Role of the endothelium in inflammatory bowel diseases

Abstract

Inflammatory bowel diseases (IBD) are a complex group of diseases involving alterations in mucosal immunity and gastrointestinal physiology during both initiation and progressive phases of the disease. At the core of these alterations are endothelial cells, whose continual adjustments in structure and function coordinate vascular supply, immune cell emigration, and regulation of the tissue environment. Expansion of the endothelium in IBD (angiogenesis), mediated by inflammatory growth factors, cytokines and chemokines, is a hallmark of active gut disease and is closely related to disease severity. The endothelium in newly formed or inflamed vessels differs from that in normal vessels in the production of and response to inflammatory cytokines, growth factors, and adhesion molecules, altering coagulant capacity, barrier function and blood cell recruitment in injury. This review examines the roles of the endothelium in the initiation and propagation of IBD pathology and distinctive features of the intestinal endothelium contributing to these conditions.

Key Words: Microvasculature; Endothelium; Inflammation; Nitric oxide; Adhesion molecules; Crohn’s disease; Ulcerative colitis; Cytokines; Chemokines; Growth factors

INTRODUCTION

Inflammatory bowel diseases (IBD) include Crohn’s disease (CD), ulcerative colitis (UC) (and indeterminate colitis), which share several inflammatory characteristics with other chronic immune disturbances including immune activation, leukocyte infiltration into tissues and increased vascular density[1]. In UC, the colon shows a continuous, superficial inflammation, while CD occurs as patchy transmural inflammation which may affect any region of the gastrointestinal tract. Genetic susceptibilities may play an important role in the development of IBD[2-6] with polymorphisms in CARD15/NOD2 haplotypes (especially in Caucasians) and HLA-DR haplotypes (especially in Asian IBD) and possible defects in interleukin (IL)-23, IL-2, and IL-10 signaling[2,7-10]. IBD is more prevalent in developed nations[11], with several mechanisms being considered to explain disease pathology including environment, hygiene and altered gut flora[11-13]. These different contributing causes may underlie divergent forms and patterns of IBD, which ultimately may lead to a redefinition of different sub forms of UC and CD.

While the mechanisms initiating and sustaining IBD may differ, both UC and CD may reflect dysfunction within antigen-presenting cells (e.g. dendritic cells) or excess activation of CD4⁺ T-cells (resembling T-cell disturbances in psoriasis). Reduced activation of T-cells in some forms of CD appear to allow gut microbiota that have breached the gut epithelium to trigger microvascular inflammation[1,5,9,14-16]. The activation of immune responses in IBD release inflammatory cytokines [e.g. tumor necrosis factor (TNF)-α] and growth factors [e.g. vascular endothelial growth factor (VEGF)-A] into gut tissues provoking gut inflammation and injury[5,17]. Antibodies produced against “normal” gut antigens (e.g. anti-colon, anti-mucin, anti-tropomyosin) have been found in IBD and are suggested to activate cytotoxic T lymphocytes, further increasing inflammation[7]. As IBD progresses, cytokine-mediated inflammation and epithelial apoptosis disturb the intestinal barrier, to allow penetration of gut flora beyond the lamina propria causing intense inflammatory responses[18] while also provoking endothelial microvascular permeability[19].

Another key event in IBD progression is the expansion of the intestinal microvasculature. Angiogenesis in IBD sustains inflammation through alterations in the endothelial lining of these vessels. The endothelium regulates recruitment of inflammatory cells, tissue damage (e.g. vasogenic edema), and production of inflammatory mediators[19-22]. In this review we describe the key roles of the endothelium in mediating and aggravating inflammation in IBD (Figure 1).

Figure 1 Inflammation triggers a change in the endothelium of the intestinal vasculature in response to the cytokines, chemokines and growth factors released by immune cells leading to increased angiogenesis, adhesion molecule expression, leukocyte extravasation, decreased endothelial barrier function and increased coagulation. TNF: Tumor necrosis factor; IL: Interleukin; iNOS: Inducible nitric oxide synthase; eNOS: Endothelial nitric oxide synthase; VEGF: Vascular endothelial growth factor; MMPs: Matrix metalloproteinases; VCAM: Vascular cell adhesive molecule; ICAM: Intracellular adhesive molecules.

ENDOTHELIAL CELLS IN IBD

Endothelial cells (ECs) are the major constituent of the microvasculature that line blood and lymphatic vessels. ECs during IBD undergo rapid and remarkable changes in response to elevated levels of cytokines and growth factors often producing injury to gut tissues. Normally ECs provide an anti-adhesive and selectively permeable exchange barrier[23]. Even though ECs have long been recognized as participants in inflammation their roles in intestinal inflammation during IBD are not yet clear. The unique physiological and molecular characteristics of gut microvessels may help explain several characteristics of IBD. The close relationships between gut metabolism, tissue perfusion, microvascular expansion and immune cell infiltration are unclear but suggest that microvascular alterations may be maladaptive in IBD. Intestinal vascular ECs basally exhibit unique properties which may contribute to IBD. Haraldsen et al[24] first described characteristics of human intestinal ECs (HIMECs) in long-term cultures and differences from ECs of different origin. For example lipopolysaccharide (LPS) only transiently increases HIMEC adhesion molecule expression, while causing long-lasting increases in human umbilical vein ECs (HUVECs)[25]. Nilson et al[26] found that HIMEC cultures produce different cytokines (IL-1β, IL-3 and IL-6) upon stimulation with inflammatory cytokines (e.g. TNF-α, IL-1) compared to HUVECs. Binion et al[27,28] have shown distinctive HIMEC properties such as constitutive inducible nitric oxide (NO) synthase (iNOS) as well as unique adhesive determinants, and that these properties were altered in IBD and may underlie endothelial dysfunction in IBD development.

ENDOTHELIAL NO IN IBD

Endothelial-derived NO reduces leukocyte and platelet adhesion to the endothelium[29,30], mediates flow-dependent and agonist-dependent vasodilatation, and couples VEGF-A signaling with NO-dependent permeability[31,32]. NO-mediated endothelial permeability involves 2 separate mechanisms: (1) increased guanylate cyclase and phospholipase C activity which increases intracellular Ca²⁺; and (2) permeability mediated by Erk1/2 via Ras/Raf/PKC causing increased actin contractility[29,33,34]. Increased p38 mitogen-activated protein kinase (MAPK) signaling, Rho-GTPase activity and increased Ca²⁺ release mediated by upregulated cytokines and growth factors may also represent possible mechanisms for increased endothelial permeability[35-37].

Endothelial nitric oxide synthase (eNOS)-derived NO is a radical scavenger not only absorbing O₂^- but also generating the potent oxidant ONOO^-. eNOS expression is reduced in IBD; eNOS deficiency in IBD is exacerbated by arginase-mediated depletion of substrate as well as eNOS uncoupling[38-40]. Decreased eNOS activity in IBD reduces endothelium-dependent vasodilation, leading to ungoverned oxidant formation, prominent in IBD[41]. Deletion of eNOS (eNOS^-/-) increases severity of experimental IBD[42,43] consistent with protective roles of NO against inflammation. NO may prevent development of endothelial inflammatory and hyper-adhesive phenotype in IBD by suppressing cytokine-induced EC adhesion molecules (ECAMs) and matrix metalloproteinases (MMPs)[44]. Increased endothelial oxidant stress (e.g. in IBD) also disturbs tight junctional organization via p38, p42/44 MAPK[45-47].

Sera from patients with CD reduce, while UC sera increase eNOS in HUVECs; both UC and CD sera increase iNOS[48]. This may reflect differences in anatomic origins of the endothelium i.e. venous vs intestinal. HIMEC iNOS expression appears to be a unique feature of gut microvessels. In HIMECs, iNOS appears at least as important a source of NO as eNOS. Binion et al[30], have shown that HIMECs persistently express iNOS, and that iNOS-derived NO limits leukocyte adhesion in normal HIMECs. Paradoxically iNOS inhibition increases binding of leukocytes. Thus, while leukocyte-derived iNOS may drive inflammation, HIMEC expression of iNOS limits the inflammatory responses (leukocyte adhesion, permeability, vasodilatation) in the gut, and decreased endothelial iNOS abundance and activity in IBD may represent an underrated basis of IBD pathology[27,30]. HIMECs derived from CD patients also show a persistent loss of iNOS expression[27]. Interestingly, iNOS can be decreased by injury to normal HIMECs (opposite to most tissues which mobilize iNOS in response to injury[27]) suggesting that during injury, reduced iNOS might trigger inflammatory responses. Even with the loss of endothelial iNOS, there is often increased NO in tissues surrounding the area of inflammation. Despite decreased endothelial iNOS derived-NO, IBD frequently exhibits increased leukocyte recruitment and activation of gut epithelial cells to increase overall NO production[44]. Krieglstein et al[49] found that tissue-derived iNOS, and to some extent leukocyte iNOS, mediate colitis injury, but could not specifically distinguish between tissue and endothelial contributions of iNOS in colitis. Aoi et al[50] have suggested that iNOS-derived NO plays an important role in gut healing after injury through induction of VEGF, necessary for angiogenesis in wound healing. We have previously shown that excess NO may play an important role in IBD exacerbation. Using STAT-6^-/- mice (which have high iNOS levels) in dextran sulfate sodium (DSS) colitis, we found more severe IBD in STAT-6^-/- mice correlate with extraordinary NO flux suggesting that excess NO may also drive gut injury[51]. Despite elevated NO abundance, downstream guanylate cyclase signaling appears to be depressed in DSS colitis leading to decreased cGMP in the inflamed intestine[52]. Under these conditions, cGMP dependent protective NO effects may be masked by pro-oxidant effects of NO metabolites. Conner et al[53] and Grisham et al[54] revealed an important role of the 26S proteasome in the regulation of endothelial nuclear factor-κB (NF-κB) and cumulative iNOS NO production and adhesion molecule expression. Cumulatively, these studies suggest that intestinal homeostasis is controlled by distinctive and compartmentalized NO sources, and that excess NO formation may support the pathophysiology of IBD.

ENDOTHELIAL TOLL-LIKE RECEPTORS AND IBD

The gut is an organ supporting a high bacterial load; despite physical and chemical barriers, some bacterial antigens will ultimately penetrate the gut wall to activate gut microvascular ECs through Toll-like receptor (TLR) signaling[18,55]. The intestinal microvascular endothelium also differs from ECs of other origins in TLR responses. For example, repeated exposure and activation of TLR4 in HIMECs leads to development of lipopolysaccharide tolerance; however HUVECs lack such a mechanism, indicating the importance in controlling endothelial-dependent inflammation and host commensal interactions[25,56,57]. Protease activated receptors activate transforming growth factor (TGF)-β to induce TLR4 and lead to increased disease severity in IBD[58,59]. TLR5 is constitutively expressed in all ECs, and is of particular interest in gut pathophysiology. TLR5, a receptor for flagellin[60], is constitutively expressed on the basolateral surface of the gut endothelial (an epithelial) layers[61]. TLR5 signaling induces endothelial intercellular adhesion molecule-1, TNF-α production and leukocyte binding and emigration[61]. Loss of TLR5 activity in murine models leads to the development of infectious as a result of deficient and improper responses to normal flora and pathological microorganisms[61,62]. Conversely, endothelial TLR3 has been shown to be protective in the DSS model of acute colitis. This process is mediated by interferon (IFN) type 1 induction of IL-10, a potent anti-inflammatory cytokine[63]. However, Heidemann et al[64] in 2007 found that IL-12 expression and its associated gene products were also induced by TLR3 signaling in addition to increased adhesion and transmigration of leukocytes and TLR functions in the gut remain complex, and requires further study.

IBD-ASSOCIATED CYTOKINES AND CHEMOKINES EFFECTS ON GUT ECs

During inflammation there is an increase in plasma levels of inflammatory cytokines, including IL-6, IL-23, IL-12 and TNF-α, in both human IBD and animal IBD models[1,2,15]. Kawachi et al[65,66] examined cytokine alterations in the adoptive T-cell transfer and the IL-10^-/- IBD models and found IL-1, IL-6, IL-18 and TNF-α were upregulated in both models. Many of the inflammatory cytokines that are upregulated in IBD are pro-angiogenic, the best examples being IL-17 (produced by invasive Th17 cells) and TNF-α produced by several tissue types, including infiltrating immune cells (macrophages and monocytes)[67,68] and the endothelium[69]. EC produce inflammatory mediators in response to activation by immune cells and alterations in the tissue microenvironment[64,70].

TNF-α is one example of a cytokine with pleiotropic effects on the endothelium in IBD, ranging from adhesion molecule induction [vascular cellular adhesion molecule (VCAM)-1 and mucosal addressin cellular adhesion molecule (MAdCAM)-1], promoting interaction of platelets with ECs and inducing expression of pro-angiogenic growth factors such as VEGF-A[25,44,71-73]. Defects in the activity of the anti-inflammatory cytokines such as IL-10 may play a role in the establishment of some IBD, and IL-10 deficient mice (IL-10^-/-) develop IBD spontaneously, while other animal models of colitis show reduced injury when treated with exogenous IL-10[2,74-76]. Interestingly Oshima et al[19] observed that pretreatment of ECs with IL-10 prevented IFN-γ mediated endothelial barrier disruption, indicating that an important role of IL-10 may be to prevent cytokine mediated EC barrier disturbances which initiate and exacerbate disease. This is supported by the finding that several EC adhesion molecules such as intercellular adhesion molecule (ICAM)-1, VCAM-1 and MAdCAM-1 are increased in IL-10^-/- mouse colitis and may mediate leukocyte recruitment in this model[66].

Over 40 chemokines in 4 separate families interact with as many as 19 receptors to regulate trafficking of leukocytes. Of these, several chemokines may mediate leukocyte trafficking to the gut and colon dysfunction in IBD. Papadakis et al[77] showed that CCL2 and CCL5^-/- mice are protected from colitis. Interestingly, Barcelos et al[78] and Wu et al[79] showed that CCL5 and CCL3 can induce inflammatory angiogenesis in a murine sponge model and promote angiogenesis in murine tumors. Eyman et al[80] have also shown that CCL5 upregulates pro-angiogenic genes. CCL25 interacting with its receptor on CCR9⁺ leukocytes plays a major role in the early stages of experimental IBD pathogenesis[81]. CXCL8 (IL-8) another pro-angiogenic chemokine, is known to be stored in EC Weibel-Palade bodies, can be rapidly secreted, and induces HIMEC proliferation in culture via binding to CXCR2[82,83]. Although angiogenesis may support injury IBD, IL-8 may be dysregulated in some forms of IBD. IL-8 seems to be downregulated in leukocytes and in the endothelium of patients with CD. There appears to be no upregulation in the endothelium of UC patients, suggesting a possible link to TGF-β1 over expression in IBD[84-86]. In contrast, Scaldaferri et al[87,88] found that intestinal fibroblasts treated with TNF-α produce IL-8 and monocyte chemoattractant protein-1 via p38/p42/44 mitogen-activated protein kinase.

CX3CL1/fractalkine is a chemokine expressed by EC, can be upregulated by TNF-α, IL-1, LPS and IFN-γ, and is highly upregulated in IBD[89,90]. CX3CL1 can function as an endothelial adhesive determinant to recruit a sub-population of dendritic cells and macrophages that have high CX3CR1 expression. CXCL1 can be shed from the surface of the ECs (in response to increased IL-1β in IBD). This form of CX3CL1 acts as a chemoattractant for CD4⁺ and CD8⁺ T-cells[90]. Sans et al[91] reported that in fact there is enhanced recruitment of CX3CR1 expressing T-cell to the gut via interactions with CX3CL1. CXCR4/SDF-1α and its ligand CXCL12 is an important chemokine/receptor pair in angiogenesis, but have received very little attention in IBD. Heidemann et al[92] reported that blocking this CXCR4/CXCL12 interaction is sufficient to inhibit migration and proliferation of HIMECs in response to VEGF-A. CXCR4/SDF-1α plays an important role in the recruitment of EC precursors to sites of angiogenesis, and may be impaired in IBD, leading to the conclusion that this pathway may be interrupted[93-95]. Midkine, another chemokine of great interest is increased in serum and is associated with tumor drug resistance and poor cancer prognosis[96-98]. Midkine is also upregulated in IBD serum, and has prognostic value like VEGF, TNF-α, sVCAM and VCAM[20,99-102]. Midkine has a pronounced angiogenic effect, like some other inflammatory factors, and also increases the levels of surface glycosaminoglycans on ECs to favor recruitment of circulating leukocytes in IBD[103].

INCREASED ENDOTHELIAL ADHESION MOLECULE EXPRESSION IN IBD

Inflammation in IBD is characterized by increases in both blood and lymphatic vessels in the intestine. This increase in endothelial surface area provides a powerful means of increasing leukocyte recruitment with the mobilization of ECAMs including selectins[28]. Animal models of IBD (IL-10^-/-, IL-2^-/-, SAMP1/Yit and T-bet^-/-), like human IBD, all show ECAM upregulation is linked to disease severity[66,104-106], allowing use of adhesion antagonists in IBD therapy[55,102,107]. Endogenous endothelial-derived inhibitors of leukocyte binding (e.g. sVCAM-1) may also be downregulated in IBD[21,108-111] and could provide new diagnostic or anti-adhesive strategies.

P and E-selectins, glycoproteins expressed on the surface of platelets and other leukocytes, are also expressed on the surface of activated or inflamed endothelium in IBD. P-selectin can interact with ECAMs such as VCAM-1/ICAM-1, as well as with O-glycans collectively referred to as peripheral lymph node addressins (PNAds) containing sialyl Lewis X moieties[112,113]. P-selectin at least partially mediates rolling and recruitment of gut-infiltrating leukocytes in IBD, with approximately 50% increase in gut P-selectin in UC vs control groups; serum levels of soluble P-selectin, an inhibitor of selectin binding, are decreased in IBD patients[109,114,115]. Increased platelet P-selectin, with the enhanced prothrombotic surface of the gut EC in IBD increases thrombus formation and tissue damage by ischemic injury[115]. E-selectin, a relative of P-selectin is expressed solely on the surface of activated ECs during inflammation and is a major contributor to leukocyte rolling injury. E-selectin is not stored in Weibel-Palade bodies and must be produced in response to inflammatory stimuli such as IL-1, TNF-α and VEGF-A[116,117]. In contrast to sP-selectin, sE-selectin is not downregulated in IBD and in CD, and actually increases in comparison to controls[109].

High endothelial venules (HEV) are specialized post-capillary venules that allow trafficking of leukocytes between immune (e.g. Peyer’s patches) and vascular compartments, and are increased in IBD[113]. L-selectin expressed on leukocytes (after activation) binds PNAd on HEV and recruits leukocytes expressing L-selectin in IBD. The gut and brain selective adhesion determinant, MAdCAM-1 is also expressed on HEV, and in UC MAdCAM-1 O-glycosylation increases, allowing greater L-selectin binding[118]. MAdCAM-1 interacts with α4β7 integrins on the surface of a subset of naive CD4⁺ T-cells[119,120]. MAdCAM-1 induction is found only in chronically inflamed gut endothelium and suggests that in IBD there is a fundamental alteration in the phenotype and gene expression pattern in the inflamed intestinal EC[28]. Mizushima et al[121] demonstrated that inhibition of angiotensin-II type 1 receptors reduced TNF-α dependent MAdCAM-1 expression and reduce the severity of DSS-induced colitis, possibly linking vasoregulation and inflammation. In HIMEC MAdCAM-1 is also expressed inversely with cell density, with proportionally greater levels of MAdCAM-1 found at low densities. This indicates that in newly formed vessels, larger amounts of MAdCAM-1 may be available to recruit leukocytes to these “leaky and permissive” vessels[122,123].

ICAM-1, another important ECAM in IBD binds LFA-1 (aLb2), Mac-1 (aMb2) and α4β2 integrins, and is expressed by inflamed ECs to mediate the firm adhesion of leukocytes to activated ECs[124,125]. ICAM-1 has a unique relationship with VEGF-A; Goebel et al[125] reported that HIMECs constitutively express ICAM-1, which is significantly upregulated following treatment with 50 ng/mL VEGF-A, linking inflammation and angiogenesis. In addition to direct activation and upregulation of ICAM-1 by VEGF, Zitterman et al [117] found that VEGF treatment also sensitizes cells to TNF-α induced ICAM-1 mobilization. Normally ICAM-1 concentrates at EC junctions, but is redistributed to apical surfaces of ECs under inflammatory conditions where it supports firm adhesion of leukocytes[25]. In the adoptive T-cell transfer model of murine IBD, Ostanin et al[126] found that T-cells that lack LFA-1, (a T-cell ICAM-1 ligand), fail to induce disease, revealing a critical role for EC modulated immune responses. ICAM-1 was the one of the first clinical targets in IBD, but an antisense IBD therapy showed limited success[21].

VCAM-1, an ECAM highly expressed on the luminal surface of activated ECs in IBD, mediates the adhesion of α4β1 expressing lymphocytes. In HIMECs, the expression of VCAM-1 is regulated by the PI3K/NF-κB signaling pathway and its stimulation by mediators can be inhibited by curcumin[127]. Like ICAM-1, VCAM-1 can also up regulated by VEGF-A via NF-κB[117,128]. Studies in the picrylsulfonic acid model of UC using radiolabeled anti-VCAM-1 antibodies show that leukocyte infiltration and histological damage are proportionate to VCAM-1 expression in the gut microvasculature[129]. In addition, in the DSS model of UC, there is an upregulation of VCAM-1 which if blocked (by specific antibodies) attenuates disease activity, while ICAM-1 and MAdCAM-1 blockade do not protect in this manner[129,130].

CD31/PECAM-1 expressed by ECs and leukocytes mediates homophilic binding between activated ECs and leukocytes especially during extravasation. CD31 is found on the endothelial surface and in endothelial junctions. Work by Romer et al[72] found that CD31 is not upregulated in response to inflammatory cytokines but is redistributed from cellular junctions. CD31 blockade inhibits leukocyte transmigration, and CD31 inhibition in IBD reduced leukocyte rolling and firm adhesion suggesting a unique role for CD31 in IBD or in the function of the gut microvasculature[131].

Originally considered a mesenchymal stem cell marker[132-134], CD146 is now described as a novel immunoglobulin super family adhesion molecule which is increased in gut tissue of IBD patients[108]. The function of CD146 in IBD is not completely understood, but has potential roles in inflammation since it supports rolling and invasion of natural killer T-cells[135]. The upregulation of CD146 in IBD, like ICAM-1 and VCAM-1, may be driven by VEGF-A overexpression during IBD[100]. Additionally, the soluble form of CD146, regulates endothelial and leukocyte CD146 interactions with their ligands, and is reduced in IBD, enhancing leukocyte extravasation[100,108,135]. Interestingly Tsiolakidou et al[100] determined that new vessels formed in IBD are disproportionately CD146⁺. Inflamed ECs from CD and UC patients show an increased ability to recruit naïve T-cells and macrophages to the intestinal immune compartment after stimulation with several inflammatory cytokines, but not with LPS[28,120]. These data are consistent with IBD not being initially driven by immune cells, but rather by the endothelial response to an increased inflammatory mediatory load.

PLATELETS AND COAGULATION IN IBD

Platelet and leukocyte aggregation as well as activation of the coagulation cascade increase during IBD, reflecting loss of the non-thrombogenic EC phenotype in IBD. Thrombi aggravate inflammation by binding of micro infarcts to the endothelial surface often leading to ischemic inflammation in the intestinal microvasculature[136]. Mesenteric venous thrombosis has been observed in a fraction of IBD cases, and thrombotic processes are being recognized in altered perfusion, inflammation and tissue injury in IBD[137]. Indeed, subclinical thrombosis is common in IBD, and is a major source of morbidity in approximately 25% of IBD deaths[136]. Increased markers of coagulation include thrombin anti-thrombin complex, tissue factor and fibrinopeptide B[55], and can be described early in IBD. Factor XIIIa, a fibrin-stabilizing coagulation factor (and agonist for VEGFR-2), is increased in IBD, while factor XIII TT has an increased number of mutations in IBD patients compared to controls suggesting links between thrombosis, angiogenesis and inflammation. However, Bernstein et al[138], Dardik et al[139] and Vrij et al[140] reported that factor XIII activity is reduced in IBD patients.

In addition to increased levels of coagulation cascade proteins in IBD, CD40, CD40L and soluble CD40L are increased in IBD. CD40, expressed on several cell types (including ECs) is involved in inflammatory and immune activation, and interacts with CD40L on T-cells. Danese et al[141] suggested that the primary source of sCD40L was from activated platelets. CD40 signaling increases production of pro-inflammatory cytokines and chemokines by ECs and surrounding tissue[142]. CD40L release also leads to binding of platelets and immune cells to ECs by increasing tissue factor, ECAM expression and pro-thrombotic phenotype in HIMECs[141-143]. Danese et al[71] suggested that a possible therapeutic benefit of TNF-α blockade was downregulation of CD40/CD40L signaling in IBD. A still unanswered question is whether coagulation is a secondary or initiating event in inflammation. It is worth mentioning that individuals with coagulation cascade disorders (e.g. hemophilia, factor V deficiency and von Willebrand disease) rarely develop IBD[55]. The opposite of the previous observation is also true; patients with IBD have an increased likelihood of having genetic pro-thrombotic disease Factor V Lieden[144]. This evidence strongly links thrombus formation as a possible trigger of IBD and suggests prognostic factors which may increase risk of IBD development.

ENDOTHELIAL BARRIER DYSFUNCTION IN IBD

The maintenance of normal vascular barrier supports nutrient and O₂ exchange, osmotic balance and leukocyte abundance in the extracellular compartment. In IBD, increased vascular permeability leads to tissue edema and damage in both human IBD and animal models of IBD[19]. This alteration in solute permeability of the vasculature is not restricted to the gut microcirculation but is widespread affecting the vasculature of other organs including the brain[145]. Several classes of mediators in IBD alter both solute permeability and angiogenic balance, including angiogenin (an angiogenic peptide with ribonuclease activity), chemokines (e.g. IL-8, IL-10), coagulation factors (thrombin), cytokines (IFN-γ, IL-13), and growth factors, most notably VEGF, the most potent and important blood vascular angiogenic growth factor and an important inflammatory mediator[19,36,37,47,146-148]. Tolstanova et al[149] found that VEGF-A inhibition by neutralizing antibodies reduced vessel permeability in the iodoacetamide model of colitis. Downregulation of anti-inflammatory cytokines e.g. IL-10 may play an equally important role in increasing endothelial permeability. Oshima et al[19] have shown increased vascular permeability in the IL-10^-/- colitis model due to loss of IL-10 inhibition of IFN-γ induced junctional degradation; also IL-10 protects against IFN-γ mediated loss of human microvascular barrier.

Leukocytes, e.g. neutrophils and monocytes, can degrade endothelial junctions through protease secretion and upregulation. Cytokines and growth factors also induce MMP-9, MMMP-3 and MMP-1[150,151], resulting in degradation of junctional and matrix targets[152]. Neutrophil elastase is elevated in IBD and can degrade vascular endothelial cadherin, important in maintaining junctional apposition, adhesion and barrier function[153-156]. Endothelial junctional adhesion molecule-A is also dysregulated in IBD, and is closely linked to disease activity in DSS colitis[37,157].

ANGIOGENESIS IN IBD

Angiogenesis (increased blood vessel density) in IBD increases the area of endothelium available for exchange, but also for extravasation of blood constituents into surrounding tissue to increase disease severity in IBD[158]. Increased vessel formation in IBD may represent recruitment of endothelial progenitor cells, vascular intussusception (splitting) and extension from existing vessels[159]. Increased angiogenesis is observed in animal (2,4,6-trinitrobenzene sulphonic acid (TNBS), DSS and iodoacetamide) colitis models and in human colitis. However, inflammatory angiogenesis in IBD does not simply match increased tissue mass. Vessels formed during inflammation are different from those formed during normal development. These vessels are immature, lacking investment with pericytes. They express ECAMs, leak, are hypoperfused, often stenose and are hyperthrombotic, with an elevated ability to respond to growth factors[160-163] actively supporting IBD progression[149,164-168]. Spalinger et al[158] and Maconi et al[169] concluded that there is an increased blood vessel density in the intestines of CD and UC patients and that increased vascular density in IBD was directly correlated with increased IBD disease severity. This is also true in animal models of IBD like TNBS- and DSS-induced colitis models[166,170].

Growth factors, especially VEGF-A, dramatically alter several aspects of the colon microvascular endothelial phenotype, resembling a de-differentiation (loss of maturity) of the vessels which can reflect changes in vascular support cells, e.g. pericytes/smooth muscle, that surrounds the capillaries. Inflamed tissues display increased vascular density resulting from the formation of new vessels during angiogenesis. These changes result in decreased perfusion, increased solute permeability (via cytokines and VEGF-A induced junction remodeling) and contractility, as well as increased leukocyte and platelet adhesiveness[161,171,172]. Ganta et al[163] have demonstrated that in angiotensin-2 knockout mice (using the DSS model of UC), loss of the pericytes around vessels resulted in diminished angiopoietin-1 signaling that destabilized the endothelial layer, increased leukocyte recruitment to the tissue, increased vessel permeability and induced vessel hyper-proliferation. Blood and lymphatic vessels are hyperstabilized by angiopoietin-2 deficiency, and show diminished inflammatory remodeling as well as decreased capacity to recruit leukocytes suggesting a link between maturity and inflammatory capacity[163].

ENDOTHELIAL CELL AND ANGIOGENIC GROWTH FACTOR INTERACTIONS IN IBD

VEGF-A is the first described and best known VEGF, which controls developmental angiogenesis, wound healing and pathology[173,174]. Bousvaros et al[175], Kapsoritakis et al[101] and Ozawa et al[176] all found elevated VEGF-A levels in plasma and tissue during active human and animal IBD, often twice normal[101,109,166,175,176]. However, Chidlow et al[166] have reported that DSS diminishes levels of VEGF-A as well as VEGF-C and VEGF-D, suggesting complex, concentration-dependent and inhibitor-regulated effects of VEGF in different animal models of IBD. Danese et al[177] and Scaldaferri et al[167] have shown that inhibition of VEGF signaling can attenuate disease activity in the DSS model of UC while overexpression of VEGF-A increases disease severity in the same model[167,177]. VEGF-A is released by several cell sources (e.g. neutrophils, platelets, macrophages, pericytes, fibroblasts, ECs, and colonic epithelial cells) and is transcriptionally activated by hypoxia through hypoxia inducible factor 1α, and message stabilization via eukaryotic translation initiation factor 4e[70,178-183]. Interestingly Birmingham et al[184] have shown that activated colonic epithelium represents an important source of VEGF-A, and injury or inflammation of the colon epithelium may provide a local stimulus for blood vessel growth. Invasive leukocytes, specifically neutrophils, granulocytes, macrophages and platelets, are increased in tissue during active IBD, and are also important sources of VEGF-A in inflamed tissues[178,179,185-187]. Salivary secretions also contain high levels of VEGF-A and VEGF-C, which have been suggested as important sources of these growth factors in IBD[188] released site-specifically during denudation. Apart from VEGFs, other angiogenic growth factors, e.g. basic fibroblast growth factor (bFGF), TGF-β and platelet-derived growth factor (PDGF) are upregulated in IBD and may be of clinical relevance[86,189].

TGF-β is an important regulator of the cell cycle and apoptosis, especially in mucosal immune cells. The expression of TGF-β and its 2 receptors (TGFR1 and TGFR2) are increased in IBD, specifically UC; however, it appears that the levels are decreased in CD[190]. In IBD either tachyphylaxis develops for TGF-β (UC), or the lack of TGF-β (CD) allows mucosal immune cells to proliferate when they would have undergone apoptosis[190,191]. Early studies on the role of TGF-β in IBD indicated a protective role; more recent studies may point to a pathological role of TGF-β signaling in IBD[191,192]. In fact, TGF-β is important in the formation of fibrosis in the colon of IBD patients by stimulating the transition of many cell types to fibroblasts[193]. Over one-third of the fibroblasts responsible for inflammatory fibrotic injury may actually originate from the transformation of ECs to fibroblasts (not counting contributions of pericytes to fibroblast formation). Therefore the vasculature may provide a significant proportion (if not the majority of fibroblasts) and associated fibrosis in IBD[194,195]. bFGF, a potent mitogen for the cells of mesodermal origin, stimulates EC proliferation, activates MMPs resulting in proteolysis of extracellular matrix, and increases cellular motility[191]. Even though levels of bFGF are elevated in IBD there is no correlation with the stage or severity of the disease. However, the contribution of bFGF in the initiation or maintenance of IBD should not be discounted[196]. PDGF is a close relative to VEGF and is upregulated in IBD. PDGF is predictive of both oxidative stress and angiogenesis in the intestine[189]. PDGF is released in response to inflammatory and thrombotic stimuli. PDGF increases P-selectin expression on ECs and induces histamine secretion which induces other effects such as increased vascular leakage[197,198]

ENDOTHELIAL PROGENITOR CELLS AND VESSEL SPROUTING IN IBD

Recruitment of endothelial progenitor cells (EPCs) may contribute to angiogenesis in IBD, although reduced numbers of VEGFR2⁺, CD34⁺, CD133⁺ cells (endothelial, bone marrow, and stem cell markers) have been reported in IBD[199], and EPCs from IBD have reduced antigenic activity[95]. These findings suggest that recruitment of EPCs is unlikely to be a source of increased vessels, however, these findings are from patients with established disease as initial angiogenesis in early stages of IBD may rely on EPCs. Apart from EPC recruitment, angiogenic sprouting is active in IBD; sprouting ECs referred to as “tip” cells, are highly motile with distinct gene expression compared to that in quiescent ECs[200]. VEGF-A induces the tip cell phenotype and also guides vessel sprouting, indicating that in IBD, VEGF might induce new vessel formation in this way[201]. Normally, not all sprouts survive, many undergoing apoptosis, (vessel “pruning”) suggesting that high levels of VEGF prevent endothelial apoptosis resulting in increased numbers of surviving sprouts in IBD[201].

INHIBITORS OF VASCULAR ENDOTHELIAL EXPANSION IN IBD

While increased pro-angiogenic growth factors increase angiogenesis, reductions in anti-angiogenic factors (seen in the DSS model of colitis) may be as important for permitting expansion of the vascular endothelium[166,167]. Angiopoietin-1 a competitive inhibitor of Ang-2, binds to the Tie-2 receptor and inhibit vascular remodeling. Angiopoietin-2 is upregulated during inflammation and angiogenesis[163,202] and competes with angiopoietin-1, to allow ECs to maximally respond to cytokines and growth factors. Work by Ganta et al[163] found that angiopoietin-2 signaling also appears to be necessary for neutrophil infiltration, and blood and lymphatic vessel proliferation in DSS colitis. Interestingly angiopoietin-2 can be upregulated by both bFGF and VEGF, potent pro-angiogenic growth factors also upregulated in IBD[203-205].

Angiostatin, a fragment of plasminogen generated by MMPs has anti-angiogenic and anti-proliferative effects on ECs and blocks vessel maturation[206]. During IBD, levels of MMPs are elevated and generate angiostatin[153]. In fact 2 models of experimental colitis (iodoacetamide, TNBS) show increased angiostatin, and may represent a feedback control for angiogenesis[207]. Interestingly, the effect of angiostatin hinges less on inhibition of EC proliferation, but more on inhibiting final vessel maturity[208]. Much like angiostatin, endostatin results from the cleavage of collagen type XVIII yielding an anti-angiogenic fragment that is upregulated in experimental colitis[207,209]. Endostatin reduces EC migration and proliferation; however like angiostatin, endostatin fails to block angiogenesis in the TNBS model, but may play a role in disease progression and maintenance by impairing vessel maturity and tissue healing by antagonizing VEGF-A induced tissue repair[207]. Interestingly Deng et al[210] showed mesalamine treatment of iodoacetamide colitis restored levels of endogenous angiogenesis inhibitors, endostatin and angiostatin helping reduce disease severity.

Soluble VEGF receptors (sVEGFRs) are truncated forms of VEGFR1 or VEGFR2 genes[211] that under normal physiological conditions maintain tissue avascularity (e.g. in the cornea) and might be dysregulated in IBD. During inflammation, sVEGFR1 inhibition seems to be lost (e.g. in the case of an alkali burn)[211,212]. sVEGFR2 seems to plays an important role in the inhibition of lymphangiogenesis compared to sVEGFR1, but sVEGFR2 blocks transplant rejection which points out its greater immunomodulatory effect[213]. Additionally, Scaldaferri et al[167] found that over expression of sVEGFR1 reduced disease severity in the DSS model of colitis, suggesting that loss of this molecule in IBD would be detrimental. Interestingly the anti-angiogenic VEGFs, alternate splice variants of VEGFs, are downregulated in several inflammatory diseases, and are linked to the alteration of the cytokine milieu in the tissues[214-217]. These inhibitory VEGFs make up a majority of the VEGF load in the normal intestinal micro-environment with approximately 20 times greater levels in the healthy gut[217]. Currently, the levels of these inhibitory VEGFs are unknown in IBD, but may provide a new avenue for anti-angiogenic therapies, we are pursuing this possibility which is currently showing great promise (unpublished data).

IBD THERAPIES

It is increasingly clear that IBD therapies affect the microvasculature, and that the microvasculature is a central target in IBD, coordinating cell infiltration, solute permeability, cytokine/chemokine production and gut immunological responses. An increasing number of drugs that show efficacy in treating IBD have now been found to affect the endothelium. Accumulating evidence suggests inhibition of angiogenesis as a secondary mechanism of action for many IBD therapies including anti-TNF-α antibodies, and some immunosuppressive agents (cyclosporine A)[218-220]. Scaldaferri et al[87] found TNF-α mediated lymphocyte adhesion and chemotaxis across intestinal microvascular ECs depends on expression of ICAM-1, VCAM-1 and fractalkine in the affected ECs mediated by p38 MAPK, p42/44 MAPK and JNK. Danese et al[71] found that anti-TNF-α therapies can reduce thrombus formation and adhesion to the endothelium by interfering with CD40/CD40L signaling. Integrin-blocking antibodies have been used in the treatment of IBD, but not without a controversial side effect. Natalizumab (Tysabri), an α4-integrin blocking monoclonal antibody originally developed for use in the treatment of multiple sclerosis, but has recently been approved for the treatment CD[21,221]. AJM300, a peptide blocker for α4 integrins, successfully blocked α4 -VCAM-1 and MAdCAM-1 adhesion and prevented exacerbation in IBD models[20,21]. However, recent preclinical trials using AJM300 failed to inhibit disease progression[20,21]. Rafiee et al[222] found that 2 drugs used in IBD, thalidomide and cyclosporine-A, are anti-angiogenic; thalidomide targets TNF-α and VEGF-A, while cyclosporin-A targets VEGF-A alone[222,223]. Studies by Ogawa et al[224] determined that HIMEC expression of the inflammatory mediators IL-6 and cyclooxygenase (COX)-2 by LPS were inhibited by butyrate, and that butyrate also inhibited HIMEC angiogenesis[224-226]. Despite its anti-inflammatory properties[223,227,228], cyclosporin-A increases leukocyte binding, unlike thalidomide which reduces leukocyte binding to HIMECs[227,228].

IBD-INDUCED ANGIOGENESIS AND COLORECTAL CANCER

The risk of developing cancer is elevated by inflammation, and the link between IBD and colorectal cancer (CRC) is convincing[229-231]. Inhibition of angiogenesis in CRC by bevacizumab (anti-VEGF monoclonal antibody) improves clinical outcomes, revealing the importance of angiogenesis in the progression from IBD to CRC[232]. As stated before, IBD in human disease and animal models is associated with an increase in vascular density, and it is possible this vascular endothelial expansion may enable CRC[158,163,169]. CRC incidence may depend on COX expression (seen in adenomatous polyposis coli, pre-cancerous lesions enriched in COX[233-235]. COX-2 is increased in human IBD and IBD models, and may promote CRC through angiogenesis[236,237]. COX-2 promotes EC proliferation by prostaglandin induction of VEGF-A, important for tumor angiogenesis[237,238]. Chan et al[239] reported that the regular use of aspirin, a non-selective COX inhibitor, significantly reduced the risk of COX-2⁺ CRC, which constitutes approximately 67% of human CRC[239,240]. Additionally COX-2 inhibition reduces tumor growth and increased tumor apoptosis, and is associated with reduced tumor angiogenesis[238,241,242]. Conversely, Ishikawa et al[243] found that COX-deficient animals were not protected from tumor formation in azoxymethane (tumor promoter)-induced colorectal cancer, and concluded that COX expression was not a major determinant of tumor formation in UC. While COX expression may not be not necessary for tumor formation in UC, COX-2 upregulation is only one mechanism for increased angiogenesis in IBD[86,166,189]. VEGF-A and other angiogenic factors are upregulated independent of COX-2 in IBD; therefore, while COX-2 may be important in CRC in the absence of IBD, expansion of the vasculature in IBD through other mechanisms may contribute to the development and growth of CRC[166,244].

CONCLUSION

A unique combination of genetic and environmental factors may contribute to development of IBD. ECs are now recognized as central and fundamental elements in IBD pathophysiology. ECs are indirectly affected by many IBD medications, which are increasingly targeting ECs directly. As treatments for IBD are developed and refined there will be an increased interest in inhibiting functions of ECs in IBD such as immune cell recruitment and inflammatory angiogenesis, and improving beneficial lymphatic function. Use of endogenous inhibitors of leukocyte binding (sVCAM) and peptides (AJM300) may become novel therapies which supplement or replace current anti-adhesion treatments. Additional studies on the interactions between the gut microvasculature, platelets and their regulation of inflammatory angiogenesis may provide new avenues for treatments that not only reduce thrombosis but also several clinical manifestations of IBD. Inhibition of inflammatory growth factors, cytokines and chemokines that promote angiogenesis by the use of “traps” or decoy receptors, alone or in combination, in addition to current treatments could provide greater anti-inflammatory effects by reducing endothelial expansion in IBD. More importantly, work in our laboratory suggests that endogenous angiogenic inhibitors (VEGF164b) have great potential in the treatment of IBD. Future studies promoting therapeutic intervention by combining anti-angiogenic, anti-immune and anti-inflammatory agents as treatment options focusing on the endothelium as core/vital for IBD pathogenesis will provide greater specificity and efficacy for treating CD and UC patients.