OUP user menu

Immaturity of microvessels in haemorrhagic plaques is associated with proteolytic degradation of angiogenic factors

Julien Le Dall, Benoît Ho-Tin-Noé, Liliane Louedec, Olivier Meilhac, Carmen Roncal, Peter Carmeliet, Stéphane Germain, Jean-Baptiste Michel, Xavier Houard
DOI: http://dx.doi.org/10.1093/cvr/cvp253 184-193 First published online: 20 July 2009


Aims We investigated the causes of microvessel immaturity and destabilization in human atherosclerotic lesions.

Methods and results Human atherosclerotic carotid plaques (n = 24) were classified as non-haemorrhagic (NH) or haemorrhagic (Hem), according to their macroscopic aspect and haemoglobin content. Plaque microvessel density and maturity were quantified by immunohistochemistry. Expression of angiogenic factors was studied by immunohistochemistry, in situ hybridization, and ELISA. Plaque-conditioned media were tested for plasmin and elastase activities and for their ability to degrade angiogenic factors and to induce smooth muscle cell migration. Microvessel density and leucocyte infiltration were increased in Hem compared with NH plaques. Plaque vasculature appeared vulnerable as indicated by the absence of α-actin-positive mural cells in most plaque vessels. Despite increased numbers of angiogenic factor-expressing microvessels and leucocytes in Hem plaques, lower levels of vascular endothelial growth factor, placental growth factor, and angiopoietin-1 were found in conditioned media from Hem plaques. However, NH and Hem plaques released similar levels of the vascular destabilizing factor, angiopoietin-2. Addition of recombinant angiogenic factors to plaque extracts showed that all factors but angiopoietin-2 were selectively degraded by plasmin and/or elastase released from Hem plaques. Furthermore, conditioned media from Hem plaques showed a reduced ability to induce smooth muscle cell migration.

Conclusion Our results provide evidence that immaturity of plaque vessels is associated with the degradation of angiogenic factors by haemorrhage-conveyed leucocytes and proteases.

  • Angiogenesis
  • VEGF
  • Angiopoietins
  • Plasmin

1. Introduction

Earlier1 and more recent biochemical2 and functional imaging3 data have shown that intraplaque haemorrhages are the main link between plaque progression and the clinical expression of human atherothrombotic disease. Microvessel density increases with plaque complexity4 and both unstable and ruptured plaques display higher microvessel density than stable and non-ruptured plaques.57 Neovessels in plaques express leucocyte adhesion molecules,8 appear leaky,4,9,10 and microhaemorrhages can be observed in their vicinity.11,12 When localized in inflammatory and haemorrhagic (Hem) areas, microvessels are dysmorphic and lack surrounding α-actin-positive mural cells.12,13 All these studies suggest that microvessel development and immaturity is involved in intraplaque haemorrhage and leucocyte infiltration, two critical determinants for plaque progression.

We proposed that intramural haematoma/thrombus could be the main source of proteases, including serine proteases, in vulnerable plaques14,15 and we demonstrated that the more complicated the plaque, the richer it is in serine proteases.16 In earlier studies, we demonstrated in vitro17 and in vivo18 that serine proteases are capable of limiting cell colonization of the proteolytic-rich haematoma/thrombus and thus, of inhibiting lesion repair.

Neovessel development and maturation require angiogenic factors, among which the components of the vascular endothelial growth factor (VEGF) and angiopoietin systems have been extensively studied. The VEGF system is involved at different steps of angiogenesis (vessel permeabilization, endothelial cell proliferation, migration and survival, and formation of neocapillaries), whereas angiopoietins appear to be involved in vessel stabilization/destabilization. Atherosclerotic vascular walls show increased levels of angiogenic factors and receptors, whose expression is associated with macrophages, neovessels, and smooth muscle cells.19,20 Similarly, angiopoietin-1 (ang-1)/-2 (ang-2) and their receptor Tie-2 are expressed by the atherosclerotic wall, with a reported overexpression of angiopoietin-2 and Tie-2.2123

The relationship between haemorrhage and immaturity of plaque microvessels in human atherothrombosis remains unclear. Sluimer et al.10 recently showed by electron microscopy that endothelial junctions were incomplete in intraplaque neocapillaries. Here, we show that immaturity of plaque microvessels in Hem plaques is associated with degradation of angiogenic factors by plaque proteases. Proteolytic degradation of angiogenic factors within Hem plaques is likely to create an environment hostile to the maturation of neovessels that impairs mural cell recruitment and potentially participates to leucocyte infiltration and further plaque destabilization.

2. Methods

2.1 Tissue samples

Human carotid endarterectomy samples (n = 24) and healthy mammary arteries (n = 10) were obtained as surgical waste in the absence of patient opposition, in accordance with the French ethical laws (L. 1211-2 to L. 1211-7, L. 1235-2 and L. 1245.2) and following Ethics Committee advice (CPP Paris Ile de France XI no 07043, Centre Hospitalier de Poissy/Saint-Germain-en-Laye). The investigation conforms with the principles outlined in the Declaration of Helsinki.

Human carotid endarterectomy samples were dissected into culprit plaques (CP, the stenosing segment causing the symptoms which led to surgery) and adjacent non-complicated areas (non-culprit plaques, NP), as previously described.14,16

Healthy mammary arteries were dissected to separate the adventitia from the remaining arterial segment. Each sample from mammary and carotid arteries was cut into small pieces (5 mm3) and separately incubated in RPMI-1640 medium (Gibco) for 24 h at 37°C (6 mL/g of wet tissue).24 The tissue-culture media containing released material were then collected and stored at −80°C until use.

2.2 Haemoglobin/haem determination

The presence of haemorrhage in CP and NP was determined by macroscopical observation and measurement of haemoglobin/haem concentration by the addition of formic acid to the tissue-culture media.14,25 Haem content was then determined by reading the absorbance at 405 nm.25

2.3 Histology and immunohistochemistry

Tissues were fixed in 3.7% paraformaldehyde, embedded in paraffin, and sectioned at 5 µm. Sections were stained with Perl's Prussian blue for haemosiderin and with Masson's trichrome for fibrous tissue and general morphology. Immunohistochemistry was performed using mouse monoclonal antibodies to CD31 (clone JC70A, Dako), α-smooth muscle actin (clone 1A4, Dako), CD68 (clone PG-M1, Dako), and CD66b (clone 80H3, Immunotech), and rabbit polyclonal IgG to PlGF (Relia Tech), and goat polyclonal IgG to VEGF (R&D systems), angiopoietin-1 (R&D systems), and angiopoietin-2 (R&D systems). A peroxidase Dako LSAB2+ kit was used for detection. Control irrelevant antibodies (Dako) were applied to assess non-specific staining. 3,3′-Diaminobenzidine (Dako) or histogreen (Histoprime) were used as peroxidase substrates. Sections were counterstained with Mayer's haematoxylin or with nuclear fast red.

Microvessel density in the shoulder and cap regions of plaques was quantified by counting CD31-positive vessels on plaque sections (three fields per plaque from eight carotid samples per group at ×40 magnification). The percentage of pericyte-covered vessels in the vascular hot spots was calculated from the number of alpha-smooth muscle actin-positive vessels over the total number of CD31-vessels in adjacent serial sections.

Quantification of macrophage and neutrophil infiltration in culprit atherosclerotic plaques was estimated by counting the number of CD68- and CD66b-positive cells in plaque sections (three to five fields per section from five carotid samples per group at ×100 magnification).

2.4 In situ hybridization

cDNAs were used for the generation of 35S-RNA probes for VEGF, VEGFR-1, VEGFR-2, angiopoietin-1, angiopoietin-2, and Tie-2.26 Sections received 50 µL of hybridization mixture containing 5 × 105 c.p.m. of 35S-UTP-labelled riboprobe and were hybridized at 50°C overnight. Slides were exposed for 4 days on X-ray film (BioMax MR, Kodak) to evaluate the signal intensity and then dipped in NTB2 liquid emulsion (Kodak). The slides were exposed for 4 weeks, developed, fixed, stained with nuclear fast red, and mounted in Eukitt.

2.5 ELISA tests

Concentrations of VEGF, PlGF, angiopoietin-1, angiopoietin-2, and soluble Tie-2 in plaque-conditioned media were determined by ELISA kits (R&D Systems), following the manufacturer's instructions.

2.6 Plasmin and leucocyte elastase activities

Tissues (∼20 mg) were incubated in 0.05 M HEPES buffer pH 7.4, 0.75 M NaCl, 0.05% NP40 with 40 µM of the synthetic plasmin substrate, MeOSuc-Ala-Phe-Lys-AMC (Bachem), or the leucocyte elastase substrate, MeOSuc-Ala-Ala-Pro-Val-AMC (Calbiochem). Substrate hydrolysis was monitored for 2 h by spectrofluorometry (Hitachi F-2000; λExc = 380 nm; λEm = 460 nm).

2.7 Proteolysis of angiogenic mediators

Recombinant PlGF (100 ng, Relia Tech), VEGF (100 ng, R&D Systems), angiopoietin-1 (200 ng, Relia Tech), and angiopoietin-2 (200 ng, Relia Tech) were incubated for 3 h at 37°C in 0.05 M HEPES buffer pH 7.4, 0.75 M NaCl, 0.05% NP40 with plasmin (33 nM, American Diagnostica) or leucocyte elastase (33 nM, Calbiochem). Ex vivo proteolytic degradation was performed by incubating recombinant angiogenic mediators (VEGF and PlGF: 250 ng; Ang-1/-2: 500 ng) with CP (n = 8) or NP (n = 8) samples for 3 days at 37°C with or without 66 µM of plasmin inhibitor (H-D-Val-Phe-Lys-CMK, Calbiochem) or leucocyte elastase inhibitor (MeOSuc-Ala-Ala-Pro-Val-CMK, Calbiochem).

2.8 Immunoblot

Samples were transferred to nitrocellulose membranes (Pierce) after resolution by 12% SDS–PAGE. Membranes were probed with anti-PlGF (Relia Tech), anti-VEGF (R&D systems), anti-angiopoietin-1 (Zymed Laboratories), and anti-angiopoietin-2 (Zymed Laboratories) and with peroxidase-conjugated secondary antibody (Jackson ImmunoResearch).

2.9 MTT assay

Adhesion and survival of vascular smooth muscle cells in response to plaque-conditioned media was assessed by the tetrazolium salt, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay. Confluent human aortic smooth muscle cells seeded in 48-well plates were incubated for 22 h at 37°C with 400 µL of serum-free SM2 medium (Promocell) supplemented with 200 µL of plaque-conditioned media or RPMI-1640 medium. Non-adherent cells were then eliminated by two washes in PBS and residual viable adherent cells were incubated for 3 h with 0.5 mg/mL of MTT (Sigma). MTT was then removed, the formazan crystals produced by the reaction dissolved in DMSO, and the absorbance was read at 540 nm.

2.10 Cell migration assay

Cell migration was assessed using a modified cell dispersion assay.27 Human aortic smooth muscle cells (5 × 105) (Promocell) were seeded inside a 3 mm glass ring placed in the middle of a collagen-coated coverslip in 24-well plates. Four hours after plating, the glass ring was removed and the cells were covered with serum-free SM2 medium (Promocell), containing or not tissue-culture medium from atheromatous carotid samples, for 22 h at 37°C in a humidified atmosphere. After staining cells with haematoxylin, cell-covered areas were quantified by morphometric analysis using Histolab software (Microvision Instruments, Evry, France). In this model, cells migrate as an outgrowth from the confluent area initially delimited by the ring (13.25 mm2). Cell migration is expressed as the difference between the surfaces covered by cells in the stimulated vs. unstimulated control conditions.

2.11 Statistical analysis

Data represent mean ± SEM and were analysed by the Wilcoxon paired non-parametric test or the Mann–Whitney non-parametric test, when appropriate. Statistical significance was accepted for P < 0.05.

3. Results

3.1 Hem plaques are characterized by a high density of alpha-smooth muscle actin-negative microvessels

Human carotid endarterectomy samples were dissected into CP and adjacent non-complicated areas (NP). All CP used in this study displayed a non-ruptured fibrous cap upon macroscopic observation (Figure 1A). Among 24 endarterectomy samples, 11 CP were non-haemorrhagic (NH) and 13 CP presented macroscopically visible intraplaque haemorrhage and significantly higher levels of haemoglobin when compared with other samples, while all NP were NH (Figure 1B).

Figure 1

Characterization of non-haemorrhagic (NH) and haemorrhagic (Hem) plaques. (A) Representative human atheromatous culprit plaques (CP) classified as non-haemorrhagic or Hem, according to their macroscopical aspect. (B) Verification of the haemoglobin content in conditioned media from NH and Hem CP and from their adjacent non-complicated segments (NP) by a chromogenic assay. (C) Masson's trichrome (upper panels) and Prussian blue (lower panels) staining of NH and Hem culprit plaques (CP/Hem). The presence of extravasated red blood cells and of haemosiderin, a degradation product of phagocytosed haemoglobin, allowed the localization of intraplaque haemorrhage (white arrows). NH plaques corresponded to early atherosclerotic lesions (type II and III) and fibrous lesions of type IV and V. Hem plaques included type V lesions with old and/or recent microhaemorrhages and VIb lesions with large intraplaque haematoma. Inset: extravasated red blood cells (black arrows) present in the vicinity of microvessels in Hem areas. Bars = 500 µm.

Recent and old intraplaque haemorrhages were identified by the presence of extravasated red blood cells and by positive Prussian blue staining, which reveals haemosiderin, a degradation product of phagocytosed haemoglobin28 (Figure 1C). In accordance with the classification established by Stary et al.,29 histological analysis revealed that NH samples corresponded to early atherosclerotic lesions (type II and III) and fibrous or calcified lesions of type IV and V. Hem plaques included non-ruptured type V lesions with old and/or recent microhaemorrhages, and non-ruptured type VIb lesions with large intraplaque haematoma (Figure 1C).

Microvessel density was increased in Hem plaques when compared with NH plaques (Figure 2A and B). Plaque microvessels predominated in the shoulder region of the plaque, at the interface between the core, the cap, and the media (Figure 2A). It is noteworthy that no difference in microvessel density was found between Hem plaques with and without haematoma (11.82 ± 6.84 vs. 12.57 ± 5.99 microvessels/mm2, P = 0.971, n = 6). The majority of microvessels in Hem plaques lacked alpha-smooth muscle actin-positive cells in their wall (Figure 2A and C). This reduced mural cell coverage of Hem plaque microvessels indicated that these vessels were structurally immature and fragile. These results suggest a link between increased neo-vascularization combined with defective blood vessel maturation and intraplaque haemorrhage.

Figure 2

Haemorrhagic (Hem) plaques have increased microvessel density and decreased microvessel pericyte coverage. (A) Representative photomicrographs of CD-31 and alpha-smooth muscle actin immunostaining of non-haemorrhagic (NH) and Hem plaques. Microvessels were mainly located in the shoulder and cap regions. Bar = 500 µm. Arrows indicate the position of the high magnification view of CD31-positive vessels and corresponding alpha-smooth muscle actin staining (insets). (B) Immunohistochemical quantification of microvessel density in NH (CP/NH) and Hem culprit plaques (CP/Hem) (n = 8–12 plaques per group). (C) Comparison of microvessel pericyte coverage between CP/NH and CP/Hem. Pericyte coverage of plaque microvessels was estimated by calculating the percentage of CD31-positive vessels which were α-smooth muscle actin-positive in adjacent serial sections (n = 8–12 plaques per group).

3.2 The intraplaque angiogenic balance is altered in Hem plaques

Microvessel growth and maturation are regulated by angiogenic factors. Differences in microvessel density and maturity between Hem and NH plaques may depend on different intraplaque angiogenic balances. Positive immunostaining for VEGF, PlGF, angiopoietin-1, and angiopoietin-2 was detected in both Hem and NH plaques (Figure 3A). Staining for angiogenic factors within plaques was mostly associated with microvessels, their surrounding smooth muscle cells, and was also present in their vicinity, where CD68-macrophages and CD66b-neutrophils were observed (Figure 3A and C). The expression of angiogenic factors and of their receptors by vessels and inflammatory cells within NH and Hem plaques was further confirmed by in situ hybridization (Figure 3B). Taken together, these results show that angiogenic factors and their receptors are expressed in NH culprit plaques (CP/NH) and Hem culprit plaques (CP/Hem) and that endothelial cells and leucocytes are the main cell source of angiogenic factors in plaques. Interestingly, in parallel to their increased microvessel density (Figure 2A and C), Hem plaques showed several-fold higher numbers of infiltrating macrophages and neutrophils than NH plaques (Figure 3C and D).

Figure 3

Immunohistochemical localization and in situ hybridization analysis of components of the VEGF and angiopoietin systems. (A) Immunohistochemical localization of angiogenic factors in non-haemorrhagic (NH) and Hem plaques. VEGF, PlGF, ang-1, and ang-2 staining was associated with microvessels and with infiltrating inflammatory cells. Arrows indicate positive immunostaining in vessels and surrounding inflammatory cells. Bar = 100 µm. (B) In situ hybridization analysis of angiogenic factor expression in NH and Hem plaques. mRNA of VEGF, VEGFR-2, ang-1, ang-2, and Tie-2 were found to be associated with vessels and inflammatory cells of NH and Hem plaques. Bar = 50 µm. (C) Immunostaining of CD68- and CD66b-positive cells in NH and Hem culprit plaques (CP). Bar = 100 µm. (D) Immunohistochemical quantification of CD68- and CD66b-positive cells in sections from NH and Hem plaques (n = 5 plaques per group).

Since microvessels and leucocytes are sources of angiogenic factors (Figure 3A), their content may be increased in Hem plaques. Quantification of angiogenic factors released into tissue-culture media from healthy mammary arteries and atheromatous carotids revealed that levels of VEGF, PlGF, and angiopoietin-1 were significantly increased in atheromatous carotid samples (Figure 4AC). Furthermore, levels of PlGF, VEGF, and soluble Tie-2 increased with plaque complexity in NH lesions, as revealed by the higher levels measured in CP compared with non-complicated ones (Figure 4AC, E). Surprisingly, levels of PlGF, VEGF, and angiopoietin-1 were significantly decreased in culprit Hem plaques when compared with CP/NH (Figure 4AC). In contrast, levels of soluble Tie-2 were increased in Hem CP, whereas angiopoietin-2 levels were similar in culprit and non-complicated parts of both Hem and NH lesions (Figure 4D). Thus, pro-angiogenic factor levels are decreased in Hem CP, while their angiopoietin-2 levels remain stable. This shows that intraplaque haemorrhage is associated with angiogenic imbalance in favour of the vascular destabilizing factor, angiopoietin-2.

Figure 4

The intraplaque content in angiogenic factors varies with plaque complexity. Tissue-culture media from the medial layer of healthy mammary arteries (Mam), non-complicated plaques (NP) and non-haemorrhagic (NH) and haemorrhagic (Hem) culprit plaques (CP) were analysed by ELISA for PlGF (A), VEGF (B), ang-1 (C), ang-2 (D), and soluble Tie-2 (E).

3.3 Angiogenic factors are degraded by plasmin and leucocyte elastase in Hem plaques

The reduced levels of angiogenic factors in Hem plaques may be due to decreased release and/or to degradation of these factors by haemorrhage-associated factors. VEGF has heparin-binding properties and can be retained in the ECM, which may result in apparently lower levels in plaque-conditioned media. However, the addition of heparinase-III (1 U/mL) during the preparation of plaque-conditioned media did not affect VEGF levels in NH or in Hem plaques (data not shown). This suggests that the decreased level of VEGF released by Hem plaques was not due to increased trapping of VEGF within the ECM.

Consistent with our previous report,14,16 a significant increase in plasmin activity was detected in Hem plaques when compared with NH and non-complicated plaques (Figure 5A). Leucocyte elastase activity was also significantly elevated in Hem culprit lesions14 (Figure 5B). Both proteases (33 nM) degraded in vitro VEGF, PlGF, and angiopoietin-1 and 2, as evidenced by the presence of proteolytic products with lower molecular weights compared with the native proteins or by the disappearance of immunoreactive species in western immunoblot after treatment of recombinant growth factor by plasmin and leucocyte elastase (Figure 5C).

Figure 5

Plasmin and elastase enzymatic activities and proteolysis of angiogenic factors in haemorrhagic (Hem) plaques. In situ plasmin (A) and elastase (B) enzymatic activities were determined using selective fluorogenic substrates in non-complicated (NP) and non-haemorrhagic (NH) and Hem culprit plaques (CP). (C and D) Western blot analysis of angiogenic factor degradation by plasmin, elastase, and atherosclerotic plaques. Recombinant VEGF, PlGF, ang-1, and ang-2 were incubated with either (C) purified plasmin and leucocyte elastase or (D) tissue samples of non-complicated (NP) or culprit (CP) areas of Hem and NH plaques. (E) Recombinant angiogenic factors were incubated with tissue samples of culprit areas of Hem plaques in the presence or absence of a plasmin inhibitor (H-D-Val-Phe-Lys-CMK), or a leucocyte elastase inhibitor (MeOSuc-Ala-Ala-Pro-Val-CMK). Plasmin and elastase inhibitors partially prevented proteolysis by Hem culprit tissue sample. Arrows indicate the molecular weights of the recombinant growth factors.

To determine whether the decreased levels of angiogenic factors in conditioned media from Hem plaques originate from the enrichment of these lesions in proteases, recombinant angiogenic factors were incubated with tissue samples and then analysed by immunoblotting (Figure 5D). No degradation was observed when recombinant factors were incubated with the non-complicated adjacent segments (NP) of CP. When incubated with CP/NH, only a slight degradation was observed for VEGF and angiopoietin-1. In contrast, almost complete disappearance of immunoreactive species was obtained when VEGF, PlGF, and angiopoietin-1 were incubated with Hem culprit plaques (CP/Hem), whereas angiopoietin-2 appeared to be degraded to a lesser extent (Figure 5D and E). Proteolysis of VEGF, PlGF, and angiopoietin-1 and -2 was partially prevented by the addition of specific plasmin and elastase inhibitors (Figure 5E). These results suggest that degradation of angiogenic factors by plasmin and leucocyte elastase may account for the angiogenic imbalance of CP/Hem. However, the partial character of this inhibition suggests that proteases others than elastase and plasmin are also involved in the degradation of growth factors in Hem plaques.

3.4 Hem plaques have a reduced capacity to induce smooth muscle cell migration

Decreased levels of growth factors may affect mural cell recruitment and survival leading to immaturity of neovessels and subsequent bleeding. For this reason, we compared the effect of tissue-culture media, conditioned by NH and Hem plaques, on vascular smooth muscle cell spreading, migration, and survival.

Measurement of the area covered by the migrated smooth muscle cells in response to plaque extracts showed that both NP and CP were able to induce migration of smooth muscle cells (Figure 6A and B). Interestingly, whereas non-complicated and CP/NH displayed a similar ability to induce smooth muscle cell migration, the presence of haemorrhage in CP decreased this ability (Figure 6A and B). Differences in the area covered by smooth muscle cells in response to plaque-conditioned media may also result from detachment and cell death of smooth muscle cells. We thus investigated the effect of plaque-conditioned media on the adhesion and survival of smooth muscle cells. Incubation of confluent vascular smooth muscle cell cultures with conditioned media from NH or Hem plaques did not affect their adhesion and induced a similar increase in smooth muscle viability when compared with unstimulated control cells, as assessed by MTT assay (Figure 6C). Thus, the reduced area covered by smooth muscle cells in response to Hem plaque-conditioned media was not due to decreased cell death or adhesion in these conditions. Taken together, these results show that Hem plaques have a reduced capacity to induce spreading and migration of plaque-stabilizing smooth muscle cells.

Figure 6

The ability of atherosclerotic plaques to induce smooth muscle cell migration is reduced by intraplaque haemorrhage. (A) Representative photographs of migrating smooth muscle cells stained with H&E after a 22 h incubation with tissue-culture medium from non-haemorrhagic (NH) culprit plaque (CP/NH, left panel) or haemorrhagic (Hem) complicated plaque (CP/Hem, right panel). Bar, 1 mm. (B) Comparison of smooth muscle cell migration among cells stimulated with supernatant from non-complicated plaques (NP) and NH and Hem culprit plaques (CP). Results are expressed as the difference between cell-covered areas in stimulated and unstimulated control conditions. The increase in surface in the control group was 6.65 ± 0.39 mm2. (C) Effect of plaque-conditioned tissue-culture media on smooth muscle cell adhesion and survival assessed by the MTT assay.

4. Discussion

Immaturity of plaque microvessels is a source of intraplaque haemorrhage, a critical factor in atherosclerotic plaque progression and destabilization.13,30 In the present study, we investigated the mechanisms interfering with plaque microvessel maturation. In agreement with recent studies,10,13 we show a link between haemorrhage and the growth of immature microvessels within human carotid atherosclerotic plaques. In fact, we found that haemorrhage occurred in plaques characterized by a high density of immature microvessels that lack α-actin-positive mural cells.

The extent and quality of angiogenesis depend on several factors including the expression of angiogenic factors. We show that VEGF, PlGF, ang-1, ang-2, VEGFR-2, and Tie-2 are expressed in microvessels from both Hem and NH plaques. Consistent with previous reports,1921,23 positive immunostaining for VEGF and PlGF was also associated with inflammatory cells. Higher levels of VEGF, PlGF, and angiopoietin-1 were released by carotid plaques than by healthy mammary arteries and, among plaques, the highest concentrations were released by culprit NH plaques. Altogether, these data are consistent with the reported angiogenic potential of plaques.31,32 It is noteworthy that, while Hem plaques were highly vascularized, CP/NH were less invaded by neovessels. This paradoxical result suggests that the density of microvessels in the plaque is not only defined by the levels of released angiogenic factors, but most likely also involves additional factors. For instance, it was recently shown that CD40 ligand-positive microparticles from human atherosclerotic plaques stimulate endothelial proliferation and angiogenesis.33 Moreover, angiogenesis is a complex process that requires destruction of the basement membrane and local degradation of the extracellular matrix (ECM) to allow migration and proliferation of endothelial cells and the release of ECM-bound growth factors. Thus, proteolytic activities may promote the growth of plaque microvessels and the associated risk of haemorrhage. Supporting this hypothesis and confirming our previous observations,14,16,34,35 we show that Hem plaques are enriched in leucocytes and in elastase and plasmin activities.

However, a highly proteolytic environment may also limit the scarring process and vessel maturation. Hemo-thrombus-associated protease activities were indeed shown to impede mesenchymatous cell spreading, proliferation, and survival.17,18,36 Therefore, proteolysis of growth factors by intraplaque proteases may prevent the maturation of plaque microvessels. Our results indicate that degradation of angiogenic factors by plasmin and leucocyte elastase accounts for the decreased levels of VEGF, PlGF, and ang-1 and for the increased solubilization of Tie-2 in Hem plaques compared with NH plaques. It is noteworthy that plasmin and elastase inhibitors produced only a partial inhibition of growth factor degradation, suggesting that other intraplaque proteases may also be involved in this phenomenon. Indeed, it has been shown previously that Hem plaques are also enriched in MMPs and mast cell tryptase.10,14

Interestingly, the levels of the vascular destabilizing protein ang-2 were similar in NH and Hem plaques. Therefore, it appears that proteases alter the angiogenic balance in Hem plaques, in particular degrading ang-1 and its receptor Tie-2, while ang-2 levels remained unchanged. These modifications in the angiogenic balance of Hem plaques corroborate recent findings showing that the balance between ang-1 and ang-2 is in favour of ang-2 in atherosclerotic plaques with high microvessel density.23

Ang-1 promotes endothelial cell barrier integrity and it is conceivable that the reduction in ang-1 levels in Hem plaques favours microvessel permeability. Furthermore, imbalance between angiopoietin-1 and angiopoietin-2 is thought to play a role in brain arteriovenous malformations,37 which lead to recurrent cerebral haemorrhages, and in the impairment of pericyte recruitment and vessel maturation after myocardial ischaemia in diabetic patients.38 We show that Hem plaques had a reduced potential to induce smooth muscle cell migration when compared with NH plaques. Pericytes and periendothelial intimal smooth muscle cells share common markers such as smooth muscle actin, suggesting a common mesenchymatous origin.39 Proteolytic degradation of growth factors in Hem plaques may impair the spreading and recruitment of stabilizing smooth muscle cells and pericytes and contribute to the compromised structural integrity of microvascular endothelium in atherosclerotic arteries.10 Since VEGF, PlGF, and angiopoietins are angiogenic factors that mainly relate to endothelial cell function, it is likely that the reduced capacity of Hem plaques to induce smooth muscle cell migration also involves degradation of other growth factors. In addition, plaque proteases may also destabilize plaque vasculature by directly damaging microvessels. Indeed, extravasation of red blood cells occurs by passage not only across the endothelium but also through the basement membrane, which implies degradation or rupture of this structure surrounding the vessels.

Of note, in this study, plaques that displayed signs of acute rupture were excluded. Thus, the presence of haemorrhage did not result from recent acute plaque rupture, fissure, or erosion but most likely from angiogenesis as it is now well established.30 However, a role of old plaque ruptures and thrombi in the mechanisms leading to the degradation of intraplaque growth factors and to immaturity of plaque microvessels cannot be excluded.

In conclusion, while immaturity of plaque vessels has been proposed to cause intraplaque haemorrhage,10,30 our results suggest that, on the other hand, intraplaque haemorrhage may further destabilize the plaque vasculature by accelerating the deposition within the lesion of leucocytes and proteases, which alter the angiogenic balance. Our findings designate proteases as potential targets for the normalization of plaque vasculature, a clinical strategy that has been proposed for plaque stabilization.40 The present study also provides new insight into the role of intraplaque haemorrhage in creating a self-perpetuating mechanism of lesion progression and may explain the reported recurrence of intraplaque haemorrhages after a first Hem event.3


X.H. is supported by the Fondation Lefoulon Delalande. This study was supported by INSERM and by grants from the Leducq Foundation (Leducq Transatlantic Network on Atherothrombosis).


We thank Dr Uriel Sebbag and Dr Tonino Palombi from the Centre Cardiologique du Nord for providing us with carotid samples and Dr Mary Osborne-Pellegrin for editing this manuscript.

Conflict of interest: none declared.


View Abstract