Endothelial cells, fibroblasts and vasculitis

Abstract

One of the most important questions in vasculitis research is not why inflammation of blood vessels occurs but why it persists, often in a site-specific manner. In this review we illustrate how stromal cells, such as fibroblasts and pericytes, might play an important role in regulating the site at which vasculitis occurs. Smooth muscle cells and fibroblasts directly influence the behaviour of overlying vascular cells, amplifying the response of the endothelium to proinflammatory agents such as TNF-α and allowing enhanced and inappropriate leucocyte recruitment. An abnormal local vascular stromal environment can therefore influence local endothelial function and drive the persistence of local vascular inflammation. However, such local vascular inflammation can have distant effects on the systemic vascular system, leading to widespread endothelial cell dysfunction. Vascular endothelial dysfunction is common in a range of immune-mediated inflammatory diseases, is seen in multiple vascular beds, and is reversible following the induction of disease remission. The mechanisms that drive such systemic vascular endothelial dysfunction are unclear but factors such as TNF-α and CRP may play a role. Persistence of such widespread endothelial dysfunction in systemic vasculitis appears to have long-term consequences, leading to the acceleration of atherosclerosis and premature ischaemic heart disease. It may also underlie the accelerated atherosclerosis seen in other immune-mediated rheumatic diseases, such as rheumatoid arthritis.

Cells of the vasculature play a central role in physiology and pathology. Endothelial cells control platelet adhesion, maintain a balance of fibrinolytic and prothrombotic activity, regulate vascular tone and play a critical role in regulating the recruitment of leucocytes into inflammatory sites [1]. Regulation of vascular tone is dependent on the production of vasodilators, such as nitric oxide (NO) and prostacyclin, and vasoconstrictors, such as endothelin. While cells of the vasculature (such as endothelial cells and platelets) have been proposed to be the major producers of vasoconstrictors and vasodilators, it is now becoming clear that stromal cells such as fibroblasts and smooth muscle cells can also produce biologically active proteins that act in a paracrine manner to regulate endothelial cell function [2, 3].

In addition to regulating vascular tone, endothelial cells act as the gatekeepers for tissues, regulating the quality and quantity of leucocytes entering organs during both basal immune surveillance and inflammation [4]. In order to do this they respond to chemokines and other proinflammatory agents that modify the expression of adhesion receptors on the endothelium, leading to the ordered capture and transendothelial migration of leucocytes from flowing blood [5]. However the factors regulating the sites at which transendothelial migration occurs remain unclear [6, 7].

The gene expression profiles and functional properties of endothelial cells vary between vascular beds [8, 9]. For example, the gene expression profiles of arterial, venular and lymphatic endothelial cells are very different, reflecting their different physiological roles, secretory potential and permeability [8]. This variation in function extends to their ability to support transendothelial migration. Whereas endothelial cells in the high endothelial venules of primary and secondary lymphoid organs are particularly adapted to supporting the migration of naive lymphocytes, postcapillary venules in peripheral organs can support more mixed leucocyte traffic, including monocytes and neutrophils [10, 11]. Thus, endothelial cells display considerable diversity in structure and function, both between and within different organs. In this review we examine the consequences of endothelial cell diversity for the site specificity, persistence and priming of different vascular beads to systemic inflammation in vasculitis.

A key role for the local stromal microenvironment in regulating endothelial function

Endothelial cells exist in very close proximity with stromal cells such as pericytes [2]. These interactions help define the physiological properties of specific vascular beds such that the characteristics of endothelial cells in different organs are influenced by the nature of the underlying stromal cells [3, 11]. For example, the glomerular podocytes in the kidney help dictate the filtering function of this high-flow vascular bed. In contrast, the supporting astrocytes in the brain help define and maintain the blood–brain barrier. The molecular mechanisms by which this occurs remain obscure but there is a growing awareness that matrix proteins, which are laid down and matured during development of vascular beds and which help define organ specialization, play a role [11]. Thus the growth characteristics, angiogenic potential and function of endothelial cells in vitro can be directly modified by the type of matrix on which they are grown [12]. Furthermore, recent studies in which endothelial cells are grown in co-culture with different stromal cells, such as smooth muscle cells and fibroblasts, have clearly demonstrated that varying the stromal cells can have dramatic effects on endothelial cell function [13]. Indeed, recent studies examining how stromal fibroblasts support epithelial cell function have shown that cross-talk between stromal cells and their overlying tubular structures is a general feature by which site specificity is defined and maintained [14].

The role of dysregulated endothelial–stromal interactions in pathology

While it remains difficult to unequivocally demonstrate in vivo that changes in the underlying stroma can directly affect endothelial cell function, it is now clear that targeting both stromal cells (pericytes) as well as endothelial cells in the tumour vasculature is required to obtain significant anti-tumour effects [12]. In addition, targeting the paracrine signalling between tumour cells and stromal fibroblasts by disrupting plateletderived growth factor-receptor α signalling significantly reduces tumour growth [15]. In vitro models of atherosclerosis have shown that secretory smooth muscle cells can prime co-cultured endothelial cells for TNF-α-induced leucocyte adhesion via a mechanism that involves TGF-β [13]. We have recently found that stromal fibroblasts can also modify the behaviour of endothelial cells in co-culture models of chronic synovial inflammation, leading to increased leucocyte recruitment, migration and retention within inflamed tissues. Therefore, an important and evolving concept is that endothelial cell function is pliable and closely regulated both in physiology and pathology by paracrine signals from underlying stromal cells [3, 11]. Moreover, progression to chronic persistent inflammation appears to arise because of changes in stromal fibroblasts, which lead to aberrant leucocyte accumulation within the inflamed tissue [16]. While changes in leucocyte survival, retention and differentiation within inflamed tissue undoubtedly contribute, it is likely that changes in the cross-talk between vascular cells and their underlying stroma also contribute significantly to the location and persistent recruitment of leucocytes into the inflamed tissue.

Thus, in patients with vasculitis, stromal cell/matrix interactions with endothelial cells at the site of inflammation are likely to be important in the initiation and maintenance of the local inflammatory response and the determination of site specificity. In addition, in parallel with these local endothelial changes, and probably secondary to the accompanying systemic inflammatory response, endothelial function is altered at distant sites; an effect which may lead to accelerated atherosclerosis (Fig. 1).

Fig. 1.

Vascular endothelium at the site of the primary inflammatory response and at distant sites. Local endothelial activation, regulated by stromal cells and matrix, drives the vasculitic process with leucocyte recruitment, thrombosis and platelet plugging. Such local inflammation has distant effects on the vascular endothelium. This systemic endothelial dysfunction may be a consequence of circulating inflammatory mediators such as TNF-α and CRP.

Systemic vasculitis and global endothelial cell dysfunction

Endothelial function can be measured in vivo by assessing the vasodilatory response to endothelial-dependent stimuli such as increased shear stress or acetylcholine (ACh) infusion, which lead to the release of endothelial NO. Impairment in this vasodilation represents endothelial dysfunction—a central early event in atherosclerosis. Although NO-mediated vasodilatation is the readout used in many in vivo assays of endothelial function, in fact NO is central to the regulation of other endothelial and vascular functions, such as leucocyte adhesion, platelet aggregation and thrombosis, and the growth and migration of vascular smooth muscle cells [17]. Impaired release of NO by vascular endothelium is thus suggestive of a generalized perturbation of endothelial function.

Transient endothelial dysfunction during a transient systemic inflammatory response has been shown after Salmonella typhi vaccination [18]. Similar endothelial dysfunction has been found in the persistent inflammation that characterizes systemic vasculitis. This was first demonstrated in a cross-sectional study which showed impaired flow-mediated vasodilatation at the brachial artery in vasculitis patients [19]. A subsequent larger study confirmed these findings and demonstrated endothelial dysfunction both in patients with ANCA-associated vasculitis (Wegener’s granulomatosis and Churg–Strauss syndrome) and in patients with polyarteritis nodosa [20]. The occurrence of endothelial dysfunction at the brachial artery in patients with small-vessel vasculitis, without evidence of brachial artery involvement, suggested that the dysfunction was occurring distant to the site of primary vascular inflammation. An impairment in the cutaneous microvascular vasodilatory response to ACh supported the concept of diffuse endothelial dysfunction in patients with systemic vasculitis [20]. In a separate study, endothelial function, assessed in terms of forearm blood flow response to ACh, was similarly found to be impaired in systemic vasculitis [21]. Data regarding the relationship between disease activity and endothelial function have been conflicting. In the largest cross-sectional study of patients with systemic vasculitis there was no correlation between endothelial function at the brachial artery or cutaneous microvasculature and either the CRP or the clinical Birmingham Vasculitis Activity Score [20]. In a subsequent crosssectional study of patients with ANCA-associated systemic vasculitis, arterial stiffness (a phenomenon regulated by both structural and functional factors, including endothelium-derived NO) was significantly impaired in patients with vasculitis and correlated with the CRP [21]. Despite conflicting observations in cross-sectional studies, the concept that endothelial function is related to vasculitis disease activity is supported by longitudinal studies showing an improvement in endothelial function following the induction of disease remission. Thus, endothelial function significantly improved following successful treatment with steroid and cyclophosphamide [19] or with infliximab, used either alone or in combination with steroid and cyclophosphamide [22]. This improvement in endothelial function in patients with active disease is important as it suggests that the defect does not represent fixed and irreversible damage.

Although these data suggest that diffuse endothelial dysfunction is a consequence of active inflammation at a distant site, the mechanisms for this remain unclear. Systemic TNF-α may be an important mediator. TNF-α impairs endothelium-dependent vasodilatation both in vitro and in human arteries in vivo [23, 24]. Interestingly, whilst the local infusion of TNF-α into the brachial artery reduced endothelial-dependent vasodilatation, no such effect was seen with IL-6 [24]. A role for TNF-α is further supported by serial observations in patients with active vasculitis, which have shown a transient improvement in vascular endothelial function following infliximab therapy which deteriorates 10 days after the infusion [25]. Whether TNF-α is having direct or indirect effects on vascular endothelium is unclear. Several direct effects of TNF-α have been described. TNF-α modulates endothelial nitric oxide synthase (eNOS) gene expression at transcriptional and post-transcriptional levels, inhibiting activity of the eNOS gene promoter [26] and destabilizing eNOS mRNA [27]. In addition, TNF-α stimulates the production of reactive oxygen species in vascular endothelial cells [28]. This can reduce endothelial NO availability through the direct inactivation of NO by superoxide and the oxidative degradation of tetrahydrobiopterin, a critical eNOS cofactor [17]. In addition, the oxidative stress which characterizes inflammatory disease can affect endothelial function through a number of other routes, including the generation of oxidized low-density lipoprotein. The inhibitory effects of TNF-α on endothelium-dependent vasodilatation are not only mediated by effects on NO production. TNF-α also inhibits the vasodilatory response to arachidonic acid (the precursor of prostanoid synthesis) [29]. In this study, IL-1 synergized with TNF-α in impairing endothelium-dependent dilatation but IL-6 had no effect. Though TNF-α is an attractive candidate, it is likely that the endothelial dysfunction in systemic vasculitis is multifactorial. CRP, for example, has direct effects on arterial endothelium, reducing eNOS expression and bioactivity and reducing prostacyclin release [30-32]. A direct role for CRP in the initiation of vascular injury is supported by the accelerated atherosclerosis seen in apolipoprotein E-deficient mice expressing the human CRP transgene [33]. In addition to these direct effects on the vascular endothelium, cytokines have a range of metabolic effects on adipose tissue, skeletal muscle and the liver, leading to disturbances which include insulin resistance and dyslipidaemia, which themselves affect the vascular endothelium [34, 35].

Diffuse endothelial dysfunction in patients with systemic vasculitis suggests that they are at increased risk of atherosclerosis. Indeed, inflammatory rheumatic diseases in which secondary vasculitis is common, such as rheumatoid arthritis and SLE, are associated with both endothelial dysfunction [36-38] and accelerated cardiovascular disease [39-41]. Historically, the significant morbidity and mortality associated with acute episodes of vasculitis has meant that less attention was paid to long-term consequences of the disease. However, as treatments for acute events improve, the long-term complications, such as accelerated atherosclerosis, become more relevant. To date few groups have addressed this issue and data have only been reported in preliminary form. However, it would appear that in patients with Wegener’s granulomatosis the thickness of the carotid artery intima media is increased compared with controls [42] and that the risk of cardiovascular and cerebrovascular disease is increased in patients with ANCA-associated systemic vasculitis [43].

Patients with focal vascular inflammation have perturbations in endothelial function both locally and at distant sites. Local endothelial activation, regulated by stromal cells, drives the vasculitic process with leucocyte recruitment, thrombosis and platelet plugging. Such local vascular inflammation has distant effects on the vascular endothelium. This systemic endothelial dysfunction may be a consequence of circulating inflammatory mediators such as TNF-α and CRP. In the long term this may lead to accelerated atherosclerosis.