By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Please be advised that we experienced an unexpected issue that occurred on Saturday and Sunday January 20th and 21st that caused the site to be down for an extended period of time and affected the ability of users to access content on Wiley Online Library. This issue has now been fully resolved. We apologize for any inconvenience this may have caused and are working to ensure that we can alert you immediately of any unplanned periods of downtime or disruption in the future.
Marijke J. E. Kuijpers, Department of Biochemistry, Maastricht University, PO Box 616, 6200 MD Maastricht, the Netherlands. Tel.: +31 43 3881537; fax: +31 43 3884159. E-mail: Marijke.Kuijpers@bioch.unimaas.nl.
Summary. Background: Atherothrombosis is a major cause of cardiovascular events. However, animal models to study this process are scarce. Objectives: We describe the first murine model of acute thrombus formation upon plaque rupture to study atherothrombosis by intravital fluorescence microscopy. Methods: Localized rupture of an atherosclerotic plaque in a carotid artery from Apoe−/− mice was induced in vivo using ultrasound. Rupture of the plaque and formation of localized thrombi were verified by two-photon laser scanning microscopy (TPLSM) in isolated arteries, and by immunohistochemistry. The thrombotic reaction was quantified by intravital fluorescence microscopy. Results: Inspection of the ultrasound-treated plaques by histochemistry and TPLSM demonstrated local damage, collagen exposure, luminal thrombus formation as well as intra-plaque intrusion of erythrocytes and fibrin. Ultrasound treatment of healthy carotid arteries resulted in endothelial damage and limited platelet adhesion. Real-time intravital fluorescence microscopy demonstrated rapid platelet deposition on plaques and formation of a single thrombus that remained subocclusive. The thrombotic process was antagonized by thrombin inhibition, or by blocking of collagen or adenosine diphosphate receptor pathways. Multiple thrombi were formed in 70% of mice lacking CD40L. Conclusions: Targeted rupture of murine plaques results in collagen exposure and non-occlusive thrombus formation. The thrombotic process relies on platelet activation as well as on thrombin generation and coagulation, and is sensitive to established and novel antithrombotic medication. This model provides new possibilities to study atherothrombosis in vivo.
Atherothrombosis is a life-threatening disease, which is caused by thrombus formation at the surface of ruptured or eroded atherosclerotic plaques in coronary or carotid arteries. Subsequently, a combination of acute and chronic events, such as myocardial infarction or stroke, can occur [1,2]. As the rupture of a vulnerable plaque is an unpredictable event, the ensuing process of thrombus formation is difficult to monitor in man.
In recent years, several murine thrombosis models have been developed to study thrombus formation in microvessels of the mesentery, but also in larger vessels such as the carotid artery [3–6]. In this work, intravital microscopic imaging has become a highly valuable tool to identify the dynamics of the thrombotic process. In general, it has been shown that arterial thrombosis is triggered and driven by activation of both platelets and the coagulation system, although the relative contribution of either may depend on the vessel type and the method of vascular damage. However, in all these models, healthy blood vessels were damaged rather than diseased, atherosclerotic vessels. Research into atherothrombosis thus lacks a suitable animal model of acute plaque rupture.
Upon feeding with high fat diet, Apoe−/− mice progressively develop atherosclerotic lesions in the aortic arch and the carotid arteries. Because spontaneous rupture of these lesions is very rare , various methods have been explored to provoke this event. These include manual squeezing of the plaque-containing aorta between forceps , ligation of the carotid artery and cuff placement , collar placement around the carotid artery followed by adenovirus p53 transfection , generation of double knock-outs such as Apoe−/−Npc1−/− mice , and photochemical thrombus induction with rose Bengal . However, these methodologies have mostly led to incomplete or non-acute rupture of plaques or to non-predicted localization of the thrombotic process.
Using intravital microscopy, earlier used to study thrombus formation in healthy murine carotid arteries [12,13], we aimed to develop a model for plaque rupture-evoked thrombus formation in the same vessel type. We report that targeted ultrasound treatment of a plaque shoulder in carotid arteries of Apoe−/− mice acutely produces a non-occlusive thrombus, the formation of which relies on both platelet activation and thrombin generation.
Materials and methods
Apoe−/− mice on C57Bl/6 background of either sex (four weeks old) were obtained from Charles River (Maastricht, the Netherlands). Mice with the same background deficient in both Apoe and CD40L were obtained as described . Experiments were approved by the local animal care and use committee.
Carboxyfluorescein diacetate succinimidyl ester (CFSE) and Syto41 were from Molecular Probes (Leiden, the Netherlands). Biotinylated CNA35 was conjugated with quantum dot (QD)585-streptavidine as described . Hirudin was from Pharmion (Cambridge, UK), cangrelor from the Medicines Company (Parsippany, NJ, USA), and TGX221 from Baker (Melbourne, Victoria, Australia). Antiglycoprotein (GP) VI antibody JAQ1 was obtained as described . Sources of other materials are given elsewhere .
Plaque rupture and measurement of acute thrombus formation
Apoe−/− mice were fed a Western-type diet with 0.15% cholesterol for 18–20 weeks. One or more plaques of a length of 0.5–1.0 mm were then present often in both carotid arteries (diameter 500–550 μm) near the bifurcation. Animals were anesthetized by subcutaneous injection of ketamine and xylazine (0.1 and 0.02 mg g−1 body weight); body temperature was held at 37 °C. Carotid arteries were carefully dissected free from surrounding tissue, and the animal was injected intravenously with CFSE-labeled platelets , obtained from a donor mouse with the same genetic background. Mice were subsequently injected with vehicle solution, anticoagulant or antiplatelet drug, as indicated.
Using intravital fluorescence microscopy, a plaque was selected in one of the carotid arteries near the bifurcation under bright-field illumination. The tip (0.5 mm diameter) of a titanium ultrasound probe was placed at the shoulder region of a plaque. Rupture was induced by ultrasound application during 10 s at 6 kHz using a VibraCell VCX130 processor (Sonics, Newtown, CT, USA). Thrombus formation was recorded as quickly as possible by capturing of 12-bit fluorescence images at 33 Hz during at least 10 min, using a back-thinned electron multiplier C9100-12 EM-CCD camera (Hamamatsu, Hamamatsu City, Japan) at fixed gain settings. Where indicated, the effects of ultrasound treatment on carotid blood flow were monitored with a Doppler flow probe (0.5 mm V series; Transonic Systems TS420, Ithaca, NY, USA).
Time series of images were analyzed using Wasabi (Hamamatsu) software. Per region of interest (ROI) representing the site of thrombus formation, total pixel intensity was calculated and corrected for background. To quantify thrombus size after specific time intervals, fluorescence images were processed using ImagePro software (Media Cybernetics, Silverspring, MD, USA). Within the image of a carotid artery, two similar ROIs were defined, representing the thrombus area and an adjacent area representing the background. Subsequently, a threshold level was set by eliminating all pixels with intensity lower than 99.0% of the pixels of the background ROI. Intensities (gray levels) of all pixels in the thrombus ROI were then integrated. No image processing was applied. Masks are shown in 10-color level.
Isolation of mouse plasma and platelets
Anesthetized mice were bled for isolation of platelet-rich plasma (PRP) for measurements of tissue factor-induced thrombin generation , and of aggregation and P-selectin expression following platelet stimulation .
Two-photon laser scanning microscopy
After in vivo treatment, carotid arteries containing the bifurcation were isolated, and mounted into a perfusion chamber at 60 mmHg fluid pressure, as described . Vessels were perfused with QD585-labeled CNA35 (50 nm) to detect luminally exposed collagen, and counter-stained with the nuclear probe Syto41 (2.5 μm). Mounted vessels were scanned near the plaque–lumen interface (500 lines s−1, 3× Kalman averaging) at 50% pulse-laser intensity (800 nm excitation, 100 fs pulse-width). Two-photon fluorescence was recorded at 450–470 nm (blue), 500–550 nm (green), and 560–600 nm (red), using a Biorad 2100MP system (Zeiss, Hemel Hempstead, UK) [13,15].
Carotid arteries including bifurcations were fixed and embedded into paraffin. Sections were stained with hematoxylin and eosin , or Carstair’s stain .
Groups were compared with the non-parametric Mann–Whitney U-test (one-tailed). A P-value below 0.05 was considered significant; a P-value below 0.1 was considered as borderline significant. Data are presented as mean ± SEM.
Results and discussion
Targeted ultrasound treatment of a carotid plaque results in local collagen exposure and thrombus formation
Apoe−/− mice were fed with a cholesterol-enriched diet for 18–20 weeks to achieve formation of one or more plaques around the bifurcation mostly in both carotid arteries (Fig. 1A). After dissection from surrounding tissue, these plaques are detected as a white mass by bright-field light transmission microscopy (Figs 1B–D). We studied the susceptibility of these plaques to rupture by local ultrasound treatment. Injection of CFSE-labeled Apoe−/− platelets and fluorescence microscopy enabled visualization of thrombus formation. After testing various ultrasound systems at different settings, most reproducible results were obtained with a VibraCell VCX130 sonicator system. This sonicator was equipped with a miniaturized ultrasound probe (0.5 mm wide titanium tip) for local application of energy. Guided by the microscope, the titanium tip was placed at the downstream shoulder region of a suitable plaque (length 0.5–1.0 mm) of the left or right carotid artery (Figs 1B and 1E). After 10 s of ultrasound application (frequency of 6 kHz), platelets immediately adhered, after which a thrombus was formed in 95% of the cases (see below). However, the treatment did not change blood flow in the artery (Fig. 1F). As estimated from the site of platelet adhesion, a circular area was damaged of ∼300 μm in diameter.
Bright-field intravital observations of the outside of the vessel showed that the overall appearance of the plaque remained similar, thus demonstrating that the treatment had not resulted in removal of the plaque. In contrast, histological analysis of paraffin-embedded sections revealed clear morphologic changes inside the vessel, at the region where the ultrasound probe was applied. Whereas untreated control plaques had a smooth cap covered with endothelial cells (Fig. 2A), plaques exposed to the ultrasound treatment had a frayed appearance and part of the core content was exposed to the vessel lumen (Fig. 2B). On top and in between the disturbed plaque tissue, loose thrombi were present, consisting of platelets and erythrocytes. Carstair’s stain detected exposure of collagen and formation of fibrin networks (Fig. 2C). No structural changes were seen in the tunica media and adventitia.
In control experiments, the same ultrasound treatment was applied to the healthy carotid arteries from wild-type mice. This resulted in damage but not in complete removal of the endothelial cells. Injection of CFSE-labeled autologous platelets showed adhesion of mostly single platelets and microaggregates at the treatment side (Fig. S1).
Besides structural cap injury, a key marker of plaque rupture is intrusion of erythrocytes into the plaque core region [21,22]. To demonstrate this, we analyzed serial paraffin sections of a representative plaque-containing vessel, starting downstream and ending at 400 μm upstream of the treatment site. This indeed showed intra-plaque accumulation of erythrocytes and fibrin starting at the ruptured site and continuing into sections of the upstream region (Figs 2D and 2E). Carotid vessels were also checked for the presence of intraplaque hemorrhages and neovessels, but these were not detected. Given the fact that spontaneous plaque rupture is very rare in mouse carotid plaques , these results thus provide additional evidence that the ultrasound treatment indeed leads to rupture of the plaque.
The presence of fibrin indicated that coagulation had been activated upon plaque rupture. Earlier in vitro studies with isolated atherosclerotic plaque material demonstrated a role of plaque-derived tissue factor in the triggering of coagulation [23,24]. To verify a role of tissue factor, we determined tissue factor activity in plaque tissue obtained from Apoe−/− mice as described previously . We observed 10 pm active tissue factor per mg of protein, indicating that substantial amounts of tissue factor are present within the plaques.
To further characterize the extent of vascular damage, intact carotid arteries were scanned by two-photon laser scanning microscopy (TPLSM), a technique that allows imaging of the morphological details of the vessel wall and plaque shoulder (Fig. 3A). Mice were injected with CFSE-labeled platelets, and plaques in the carotid arteries were treated or not by ultrasound. Intact arteries were then carefully isolated, mounted without fixation into a flow chamber, and slowly perfused with QD585-CNA35, a labeled Staphylococcus aureus protein that specifically binds to exposed collagen . Optical cross-sections of ultrasound-treated vessels showed bright staining of collagen and fluorescent platelets at the plaque shoulder region near the lumen (Fig. 3B). Post-staining of cellular nuclei with Syto41 revealed aligned strings of smooth muscle cells in the vessel wall, whereas endothelial cells had mostly disappeared (Fig. 3C). In contrast, in untreated control arteries, the endothelial cells of the plaque shoulder region still formed an intact monolayer, while hardly any collagen or platelet staining was visible (Fig. 3D). Three-dimensional image reconstructions illustrate the large exposed collagen sheet in the ultrasound-treated vessel (Figs 3E and 3F). Together, these results indicate that the targeted use of ultrasound provoked local plaque damage, exposure of collagen, and assembly of a thrombus with platelets, fibrin and erythrocytes.
Targeted ultrasound treatment results in acute formation of subocclusive thrombi
Thrombus formation in damaged carotid arteries can be monitored in a sensitive way by intravital microscopic detection of CFSE-labeled platelets . We used this method to quantify the thrombus-forming process following ultrasound plaque treatment. Mice were injected with CFSE-labeled Apoe−/− platelets, to an extent that 9 ± 2% of the circulating platelets were fluorescently labeled. At baseline fluorescent platelets flowed through the carotid artery, and incidentally adhered at the plaque region (Fig. 4A). This confirms the earlier observation that murine platelets can stick to plaques and thereby contribute to atherosclerosis development .
In the large majority of atherosclerotic arteries (14/15), targeted application of ultrasound (6 kHz for 10 s) at the plaque shoulder resulted in local formation of a single fluorescent thrombus (Fig. 4A). Only in one case did the ultrasound treatment lead to formation of one larger thrombus with adjacent smaller thrombi. At lower ultrasound energy levels, no thrombi were formed (not shown). The maximal thrombus size was reached after a variable time of 30–60 s, after which the thrombi declined (Fig. 4B). As a consequence of this variation in growth, calculated integrated fluorescence intensities after 30 and 60 s were variable: 2192 ± 870 and 2078 ± 780 (×103, mean ± SEM, n = 8). In the declining phase, only few single platelets and (very rarely) platelet aggregates left the thrombus. Also, fluorescence bleaching was relatively low (<2%/min), suggesting that the declining phase had a different cause (e.g. contraction of platelets and fibrin). From analysis of images from multiple vessels we learned that the secondary size reduction was a general phenomenon and extended to a period of 10 min (Fig. 4C). Remarkably, thrombi never occluded the vessel. Also, a second and third ultrasound treatment of the same plaque shoulder did not result in full occlusion. This biphasic pattern of thrombus formation resembles the pattern earlier described for laser-induced thrombus formation in murine arterioles, where platelet loss was associated with increased cAMP levels . Altogether, our results show that, in the carotid arteries of Apoe−/− mice, thrombus formation induced by plaque rupture is an acute, well-regulated and partly transient process, which by itself does not lead to vascular occlusion.
Platelet and thrombin inhibitors affect thrombus formation on ruptured plaques
Various laser- and FeCl3-induced models of thrombus formation demonstrate that thrombus formation relies on the combined activation of platelets and the coagulation system [4,6,20,27]. Also, in vitro studies, using isolated atherosclerotic plaques, point to a combined prothrombotic effect of platelet- and coagulation-stimulating factors [23,24]. It was therefore of interest to use inhibitors of both processes in the present plaque rupture model.
To study the role of thrombin, Apoe−/− mice were injected intravenously with a high-affinity thrombin inhibitor, hirudin (5 mg kg−1). This resulted in complete eradication of thrombin generation in mouse plasma ex vivo (Fig. S2A). In vivo, hirudin injection strongly suppressed the formation of thrombi following plaque rupture. Only small platelet aggregates assembled in the first minutes (Figs 4A and 4B). These aggregates did not grow in size over a time period of 10 min (Fig. 4C) or even 30 min (not shown). This pointed to a potent contribution of the coagulant system.
In other mouse models, the platelet collagen receptor GPVI and its co-receptor FcRγ-chain have been shown to play a key role in arterial thrombosis [3,12]. A suitable tool to study involvement of this pathway is by injection of the anti-GPVI mAb, JAQ1 (100 μg). This results in mild and transient thrombocytopenia (maximal platelet drop of 30%), but also in complete abolition of the platelet responses to collagen and other GPVI-activating agonists, such as convulxin . This was confirmed in the present study; that is, platelets from mice injected with JAQ1 were unable to expose P-selectin in response to convulxin (Fig. S2B). In vivo experiments indicated that JAQ1 injection significantly impaired the thrombus formation evoked by plaque ultrasound treatment (Figs 5A and 5B). Fluorescence recordings after treatment showed normal numbers of circulating labeled platelets but greatly reduced adhesion, pointing to a diminished platelet interaction with exposed collagen.
Other Apoe−/− mice were pre-injected with agents interfering with the ADP/P2Y12 signaling pathway, which is crucial for stable platelet aggregation [28,29]. Platelet P2Y12 receptors were blocked by continuous infusion with the specific inhibitor cangrelor . Alternatively, TGX221 was injected as a single dose to block activity of phosphoinositide 3-kinase β (i.e. a key signaling enzyme downstream of P2Y12 and GPVI). Control experiments showed that P2Y12 receptor treatment abolished all ADP-dependent stimulation of thrombin generation in mouse PRP (Fig. S2A), which is a unique P2Y12-mediated effect . Other controls indicated that injection of TGX221 affected collagen- and ADP-evoked platelet aggregation as well as GPVI-dependent P-selectin exposure (Figs S2B and S2C). In vivo, both cangrelor and TGX221 reduced the thrombus formation caused by ultrasound treatment, while TGX221 provoked incidental embolization (Figs 5A and 5B). Together, these findings point to a contribution of both collagen and ADP receptor pathways to the thrombotic process induced by plaque ultrasound treatment.
Both thrombin and collagen have been reported to contribute to the progression of atherosclerotic lesions in Apoe−/− mice. For instance, either thrombin inhibition with melagatran  or deficiency in GPVI/FcRγ  can result in diminished lesion progression. These changes have been explained by anti-inflammatory effects of thrombin inhibition or of FcRγ deletion, respectively. However, because local platelet activation can contribute to lesion progression and because activated platelets abundantly secrete pro-inflammatory mediators [25,33], the present results suggest that thrombin and collagen are involved in platelet interactions with a damaged or ruptured plaque (e.g. by stimulating the formation of transient, non-occlusive thrombi).
Subocclusive thrombus formation on carotid plaques in another genetic mouse model
The effect of ultrasound treatment was also studied in mice with a different plaque phenotype. Mice lacking Apoe and CD40L on platelets and leukocytes develop plaques with a reduced lipid and increased collagen content . In addition, platelets from CD40L−/− mice form thrombi with reduced stability . Provoking plaque rupture in the carotid arteries of Apoe−/−/CD40L−/− mice resulted in the formation of mostly two or more adjacent thrombi (5/7), which had a tendency to embolization (3/7) (Figs 5A and 5C). Histological analysis confirmed the presence of multiple thrombi at one damaged plaque (Fig. 5C). Image analysis further showed that the CD40L−/− thrombi were similar in size to the control thrombi. After 30, 60, and 120 s of rupture, integrated fluorescence intensities of thrombi were 1267 ± 371, 824 ± 261, and 594 ± 207 × 103, respectively (mean ± SEM, n = 7). Although this requires further investigation, this indicates that the absence of CD40L may increase the probability of multiple plaque rupture (possibly because of the increased collagen content) while it destabilizes platelet–platelet interactions in the thrombi.
In this paper, we present a first validated animal model of acute plaque rupture, which allows the study of the ensuing process of thrombus formation in real time. Rupture is induced by ultrasound treatment of the shoulder region of a murine carotid plaque. This results in local platelet adhesion and formation of a thrombus, which remarkably remains subocclusive. The present atherothrombosis model thereby is different from other murine arterial thrombosis models, where larger parts of the vessel are damaged (e.g. by treatment with rose Bengal or FeCl3) or where the blood flow is temporarily interrupted (e.g. by ligation). In such cases, thrombus formation generally leads to occlusion . Differences of the current model in comparison to other models in particular are: the restricted area of plaque damage evoked by the small ultrasound tip; the limited extent of exposure of thrombogenic substances at the ruptured site; and/or the local high blood flow causing rapid dilution of soluble agonists. Nevertheless, all evidence indicates that both exposed collagen and tissue factor contribute to the thrombus formation, and this is the result of a multifactorial process, involving platelet collagen and ADP receptors as well as thrombin and fibrin generation. The present results thus suggest that, at least in part, similar mechanisms play a role in thrombus formation on a ruptured plaque and on a damaged, non-diseased vessel wall. However, atherosclerotic plaques contain many components and cells with a thrombogenic potential (e.g. oxidized LDL, lysophosphatidic acid, macrophages) that are absent in healthy vessels. This plaque rupture model now presents the opportunity to study the importance of these components under conditions of in vivo thrombus formation.
The processes of plaque development and arterial thrombus formation are fundamentally different between man and mouse, if only because of the much smaller size of the murine vessels and hence of the lesions and thrombi formed in these vessels. The miniscule plaques in Apoe−/− murine arteries have thin fibrous caps, are more extensively exposed to the blood stream, and, when injured, can accommodate only small thrombi . It is also clear that mouse plaques are notoriously resistant to rupture . Also, spontaneous formation of large luminal thrombi is only rarely observed in murine vessels; this is only reported under specific conditions (e.g. in Apoe−/−Npc1−/− double knockouts) . The relative stability of mouse plaques is also apparent from the present experiments, where a high-energy input (6 kHz during 10 s) was required to provoke rupture.
Autopsy studies suggest that rupture of human plaques in the majority of cases results in formation of relatively small thrombi rather than in large occlusive thrombi. Often, multiple ruptures are required for a symptomatic thrombotic event [35,36]. Reports have shown that in man only 11% of first plaque ruptures result in coronary death (so-called virgin rupture) . This indicates that non-occlusive thrombus formation, as in the present model, is a relevant condition for the human situation, perhaps even more than the enormous occlusive thrombi that are formed in healthy vessels in other experimental models.
In summary, we have developed a new model of acute plaque rupture, which allows real-time measurement of the atherothrombotic process in mice. Plaque ultrasound treatment results in endothelial denudation, collagen exposure, and luminal thrombus formation, accompanied by intra-plaque intrusion of erythrocytes and fibrin formation. The thrombi are subocclusive and are sensitive to inhibition of thrombin and platelet ADP and collagen receptors. This model may be useful for unraveling the mechanisms of in vivo thrombus formation by plaque components and for evaluation of antithrombotic treatments.
M. J. E. Kuijpers performed and analyzed/interpreted experiments and shared writing with J. W. M. Heemskerk, who was also involved in concept and design of experiments and interpretation of data. K. Gilio and R. Nergiz-Unal assisted with IVM experiments and data analysis. S. Reitsma, L. Prinzen, and M. A. M. J. van Zandvoort assisted with TPLSM experiments and interpretation, and provided fluorescent labels. S. Heeneman and E. Lutgens were involved in interpretation of histology data. E. Lutgens was also involved in experiments with Apoe−/−/CD40L−/− mice. B. Nieswandt provided JAQ1 antibody and revised the manuscript. M. G. A. oude Egbrink was involved in IVM experiments and analysis of data. All authors approved the final version of the manuscript.
We acknowledge H. Spronk for tissue factor activity measurements. This study was supported by the Netherlands Heart Foundation (2005B079) and the Netherlands Organization for Scientific Research (902-16-276).
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.