalyx integrity by wheat germ agglutinin staining and analyzed the integrity of the interendothelial junctions by electron microscopy. Staining for wheat germ agglutinin in vivo showed an irregular staining pattern in the vasa vasorum (Figure 2D), contrary to an evenly distributed expression of glycocalyx throughout control microvessels in the sternohyoid muscle (Figure 2E; Figure III in the online-only Data Supplement). The interendothelial junctions were intact as shown by electron microscopy (Figure 2F and 2G; arrows), and occasionally, pericytes (P) were found covering the microvascular endothelium. The endothelial cells did, however, show initial signs of membrane blebbing, indicative of early endothelial dysfunction.

Plaque-Associated Vasa Vasorum Showed Decreased Blood Flow, Yet an Increased Adhesion and Extravasation of Leukocytes

Leukocytes were observed within the vasa vasorum after in vitro (Figure 3A) and in vivo (Figure 3B, Figure II in the online-only Data Supplement) labeling. Using labeled leukocytes and platelets, blood flow velocity and blood flow were assessed in the carotid artery, vasa vasorum, and control microvessels. Both blood flow velocity (2.1-fold decrease, P<0.001; Figure 3C, Table I in the online-only Data Supplement) and blood flow (7.6–9.8-fold decrease, P<0.001; Figure 3D) were significantly reduced in vasa vasorum compared with control microvessels. Changes in blood flow and flow velocity were independent of microvessel diameter. Blood flow direction was identical to the main artery.

• Download figure
• Download PowerPoint

Additionally, adhesion and transmigration of ex vivo and in vivo labeled leukocytes were significantly increased compared with control microvessels (Figure 4). Increased adhesion (Figure 4A, 1.3×10⁻²±0.3×10⁻² versus 0.2×10⁻²±0.1×10⁻² cells/mm² per minute, P<0.05) and extravasation (Figure 4B, 1.8×10⁻²±0.4×10⁻² versus 0.3×10⁻²±0.2×10⁻² cells/mm² per minute, P<0.05) of leukocytes were observed in plaque-associated vasa vasorum and control microvessels, respectively. Leukocyte adhesion and extravasation were examined in more detail ex vivo, using 3-dimensional image analysis. Cell morphology suggested transmigration of different leukocyte subtypes, including pseudopodia-rich cells resembling a dendritic morphology, from the vasa vasorum toward the plaque (Figure 4C and 4D, Movie III in the online-only Data Supplement).

• Download figure
• Download PowerPoint

Ex Vivo MPLSM Confirmed Presence of Vasa Vasorum and Intraplaque Microvascular Structures

Plaque-associated vasa vasorum were also imaged ex vivo, allowing imaging at higher resolution and signal-to-noise ratio without motion artifacts. For ex vivo imaging, the carotid artery or abdominal aorta was excised, mounted in a perfusion chamber and imaged under a transmural pressure of 80 mm Hg to mimic physiological conditions.²⁴ Ex vivo imaging was performed up to 30 μm from the outer elastic lamina. Vasa vasorum were detected both in the carotid artery and abdominal aorta (Table 1).

Ex vivo imaging clearly visualized intraplaque microvascular structures (Figure IVA and IVB in the online-only Data Supplement), yet intraplaque microvessels were only found in 6 of 20 plaques (Table II in the online-only Data Supplement). Surprisingly, none of these microvessels derived from the vasa vasorum. Rather, data suggested that these microvessels originated from endothelial sprouting from the luminal endothelial layer (3 of 6 plaques; Figure IVC in the online-only Data Supplement). No functional characteristics could be determined for intraplaque vessels, arguing a limited role for intraplaque microvessels in this model. Also, occurrence of intraplaque hemorrhage (IPH) was determined in aged ApoE^−/−, and was found in 4 of 20 (20%) mice examined. Although all mice that showed IPH also showed vasa vasorum, only a minor fraction of vasa vasorum–containing plaques showed features of IPH. Occurrence of IPH could not be linked to occurrence of luminal microvascular structures.

Discussion

The presence and role of plaque-associated microvessels in murine plaques have been extensively debated in literature. Previous attempts to detect these microvessels yielded varying results, in part depending on the animal model and visualization techniques used. Here, we used in vivo MPLSM to show that plaque-associated vasa vasorum (1) are present in ApoE^−/− mice; (2) show angiogenic activity; (3) are actively perfused and hyperpermeable; (4) have significantly reduced blood flow and blood flow velocity compared with control microvessels with similar diameter, and (5) show enhanced leukocyte adhesion and transmigration at the site of the atherosclerotic plaque. Taken together, we show that plaque-associated microvessels exhibit functional features that could facilitate plaque development and progression.

The increased adhesion of leukocytes may, in part, be a consequence of the reduction in blood flow velocity, because this would cause a reduction in local shear stress at the microvascular wall. Consequently, low shear stress is deemed to cause endothelial cell dysfunction and upregulation of adhesion molecules like vascular cell adhesion molecule-1 and intercellular adhesion molecule-1.^33–35 The increased leukocyte adhesion was accompanied by augmented transmigration to the adventitial tissue, facilitating a second site of ingress of leukocytes, besides migration through the luminal endothelium. These cells may contribute to further plaque development, either by functioning in the adventitia itself, or by infiltration into the plaque after migrating through the media. As such, the vasa vasorum could contribute to exacerbation of atherosclerosis, by infiltration of new inflammatory cells.³⁶

Additionally, vasa vasorum hyperpermeability may be another important contributor to plaque progression, as it favors extravasation of lipids and plasma proteins at sites of plaque development.³⁷ Increased permeability may be attributed to disruption of the endothelial barrier function.¹⁰ Our finding of augmented microvascular extravasation in the vasa vasorum, at least for molecules with a molecular weight of up to 70 kDa, suggests that abundant plasma proteins, such as albumin, can leak into the adventitia. Indeed, extravascular albumin has been detected in both human and mouse atherosclerotic lesions.^38,39 Here, we extend these findings, demonstrating an apparent size limit for endothelial leakage in vasa vasorum of 70 to 150 kDa in vivo in aged ApoE^−/− mice. The origin of the microvascular hyperpermeability remains to be further elucidated. Electron microscopy did not show any aberrations in endothelial cell junctions, concordant with a recent study from Eriksson.³¹ However, the endothelial glycocalyx of plaque-associated vasa vasorum was disrupted, indicating a possible cause for the increase in vascular permeability, and more specifically, permeability for smaller molecules. Both the increased leukocyte extravasation and hyperpermeability will affect plaque development and progression, especially when plaques become more advanced and vasa vasorum has expanded by increasing adventitial inflammation. As such, these processes could enhance plaque progression independent of microvascular disruption and hemorrhage.

Vasa vasorum were abundantly present at sites of atherosclerotic plaques, sites at which we also observed an increase in vessel wall thickness (not shown). Thickening of the vessel wall may infer an increase in vascularization, because oxygenation of the entire tissue is hindered by vessel wall thickening, potentially increasing local hypoxia. Remarkably, although the entire carotid artery was examined, we did not observe any vascularization of nonatherosclerotic segments, neither in control animals nor in the young or aged ApoE^−/− mice. This is consistent with reports by, for example, Gräbner et al⁴⁰ or Eriksson.³¹ Given the fact that vasa vasorum were absent in nonatherosclerotic segments, we chose the sternohyoid microvasculature as a substitute model for general microvascular blood flow. Taken together, development of vasa vasorum in ApoE^−/− mice appears to be an adaptive response to pathological, atherosclerosis-induced vessel wall changes.

Many studies have suggested a role for angiogenesis in atherosclerosis. Proatherogenic effects of increased angiogenesis and antiatherogenic effects on decreased angiogenesis have been described in models of pharmacological stimulation or inhibition of angiogenesis,^18,19,41 as well as in certain animal models (eg, the collagen-XVIII/ApoE double knock-out mouse).¹⁵ Here, presence of CD13⁺/CD31^low sprouts in the adventitia substantiates the presence of ongoing angiogenesis. The reduced expression of CD31 is in accordance with studies that have shown that both embryonic stem cells and mature endothelial cells show reduced expression of CD31 during endothelial differentiation, migration, and proliferation.^42,43 It is, however, surprising that the plaque-associated vasa vasorum showed an almost exclusively arterial phenotype, which is in contrast to the model suggested by Eriksson.³¹ Possibly, this is attributable to differences in experimental setup, as the current study was performed on mice without a diet and at a younger age than the mice used by Eriksson, resulting in a less developed vasa vasorum. On the other hand, it should be noted that Ephrin-B2 is known to be expressed on angiogenic endothelium and, therefore, could account for the increased Ephrin-B2 staining. However, because there was no response to venous ligation, it is conceivable that the vasa vasorum in fact may drain on the carotid artery, acting comparable with an arterial shunt. This would explain the low flow, which could be a consequence of a small pressure gradient over the vasa vasorum, although the latter may also be influenced by pulsation caused by systolic and diastolic phases of the cardiac cycle.

Whereas plaque-associated microvessels were primarily located in the adventitia surrounding the atherosclerotic lesion, vessel-like structures were sporadically detected within the plaque itself (Table II in the online-only Data Supplement). Surprisingly, these structures did not originate from the vasa vasorum, because microvessels did not traverse the media. These findings oppose observations in human plaques.⁸ Of note, incidental sprouting from the luminal endothelium was observed, yet these intraplaque vascular structures did not seem to be functionally perfused in vivo. Lucerna et al¹³ showed that overexpression of vascular endothelial growth factor in mice led to more luminal invaginations, accompanied by an increased leukocyte influx into the plaque. Therefore, intraplaque microvascular structures, even when originating from the lumen, may contribute to the progression of atherosclerotic plaques in mice. Additionally, intimal microvessels are suggested to be causal for IPH, although no causal link between IPH and plaque vascularization could be discerned. The observed occurrence of IPH was within the range described in, for example, the studies from Rosenfeld,⁴⁴ Calara,⁴⁵ and Johnson,⁴⁶ which however show a large variation in the occurrence of IPH (3.6%–75%). Generally, proof of a causal relationship between vascularization and IPH is lacking in murine models, which is substantiated by our current study. The data, however, are suggestive of a contribution of vasa vasorum to IPH, or plaque instability in general, although the presence of vasa vasorum does not infer occurrence of IPH.

Although this murine model does not completely mimic human atherosclerosis, because it only shows a limited, yet consistent occurrence of (intraplaque) vascularization, it provides new insights into the potential role of vasa vasorum in human atherosclerosis. Importantly, a number of features found in our study are consistent with previous studies in human atherosclerosis. For example, human plaque microvessels have been associated with ingress of leukocytes and erythrocytes and with leakage of albumin.² In this study, we established increased leukocyte adhesion and extravasation from adventitial microvessels, as well as an increased permeability of microvessels for 70-kDa dextrans, resembling albumin leakage in human atherosclerosis. Interestingly, and in contrast to our model, human plaque microvessels generally seem to have a more instable microvascular phenotype and show disrupted interendothelial junctions,¹⁰ especially in intraplaque microvessels. Hence, it would be interesting to investigate functional features of plaque neovascularization in other murine models, which have already described intraplaque microvessels or more extensive vasa vasorum expansion.^15,47

In summary, using a novel in vivo imaging strategy, we were able to study plaque-associated microvessels in mice and quantify its functional characteristics. The plaque-associated vasa vasorum demonstrated a significantly reduced blood flow, blood flow velocity, as well as an increase in leukocyte adhesion, leukocyte transmigration, and endothelial permeability compared with control microvasculature. In conjunction with the ongoing angiogenic activity, these features would support progression of atherosclerotic lesions. Further studies will help unravel by which means microvascular function in the vasa vasorum may be normalized.

Footnotes