Estradiol et progéniteurs endothéliaux

Plusieurs études récentes ont permis de démontrer un effet direct de l’E2 sur la physiologie des progéniteurs endothéliaux (EPC). A partir de culture d’EPC dérivant de cellules extraites du sang, l’équipe de Asahara démontre que l’E2 augmente la migration, la prolifération et diminue l’apoptose de ces cellules. De plus, des expériences in vivo, démontrent une augmentation de l’incorporation d’EPC différenciés en cellules endothéliales matures au niveau de l’utérus (Masuda, Kalka et al. 2007)). Ces résultats confirment les résultats de Losordo et al (Hamada, Kim et al. 2006). Cette équipe a mis en évidence sur des EPC issus de souris ERα KO, que l’E2 perdait toute capacité à induire une augmentation de la migration, de la formation de structures tubulaires in vitro et de l’expression du VEGF. De plus, l’utilisation d'un modèle d’ischémie cardiaque, sur des souris sauvages greffées avec de la moelle osseuse de souris déficientes en ERα ou ER, démontre chez ces dernières une perte importante de la localisation des EPC au niveau du site ischémié. Ce phénomène est plus important chez des souris déficientes en ERα d’origine médullaire. Ces données démontrent donc que l’E2 a un effet important sur l’activité des EPC via ERα (Hamada, Kim et al. 2006).

De plus, deux études, ont confirmé un effet bénéfique de l'E2 sur la réendothélialisation et sur la diminution de l’épaississement intimal. Par l’utilisation de coupes histologiques, ces auteurs démontrent que l’E2 augmente l’intégration des EPC au sein de la zone traumatisée (Iwakura, Luedemann et al. 2003; Strehlow, Werner et al. 2003). Sous l’influence de l'E2, le nombre d’EPC circulants augmente précocement chez la souris (pic sanguin vers le troisième jour). De plus, le nombre d’EPC adhérant dans la zone traumatisée passe de 2 à 6 cellules par section de carotide. Cet effet disparaît chez les souris déficientes en NO synthase endothéliale.

COMPLEMENTARITY BETWEEN MOUSE AND LARGER ANIMAL MODELS

CE. Toutain, E. Grunenwald, F. Lenfant, J-F. Arnal

Institut National de la Santé et de la Recherche Médicale (INSERM), U858, Institut de Médecine Moléculaire de Rangueil, IFR31 Toulouse, France

I Lessons from medium and large animal studies... 4 II Models to study endothelial regeneration... 6

II.1 Models of vascular injury... 6 II.1.1 Endovascular models... 6 II.1.2 Perivascular models... 7 II.2 Quantitative and qualitative analysis of reendothelialization... 8

III Acceleration of endothelial repair ... 9

III.1 Pharmacological approach... 9 III.1.1 Statins ... 9 III.1.2 Estrogens and SERMs ... 10 III.1.3 ACE inhibition ... 11 III.2 Local cellular actors... 11 III.2.1 Endothelial cells ... 11 III.2.2 Smooth muscle cells ... 11 III.3 Circulating cellular actors... 12 III.3.1 Endothelial progenitor cells... 12 III.3.2 Platelets ... 14 III.4 Molecular actors: growth factors and local messengers... 15 III.4.1 FGF2... 15 III.4.2 VEGF ... 15 III.4.3 HGF ... 16 III.4.4 Nitric Oxide (NO)... 16

Endothelial healing

ABSTRACT

Drug-eluting stents (DES) have reduced rates of restenosis in patients with obstructive artery disease, compared to bare-metal stents. However, DES implantation may lead to vascular injury, mainly deendothelialization and mechanical damage to vascular layers. Delayed endothelial healing may cause deleterious events such as neointimal thickening and late thrombosis. Drugs released from DES aimed at preventing vascular smooth muscle cell proliferation and migration and thereby restenosis, may also impair endothelial healing and thus, favour a prothrombogenic environment.

The present review will focus on our understanding of macrovascular endothelial healing. To this aim, we will summarize artery injury models available to study reendothelialization and our learning from mouse studies. Finally, we will discuss the clinical implications of reendothelialization as a possible treatment for patients with obstructive artery disease undergoing mechanical revascularization.

INTRODUCTION

Treatment of obstructive coronary or peripheral artery diseases is often associated with stent implantation. Many patients undergoing percutaneous transluminal angioplasty with bare metal stents (BMS) developed exaggerated neointimal hyperplasia, namely in-stent restenosis. Drug eluting stents (DES), designed to release antiproliferative drugs, have emerged as a potential solution to limit restenosis (Sousa, Serruys et al. 2003). Although primarily aimed at preventing vascular smooth muscle cell (SMC) migration and proliferation (and thereby restenosis), DES also impair endothelial cells proliferation, leading to incomplete reendothelialization. (Finn, Nakazawa et al. 2007). This delayed and altered arterial healing has been associated with an increased risk of late stent thrombosis, a rare but life-threatening complication (Finn, Joner et al. 2007). Several factors are also associated with an increased risk of stent thrombosis, including the procedure itself (stent malapposition and/or underexpansion, number of implanted stents, stent length, persistent slow coronary blood flow and dissections), the patient and lesion characteristics, stent design and premature cessation of antiplatelet therapy. Furthermore, DES may exert a direct prothrombogenic effect, as drugs released from DES (sirolimus, paclitaxel) may increase tissue factor expression and inhibit endothelial regeneration (Luscher, Steffel et al. 2007).

Favoring integrity and regeneration of the endothelial cell monolayer appears to be of crucial importance, as endothelium plays a pivotal role in the control of permeability, trafficking of inflammatory cells, coagulation, vessel tone and, importantly, vascular smooth muscle migration and proliferation. These aspects are crucial in the problematic of restenosis and understanding the cellular and molecular mechanisms of reendothelialization will help to optimize treatments based on percutaneous angioplasty followed by stenting.

Endothelial healing after injury has been extensively studied in animal models. More than 20 yearsago, a series of pioneer studies unravelled several cellularmechanisms involved in endothelial regrowth after injury, inrats and rabbits.

The objective of the present review is to summarize the recent mechanistic insights provided by mouse studies. We will first summarize information brought by larger animal models. We will then present artery injury models available to study reendothelialization with special emphasis on mouse studies contributions, as this species offers a large panel of genetic manipulations and is thereby a model of choice to gain insight intothe cellular and molecular mechanisms. Finally, we will discuss the clinical implications and means to favour reendothelialization as a possible treatment for patients with obstructive artery disease undergoing mechanical revascularization.

Endothelial healing

I LESSONS FROM MEDIUM AND LARGE ANIMAL STUDIES

Before the advent of transgenic mice, larger animal models, mainly rat models, allowed to describe precisely the morphological features of arterial healing. Mechanical injury, mainly inflated balloon catheter, has been the most extensively used technique to abrade arterial endothelium. In 1975, Schwartz and co-workers developed a mechanical denudation of the rat aortic intima using an embolectomy catheter, via the common carotid artery. The balloon was inflated with controlled pressure and withdrawn from the distal (abdominal) aorta to the origin of the left carotid artery. This process caused endothelial denudation, associated to variable medial damage and produced both an endothelial and a smooth muscle cell proliferative response (Schwartz, Stemerman et al. 1975). As reendothelialization occurred from the numerous intercostals arteries, the common carotid artery, devoid of collateral vessels, was preferentially chosen to study reendothelialization, occurring from each end of the carotid (Clowes, Reidy et al. 1983). The same technique has been applied to study endothelial regrowth in nonhuman primates, swine, dogs and rabbits (Schwartz, Reidy et al. 1995). To determine the influence of the size of wound and the resulting velocity of endothelial regeneration, Reidy developed a model with a nylon wire to inflict a small defined injury on the endothelium (row of three to five cells), with various direction of the injury, like circumferential or longitudinal injury (Reidy and Schwartz 1981).

A few minutes after vascular injury, the denuded intimal area was covered by a platelet monolayer. Platelets were found at site of injury as long as endothelium was absent (Schwartz, Stemerman et al. 1975).

In the rat aorta, endothelial regeneration started from the leading edges of the denuded area and from the ostia of collateral arteries. Endothelial cells migration begins 8 to 16 hours after injury, prior to the onset of cell proliferation (using 3H-thymidineincorporation), by 20 hours. The regenerative process was not limited to the cells at the front of injury, but involved the commitment of numerous layers of endothelial cells behind this front line (60 to 100 rows of endothelial cells) thereby constituting a large regenerative zone recruited to migrate and proliferate (Schwartz, Haudenschild et al. 1978). Similar results were obtained in the rat common carotid artery (Clowes, Collazzo et al. 1978).

by blood flow. Reendothelialization in the axial direction (i.e. in blood flow direction) is six times faster than around the aortic circumference (Schwartz, Haudenschild et al. 1978; Haudenschild and Schwartz 1979). After a limited injury along the longitudinal axis of the vessel, the wound healed in 48 hours by replication from the adjacent cells. In contrast, after a circumferential injury, reendothelialization is much faster and is completed in only 8 hours by cell spreading, without proliferation, in the axial direction (Reidy and Schwartz 1981). Thus, reendothelialization of small denuded areas can be achieved rapidly and entirely (Reidy and Schwartz 1981; Lindner, Reidy et al. 1989). In contrast, extensive balloon denudation in the rabbit aorta or the rat common carotid artery resulted in an incomplete endothelial healing (Reidy, Standaert et al. 1982; Reidy, Clowes et al. 1983). The rate of endothelial regrowth is also related to animal models and experimental conditions. In a balloon catheter injury of the entire common carotid artery, the endothelial regrowth stopped at 6 weeks in the rat (10.0 ±2.6 mm of final reendothelialization) but only at 2 weeks in the rabbit (3.0 ±0.9 mm of reendothelialization). However, this growth arrest was not due to cell senescence as shown by persistence of endothelial cell capacity of replication after a second injury 12 weeks later, but rather by presence of smooth muscle cells on the luminal surface (Reidy, Clowes et al. 1983). Indeed, smooth muscle cells migrate, proliferate and cover the intima surface persistently devoid of endothelium, leading to intimal thickening. In carotids lacking endothelium for a long time, smooth muscle cells proliferation reached a maximum at 96 hours in the intima (73%) and persisted in regions lacking endothelium as late as 12 weeks after injury (Clowes, Reidy et al. 1983). Moreover, smooth muscle cells, when forming a pseudoendothelium, inhibit endothelium regrowth (Reidy, Clowes et al. 1983; Reidy 1988).

Despite endothelial healing, morphology of endothelial cells became abnormal with altered shape, size and orientation (Clowes, Collazzo et al. 1978; Haudenschild and Schwartz 1979; Reidy, Clowes et al. 1983), associated to an impaired endothelium-dependant relaxation of balloon-injured vessels 2 and at 4 weeks after injury (Weidinger, McLenachan et al. 1990). The regenerating endothelium showed a marked decreased in density of gap junctions and loss of tight junctions (Spagnoli, Pietra et al. 1982).

Thus, rat, rabbit or larger animal models previously allowed investigating the action of drugs as well as biological molecules through their pharmacological administration or their transgenic overexpression. Although the blockade of endogenous molecules can be possible when appropriate pharmacological inhibitors are available, only mouse species offers the

Endothelial healing

possibility of temporo-spatial gene inactivations. Transgenic mice with targeted gene modification (deletion or overexpression) in the whole organism or in a specific cell type, provide tools to further unravel cellular and molecular mechanisms of vascular regeneration. Moreover, the mouse has a small size and a short gestation period (19-20 days), allowing a cost-effective maintenance. A large panel of mouse strains is already available, offering thereby the possibility to explore in vivo the functional role of genes of interest.

II MODELS TO STUDY ENDOTHELIAL REGENERATION

II.1 MODELS OF VASCULAR INJURY

II.1.1 Endovascular models

Dry air was initially applied on a segment of the rat common carotid artery to denude the endothelium through desiccation, without causing any significant damage to the media (Fishman, Ryan et al. 1975; Schwartz, Stemerman et al. 1975; Clowes, Collazzo et al. 1978). This technique, using a brief drying with a gentle stream of compressed air along the lumen of the vessel, also allowed to achieve both endothelial and medial injury using an endovascular approach that results in a progressive intimal thickening over time (Simon, Dhen et al. 2000; Tan, Jiang et al. 2006)

As described above, mechanical injury has been the most extensively used technique to obtain deendothelialization in rats as well as in larger animals. An adaptation of this mechanical injury model to the mouse was proposed in 1993 (Lindner, Fingerle et al. 1993). A curved flexible wire (0.35 mm diameter) was introduced into the common carotid artery via an incision between two ligatures on the external carotid artery. The wire was passed three times along the vessel, under rotation and then removed. This procedure produced a complete endothelial denudation with little damage to the media. Then, the external carotid artery was tied off with a ligature. Complete reendothelialization (approximately 9 mm) was achieved within 3 weeks. This model was adapted by the group of Mendelsohn with minor modifications, to study estrogen action on the media after injury, in ovariectomized mice (Sullivan, Karas et al. 1995; Iafrati, Karas et al. 1997; Karas, Hodgin et al. 1999). It mainly results in a significant thickening of both intimal and medial area. Roque also adapted this model to the femoral artery (Roque, Fallon et al. 2000). However, transluminal injury models

are laborious, due to the very small size of the mouse arteries (less than 500µm, rendering the introduction of any intravascular device difficult), time-consuming and the frequency of complications (haemorrhage and thrombosis) largely depends on the experience of the surgeon. To circumvent this difficulty, perivascular models, quicker and easier to perform, were proposed.

II.1.2 Perivascular models

Perivascular injury models have been proposed to simplify surgical procedures. Matsuno and al. developed a model of photochemically induced thrombosis in the rat femoral artery (Matsuno, Uematsu et al. 1991; Matsuno, Takei et al. 2004). This technique was initially described by Watson et al. to induce a brain infarction (Watson, Dietrich et al. 1985; Watson, Prado et al. 1986). The vessel was irradiated by a green light (540 nm) after Rose Bengal injection. Reendothelialization of the common carotid artery was studied using this model, both in mouse and hamster (Matsuno, Ishisaki et al. 2003; Matsuno, Takei et al. 2004)

Carmeliet proposed a model of perivascular electric injury, destroying both endothelial and smooth muscle cells through an heat injury (Carmeliet, Moos et al. 1997). Femoral arteries were injured by electric current, delivered by a bipolar microcoagulator through the tips of a microforceps (200 µm wide), every millimetre during 2 seconds, with a current pulse of 160 µA. Complete reendothelialization was achieved within 2 weeks. This injury model allowed observing both reendothelialization and neointima formation.

We proposed an adaptation of this electric injury model to the common carotid artery, in order to deendothelialize a longer segment of artery (Brouchet, Krust et al. 2001). To standardize the heat delivered to the vessel wall, we proposed to apply an electric injury using: (i) a forceps with larger tips (1 mm) over a total length of 4 mm, with the help of a size marker placed parallel to the long axis of the carotid and (ii) a generator microprocessor allowing to deliver electric energy within a low range of resistance and able to disrupt the electric current when the resistance reaches a certain level (2W applied for 2 seconds), because the resistance increases as a consequence of temperature increase. Under these optimized conditions, the risk of desiccation and coagulation of the arterial wall leading to thrombosis was minimized (less than 10%).

Recently, we compared the kinetics of reendothelialization of mouse carotid artery after endovascular (where smooth muscle cells are preserved) and perivascular injury (where

Endothelial healing

they are destroyed). We found that both basal and estradiol-stimulated reendothelialization were similar in both models. Rather unexpectedly, smooth muscle cells destroyed by the electric injury do not regenerate at all during the first week, allowing to precisely define the injury limit (Filipe, Lam Shang Leen et al. 2008).

This perivascular model offers various major advantages, such as absence of blood flow modification in the carotid artery, as no external carotid artery ligation is required, rapidity and reliability (thrombosis rate < 10%, vs. 30% or more in the endovascular model). Moreover, perivascular electric injury can be easily transferred from one investigator to another and requires less operator skills than the endovascular injury. Thereby, perivascular electric injury represents a valid and reproducible model to study endothelial injury and healing.

II.2 QUALITATIVE AND QUANTITATIVE ANALYSIS OF REENDOTHELIALIZATION

Endothelial cell monolayer can be studied on transversal sections using classical staining, immunohistochemistry or electron microscopy. However, transversal sections of endothelium are limited for the following reasons. (i) As endothelium is a very thin and fragile cell monolayer, it can be easily lost during microtome section leading to conclude incorrectly to an absence of endothelium. (ii) It gives access to a limited number of endothelial cells, without any indication on cell shape, that is highly influenced by and indicative of the level of shear stress. (iii) There is no way to define the limit of the injury and thereby the endothelial zone participating to reendothelialization.

In contrast, “en face” examination of the endothelium allows avoiding most of the previous limitations. In 1953, Lautsch described ”en face” vessel examination by incident light microscopy. This procedure consisted of longitudinally opening, spreading and mounting a segment of vessel on a glass slide, after staining the luminal surface, providing a direct access to cell size, shape and orientation (Lautsch, Mc et al. 1953). Haematoxylin is usually used to stain endothelial cells (Schwartz, Haudenschild et al. 1978), whereas the extracellular matrix of the deendothelialized area is stained by Evans Blue (Haudenschild and Schwartz 1979). Scanning electron microscopy allowed precisely defining the morphology of both endothelial cells and platelets covering the deendothelialized area.

Evans blue dye infusion by intravenous or intracardiac route, allows staining the denuded areas. After transparency scanning and numerization, Evans Blue stained and total

carotid artery areas are quantified by planimetry with an image analyzer. Then, the ratio of the area stained in blue on the total carotid artery area is calculated, in order to minimize the area changes due to elasticity and flattening of the vessel. To limit these area changes, two options were proposed : a complete deendothelialization of the whole common carotid artery, or a limited deendothelialization to a controlled length (Simon, Dhen et al. 2000; Brouchet, Krust et al. 2001; Goukassian, Kishore et al. 2003; Iwakura, Luedemann et al. 2003).

“En face” confocal microscopy (EFCM) coupled to immunohistochemistry applied to the electric injury model allowed recently to solve most of the previously stated limitations (Yeh, Rothery et al. 1998; Yeh, Lai et al. 2000; Filipe, Lam Shang Leen et al. 2008). Furthermore, serial sections can be reconstructed to obtain 3-D images using image analysis software. This approach gives access to a large number of cells of each layer (intima, media or adventitia) through reconstitution of the whole carotid by association of images. At variance with the endovascular model, the electric model allow to precisely define the injury limit using EFCM, as smooth muscle cells proliferation does not occur during the first week after injury (Filipe, Lam Shang Leen et al. 2008). This unprecedented access to both early and late reendothelialization allows precise studying of the cellular and molecular actors of this healing process.

III ACCELERATION OF ENDOTHELIAL REPAIR

III.1 PHARMACOLOGICAL APPROACH

The absence of endothelium after vascular injury may lead to deleterious events, mainly neointimal thickening, leading to restenosis and late stent thrombosis. Three classes of pharmacological agents are principally recognized to accelerate endothelial regeneration.

III.1.1 Statins

3-hydroxy-3-methylglutaryl (HMG)-CoA reductase inhibitors (statins) were first developed to lower cholesterol levels in hyperlipidemic patients, at risk for coronary atherosclerotic disease. However, statins appear to be potent drugs with pleiotropic effects, at least in part independent of cholesterol reduction. Indeed, statins exert direct effects on endothelium, as improvement of endothelium dysfunction or reduction of vascular inflammation (Walter, Dimmeler et al. 2004; Ii and Losordo 2007). Moreover, statins

Endothelial healing

accelerate reendothelialization and reduce neointimal hyperplasia in rats (Walter, Circulation, 2002), as well as after stent implantation (van der Harst, Groenewegen et al. 2008). Statins can enhance endothelial cells migration and proliferation in hamster carotids, mainly explained by VEGF secretion (Matsuno, 2004). Statins have been shown to mobilize bone marrow-derived endothelial progenitor cells (EPC) as well as EPC incorporation into rat injured carotid artery and to upregulate adhesion molecules expression (Werner, Priller et al. 2002)

III.1.2 Estrogens and SERMs

Beside its role in reproduction, estradiol (E2), the main estrogenic hormone, exerts several beneficial effects at the level of the endothelium: it promotes endothelial NO production, prevents VCAM-1 expression and endothelial apoptosis (refs in (Mendelsohn and Karas 2005; Arnal, Scarabin et al. 2007)). In addition, E2 stimulates endothelial regrowth after endothelial denudation in rats (Krasinski, Spyridopoulos et al. 1997) and in mice (Brouchet, Krust et al. 2001; Iwakura, Luedemann et al. 2003; Strehlow, Werner et al. 2003). Interestingly, inhibition of neointimal proliferation and acceleration of reendothelialization with local administration of E2 following balloon angioplasty in a pig model has been recently reported (Chandrasekar and Tanguay 2000; New, Moses et al. 2002). The accelerative effect of E2 on reendothelialization is mediated by ERα (Brouchet, Krust et al. 2001) and endothelial NO synthase (Iwakura, Luedemann et al. 2003; Billon, Lehoux et al.