OUP user menu

Mouse models of angiogenesis, arterial stenosis, atherosclerosis and hemostasis

Peter Carmeliet , Lieve Moons , Désiré Collen
DOI: http://dx.doi.org/10.1016/S0008-6363(98)00108-4 8-33 First published online: 1 July 1998


The development of efficient transgenic technologies in mice has allowed the study of the consequences of genetic alterations on cardiovascular (patho)physiology, although the development of such models have been hampered by size limitation of species resulting in time-consuming, labor-intensive and costly analyses. This overview summarizes the murine models currently available for studying or manipulating angiogenesis, arterial stenosis, atherosclerosis, transplant arteriopathy, thrombosis, thrombolysis and bleeding and addresses techniques to evaluate vascular development during embryogenesis.

  • Angiogenesis
  • Atherosclerosis
  • Restenosis
  • Bleeding
  • Thrombosis
  • Transplantation
  • Adenovirus

Time for primary review 10 days.

1 Blood vessel formation

Blood vessels are among the first organs to develop during embryogenesis [1]. Vascular development proceeds in the mouse during the early postnatal period in the lung [2], the heart [3], the brain [4, 5]and the retina [6]. During adulthood, blood vessel formation resumes during reproduction and wound healing. Excess formation of blood vessels may contribute to hypertrophic scar proliferation, chronic inflammatory disorders, diabetic retinopathy, or tumor growth [7]. On the contrary, insufficient blood vessel formation is responsible for tissue ischemia, delayed wound healing and chronic ulcers. Blood vessels initially develop as endothelial cell-lined channels by in situ differentiation of angioblasts and their subsequent assembly into a primitive vascular plexus (vasculogenesis) (Fig. 1a). Subsequently, the primitive vascular network expands and remodels via sprouting angiogenesis (budding of new sprouts from preexisting vessels), nonsprouting angiogenesis (intercalated growth of endothelial cells) or intussusceptive growth (whereby new intercapillary pillars split preexisting vessels). Investment of the endothelial tubes by periendothelial mural cells (pericytes around capillaries, smooth-muscle cells around arteries) is essential for their structural integrity and vasomotor control [8–10](Fig. 1b, c). Increased vascular supply can also be mediated by remodeling of preexisting vessels such as of the coronary arteries [11]. As recent discoveries have identified several candidate angiogenic or angiostatic factors, increasing interest has arisen to study their biological role or to test their potential therapeutic value. Consequently, the need for quantitative angiogenesis assays has been growing during recent years. Whereas previous reviews have discussed the use of angiogenesis assays in other species [12–14], this section will review the use of the mouse as a model to study blood vessel formation and function.

Fig. 1

Different phases of embryonic vascular development. (a) Endothelial precursors (angioblasts) differentiate to early endothelial cells, which become assembled into a primitive capillary plexus (vasculogenesis). This emerging network expands via intussusceptive growth, intercalated growth and sprouting angiogenesis, after which it becomes remodeled via pruning, fusion and regression of preexisting vessels into a tree of arteries, capillaries and veins. Endothelial cells further differentiate and acquire specific properties such as the formation of a tight barrier in the brain, or the formation of fenestrations in exocrine glands. (b) Mesenchymal cells differentiate to pericytes (PC) and primitive smooth-muscle cells (SMC), which become recruited around the endothelial tubes. Mural cells migrate longitudinally alongside the endothelial tubes or sprout from preexisting muscularized vessels. Both endothelial and mural cells constitute the essential cellular components of a mature vascular network.

1.1 Vascular development in the mouse embryo

The study of the embryonic vasculature has yielded novel insights in the molecular mechanisms of blood vessel formation [1, 8, 10]. Histological analysis, immunostaining and in situ hybridization for endothelial, periendothelial or basement membrane markers allows the study of embryonic vessels on tissue sections or in whole mount embryos [15–17]. Specific markers can be used to identify the different vascular beds; for example, GLUT-1 marks the developing blood–brain barrier [18], whereas FLT-4 identifies lymphatic vessels (at least beyond midgestation) [19]. The embryonic vasculature can also be visualized by inter-crossing the target mice with other transgenic mouse strains, expressing a ß-galactosidase marker gene in endothelial or mural cells [20–23](Fig. 2a, b). Computer-assisted image analysis can be used to reconstruct the three-dimensional pattern of the embryonic vasculature [15]. In addition, barium-gelatin angiography, and scanning electron or light microscopy of vascular casts in embryos aged 9 to 20 days have been used to visualize their three-dimensional architecture [2]. Blood vessel function and blood flow can be studied in embryos, freshly dissected and still attached to their yolk sac during the early stages of embryogenesis (<12.5 days of gestation), or in cultured embryos during the initial period of vascular development (8.0 to 10.5 days of gestation) by transillumination of the transparent yolk sac and embryo [24, 25], or after injection of fluorescent dyes (microangiography) [26]or other tracers (ink) [27](Fig. 2c, d). Injection of trypan blue or horse radish peroxidase has been used to study the development of the blood brain barrier [18]. Pulsed Doppler has been used to measure blood velocity in 10.5- to 14.5-day-old embryos in utero while maintaining uteroplacental continuity [28]. Noninvasive high-resolution nuclear magnetic resonance imaging may provide future opportunities to monitor vascular development.

Fig. 2

Angiogenesis in the mouse embryo. (a) Whole mount staining for β-galactosidase of a wild-type 9.5 day-old embryo expressing the blue marker in the entire vasculature after interbreeding onto a TIE1:LacZ background. (b) Section through a 9.5 day-old embryo (the same as in panel a), revealing the blue endothelial cells throughout the entire vasculature. (c) Ink injection into the heart of a 9.5 day-old embryo, visualizing the embryonic vascular network. (d) Whole mount view of a cultured 9.5 day-old embryo (still attached to the transparent yolk sac), revealing the blood-filled vitelloembryonic vessels.

The developing embryonic vasculature (like other neovessels) is sensitive to (anti)-angiogenic factors. Immunoglobulins (type IgG) and only a few growth factors (including TGF-ß1) will be transferred across the yolk sac or placental barrier after intravenous injection into the maternal plasma of pregnant females or upon supplementation into the medium of cultured embryos [29, 30]. Other molecules (growth factors, antibodies, or drugs) with (anti)-angiogenic properties have to be injected into the extracoelomic cavity or, directly, into the heart of the embryo in utero or in culture [25]. Intracardiac injection of recombinant adenoviruses, resulting in the preferential expression of the transferred genes in the embryonic vascular endothelium, allows visualization (ß-galactosidase marker gene) or genetic manipulation (anti-angiogenic genes) of the embryonic vasculature [26, 31].

1.2 Vascular development in the neonatal mouse

Physiological development of blood vessels in the mouse proceeds during the first postnatal weeks. As growing and remodeling blood vessels are more sensitive to modulation by (anti)-angiogenic factors than mature, quiescent vessels, and since vascular growth in the adult mouse is restricted to reproduction or to pathological conditions (tumorigenesis, inflammation etc.), these neonatal angiogenesis models offer a unique opportunity to study the molecular mechanisms of physiological vessel formation.

The myocardium becomes vascularized during embryonic development by an interconnected network of coronary arteries, capillaries and veins [32]. Immediately after birth, the pulmonary pressure drops (as a result of lung inflation), concommittant with a rise in the systemic pressure. The heart adapts to these hemodynamic changes by a two- to three-fold hypertrophy (and, to a limited extent, hyperplasia) of the cardiomyocytes [3]. The increased metabolic demands of the postnatal heart are met by a significant angiogenic response during the first two postnatal weeks, resulting in a two-fold increase in the number of myocardial capillaries. Methods for morphometric quantitation of the capillary density, the capillary-to-cardiomyocyte fibre ratio, the capillary length, surface or volume densities and proliferation have been described [3]. The tree of larger coronary vessels, which are mature at birth, can be visualized by corrosion casting or microangiography [33].

The retina is largely avascular at birth, and only becomes vascularized during the first three weeks of life. Natural development of the retinal vasculature is considered to be regulated by oxygen and mediated by vascular endothelial growth factor (VEGF) [6]. Physiological levels of hypoxia, caused by increasing demands for oxygen at the onset of neuronal activity, are detected by strategically located populations of neuroglia that secrete VEGF and induce the formation of retinal vessels. As the vessels become patent, the hypoxic stimulus is relieved, so vessel formation is matched to oxygen demand. Vessel regression is a natural response to oxygen surplus, resulting in capillary remodeling and formation of capillary-free zones [34]. Initially, a capillary plexus consisting of endothelial cell-lined channels is formed, which subsequently becomes invested by pericytes. The retina is particularly suited to study pericyte biology as it contains the highest ratio of pericytes versus endothelial cells [35]. The retinal vasculature is readily accessible for qualitative and quantitative analysis via angiograms using fluorescent markers, whole mount staining (using endothelial or pericyte-specific markers), or histological analysis [36, 37]. As the immature retinal vasculature is more sensitive to changes in oxygen concentrations than the mature adult network, a model of ischemic retinopathy was developed whereby hyperoxia induces irreversible damage to immature retinal vessels of the neonate, resulting in intense retinal ischemia [37]. When the neonate is returned to normoxia, a second phase is initiated, distinguished by excessive revascularization with abnormally leaky vessels. This model, which mimicks to a certain extent the vascular response during retinopathy of prematurity or diabetic retinopathy, may be useful to test the efficacy of (anti)-angiogenic molecules [38].

Another vascular network in the eye which undergoes significant remodeling shortly after birth is the hyaloid vasculature, which grows into the vitreum and onto the surface of the lens [39, 40]. The hyaloid system consists of the tunica vasculosa lentis (on the posterior lens), the vasa hyaloidia propria (in the vitreum), and the pupillary membrane (on the anterior surface of the lens in the anterior chamber). Regression of this system, which disposes cells that would disturb the passage of light to the retina, begins with the vasa hyaloidia propria at about 5 days after birth and is completed by about 21 days. The pupillary membrane has proven useful for analysis, mainly because it can be removed from the eye by dissection and visualized in toto. Vital and histological analyses can be used to determine the blood flow through and the fate of whole capillary segments in vivo [40].

The growth and development of the brain and its blood vessels are intimately linked [5]. The somatosensory cortex is an attractive model for following this development because its topographic neuronal organization and its integrated vasculature are largely absent at birth and develop postnatally. Neonatal mice have dense surface capillary networks with numerous anastomoses between arterioles and relatively small and irregular veins. Intraparenchymal blood vessels grow from endothelial sprouts that penetrate the embryonic brain from the pial vascular plexus. Morphological changes over the first two postnatal weeks include a three- to four-fold increase of the intraparenchymal capillary density, and an associated increase in the intraparenchymal capillary length (likely reflecting an adaptation to increased neural activity and metabolic rate). These angiogenic changes appear to be mediated in part by VEGF [4]. In addition, a reduction in the density of the pial vascular network occurs, which contains venules with increased diameter, length and less tortuous trajectory, and fewer arteriolar anastomoses. In vivo videomicroscopy with fluorescent tracers in cranial windows has been used to image the dynamic changes in patterning and blood flow in anesthetized mice [5]. Perfusion with photographic emulsion allows reconstruction of cortical blood vessels, whereas horse radish peroxidase injection has been used for the measurement of permeability [5].

The rate of postnatal growth of the lung is most rapid in the neonatal period, resulting in a 10-fold increase in the number of alveoli, and a doubling in the alveolar size [41]. At birth, the distribution of pre-acinar arteries is completed, but with increasing age, there is a significant growth of many small intra-acinar arteries as the lung adapts from a high-pressure/low-flow system to one with low pressure/high flow. Structural remodeling of the walls of pre- and intra-acinar arteries and veins is a striking feature of normal lung growth. External (pathological) stimuli to the lung and its vessels (including changes in oxygen or blood pressure) during this period of rapid expansion induce responses greater in magnitude than stimuli occurring after growth has ceased.

1.3 Angiogenesis in the adult mouse

Neovascularization in the adult mouse results in most instances from angiogenic sprouting of new capillaries from preexisting vessels [8], although colonization of peripheral tissues by bone marrow-derived angioblasts has recently been proposed as an alternative mechanism [42]. In addition, remodeling of larger preexisting vessels (such as of the coronary arteries) is an alternative mechanism whereby the increased demands for vascular supply of nutrients and oxygen are met [11]. Lymphangiogenesis is a poorly characterized process that only recently has become accessible to molecular analysis [43]. The following discussion reviews the different characteristics of assays, categorized according to their use in studying angiogenesis induced by (i) application of angiogenic factors; (ii) implantation of tumors; (iii) wound healing; (iv) tissue ischemia; or (v) lymphangiogenesis (Fig. 3).

Fig. 3

Overview of the different angiogenesis models in the mouse.

The effect of (anti)-angiogenic molecules has been frequently studied in the corneal micropocket assay [44]. Implantation of hydron pellets containing angiogenic factors in the stroma of the avascular mouse cornea allows screening of the angiogenic response in the absence of inflammation. As the cornea in the mouse is thinner than in other species, growth of new blood vessels largely occurs in a two-dimensional plane. Implantation of pellets on the iris into the anterior chamber of the eye is technically more challenging due to its small size. This model (which can also be used to study vascularization induced by implantation of tumor cells, or by thermal or chemical injury) is strain-dependent, as nude mice, spontaneously develop a higher degree of vascular channels within the peripheral stroma [45]. Other mouse strains, including Corn1 mice, spontaneously develop irregular thickening of the corneal epithelium and significant stromal neovascularization [46].

Chronic transparent chambers in the brain and the dorsal skin offer the advantage of quantitating the length, surface area, volume and number of vessels of the network, as well as the dynamic changes of blood flow, permeability, and shear stress using videomicroscopy and computer-assisted image analysis [47, 48]. The skin model allows visualization of blood vessels by transillumination, whereas epi-illumination and injection of contrast agents (fluorescent dyes, photographic emulsion, corrosion casts) are required in the cranial model.

An alternative method to study new vessel growth involves the implantation of a polymer matrix (gel, sponge, capsules etc) containing (anti)-angiogenic factors [49–52]. Although analysis of blood vessel formation in the latter model largely relies on histological analysis, insights in the angiogenic response can be obtained through quantitation of the hemoglobin and red blood cell counts, measurements of blood-flow rates using radioactive tracers, or analysis of biochemical parameters, characteristic of angiogenesis (such as for example extracellular matrix turnover).

In contrast to the significant angiogenic response in other rodents [53], neovascularization in the murine peritoneum after intraperitoneal injection of angiogenic factors is minimal (V. Laux, personal communication). However, because of their transparency (∼120 μm thin), the exteriorized peritoneum or cremaster muscle offer the advantage of evaluating in situ, the microvascular changes in permeability, flow-rates, vasomotor activity and leukocyte rolling after topical application of angiogenic factors [54–56].

The hairless mouse, although similar in appearance to the nude mouse, is immunologically competent, with a normally functioning thymus and T-cells [57, 58]. The ear of the hairless mouse comprises a central cartilaginous sheet, sandwiched between two full-thickness dermal layers (together ∼300 μm thick), and measures 10 to 13 mm in width and length. Its nourishment is supplied by three to four neurovascular bundles, which can be directly observed through vital microscopy. Neovascularization after injection of angiogenic factors (or implantation of tumors, or full thickness wounds) can be quantitatively evaluated [59], although the model suffers from the limited ability to keep the grafted tissue undisturbed and from the poor optical quality after graft implantation.

Tumor vascularization, and its modulation by (anti)-angiogenic factors [60], has been studied using chronic transparent chambers in the brain and the skin [47, 48]. Videomicroscopy of the vascularization of tumors implanted in the chambers allows easier visualization than in other tissues without a chamber [61]. Xenografts can be implanted in chambers placed in immunodeficient recipients. Ingrowth of human vessels into human tumor grafts can be studied by prior transplantation of full-thickness human foreskin grafts onto immunodeficient mice, and subsequent grafting of human cancer cells into the human skin [62]. However, after 3 to 4 weeks, murine blood vessels progressively vascularize the human tumor. Matrigel, a reconstituted extract of basement membrane, induces an angiogenic response, enhancing thereby the growth of tumors which cannot be grown in xenogenic immunodepressed hosts, even when inoculated at high cell doses [63]. Agarose microencapsulation of tumor cells, attached to microcarrier beads, has been used to isolate the tumors from the immune system of nonsyngenic hosts [64]. Tumor vascularization can also be evaluated by implantation of the tumor cells in the stroma of the mouse cornea, but the angiogenic response is confounded by inflammation. As the role of the (orthotropic) microenvironment of the host tissue is becoming increasingly recognized, other tissues such as the mouse liver have been used to study tumor neovascularization [65].

In contrast to the angiogenic response induced by tumors implanted into (immunodeficient) hosts, vascularization can be also studied in tumors spontaneously developing in situ in transgenic mice carrying dominant oncogenes (bovine papilloma virus-1 genome [66], Simian virus 40 large-T antigen [67], Fps/Fes tyrosine kinase [68], Polyoma middle-T antigen [69, 70]) or in tumor suppressor-knockout mice (p53) [71]. Such transgenic models more closely mimics the multistage growth of carcinogenesis with the onset of an angiogenic switch as a hallmark of malignant tumor progression. In addition, they allow three-dimensional tumor growth, required for reproduction of (hypoxic/hypoglycemic) stress-induced tumor angiogenesis [72, 73]. Unfortunately, direct visualization of the dynamic aspects of the microcirculation via microvascular techniques is limited. Noninvasive high-resolution nuclear magnetic resonance imaging may provide a means for monitoring the development of the tumor vascularization over its entire time course [72].

Vascularization of healing wounds is a natural process, which can be studied using the corneal chemical or thermal injury model, the dorsal skin or cranial window models, the hairless mouse ear model, or the models relying on implantation of gels, sponges or perforated capsules, containing (anti)-angiogenic factors [74].

Models of angiogenesis, induced by tissue ischemia, have been recently developed in the mouse. Limb ischemia due to permanent ligation of the femoral artery, induces the formation of collateral vessels over a time course of several weeks [75]. Revascularization of the ischemic extremities can be monitored noninvasively using Laser Doppler perfusion imaging, which has also been used to measure microregional erythrocyte flux in tumor implants in mice [76]. As synchrotron radiation microangiography (with a resolution limit of 30 μm) has been recently used to visualize small collateral arteries in the rat [77], it may provide a means to detect changes in the microcirculation in the mouse as well. Permanent ligation of the left anterior descending coronary artery in the murine heart induces myocardial ischemia, resulting in the infiltration of the ischemic regions by a highly vascularized granulation tissue as well as in the remodeling of the coronary arteries ([33]and unpublished observations). Microangiography and corrosion casts have been used to visualize the myocardial vascular network [33]. A mouse model of diabetic retinopathy based on the prolonged feeding of galactose-rich diets has been developed [78]. Although this model is characterized by the typical saccular microaneurysms, thickening of the capillary basement membrane and pericyte drop-out, neovascularization does not occur in this model. The model of ischemic retinopathy in neonates [37]has been described above.

Fluorescence microlymphangiography allows the examination of the endogenous or tumor-associated lymph circulatory system in the mouse [79, 80]. Although lymph capillaries can be histologically distinguished from vascular channels by their underdevelopment of a basement membrane, absence of red blood cells (except during conditions of injury or infection), and lack of tight junctions, the identification of a novel lymph-angiogenic marker, FLT-4 [19], and the generation of mice expressing the ß-galactosidase marker under control of the endogenous FLT-4 promoter (K. Alitalo, personal communication) might aid in the visualization of these lymphatic structures.

1.4 Gene transfer in the vasculature

Gene transfer offers the opportunity to genetically manipulate the endothelium (smooth-muscle cell gene transfer in the mouse is discussed in the section on arterial restenosis). Due to its small size, only systemic instead of local gene transfer in the mouse has been successfully employed thus far. Recombinant adenoviruses, injected intracardially into early stage embryos, uniformly infect the entire vascular bed. Use of viruses expressing a ß-galactosidase marker gene allows visualization, whereas use of viruses encoding (anti)-angiogenic genes allow genetic manipulation of the endothelial or periendothelial cells in the embryonic vasculature [26, 31]. Later during embryonic development and after birth, systemic injection of recombinant adenoviruses results in the preferential infection of the liver parenchyme with only minimal infection of the systemic endothelium. Cationic liposome–DNA complexes may be better vehicles for targeting transgenes to the endothelium in an organ-specific pattern [81]. Endothelial cells within certain organs [82]or tumors [83]express specific markers, a property that can be exploited for targeting (anti)-angiogenic factors.

1.5 Technical considerations

The number, length and surface area of blood vessels, and their constituent cells (endothelium and pericytes/smooth-muscle cells) or extracellular matrix can be analyzed by standard histological and immunocytochemical methods. Endothelial cells in sprouting vessels fail to stain for von Willebrand factor, but are immunoreactive for CD31, CD34, lectin, VE-cadherin, thrombomodulin or the αvß3 [37, 84–87]integrin. Other endothelial markers include FLK-1 FLT-1, TIE-1 or TIE-2 [88, 89], whereas FLT-4 identifies lymphatic endothelial cells [19]. Smooth-muscle α-actin, desmin, vimentin, smooth muscle myosin, tropomyosin, or the PDGFR-ß mark pericytes but their expression pattern varies in different vascular beds and, in addition, these markers frequently identify vascular smooth-muscle cells [90–94]. Morphological and functional characteristics of the endothelial cells in vivo (e.g.the size, shape, border and number of endothelial cells; the number and distribution of endothelial gaps; the sites and degree of the permeability; or the attachment of leukocytes) can be studied using silver nitrate, Monastral blue B, Evans blue, India ink, radioactively labelled tracers or fluorescent microspheres [95–101]. The three-dimensional pattern of a vascular network can be visualized by whole mount immunostaining or in situ hybridization [16], by injection of intravascular markers (colloidal carbon, India ink, radioactively labeled red blood cells or albumin, fluorescent high-molecular weight tracers or liposomes) [27, 36, 99], by barium-gelatin angiography [2], by vascular casting [33, 96], or, hopefully in the near future, by magnetic resonance imaging. Its pattern can be reconstructed via computer-assisted imaging [15, 102]. Blood flow can be monitored noninvasively using Doppler ultrasound [103], laser Doppler [76], microvascular videomicroscopy [12], magnetic resonance imaging [104], or, invasively, by colored microspheres or radioactive tracers [105, 106]. Shear stress, direction and speed of the blood flow, capillary perfusion, and cell–cell interactions (leukocyte rolling, extravasation) can be measured using intravital videomicroscopy [12]. Interactions between blood-borne cells (leukocytes, tumor cells) and vascular endothelium can be studied by prelabeling the circulating cells with fluorescent dyes [107], or using fluorescent dyes which selectively attach to neutrophils (acridine orange) [108]. Endothelial cell proliferation can be evaluated by quantitative autoradiography of incorporated 3H-labeled thymidine, or by histological analysis of proliferation markers (Ki67, PCNA) or of incorporated 5′-bromo-2′-deoxyuridine (BrdU) [109]. Endothelial cell death can be analyzed in situ using vital dyes, routine light microscopy or electron microscopy, the TUNEL method [34], or biotin-labeled Annexin V [110].

1.6 Transgenic mouse models

Several transgenic mouse models of perturbed vascular development have been developed over the last years. Abnormal embryonic vascular development, resulting from defects in the formation of a primitive capillary plexus, has been observed in mice lacking VEGF [27], FLT-1 [20], FLK-1 [22], TGF-ß [111], fibronectin [112], or VE-cadherin [113]. Defects in the expansion and remodeling of the embryonic vasculature occur in mice deficient in TIE-1 [21], TIE-2 [114], angiopoietin-1 [115], Braf [116], HIF-1α (unpublished observations), ARNT [117, 118], VHL [119], VCAM-1 [120], the α4 integrin [121], or in mice overexpressing neuropilin [122], or angiopoietin-2 [23]. Impaired recruitment and investment of mural cells has been observed in mice with disruption of the genes encoding PDGF-B [91], PDGF-B receptor [123], tissue factor [24], and LKLF [124], whereas reduced formation of extracellular matrix occurs in mice lacking collagen type I [125].

Abnormal blood vessel development or function after birth occurs in mice lacking collagen type III [126](vessel fragility with bleeding), P-selectin [54](reduced leukocyte rolling), or fibrillin-1 [127](aneurysmal dilatation and rupture). Mice overexpressing VEGF in the retina [128], or mice with sickle cell disease [129]develop retinal and choroidal neovascularization, whereas overexpression of the Fps/Fes tyrosine kinase [68]or the Polyoma middle-T antigen [69, 70]results in the formation of hemangiomas. Overexpression of VEGF-C in the skin induces lymphangiomas [130], while overexpression of the bovine polyoma virus-1 genome [66], the simian virus large T-antigen [67], or conditional overexpression of VEGF [131]results in the formation of vascularized tumors.

2 Arterial stenosis and reendothelialization

Arterial injury such as occurs after balloon angioplasty or placement of intravascular stents frequently results in chronic narrowing of the lumen due to constrictive remodeling of the vessel wall and intimal accumulation of smooth-muscle cells and matrix [132, 133]. Few long-term treatments are available, mandating a better understanding of this process at the genetic level. To date, only a few mouse models of arterial injury have been developed, despite significant efforts (Table 1). The mouse differs from other species by the thinness of its arteries (only 2 to 3 smooth-muscle cell layers and elastic laminae are present in the media), by the absence of smooth-muscle cells in the intimal layer, by the lack of side-branches in the carotid artery, and, last but not least, by its small size (the lumen of the carotid artery is ∼300 μm wide).

View this table:
Table 1

Overview of arterial stenosis used in the mouse

Mechanical injuryElectric injuryRemodeling modelPerivascular cuff
Medial necrosis25–50%100%60%ND
  • ND: Not determined.

2.1 Mechanical injury models

Based on a widely used mechanical injury model in the rat [134], a flexible nylon filament wire was used to mechanically injure the murine carotid artery [135]. This results in complete denudation of the endothelium, disruption of the internal elastic lamina and focal injury of the medial smooth-muscle cells (∼25% cell death). Thrombosis or inflammation are minimal to absent. Within 48 h, residual medial smooth-muscle cells start to proliferate and to migrate across the elastic laminae, with the first smooth-muscle cells accumulating in the neointima by day 4 after injury. Proliferation of smooth-muscle cells in the media is maximal (∼10%) within 5 days after injury, and in the intima (∼60%) within 8 days after injury. Despite these significant proliferation rates, the (frequently eccentric) neointima contains only two to three layers of smooth-muscle cells (∼120 cells per cross-section), covering an area of ∼0.012 mm2 within 2 weeks after injury. It is unknown whether arterial remodeling occurs in this model. Reendothelialization of the denuded area is complete within 3 weeks after injury.

This model was subsequently modified by using different mechanical devices to denude the endothelium and to injure the medial cells [136–139]. In our hands, the use of a guidewire (∼350 μm) induces a significant neointimal response containing several layers of smooth-muscle cells (∼250 cells per cross-section) within 3 weeks after injury (Fig. 4a, b). This more vigorous response appears to be, at least in part, due to a more severe injury (∼50% medial cell death). Initial analysis indicates a significant influence by the genetic background of the mouse, as C57Bl/6 mice appear to be less responsive than mice with a mixed genetic background of C57Bl/6 and 129. In aggregate, this model allows the study of the molecular basis of the smooth-muscle cell response to endothelial denudation and medial injury in the presence of only minimal thrombosis and inflammation. In addition, the type of injury mimics balloon angioplasty, frequently used in humans.

Fig. 4

Arterial stenosis models in the mouse. (a) Histological section through a normal femoral artery. (b–d) Within 2 to 3 weeks after injury, a neointima develops in the femoral artery after mechanical injury (b), perivascular injury (c) or in the carotid artery after perivascular cuff application (d). (e, f) Neointima formation in a carotid artery transplant, stained with hematoxylin–eosin (e) or smooth muscle α-actin (f).

2.2 Perivascular electric injury

To circumvent the technical challenge of inserting intravascular devices, a perivascular injury model was developed whereby an electric current is passed through the wall of femoral or carotid arteries using an electromicrocoagulator [109]. Electric injury induces a severe injury, destroying all cells across the vessel wall, including the smooth-muscle cells in the media, the endothelial cells in the intima and the fibroblasts in the adventita. This results in transient platelet- and fibrin-rich thrombosis, and removal of the necrotic cell and matrix debris by infiltrating leukocytes in the intima, media and adventitia during the first week after injury. Because of the complete cell necrosis across the injury, healing of the vascular wounds initiates from the adjacent uninjured borders. Smooth-muscle cells in the adjacent uninjured borders start to proliferate and to migrate across the internal elastic lamina into the intima (Fig. 4c). Subsequently, they migrate alongside the lumen and within the media, thereby repopulating the necrotic wound. A similar pattern of healing also occurs in end-to-end microvascular anastomoses [140], venous or prostethic grafts [141], or laser-induced thermal injury [142]. Sequential analysis of the topographic pattern of cell accumulation in the wound during the entire time-course of healing allows quantitatation of smooth-muscle cell migration. Beyond two weeks after injury, the intima contains on average ∼13 layers of α-actin immunoreactive smooth-muscle cells (∼200 cells per cross-section), covering an intimal area of ∼0.03 mm2 and stenosing the lumen for ∼30%. Vascular remodeling (e.g. the adaptive outward displacement of the vessel wall in response to intimal thickening so that the vascular lumen is preserved) prevents progressive stenosis of the lumen, despite increasing neointima formation. A marked neoadventitial response occurs, characterized by accumulation of leukocytes and fibroblasts, and secondary matrix deposition. Reendothelialization (resulting from both proliferation and migration of endothelial cells) in this model is rapid and complete within two weeks after injury. This model has allowed us to evaluate the specific role of the different components of the plasminogen system [136, 137, 143, 144]. Notably, qualitatively similar genotypic differences were obtained when the mechanical injury model was used. In addition, this model has been used successfully to test the suppressive effect of adenoviral gene transfer of PAI-1 (the primary inhibitor of the plasminogen activators) on neointima formation [137]. When the replication-defective PAI-1 adenovirus was injected intravenously, infected parenchymatous liver cells produced large amounts of PAI-1, resulting in ∼1000-fold increased plasma PAI-1 levels. Secondarily, PAI-1 was deposited into the developing neointima, preventing smooth-muscle cell migration and the formation of a neointima in situ. The wounded artery was not directly transduced by the PAI-1 adenovirus. Thus, this model offers the opportunity to study the molecular mechanisms of smooth-muscle cell migration during conditions of thrombosis, inflammation and matrix remodeling, that are more severe than in the mechanical injury model (see above) or in the other injury models (ligation model, perivascular collar, etc; see below). It should be noticed that fibrin-rich clots, matrix deposits and leukocytes are frequently present in atherosclerotic arteries subjected to vascular interventions of angioplasty or stenting [133, 145]. In addition, this model allows the study of the molecular basis of the physiological (outward) remodeling in response to intimal thickening.

2.3 Arterial remodeling models

Vascular remodeling is an adaptive process that occurs in response to chronic changes in hemodynamic conditions [146]. It involves changes in cell growth, cell death, cell migration and in extracellular matrix composition, that lead to a compensatory adjustment in vessel diameter and lumen area. The blood vessel essentially remodels such that the lumen area is modified to maintain shear stress (a reduction in blood flow increases intimal lesion formation in vascular grafts and balloon-injured vessels) and/or wall stress (determined by the diameter of the vessel and blood pressure, and inversely by the vessel wall thickness). Constrictive remodeling during restenosis may result in late luminal area loss [147]. Remodeling during atherosclerosis initially involves outward displacement of the vessel wall so that the vascular lumen is relatively preserved, but once this adaptive mechanism of remodeling is exhausted, luminal narrowing occurs during advanced atherosclerosis [148].

To study the molecular mechanisms of arterial remodeling during arterial stenosis, a mouse model was developed, in which the blood flow (but not the pressure pulsations) in the common carotid artery is disrupted by ligating the vessel near the distal bifurcation [149]. The endothelium in the ligated vessels is not denuded, but becomes detached from the underlying internal elastic lamina, thereby forming spaces that are filled with red blood cells. Within two days after ligation, a marked loss of smooth-muscle cells in the media is evident (∼60% cell death). Luminal narrowing (∼80% reduction) occurs in part by formation of a concentric neointima, containing on average ∼100 α-actin immunoreactive smooth-muscle cells per cross-section and covering an intimal area of ∼0.035 mm2. In addition, the vessel becomes constricted as evidenced by the ∼25% reduction in the outer circumference. Thrombosis only occurs within the first 2 mm of the ligature. Leukocytes and platelets penetrate through discontinuities in the endothelial sheet into the subendothelial space, and are detected during the initial postligation period in all layers across the vessel wall. Possibly, the altered flow conditions influence the expression of endothelial-leukocyte adhesion molecules and chemokines, with a subsequent influx of inflammatory cells. The latter presumably play an essential role in this model, as ligated P-selectin-deficient arteries are not infiltrated by leukocytes, and only develop a minimal neointima without luminal narrowing [149]. This model thus offers the opportunity to study the molecular mechanisms (in particular the role of leukocyte–platelet–smooth-muscle cell interactions) contributing to the formation of intimal lesions in the absence of noticeable endothelial denudation. In addition, it allows us to unravel the molecular basis of the vascular constriction, accompanying conditions of perturbed flow and turbulence.

Vascular remodeling during atherosclerosis was examined in mice lacking the apolipoprotein E (Apo-E) alone [107], or in combination with the low-density lipoprotein (LDL) receptor [150]. Despite significant intimal thickening, the lumen in the atherosclerotic aorta was not smaller because of outward displacement of its vessel wall. However, in the femoral and carotid arteries, the lumen is constricted, possibly related to the media necrosis and adventitial inflammation [107]. Endothelium-dependent relaxation to receptor- and nonreceptor-mediated agonists was significantly impaired in the atherosclerotic vessels [150]. We have also observed that the vascular lumen in the aorta from mice lacking ApoE, ApoE and the plasminogen activator (PA), tissue-type PA (t-PA) or urokinase-type PA (u-PA) remains relatively preserved because of adaptive remodeling, despite large intimal plaques [151]. However, vascular remodeling in the ApoE- and ApoE:t-PA-deficient but not in the ApoE:u-PA-deficient aorta was accompanied by additional destruction of the media, resulting in aneurysm formation (see below) [151]. Why media necrosis results in luminal stenosis in peripheral arteries, in contrast to the aneurysmal dilatation in the aorta, remains to be further determined.

2.4 Other arterial stenosis models

Other models have also been used to induce intimal hyperplasia, but have not been characterized in detail or are still under investigation. A perivascular cuff model based on the use of a nonconstricting hollow polyethylene tube (similar to that used in the rabbit [152]), induces a significant neointima in mice within three weeks [153]. In this model, the endothelium is not denuded, there is only minimal thrombosis and smooth-muscle cell death, and the elastic laminae remain structurally intact. A concentric neointima, consisting of 8 to 10 smooth-muscle cell layers and a significant neoadventitia, characterized by active neovascularization, develop within 3 weeks (Fig. 4d). Although the precise mechanisms responsible for intimal thickening in this mouse model remain undetermined, previous studies in other species have suggested that this may result from endothelial activation, kinking of the artery, vasospasms, increased flow velocity, blockage of the lymphatic drainage, damage of the adventitial nerves of vasa vasorum (resulting in hypoxia of the media), or chemical toxicity of the collar [152]. Leukocytes presumably play a central role in triggering the smooth-muscle cell response.

Another model involves the surgical removal of an intimal/medial fragment of the carotid artery using an intraluminal tungsten wire to induce an intimal response (P. Yiu and J. McEwan, personal communication). Smooth-muscle cells migrate through the breaks in the elastic laminae into the intima to form an eccentric intimal lesion, which is proportional to the depth of the injury. In a venous graft model, a fragment of an autologous vein is surgically sutured into an arteriotomy in the carotid artery (V. Shi, personal communication). Activation or dysfunction of the grafted venous endothelium by the arterial blood pressure may play a role in the induction of the smooth-muscle cell response and intimal thickening of the graft. Although reproducible, this model is technically challenging and the lack of clearly visible elastic laminae in the venous graft may hamper morphometric quantitation. Other models relying on the use of external crush, air desiccation or cryoinjury have been attempted without much success. Photochemical dyes (such as Rose Bengal) induce controllable endothelial denudation and medial injury, once they are activated by transluminal green light, and have been used to induce an intimal response in other species [154]. Their usefulness in mice is currently being tested.

2.5 Gene transfer in murine arteries

Local gene transfer in the vessel wall of larger species has been based on the use of recombinant adenovirus [155], adeno-associated virus [156], liposomes [157, 158], naked DNA [159, 160], or on the seeding of retrovirally-transduced endothelial [161]or smooth-muscle cells [162]. Due to its small size, local gene transfer in the murine arteries has not been achieved yet, and systemic injection of adenovirus does not result in significant transduction of the quiescent or injured murine artery [137]. However, systemic gene transfer may be successfully used in the mouse. Indeed, when recombinant adenovirus is systemically injected in the mouse, it preferentially transduces parenchymatous liver cells which subsequently produce large amounts of the transgene. This results in plasma levels of the expressed protein of 10 to 100 μg per ml for at least 7 to 10 days [137]. An advantage of such systemic gene transfer is the absence of local inflammation or perturbed gene expression in the artery, as occurs after local adenovirus-mediated gene transfer in larger species [163]. The feasibility of such a strategy was demonstrated by the inhibitory effect of adenovirus-mediated gene transfer of PAI-1 [137]or ApoA-I [139]on neointima formation in mice. Use of the second generation adenovirus vectors, which prolongs expression of the transgene up to several months, may even further improve the applicability of this strategy [164].

2.6 Transgenic mouse models

Mice with genetic deficiencies of the different components of the plasminogen or matrix-metalloproteinase system have been used to analyze the wound-healing response of mechanically or electrically injured arteries [136, 137, 143, 144, 165]. Notably, both injury models revealed a similar requirement for urokinase-generated plasmin proteolysis in migration of smooth-muscle cells and tissue remodeling by infiltrating leukocytes. Furthermore, systemic overexpression of their inhibitor PAI-1 by intravenous injection of a recombinant adenovirus expressing PAI-1, reduced arterial neointima formation [137]. The response to mechanical injury has also been studied in mice lacking the estrogen receptor-α [138, 166]or apoliprotein E [139]. In the latter study, a significant suppression of neointima formation by adenovirus-mediated transfer of the apolipoprotein A-I gene was reported [139]. The perivascular cuff model has been used to characterize the intimal response in mice lacking endothelial nitric oxide synthase [153]. The ligature model was used in P-selectin-deficient mice [149].

Taken together, the transgenic mouse offers a unique opportunity to unravel the molecular basis of the intimal response at the genetic level. Despite significant progress, the currently available mouse models suffer shortcomings and require further optimization. (i) Uninvolved arteries have thus far been selected for injury, and it is likely that the arterial wound-healing response in diseased (atherosclerotic) arteries, such as in the atherosclerotic apoE- or LDLr-deficient mice, will differ significantly. (ii) Arterial stenosis in patients does not only result from intimal thickening but also from arterial wall remodeling (except during stent stenosis, when shrinkage of the vessel wall is prevented). Arterial remodeling was only characterized in the perivascular electric injury model (in which it results in lumen expansion instead of narrowing) and in the ligation model (in which remodeling induces constriction of the lumen). No information is available on the type or degree of remodeling in the other models. (iii) Local (gene-) transfer in the mouse artery remains a technical challenge. (iv) None of these models, which differ significantly in their mechanisms, ideally mimic the human disease process. A more in-depth analysis and comparison between the various injury models will increase our understanding of the role of candidate disease molecules in the various aspects (cellular migration, proliferation, matrix remodeling etc.) of the arterial-healing process. Indeed, the mechanical injury models are suited to the study of the intimal response in the absence of thrombosis and inflammation, whereas the perivascular electric injury model allows the study of intimal thickening induced by a more severe injury associated with thrombosis and inflammation. The ligation and collar models, on the other hand, allow the study of the role of platelet–leukocyte interactions in the intimal response in the absence of endothelial denudation.

3 Transplant-induced arteriosclerosis

Transplant-associated arteriosclerosis is the major cause of cardiac allograft failure after the first postoperative year, and it appears to be a significant problem in the long-term survival of other solid organ transplants [167]. In contrast to the hypercholesterolemia-induced atherosclerosis, the lesion associated with transplant arteriosclerosis involves the artery in a concentric rather than an eccentric fashion, lipid accumulation is less common in its early development, and progression of the disease occurs at a faster tempo. It has been proposed that the parenchymal rejection and vasculopathy are caused by the recipient's immune response to foreign antigens, presented on the cells of the allograft.

3.1 Carotid artery transplant model

A murine model of allograft transplantation has been developed in which carotid arteries are transplanted between a B.10A(2R) (H-2h2) donor mouse and a C57BL/6 (H-2b) recipient mouse, without the use of immunosuppressive drugs [168]. Grafts are paratopically transplanted in an end-to-side anastomosis in two longitudinal arteriotomies, made in the recipient's carotid artery. Only grafts with prominent pulsations (indicative of patent flow and lack of thrombosis) are studied. The endothelium remains structurally intact and apposed to the intima throughout the entire period after transplantation. Within 7 days, the allografted carotid artery forms a three- to six-cell-thick neointima composed of recipient-derived leukocytes (predominantly monocytes or macrophages, and to a lesser extent, CD4+ helper and CD8+ cytotoxic lymphocytes). At 15 days, a third of the neointimal cells are CD45 immunoreactive leukocytes, but by 30 days, the concentric neointima contains predominantly smooth-muscle cells (∼800 cells per cross-section) (Fig. 4e, f). The neointimal accumulation of smooth-muscle cells occurs coincidently with their replacement in the media by nuclear debris and leukocytes which infiltrate through breaks in the elastic laminae. This sequence of events suggests that leukocytes degrade the elastic laminae and infiltrate in the media, activate the medial smooth-muscle cells (presumably by producing cytokines, chemotactic agents and growth factors such as for example PDGF-B and FGF-2), and induce their migration into the intima. Plasmin proteolysis appears essential, as the elastic laminae fail to become degraded, leukocytes are not infiltrating into the media, and neointimal smooth-muscle cell accumulation is significantly impaired in wild-type carotid arteries, transplanted into plasminogen-deficient recipients (unpublished observations).

Subsequent studies in mice deficient in RAG-2 (which lack an antigen-specific cellular and humoral immune response), in μMT (in which B cells are depleted and the humoral immune response is deficient), in CD4 (lacking CD4+ T cells), in MHC class II (depleted of CD4+ T cells), in MHC class I (depleted of CD8+ T cells), or in the mutant beige mice (bg/bg; in which the natural killer cells are depleted) or osteopetrosis mice (op/op; lacking macrophage colony-stimulating factor and depleted of macrophages) revealed that transplant arteriosclerosis depends on T-cell receptor and immunoglobulin gene rearrangement (both consequences of active immunization), but not on CD8+ T cells and natural killer cells [169]. Development of the neointima requires the interaction of CD4+ T cells, B cells and macrophages, as a reduction in the number of mature CD4+ T cells and macrophages, and of the probable absence of allospecific antibodies results in the inhibition of smooth-muscle cell proliferation and/or migration. In another study, luminal occlusion and cross-sectional neointimal area were found to be greater in arteries allografted into hypercholesterolemic ApoE-deficient recipients at 15 and 30 days after transplantation. This appears to be attributable to an increased accumulation of smooth-muscle cells and, to a much lesser extent, of leukocytes, collagen or lipid in the intima [170]. Possibly, the greater smooth-muscle cell response results from an increased production of growth factors by macrophage-derived foam cells than by nonlipid-containing macrophages. Appreciable outward remodeling of the arteries, allografted in the ApoE-deficient recipients, occurs, presumably in response to the excessive neointima formation.

A model of chronic rejection using aortic transplantation from C3H, B10.BR or C3H.SW donors into C57BL/10 recipients has been developed [171]. Mice, free from complications 12-h postoperatively, had long-term survival and developed a marked intimal thickening, rich in smooth-muscle cells, in the aorta transplant after 2 months [171].

3.2 Cardiac transplant model

In order to study the myocardial rejection in addition to the graft coronary vasculopathy, a model has been developed in which mouse hearts are transplanted heterotopically from B10.1 to B.10.BR mice (class I MHC antigen disparity), from bm12 to C57BL/6 mice (class II MHC antigen disparity), or from 129 to C57BL/6 (nonMHC antigen disparity) [172]. This involves ligation of the inferior and superior vena cavae and the pulmonary veins of the donor heart, and anastomosing the donor aorta and pulmonary artery to the recipient abdominal aorta and vena cava, respectively. Transient immunosuppression with anti-CD4 and -CD8 antibodies permits graft survival. Donor/recipient antigenic differences may be either class I or class II major histocompatibility antigens (H-2) or nonH-2 antigens. Initially, an infiltrate of CD4+ and CD8+ T lymphocytes and macrophages of recipient origin concentrates in the intima and adventitia of larger coronary arteries, with little involvement in the myocardium. Subsequently, the intima expands with (primarily) smooth-muscle cells of donor origin, and the media becomes infiltrated by leukocytes. It appears that T cells and heterogeneous antigens in the endothelium and/or the media play an essential role in the induction of graft arteriosclerosis in this model. Transplants between strains that produce antibodies to donor cells (B10.A to B10.BR), develop coronary lesions exceeding those in the reverse combination, in which no detectable antibodies are detected, even though their histoincompatibility is similar [173]. This suggests that humoral immunity is also implicated in the graft arterial disease. Transplantation of 129 allografts into ApoE-deficient C57BL/6 recipients results in an accelerated graft arteriosclerosis, revealing a significant proatherogenic effect of the hyperlipidemic environment [174]. This model has been used to determine the suppressive effects of IFN-γ- [175], ICAM-1- or LFA-1-specific antibodies [176]on transplant arteriosclerosis. A stimulatory effect of IFN-γ was also observed in another study, in which hearts from C-H-2bm12 KhEg (H-2 bm12) donor mice were transplanted into C57BL/6 (H-2b) IFN-γ-deficient recipients after immunosuppression with anti-CD4 and anti-CD8 antibodies [177]. In wild-type recipients, myocardial rejection peaked at 4 weeks, and by 8 to 12 weeks, coronary arteriopathy evolved. In contrast, graft arterial disease did not develop in IFN-γ-deficient recipients, despite a marked myocardial rejection, indicating that development of graft arterial disease, but not parenchymal rejection, requires IFN-γ.

In another model, hearts are heterotopically transplanted between DBA/2 (H-2d) donor mice to B10.D2 (H-2d) recipient mice, which share major histocompatibility antigens but differ in their minor antigens [178]. Significant survival of allografts (70%) was achieved without immunosuppression. Concentric intimal hyperplasia, comprising ∼40% of the graft arterial wall, together with interstitial and perivascular fibrosis were present. The endothelium appeared intact. The grafts that failed to survive, exhibited features of interstitial edema, haemorrhage, myocardial necrosis and severe mononuclear cell infiltration with diffuse fibrosis, indicative of acute rejection. There is no apparent correlation between the severity of rejection and the degree of vessel disease.

4 Hypercholesterolemia-induced atherosclerosis

Atherogenesis is a complex process in which the lumen of a blood vessel becomes narrowed by cellular and extracellular substances [179]. Lesions progress mainly through three stages. The first stage is the fatty streak lesion, which is characterized by the presence of lipid-filled macrophages (foam cells) in the subendothelial space. The second stage is the fibrous plaque, which consists of a central acellular area of lipid, derived from necrotic foam cells, covered by a fibrous cap containing smooth-muscle cells and collagen. The final stage is the complex lesion, in which plaque rupture triggers the clinical event of thrombus formation with deposition of fibrin and platelets, and in which media destruction may lead to aneurysm formation and fatal bleeding after rupture.

The wild-type laboratory mouse is highly resistant to the development of atherosclerotic plaques, presumably related to a different distribution of cholesterol among its lipoproteins [180]. Other mouse-specific differences include absence of the cholesteryl ester transfer protein (CETP) and of lipoprotein (a), and different editing of apolipoprotein (Apo)B-100 [180]. Initial studies revealed that significant strain-dependent differences determined the susceptibility to atherosclerosis [180]. The first available mouse models of atherosclerosis, including the C57BL/6 mice fed a cholesterol- and cholate-rich diet, suffered significant shortcomings. Indeed, even after prolonged feeding on an unphysiological diet, containing 10 times the cholesterol of a Western-type diet and the unnatural dietary constituent cholic acid (which regulates removal of cholesterol from the body), these mice only developed atypical and immature early fatty streak lesions with a restricted distribution, casting doubt on their overall relevance to human disease. Furthermore, as this diet caused a chronic inflammatory state, it might have induced perturbation of the interplay of immune cells and cytokines involved in atherogenesis. With the advent of transgenic mice, over- or under-expressing other candidate transgenes, and the further manipulation of gene expression via bone marrow transplantation or adenovirus-mediated gene transfer, significant progress has been achieved in unraveling the molecular mechanisms of this disease process (Fig. 5).

Fig. 5

Overview of the different atherosclerosis models in the mouse. The most widely used models (apoE- or LDLr-deficient mice) can be crossbred into other transgenic mouse strains with over- or under-expression of other atherosclerosis modifier genes.

4.1 Lipid metabolism

The most widely studied mouse model of atherosclerosis is a transgenic mouse strain, deficient in apolipoprotein E (ApoE) [181, 182]. This molecule is a ligand that mediates low-density lipoprotein (LDL) receptor clearance of chylomicrons, very-low-density lipoproteins (VLDLs) and other serum lipoproteins. Consequently, ApoE-deficient mice spontaneously develop hyperlipidemia (predominantly VLDL) and elevated cholesterol levels (∼500 mg/dl) on a low cholesterol/low fat diet. They develop lesions of all phases with morphological features closely resembling human lesions at vascular sites typically affected in human atherosclerosis, e.g. at the base of the aorta, the lesser curvature of the thoracic aorta, at several branch points of the carotid, intercostal, mesenteric, renal and iliac arteries, and in the proximal coronary, carotid, femoral, subclavian and brachiocephalic arteries [183, 184]. Adherent monocytes are present on the surface and migrate across the endothelium before the development of foam cell-rich fatty streaks. As the lesions progress, smooth-muscle cells form a fibrous cap rich in collagen and elastin fibers over the foam cell-rich areas, and cholesterol clefts, calcifications and necrotic areas appear within the core regions of the fibrous plaques. Lumen narrowing and occlusion with associated myocardial fibrosis (indicative of ischemia) are observed in the most advanced stages. Notably, this phenotype does not depend on complete absence of ApoE, as heterozygous ApoE-deficient mice also develop fibrous lesions on a high-fat/high-cholesterol diet [185, 186].

Cell surface LDL receptors (LDLr) play a fundamental role in regulating plasma cholesterol levels by mediating cellular uptake of LDL and intermediate-density lipoproteins (IDLs). Defects in the LDLr gene are the best documented genetic causes of premature atherosclerosis in humans. LDLr-deficient mice on a normal chow diet have only mild pathological lesions, with slightly elevated plasma cholesterol (IDL/LDL fraction), a phenomenon attributed to the presence of an alternative ApoB-II pathway for LDL clearance in the mouse [187, 188]. However, when fed an atherogenic diet, these mice develop significant fatty streak lesions containing a lipid-filled necrotic core. Fibrous lesions are, however, absent. Combined ApoE:LDLr-deficient mice suffer more severe hyperlipidemia and lesion formation [150].

Several additional transgenic strains have been generated that over- or under-express modifier genes, able to modulate the lipoprotein metabolism and the development of atherosclerotic lesions (Table 1). These include transgenic mice overexpressing ApoE Leiden [189, 190], ApoER142C [191, 192], ApoA-II [193], CETP [194], lecithin cholesterol acyltransferase (LCAT) [195], human apolipoprotein (a) [196], or human ApoB-100 together with the apolipoprotein (a), which results in the formation of lipoprotein (a) or Lp (a) [197]. Other modifiers have been evaluated by overexpression on a wild-type or on a ApoE- or LDLr-deficient background, including ApoA-I (ApoE-deficient background [198–200]), ApoA-IV (wild-type background [201]; apoE-deficient background [202]), or Apo C-III (wild type background [203]; LDLr-deficient background [204]). Additional transgenic mice, generated via gene targeting, lack ApoA-I [205], LCAT [206], ApoC-I [207], hepatic lipase [208], the macrophage type I and type-II class-A scavenger receptor [209], or Apobec-1 [210, 211]. Other targeted mice exclusively express ApoB48 or ApoB100 [212], or a human ApoE3 isoform instead of the normal ApoE allele [213].

Bone marrow transplantation studies offer the opportunity to further manipulate the lipid metabolism. Indeed, bone marrow from a wild-type donor has been shown to correct the hyperlipidemia and to prevent atherosclerosis in ApoE-deficient hosts [214–216]. In addition, local production of ApoE by expression of a transgene in macrophages or in the blood vessel wall diminishes atherosclerosis, independent of its cholesterol-lowering effect, revealing the importance of local ApoE production [217]. Use of recombinant adenoviruses expressing ApoE results in normalization of the lipid and lipoprotein profile with markedly decreased total cholesterol, VLDL, IDL and LDL, and increased HDL [218]. Adenoviral gene transfer of ApoE isoforms in ApoE-deficient hosts reveals a significantly impaired ability of ApoE2 to mediate clearance of remnant lipoproteins as compared to ApoE3 or ApoE4 [219]. Notably, use of second generation adenovirus expressing ApoA-I results in significantly higher levels of ApoA-I for a longer time in LDLr- than in ApoE-deficient mice [220].

4.2 Proteinases and extracellular matrix

Stable atherosclerotic lesions can develop insidiously for prolonged periods without medical threat. However, when plaques become unstable, they may rupture and trigger the onset of myocardial ischemia because of occluding thrombus formation [145]. Unfortunately, the currently available models of atherosclerosis in the mouse (and in other species) fail to progress to spontaneous plaque rupture. In the mouse, this failure may be attributable (at least in part) to the lower arterial wall stress, and/or to the absence of other risk factors (hypertension, smoking, diabetes etc). Nevertheless, the mouse may be a valuable model to study some of the mechanisms contributing to plaque instability, as evidenced by the observation that arterial calcification is under genetic control [221, 222]. Widespread calcified cartilaginous metaplasia occurs within spontaneous atherosclerotic lesions in ApoE-deficient mice [222]. Spontaneous calcification of all elastic and muscular arteries (but not the arterioles, capillaries or veins) was also observed in mice lacking the matrix-GLA protein (Mgp), a protein synthesized by vascular smooth-muscle cells [223]. Mgp-deficient mice develop till term but die within two months as a result of rupture of the thoracic and abdominal aorta. Noticeably, arterial calcification occurs in the absence of atherosclerotic plaque formation, indicating that lipid accumulation and calcification are regulated by independent genetic mechanisms. Another modifier of calcification is the macrophage colony-stimulating factor (M-CSF), as evidenced by the finding that M-CSF:ApoE-deficient mice develop significant arterial calcification, even in the absence of advanced plaques [224]. Urokinase-type plasminogen activator (u-PA) may also be implicated, as ApoE:u-PA-deficient mice develop plaques with an increased number of microcalcifications (unpublished observations).

Little information exists on the amount or nature of the extracellular matrix in atherosclerotic plaques in the various available transgenic mouse models. Mice deficient in ApoE and u-PA on a high-fat/high-cholesterol/cholate diet have similar plaque growth as ApoE-deficient mice, but an increased deposition of collagen-rich matrix at the ultrastructural level (unpublished observations). Accelerated plaque progression has been observed in mice with a combined deficiency of ApoE and plasminogen (Plg) [225]. Although it was anticipated that deficiency of Plg would result in an increased deposition of matrix, similar amounts of fibrin were present in ApoE- and in ApoE:Plg-deficient mice. The precise mechanism of the increased plaque growth remains thus unclear. A particular problem with this study is, however, that deficiency of Plg lowers the HDL levels by 75%. Whether this relates to the poor general health or altered immune/inflammatory status of these mice remains to be determined. As HDL is an important determinant for plaque growth in mice, the accelerated growth in the mutant mice may have resulted from an indirect effect on plasma lipoproteins rather than a direct local effect within the plaques.

Another life-threatening complication of atherosclerosis is the development of media destruction and aneurysm formation with resultant fatal bleeding. Although a major health threat in the elderly, its pathogenetic mechanisms remain poorly understood [226, 227]. Media destruction has been occasionally detected in advanced atherosclerotic lesions in LDLr- or ApoE-deficient mice on a high-fat/high-cholesterol diet [183]. The incidence and severity of these destructive lesions, and their progression to rupturing aneurysms appears to be increased by cholate supplements to the diet [151]. Indeed, media destruction frequently occurs in lesions beyond 15 weeks of age, progressing in ∼10% of the cases to aneurysmal dilatation and pseudoaneurysm formation [151](Fig. 6). Despite rupture of the entire aortic wall, bleeding is prevented by compensatory formation of a fibrous cap in the adventitia. Aneurysmal dilatation is more severe and frequent at the infra-renal level in the abdominal aorta than in the thoracic aorta, similar to the human pathology. Proteinases, produced by macrophages, play an important role in this process as loss of urokinase gene function in ApoE-deficient mice prevents media destruction and aneurysm formation, likely via reduced activation of matrix metalloproteinases [151].

Fig. 6

Atherosclerotic aneurysm formation in the mouse. (a–c) Histological sections through the aorta from an apoE-deficient mouse (fed a cholesterol-rich diet), revealing fragmentation of the elastic laminae (a), thinning of the media and dilatation of the aortic wall (b), rupture of the aneurysm (c). (d) Loss of medial α-actin immunoreactive smooth-muscle cells associated with media destruction.

4.3 Immune modulation

It is becoming increasingly evident that immune mechanisms influence atherosclerosis [228]. Human atherosclerotic lesions consistently contain macrophages and T lymphocytes, two cell types that interact with each other to promote cell-mediated immune responses. Macrophages may present antigen to the T lymphocyte within the vessel wall, whereas T-lymphocyte-derived cytokines may activate macrophages. In addition, antibody responses to heat-shock proteins or to modified lipids found in atheromas are characteristic of the disease. It has been proposed that T cells recognize peptides derived from oxidatively modified low-density lipoprotein (oxLDL), thereby promoting B-cell activation and production of anti-oxLDL antibodies.

Studies in C57BL/6 mice with antibody immunodepletion of CD4 T lymphocytes [229], and various genetic immunodeficiencies, including SCID mice, athymic nude mice and mice with class I or II MHC deficiencies [230], all developed typical fatty streak lesions when fed diets high in saturated fats, cholesterol and cholate. An increase in lesion area was observed in the class I MHC-deficient mice as compared to normal mice. However, the use of cholate and the absence of T-lymphocytes in the small lesions in the aortic root of C57BL/6 mice confound the interpretation of these studies with regard to the impact of the immune system on atherogenesis.

Recombinase activator gene (RAG)-2 deficiency results in a total deficiency of B and T lymphocytes. RAG-2:ApoE-deficient mice have no serum autoantibodies, no T lymphocytes, and a markedly reduced number of MHC class II-positive macrophages in their atherosclerotic lesions (T lymphocyte-derived IFN-γ and IL-4 increase the expression of MHC class II on macrophages) [231]. Despite these differences, atherosclerotic plaque size is comparable to that in immunocompetent littermates, indicating that the absence of autoantibodies and T lymphocytes does not influence the extent of aortic lesions in ApoE-deficient mice. Similar results have been reported in RAG-1:ApoE-deficient mice with immature, dysfunctional B and T lymphocytes [232].

The role of IFN-γ in atherosclerosis has remained controversial. On the one hand, it stimulates the expression of VCAM-1 on endothelial cells, MHC-II on macrophages and smooth-muscle cells and lipoprotein receptors in smooth-muscle cells, all potentially proatherogenic properties. On the other hand, it decreases lipoprotein receptor expression on macrophages, decreases collagen synthesis in smooth-muscle cells and blocks smooth muscle proliferation, all potentially antiatherogenic effects. Mice with a combined deficiency of IFN-γ and ApoE exhibit a substantial reduction in atherosclerotic lesion size and a 60% reduction in lipid accumulation, presumably resulting from an increase of atheroprotective phospholipid/ApoA-IV-rich particles [233]. In addition, the cellularity of the atherosclerotic lesions is significantly reduced, with a concomittant increase in extracellular collagen content. Whether these changes result from alterations in smooth-muscle-cell collagen synthesis, collagenase activity, or cell survival remains to be determined. Whatever the mechanisms, IFN-γ promotes atherosclerosis through both local effects in the vessel wall as well as via a systemic effect on plasma lipoproteins.

Macrophage colony-stimulating factor (M-CSF) regulates the differentiation, proliferation and survival of mononuclear phagocytes, functions as a chemotactic agent for monocytes, and influences the effector functions of mature monocytes and macrophages. Its expression is induced in the atherosclerotic wall. Crossbreeding of ApoE-deficient mice with a mouse line carrying a spontaneously mutated M-CSF gene (op) (resulting in fewer monocytes and tissue macrophages) yields mice with higher cholesterol levels than the ApoE-deficient line, yet with significantly smaller and more immature, but more calcified, atherosclerotic lesions in the aorta [224, 234]. Although these data indicate an important proatherogenic effect of macrophages, it remains unsolved whether the effects of the op mutation result from decreased circulation monocytes, reduced tissue macrophages, or diminished arterial M-CSF.

Tumor necrosis factor (TNF) is a monocyte/macrophage-derived cytokine, able to alter lipid metabolism by decreasing the activity of adipocyte-derived lipoprotein lipase and by increasing the production of hepatic VLDL in response to endotoxin, both being potentially proatherogenic effects. However, TNF induces nitric-oxide synthase production and inhibits lipoprotein lipase in the arterial wall, both potentially anti-atherogenic effects. C57BL/6 mice lacking the tumor necrosis factor receptor p55, develop aortic sinus fatty streak lesions, which are three-fold larger than in wild-type mice, despite comparable plasma lipid levels [235]. The accelerated atherosclerosis in p55 null mice appears attributable to increased expression of the scavenger receptor with resultant increased uptake and degradation of acetylated LDL.

4.4 Glucose metabolism

Individuals with diabetes are at increased risk for developing cardiovascular disease, but the pathogenetic mechanisms remain poorly understood. Genetically obese mice with differing severity of obesity, hyperglycemia, hyperinsulinemia, insulin resistance and islet hyperplasia and atrophy (fat, obese, tubby, diabetes, and lethal yellow strains) developed similar or reduced fatty streak lesion formation on a high-fat/high-cholesterol diet [236]. Notably, protection against accelerated atherosclerosis in these genetically obese mice appears to result from a concomittant increase in anti-atherogenic plasma HDL-C levels. This is the case, regardless whether these mice were congenic on a C57BL/6 or on the more diabetogenic C57BL/Ks background. The importance of HDL in determining lesion development in the mouse is further illustrated by chronic alcohol feeding of C57BL/6 mice, which markedly inhibits the development of fatty streak lesions, concomittant with a reduction in plasma HDL cholesterol [237].

More recently, a murine model of accelerated atheroslerosis in diabetic LDLr-deficient mice has been developed. LDLr-deficient mice, rendered hyperglycemic (glucose>300 mg/dl) by streptozocin treatment and fed a regular chow diet, have similar cholesterol and triglyceride levels as normal mice [238]. In hyperglycemic mice, accelerated formation of irreversible Advanced Glycation Endproducts (AGEs) occurs, with concomittant induction of their cellular receptor (RAGE) in the aortic wall. At 6 weeks of chow diet, lesion area is 4-fold increased in diabetic versus normal LDLr-deficient mice, a process that can be suppressed by treatment of these mice with recombinant soluble RAGE (a competitive antagonist of RAGE) [239].

4.5 Technical considerations

Several methods have been developed to quantitate lesion development in the murine aorta. The extent of atherosclerosis in murine models has usually been determined from cross-sections at a single predilection site (the aortic root), using modifications of a technique, originally described by Paigen [240]. This method uses sequential sections through the heart and aortic origin, and determines the lesion area in a defined number of sections, beginning from an anatomical reference point such as the aortic valve leaflets. An improved variation of this method using computer-assisted evaluation of 60 progressive sections covering the initial 1.2 mm of the aortic region has been reported [241]. An advantage of this model is that the lesion volume can be calculated from the lesion areas, providing the basis for three-dimensional reconstruction of the lesions. However, a disadvantage is the atypical site of atherosclerosis at the aortic root, influenced by the hemodynamic conditions which are different from those in the thoracic or abdominal aorta. This model is useful in evaluating differences in atherogenicity in those strains or under experimental conditions in which lesion formation is predominantly found in the aortic root.

Another morphometric method quantitates the extent of atherosclerotic lesions in the entire murine aorta. This method utilizes computer-assisted image analysis of color images of Sudan-stained flat (‘en face’) preparations of the aortic tree (from the heart to the iliac bifurcation), and determines the percentage of surface area affected by atherosclerosis [242, 243]. The method is rapid and relatively accurate with limited operator dependence. When compared to the cross-sectional analysis, this method provides additional information on the shape and the distribution of these lesions, and, when complemented with analysis in longitudinal sections [244], on lesion areas and volumes (at least through the thoracic and abdominal aorta). This method may be preferred for intervention studies on large numbers of animals, or in murine models that develop significant aortic atherosclerosis. Alternative methods to determine the extent of atherosclerosis in large groups of mice include the measurement of the unesterified and esterified cholesterol content in aortic tissue extracts [245], or the use of 125I-labeled antibodies specific for plaque-restricted epitopes (such as malonaldehyde-modified LDL). These methods do not, however, provide any direct information on the histological type, the predilections site, distribution, shape or the volume of the lesions.

5 Hemostasis

Hemostasis essentially depends on (i) the coagulation and fibrinolytic systems, which mediate formation and dissolution of plasma clots, respectively [246, 247]; (ii) platelets which are an essential component of thrombi and contribute to thrombus formation; and (iii) the structural integrity of the vessel wall. Defects or perturbations in any of these mechanisms may contribute to an increased risk for bleeding or thrombosis. Recent insights deduced from gene-targeting experiments suggest that hemostasis during the earliest stages of embryonic development may be less dependent on fibrin formation and platelet function (and more on vascular wall integrity) than anticipated, and that fibrin formation and platelet function become progressively more important in the control of hemostasis later during development and after birth [1].

5.1 Thrombosis

5.1.1 Arterial thrombosis

It is somewhat surprising that the mouse has not been more widely used as a model to test the possible therapeutic usefulness of antithrombotic agents. This may relate to the absence (at least until recently) of reliable models to induce arterial thrombosis in the mouse in a controllable and quantitative manner. In addition, the small size of the murine arteries limits the manipulation of isolated vessel segments, a requirement of many thrombotic models in larger species. Furthermore, the currently available experimental models of thrombosis in the mouse differ in their extent and location (venous or arterial) of thrombus formation.

A model of fatal thromboembolism in mice has been developed by intravenous injection of ADP, collagen or thrombin, alone or in conjunction with the potent vasoconstrictor epinephrine [248–250]. Thromboembolic death results from intravascular platelet aggregation, inducing thrombocytopenia with redistribution of platelets primarily in the lung vasculature (as evidenced by the use of 51Cr-labeled platelets) [251]. In addition, vasoconstriction secondary to the production of thromboxane A2 and prostaglandin F2α by the aggregating platelets and damaged endothelium may also contribute to the lethality. The resultant widespread systemic arterial thrombosis induces respiratory distress, hindlimb paralysis and death. Typically, the number of mice with systemic thrombosis and resultant death within a defined period, or their survival time are compared between control and drug-treated animals. This model is different from another pulmonary microembolism model, based on the intravenous administration of fixation-hardened red blood cells, in which platelets are apparently not involved [252].

Perivascularly applied electric injury (used to study neointima formation; see above) induces the formation of platelet-rich and fibrin-containing clots in the femoral or carotid arteries, but their induction is less well standardized [109]. A more controllable thrombotic model involves the topical application of a 1×2-mm strip of filter paper saturated with 10% ferric chloride solution to the adventitial surface of the surgically exposed carotid artery for 3 min [253]. This induces endothelial damage, possibly via formation of highly reactive oxidant species and triggering tissue factor-mediated initiation of blood coagulation (specific inhibitors of the factor VIIa-tissue factor complex block ferric chloride-induced thrombosis). A thrombus is formed, which predominantly contains activated platelets, fibrin strands and entrapped erythrocytes. The extent of thrombus formation is morphometrically quantitated on evenly spaced transverse sections, using computer-assisted planimetry. The kinetics of thrombus formation can be indirectly monitored by measuring the blood flow using a miniature doppler flow probe. An initial decline in flow is followed by cyclic variations in flow (resulting from intermittent embolization of platelet aggregates from the site of injury), evolving to complete loss of flow. This model allows controllable and reproducible injury as the concentration of ferric chloride, the size of the arterial segment being injured, and the duration of injury can be precisely controlled. In addition, complete occlusion of the carotid artery is well tolerated by mice due to blood flow via the contralateral artery.

Transillumination by a filtered xenon lamp (wavelength 540 nm) of a surgically exposed artery after systemic administration of a photochemical compound (Rose Bengal) damages the endothelium, initiating the formation of a platelet-rich clot [254, 255]. In order to monitor thrombosis in larger peripheral arteries, the artery is transilluminated by a special device containing acrylic optical fibers positioned underneath the artery [256]. Systemic injection of fluorescent fibrinogen improves the in situ visualization of the formation and dissolution of fibrin strands (V. Laux, personal communication). This model offers an opportunity to monitor the kinetics of clot formation and dissolution in a dynamic circulation. Argon-laser irradiation of arterioles and venules in a modified dorsal skin chamber model has been used in the rat to induce thrombosis [257]. As dorsal skin chambers have been miniaturized for the mouse, such a model could be adapted for the mouse as well.

5.1.2 Venous or capillary thrombosis

A quantitative model of venous thrombosis has been developed by pinching the femoral vein to introduce a standardized thrombogenic injury, using the tips of a forceps with flat circular opposing surfaces (0.1 mm diameter; pressure of 1500 g/mm2) [256]. The vessel is transilluminated using acrylic optical fibers and clot formation (and dissolution) in the transilluminated vein can be recorded by vital videomicroscopy and computer-assisted image analysis. This model offers the particular advantage of analyzing the kinetics of the formation and dissolution of the thrombus, as well as of measuring the thrombus area in a dynamic circulatory system.

Intraperitoneal injection of endotoxin results in systemic activation of inflammatory cells, causing them to synthesize and release several endogenous mediators that contribute to the pathophysiological process of septic shock [258]. A procoagulant state develops after exposure to endotoxin, largely due to the conversion of the endothelium from a thromboresistant to a thrombogenic surface via increased tissue factor and plasminogen activator inhibitor-1 expression, concomittant with reduced expression of urokinase-type plasminogen activator. Within 3 h after intraperitoneal injection of endotoxin in mice, fibrin formation is detected in glomerular and peritubular capillaries in the kidney [258]. However, the presence of fibrin is transient, decreasing at 8 h and disappearing by 24 h. The thrombotic response can be visualized by immunostaining for fibrin. Injection of the proinflammatory endotoxin (10 to 50 μg) in the footpad of mice induces local inflammation and venous thrombosis, which can be quantified by counting the number of occluded vessels and their degree of occlusion [259, 260].

Clinical conditions associated with hypoxia or hyperoxia can lead to prothrombotic diatheses. Hyperoxic lung injury is characterized by a prominent intra-alveolar deposition of fibrin and cell debris, the classical feature of hyaline membrane disease, in part because of increased expression of tissue factor (initiating blood coagulation) and PAI-1 [261]. Hypoxic stress has been implicated in the recruitment of mononuclear phagocytes and the derangement of endothelial and monocyte anticoagulant properties (e.g. the induction of tissue factor and the reduction of plasminogen activator expression), all resulting in thrombus formation. Perturbations in oxygen levels (hyperoxia or hypoxia) are produced by placing the mice in isolated chambers with increased (75 to 100%) or reduced (6%) oxygen for defined periods. Fibrin deposition and platelet accumulation in the lung tissue are evaluated by immunostaining and ultrastructural analysis (revealing the 22.5-nm strand periodicity of fibrin), by immunoblots revealing fibrin-gamma chain dimers, by measuring the amount of fibrin activation products in the plasma or urine, and by deposition of 125I-fibrin and 111In-labeled platelets [261–263].

Another microthrombotic model relies on local cerebral hyperthermia, regionally applied by heating the brain surface with irrigating artificial cerebrospinal fluid in a cranial window [264]. This induces platelet aggregation on damaged endothelial cells, vasoconstriction and the formation of arterial thrombi. Dehydration increases the susceptibility to thrombosis in the pial microcirculation. This model mimics many features (e.g. the microthrombi, low platelet counts and areas of local necrosis) of the thrombotic complications found in many organs from heat stroke victims. Thrombosis can also be photochemically induced in the microcirculation in the exteriorized peritoneum or cremaster muscle, and monitored by vital videomicroscopy upon transillumination [254].

5.1.3 Transgenic models with increased thrombogenicity

Several spontaneously arisen or genetically manipulated mutations have increased the susceptibility of mice to thrombosis. Among them are mice with sickle cell disease (arterial thrombosis) [265]or Lupus syndrome, spontaneously arisen [266, 267]or acquired through passive administration of antiphospholipid antibodies (arterial and venous thrombosis; myocardial infarction). In addition, transgenic mice have been generated with targeted deficiencies of the plasminogen system (e.g. of the tissue-type and urokinase-type plasminogen activators [259], or of plasminogen [268]) or of the coagulation system (e.g. of tissue factor pathway inhibitor [269], or protein C; unpublished observations), which induce arterial or venous thrombosis. Mice with a targeted mutation of the coagulation factor V (factor V Leiden [270]) or of the anticoagulant thrombomodulin are at increased risk to develop arterial or microvascular thrombosis, respectively. Mice lacking PAI-1 (the inhibitor of t-PA and u-PA) are relatively resistant to thrombosis induced by local endotoxin injection (footpad model) [260], by hyperoxia (lung model) [261], by hypoxia (lung model), or by ferric chloride application [253].

5.2 Thrombolysis

Thrombolysis can be evaluated by quantitating the degradation of an intravenously injected 125I-fibrin plasma clot that becomes embolized in the pulmonary vasculature [259, 260, 268]. By supplementing varying amounts of platelets to the plasma clots, degradation of pulmonary plasma clots that are free, poor or enriched in platelets can be analyzed. Platelet-free pulmonary plasma clots are spontaneously lysed within 24 h in wild-type mice. Alternatively, 99Tcm-labeled fibrin-specific antibodies can be included in the plasma clots [271]. Because technetium is a strong gamma emitter, the kinetics of clot lysis in individual mice can be monitored over time by imaging mice with a gamma camera. Clot lysis is markedly impaired in mice lacking tissue-type plasminogen activator [259]or plasminogen [268], or in mice overexpressing lipoprotein (a) [271], or PAI-1 (following adenoviral gene transfer; unpublished observations). In contrast, lysis is markedly increased following adenoviral-mediated t-PA gene transfer [272]. In contrast to the pulmonary clot lysis assay (which is formed in vitro and subsequently injected into the mice), dissolution of thrombi that are formed in situ can be monitored in the venous pinch model (described above) [256].

5.3 Bleeding

Acute bleeding in neonatal or adult mice is induced by transsection of the tail 1 to 3 mm proximal to its tip, respectively, using a sharp scalpel blade. The tail is immersed in a recipient containing saline preheated at 37°C to prevent vasospasms [273, 274]. The mouse is maintained in a horizontal position in a restrainer with the tip of the tail 4 to 5 cm below the body plane. Bleeding is quantitated by determining the time required until cessation of the blood stream (varying between 2 to 4 min in normal mice). To compensate for differences in the rate of blood flow, the amount of blood collected in the prewarmed saline recipients (or absorbed on filter devices) is also determined by spectrophotometric analysis of extractable hemoglobin.

Delayed rebleeding, as typically occurs in patients with deficiency of PAI-1 or α2-antiplasmin, can be evaluated after transsection of the tailtip or nail cuticle, or by surgical interventions (such as for example by removal of the caecum) [260]. Chronic bleeding can be indirectly evaluated by determining the hemoglobin content, the number of red blood cells and reticulocytes in the peripheral blood, as well as by visualization of hemosiderin (reflecting erythrocytes phagocytosed by macrophages) at the site of bleeding.

5.3.1 Transgenic models with increased bleeding

Several transgenic mouse strains bleed spontaneously during embryogenesis. Whereas the precise mechanisms of bleeding in some strains remains uncertain (mice with targeted deficiency of factor V [275], or of tissue factor pathway inhibitor [269]), other transgenic mouse strains bleed because of defects in the formation of the vessel wall. This may be due to impaired endothelial assembly (mice with targeted deficiency of TIE-1 [21], TGF-ß [111], VCAM-1 [120], fibronectin [112], or α4 integrin [121]), to abnormal pericyte recruitment (mice with targeted deficiency of tissue factor [24], TIE-2 [114], angiopoietin-1 [23], PDGF-B [91]), or to abnormal smooth muscle function (deficiency of LKLF [124]).

The birth trauma is a common cause of perinatal bleeding, preferentially occurring in the abdomen and the brain. Bleeding resulting from lack of proper fibrin clot formation occurs in mice with deficiency of factor VII [25], fibrinogen [276], factor V [275], or from absence of platelets in NF-E2 knockout mice [277]. Mice deficient in tissue factor pathway inhibitor [269], or protein C (unpublished observations) bleed as a result of the depletion of coagulation factors secondary to the initiation of uncontrolled diffuse intravascular coagulation. Other mouse strains suffer an increased bleeding tendency later in life, but only after mechanical trauma. Some of those suffer defects in fibrin clot formation (mice deficient in factor VIII [278], factor IX [279, 280], and mutant mouse strain RIIIS/J [281]), or exhibit abnormal platelet function (mutant mouse strains brachymorphic [282], subtle gray [283], ruby-eye [284], gunmetal [285]). Other mutant mice display increased postnatal bleeding tendency, presumably because of vascular defects (brachymorphic, hemimelic extra toes, ulnaless [282]), whereas other transgenic mouse strains die because of postnatal hemorrhaging due to abnormal vascular fragility (deficiency of fibrillin-1 [127], collagen type I [125]). Bleeding in adult mice can be induced by administration of anti-factor VIII/von Willebrand Factor inhibitor plasma [286].

5.3.2 Gene transfer

Hemophilia B mice [279, 280]are a good model to test possible (gene)-transfer strategies. Adenovirus-mediated transfer of human factor IX resulted in normalization of the aPTT, and prevented bleeding. Noticeably, a marked antibody response occurred in hemophilic CD-1 but not in C57BL/6 mice, suggesting that the former may be suited to study the development of inhibitors, whereas the latter might be better suited to study gene transfer [280]. Plasma factor IX levels were also restored using adenovirus-mediated gene transfer in another model of hemophilia B [279]. It will be intriguing to test recombinant adeno-associated virus-mediated gene transfer of human factor IX in these mice [287].

6 Conclusions

The availability of different inbred genetic backgrounds and the possibility to specifically manipulate the expression of the mouse genome (via transgenesis, gene transfer, or cell transplantation) offer a unique opportunity to unravel the molecular basis of complex, multifactorial biological processes and diseases such as angiogenesis, hemostasis, atherosclerosis, and restenosis. However, the mouse differs from the human in many ways, necessitating careful and in depth analysis of these mouse models. Despite species-related or model-dependent differences, valuable insights into the molecular mechanisms can be deduced, but extrapolation from these mouse findings to clinical relevance in humans should be made cautiously. Future challenges lie in the development and better characterization of such mouse models, as they will yield insights that cannot be derived from other available animal models.


View Abstract