|Home | About | Journals | Submit | Contact Us | Français|
Recent advances in our understanding of the fundamental mechanisms of seemingly unrelated diseases, i.e., rheumatoid arthritis (RA), Type 1 diabetes mellitus, psoriasis, multiple sclerosis, and inflammatory bowel disease (IBD), have demonstrated that they share similar pathogenetic mechanisms. These diseases are immune-mediated chronic inflammatory conditions, characterized by inappropriate and sustained recruitment of inflammatory cells into affected tissues, resulting in chronic tissue damage and loss of function 1–3.
The fate of the inflammatory cells differs according to cell type. Neutrophils migrate to sites of inflammation and undergo apoptosis. Some monocytes remain in inflamed tissues for a few days and others become permanent residents. By contrast, lymphocytes not only migrate but recirculate from the blood into tissues and through the lymphatics back to lymphatic tissues and blood, constantly patrolling for foreign antigens. Once they encounter an antigen in the context of an antigen presenting cell, nai–ve T cells proliferate and acquire effector functions, along with a repertoire of surface molecules (i.e., adhesion molecules, cytokine and chemokine receptors) that allow them to recognize counter-receptors/ligands expressed in specific vascular beds. Using these surface molecules, lymphocytes are able to recirculate thousands of times back to areas with a similar microenvironment to where they first encountered their cognate antigen 4, 5. These unique capabilities (i.e., memory acquisition and recirculation) are essential for the perpetuation of chronic inflammatory processes, including IBD. The molecules involved in the recirculation of lymphocytes have therefore attracted a great deal of interest regarding their potential as therapeutic targets. Some of these have crossed from the bench into the clinical arena, being currently in clinical use 6–8.
In IBD, the inflammatory process is characterized by heavy leukocytic infiltration of the intestinal lamina propria (LP), leading to fibrosis and loss of function 9–11. Lymphocytes that produce cytokines such as IL-12, IFN-γ, tumor necrosis factor-α (TNF-α), IL-23 and IL-17 12 all play an important role in chronic intestinal inflammation 9–13.
The success of the anti-TNF-α strategy in IBD 14 has led to the systematic study of anti-inflammatory cytokines and the development of antibody-based strategies to modulate the overall cytokine balance 15, 16. Unfortunately, the therapeutic efficacy of some of these newer cytokine-targeted therapies (e.g. IL-10, IL-11 blockade) has been limited 15, 16. A neutralizing antibody against the IL-12 p40 subunit, shared by IL-12 and IL-23, has shown promise 17. Alternative therapies that target other pathways of the chronic inflammatory process may be directed at interfering with lymphocyte recirculation to the intestine by targeting specific adhesion molecules, their ligands, chemokines or their receptors 18–20. Using this approach, two monoclonal antibodies against integrin α4 and αLβ2 have been approved by the FDA for the treatment of MS, CD and psoriasis (i.e. natalizumab, efalizumab) 6, 18. However, many of the basic mechanisms that account for their clinical efficacy remain to be elucidated. This limited knowledge has likely contributed to the occurrence of serious adverse events in clinical practice 21, 22.
Leukocytes primarily migrate from the blood into the tissues across the walls of post-capillary venules. Surface molecules on specialized venular endothelial cells play a crucial role. These adhesion molecules not only serve as mechanical anchors, but also confer tissue specificity to the recruitment process through their selective patterns of expression by vascular beds 23. Myeloid cells and lymphocytes share some of the steps in the adhesion cascade, but there are also significant differences 24. This review focuses primarily on lymphocyte recruitment to the intestine and how this process has been targeted for therapeutic purposes in animal models of colitis and ileitis that mimic aspects of either UC or CD respectively.
Several major classes of leukocyte adhesion molecules are involved in leukocyte recruitment, including the selectins and their glycoprotein ligands, integrins and immunoglobulin-superfamily molecules. They are all type I transmembrane glycoproteins that span the cell membrane only once. The structural and functional aspects of these adhesion molecules have been extensively discussed elsewhere 25,26–28.
The process of leukocyte recruitment to a site of inflammation encompasses the engagement and efficient arrest of leukocytes onto the vascular endothelium and their subsequent transmigration 4, 23, 29. This sequence is composed of several major steps, capture, rolling, activation and firm adhesion (Figure 1).
Capture is defined as the formation of the first molecular bond or tether between the circulating leukocyte and the vascular endothelium. Close proximity between the two is required. Capture is distinguishable from stable rolling and is mediated by L- and P-selectins. In inflamed venules in vivo, leukocyte attachment also involves other mechanisms, such as secondary capture through leukocyte-leukocyte interaction 30, 31. Most neutrophils start rolling as they exit capillaries32, whereas nai–ve lymphocytes capture in high endothelial venules (HEV) of peripheral lymph nodes (PLN) and other lymphoid tissues 5. Little is known about capture of effector/memory lymphocytes, although selectins and CD44 may be involved 33, 34.
If leukocyte capture is followed by the formation of new molecular bonds before the initial molecular bonds dissociate, a stable rolling movement is established 35, 36. This is a flow-driven downstream movement of the cell, during which it is in continuous contact with the vessel wall. Nai–ve lymphocyte rolling in HEV is mainly mediated by lymphocyte L-selectin binding to sulfated carbohydrate-containing ligands expressed on endothelial cells. Rolling has at least two distinct consequences for the cell: (i) it facilitates stable leukocyte arrest (firm adhesion); and (ii) it drastically reduces leukocyte velocity (to between 1 and 100 μm/s), increasing the duration of exposure to the endothelial surface and to chemokines and other activating signals present there. Neutrophils arrest after gradually slowing down 37, whereas naïive lymphocytes arrest immediately upon encountering chemokines 38. Effector T cells can arrest through the chemokine receptor CXCR3 in vitro, but the molecules involved in arrest in vivo remain unknown 39.
Activation of the rolling lymphocyte can be triggered by the binding of a chemokine to a heptahelical transmembrane receptor on the leukocyte surface and can result in firm adhesion 23, 40. In flow chamber systems, this process is exceptionally rapid 41,42. Prior studies, focused on lymphocyte function-associated antigen (LFA)-1, have shown that firm adhesion requires binding of integrin receptors in their active conformation to their endothelial ligands 43. Both α4β1 (VLA-4) and α4β7 integrins have been shown to mediate firm adhesion to their respective ligands Vascular Cell Adhesion Molecule (VCAM)-1 and Mucosal Addressin Cell Adhesion Molecule (MAdCAM)-1 44, 45.
The selectins are a family of transmembrane mammalian lectins expressed on the surface of leukocytes (L-selectin), endothelial cells (P-, E-selectins), and platelets (P-selectin) 27. The selectins contain an N-terminal extracellular domain with structural homology to calcium-dependent lectins, followed by a domain homologous to epidermal growth factor; and two to nine consensus repeats (CR), similar to sequences found in complement regulatory proteins (Figure 2). Each selectin inserts into the cell membrane via a hydrophobic transmembrane domain and possesses a short cytoplasmic tail. Each of the three selectins can independently mediate leukocyte rolling, given the appropriate conditions. The three selectins share approximately 50% sequence homology among the extracellular portions of the molecule, but there is no significant homology in the transmembrane and cytoplasmic domains 27, 46.
All three selectins are involved in the trafficking of granulocytes and lymphocytes to sites of inflammation. However, the role of each selectin differs in the physiological sequence of events, and the importance of each selectin varies in different models of inflammation. Importantly, L-selectin is essential for the homing of lymphocytes to peripheral lymph nodes 47, and plays an accessory role in lymphocyte homing to Peyer’s patches (PP), a component of the gut-associated lymphatic tissue 47. High endothelial venules (HEV) in these tissues constitutively express functional ligands for L-selectin on their surface. The importance of L-selectin in lymphocyte homing is emphasized by the decreased size of lymph nodes in L-selectin-deficient mice 47 and a delay in lymphocyte homing to Peyer’s patches 47. In PP, L-selectin synergizes with β7 integrins, since absence of both L-selectin and integrins completely abrogates lymphocyte homing to these structures 48.
T cell activation results in proteolytic shedding of L-selectin 49. This pattern of surface molecule expression (CD44high, L-selectinlow) is characteristic of effector memory T cells which recirculate to effector sites, while limiting their traffic through secondary lymphoid organs 5. However, subpopulations of memory cells (i.e. central memory, TCM) 50 retain or re-express L-selectin to recirculate through inductive sites and potentially to tertiary lymphoid organs present at sites of chronic inflammation, where functional ligands for L-selectin are expressed.
P-selectin (CD62P) contains nine consensus repeats (CR) and extends approximately 40 nm from the endothelial surface (Figure 2). P-selectin is constitutively synthesized by megakaryocytes and endothelial cells, but is not expressed on the surface of resting platelets or endothelial cells. Rather, it is stored in secretory vesicles called Weibel-Palade bodies in endothelial cells, and α-granules in platelets 51, 52.
These granules are triggered to fuse with the plasma membrane by activators such as histamine, thrombin, complement C5a, calcium ionophores, and adenosine diphosphate (ADP), resulting in expression of P-selectin on the cell surface 53. Maximal levels of P-selectin are detected within minutes at the cell surface, followed by rapid internalization and degradation or recycling 51. In the mouse, activation of endothelial cells by cytokines prolongs the surface expression of P-selectin 54–56. Yet in humans, expression of P-selectin is not induced by cytokine stimulation, possibly a result of structural differences in the P-selectin promoter 57.
The transient interactions between P-selectin and P-selectin Glycoprotein Ligand (PSGL)-1 allow leukocytes to roll along the venular endothelium. P- and E-selectins tend to have overlapping functions in TNF-α-stimulated venules. Indeed, in mice deficient for P-selectin, it is necessary to block E-selectin function to significantly reduce rolling, and in E-selectin knockouts, an antibody against P-selectin must be introduced to reduce rolling. Correspondingly, no leukocyte rolling is observed in E-selectin/P-selectin double deficient mice treated with TNF-α 58. Nonetheless, although P- and E-selectins seem to have redundant functions, observations of rolling flux fractions and rolling velocities indicate that under physiological conditions, P-selectin is responsible for early rolling, while E-selectin allows slow rolling and adhesion 37, 59. In mice deficient for P-selectin, rolling is absent initially but returns after 1–2 hours 60. Endothelial cell expression of P-selectin is elicited within minutes by inflammatory mediators such as histamine, thrombin, or phorbol esters. The expression is short-lived, reaching its peak after only ten minutes. In mice, additional synthesis of P-selectin is brought about within two hours by cytokines such as IL-1 or TNF-α 54–56.
E-selectin is expressed by endothelial cells after activation by IL-1 or TNF-α 61, 62, but unlike P-selectin it is not expressed by platelets. On the surface of cultured endothelial cells, it is only expressed transiently for a period of 2 to 6 hours. Subsequently, E-selectin can either be re-internalized or shed from the endothelial surface 63. In vivo, the expression of E-selectin appears to be more prolonged in some organs 64 and is constitutive in skin microvessels 65,66. As discussed earlier, the role of E-selectin during rolling is partially redundant with that of P-selectin. E-selectin-deficient mice have a reduced number of firmly adherent leukocytes in response to local chemoattractants 67 or cytokine stimulation 68, thus E-selectin participates in the conversion of rolling to firm adhesion. E-selectin ligation of PSGL-1 triggers neutrophil activation through spleen tyrosine kinase (Syk) 69. E-selectin is of particular importance in skin inflammation, as it supports the recruitment of skin-specific T lymphocytes 70. The ligand(s) responsible for E-selectin-mediated rolling interactions include PSGL-1, CD44 and E-selectin ligand-1 (ESL-1) (Figure 2) 71–73. In addition to glycoproteins, glycolipids can also support E-selectin dependent rolling in vitro 74. E-selectin mediates much slower rolling than P-selectin, because it triggers partial LFA-1 activation 69. These rolling velocities range between less than 5 μm/s 59, 75 and 15 μm/s 67. The molecular dissociation rate or off-rate of E-selectin is very similar to that of P-selectin 76. However, the velocity of E- selectin-mediated rolling is remarkably invariant with wall shear rate, presumably because E-selectin engagement triggers LFA-1 activation 69. A role of E-selectin and its ligands has been demonstrated for lymphocyte homing, particularly to skin. Cutaneous-homing effector/memory T cells express a carbohydrate epitope (i.e. cutaneous associated lymphocyte antigen (CLA)) which decorates glycoproteins such as PSGL-1 and CD43 and is associated with E-selectin ligand activity 77.
L-selectin, a 74–100 kDa molecule also known as CD62L, is constitutively expressed at the tips of microfolds or microvilli on granulocytes, monocytes, and most circulating lymphocytes. Lymphocytes preferentially express it when in the nai–ve state. In addition, it is expressed by central memory T cells 50, and most natural killer cells 27, as well as by many immature hematopoietic cells in the bone marrow. L-selectin expression on neutrophils correlates negatively with the age of the individual cell, suggesting that neutrophils may lose L-selectin as they mature in the bone marrow 78 and as they age in the bloodstream79. Although most lymphocytes in the cortex of secondary lymphoid organs express L-selectin, the activated cells in the germinal centers do not 27. This indicates that expression of L-selectin is at least transiently reduced from activated lymphocytes. L-selectin is proteolytically cleaved from the surface of leukocytes after activation by a variety of stimuli 80–83. For neutrophils, activating agents that induce L-selectin shedding include inflammatory chemokines such as IL-8, complement factors such as C5a, bacterial peptides such as formyl-methionyl-leucyl-phenylalanine (fMLP), and lipid mediators such as platelet activating factor (PAF) or leukotriene B4 (LTB4) 84. In vitro, lymphocytes shed L-selectin when they are stimulated with phorbol esters or by cross-linking of CD3 85. In vivo, soluble L-selectin is present in the plasma at high concentrations 86, most of which is derived from lymphocytes 87. L-selectin is responsible for naïve lymphocyte homing to PLN and MLN, but is also expressed by central memory T cells, which may play a role for the maintenance of IBD and other chronic inflammatory conditions 88.
PSGL-1 is the best-characterized selectin ligand (Fig. 2) 89, 90. It is a type 1 surface dimeric glycoprotein that is expressed on the microprocesses of virtually all leukocytes. PSGL-1 is also expressed in certain endothelial cells within the walls of the small intestine 91, and inflamed colon 92 as well as in cultured umbilical vein and microvascular endothelial cells 93, 94. Certain post-translational modifications are necessary for PSGL-1 to function as a selectin ligand, including tyrosine sulfation 95, 96, sialylation 97, decoration with core 2 oligosaccharides 98 and, perhaps most importantly, fucosylation 96. Fucosylation is regulated by the specific, inducible fucosyl transferase (FTVII), which plays a key role in the generation of functional selectin ligands 98–101. T-helper 1 (Th1)-polarized T cells preferentially express functional P-and E-selectin ligands 102–105. Indeed the expression of FTVII is modulated by differentiation into Th1 and Th2 phenotypes 33. FTVII is constitutively expressed by neutrophils, monocytes and eosinophils 99, 100. The trafficking of neutrophils is severely restricted in mice that lack FTVII 106, although unlike mice that lack both E- and P-selectin, these mice do not develop spontaneous disease 58, 107. The importance of selectin ligands is dramatically illustrated by a rare human disease, Leukocyte Adhesion Deficiency-II (LAD-II). A fucosylation defect in these patients renders selectin ligands inactive, causing an inability of leukocytes to roll 108. These patients have significant inflammatory pathology 109. The binding of PSGL-1 to P-selectin and the functional implications thereof have been demonstrated in vivo 110, 111. PSGL-1 also binds L-selectin 112, 113, and appears to be responsible for most114 though not all 115 of the interactions between flowing and already adherent leukocytes. Although all lymphocytes express PSGL-1, this molecule does not only function as a selectin ligand, but also binds the chemokines CCL19 and CCL21116.
Other glycoproteins serve as ligands for L-selectin, including glycosylation-dependent cell adhesion molecule-1 (GlyCAM-1) 117, Nepmucin, Endomucin 28 CD34 118, and podocalyxin 119. Most L-selectin ligands expressed by high endothelial venules contain sulfated determinants that are identified using the carbohydrate-binding monoclonal antibody MECA-79 120, 121. MECA-79 binds to a sulfation-dependent epitope and can block L-selectin binding 120. This epitope overlaps with sialyl 6-sulfo Lewis X, an L-selectin recognition determinant 122. Both PNAd and sialyl 6-sulfo Lewis X are absent in mice deficient for two N-acetylglucosamine-6-O-sulfotransferases (GlcNAc6ST-1 and GlcNAc6ST-2) resulting in reduced lymphocyte homing and adhesion in HEV 122, 123. MECA-79 binding requires sulfation. Recent studies have identified the molecular structure of its epitope as an extended core 1 structure containing GlcNAc-6-SO4 on O-linked glycans 124. The enzymes HECGlcNAc-6-sulfotransferases (HEC-6ST also known as GST, L-selectin ligand sulfotrasferase, (LSST)) are responsible for post-translational modifications which are essential for L-selectin binding. HEC-6ST expression is highly restricted to HEV, and in humans and mice a sulfotransferase with restricted expression to the intestine has been identified 125. Little is known about its role in inflammation, but two recent studies have demonstrated its induction by TNF in human bronchial mucosa and lymphotoxin in an animal model of lymphoid neogenesis in the pancreas 126, 127.
Another ligand for L-selectin is mucosal addressin adhesion molecule-1 (MAdCAM-1), expressed on the surface of high endothelial venules of Peyer’s patches and mucosal lymph nodes 128, 129. MAdCAM-1 isolated from mesenteric lymph nodes of young mice supports L-selectin dependent lymphocyte rolling in a flow chamber assay 130. A role for L-selectin in lymphocyte rolling in Peyer’s patch high endothelial venules has also been demonstrated by intravital microscopy, but it is not clear whether all L-selectin ligand activity in this system is attributable to MAdCAM-1 44, 48.
For E-selectin, a 250 kDa potential ligand was precipitated from bovine γδ T cells 131, whereas on myeloid cells, PSGL-1, CD44 and E-selectin Ligand-1 (ESL-1), a molecule with homology to fibroblast growth factor receptor, are E-selectin ligands 71,72. In addition, CD43 might be a relevant ligand for E-selectin on skin-homing activated T cells 132,133.
Several selectin inhibitors have been developed and tested in preclinical models and clinical trials. Carbohydrate-based selectin inhibitors of the sialyl Lewisx type, which inhibit all three selectins at high concentrations, have unfavorable pharmacokinetics, low affinity and short half lives 134. Drug candidates within this family (e.g. OC229648, Efomycine and CY1503) have been tested in preclinical models of peritonitis, psoriasis and ischemia-reperfusion injury. Antibodies to selectins including antibodies that block more than one selectin have been developed and humanized. Protein Design Laboratories has tested a humanized version of an anti-L-selectin antibody (HuDREG-55), in patients with trauma (www.pdl.com) and in psoriasis. An antibody against both P- and E-selectin (Hu EP5C7) has been tested in a baboon stroke model, while specific antibodies against E- (i.e. CDP850) and P-selectin (CY1747) have been tested in psoriasis and preclinically in a model of ischemia-reperfusion injury. A recombinant truncated form of a PSGL-1-immunoglobulin fusion protein has shown promise as a selectin inhibitor 135 (mainly aimed at P- and L-selectin) (www.wyeth.com). This molecule is currently in clinical trials for liver and renal transplantation (http://www.ysthera.com/news/070502.html). Small-molecule inhibitors of selectins known as glycomimetics have also been developed 136. A small-molecule inhibitor of selectin function (i.e. TBC-1269, Bimosiamose), developed by Texas Biotechnology (www.tbc.com) has been tested as an inhaled formulation for pediatric asthma and in psoriasis 137. Positive results were reported from a phase II clinical trial in patients with chronic obstructive pulmonary disease (COPD (http://www.revotar.de/news_and_press.php).
Integrins are a large superfamily of heterodimeric transmembrane glycoproteins that allow attachment of cells to extracellular matrix proteins or to ligands on other cells. Integrins contain large (α) and small (β) subunits of 120–170 kDa and 90–100 kDa, respectively. Integrins contain binding sites for the divalent cations Mg2+ and Ca2+, which are necessary for their adhesive function. Distinct families of mammalian integrins form through the association of specific α subunits with different β subunits 25 (Figure 3). The families of greatest importance for leukocyte adhesion in inflammation and immunity are the β2, α4 and β7 families of adhesion molecules. Unlike the selectins, they establish stable protein-protein interactions and mediate arrest of cells on the vessel wall. However, some integrins have also been shown to participate in rolling 138. Integrins are also important signaling molecules involved in many cellular processes 139. In addition, several β1 integrins are involved in lymphocyte interaction with ECM proteins25.
Integrins containing the β2 subunit (CD18) are exclusively expressed on bone marrow-derived cells. β2 integrins undergo a conformational change upon cellular activation (inside-out signaling), which is necessary for ligand binding 139, 140. Activation is initiated by the binding of a chemokine or other chemoattractant to its heptahelical G-protein coupled receptor (GPCR) on the leukocyte surface 41, 141. Ligation of the GPCR results in integrin activation within milliseconds 139.
β2 integrins include four different heterodimers: CD11a/CD18 (LFA-1) is the predominant β2 integrin on lymphocytes and neutrophils; CD11b/CD18 (Mac-1) is expressed by monocytes, granulocytes and some NK cells; CD11c/CD18 (p150, p95) is present on monocytes and dendritic cells. CD11d/CD18 is expressed primarily by myelomonocytic cell lines and is predominantly found on macrophages and granulocytes in the red pulp of the spleen 142. In humans 143 and mice 144, null or hypomorphic mutations in the Itgb2 gene encoding the β2 (CD18) integrin subunit result in the genetic disorder leukocyte adhesion deficiency-1 (LAD-1) 143. Patients with LAD-1 have recurrent bacterial infections due to ineffective recruitment of granulocytes in response to infections 143. This spontaneous human disease dramatically illustrates the important role played by CD18 integrins in innate immunity.
LFA-1 binds to ICAM-1 and ICAM-2 on the endothelium 145, 146. Its crucial role has been demonstrated in many cellular processes, such as migration, antigen presentation, and cell proliferation. Mice homozygous for a null mutation in the gene encoding the LFA-1 α chain have, amongst other defects, a mild defect in lymphocyte homing to secondary lymphoid organs 147. Targeted blockade of the integrin αL subunit (CD11a) of LFA-1 (i.e. efalizumab) has been approved by the FDA for the treatment of chronic plaque psoriasis 7. LFA-1 shares several characteristics and functions with Mac-1. LFA-1 has a predominant role for adhesion, Mac-1 is necessary for respiratory burst and certain forms of phagocytosis148, 149. Mac-1 has no known function on lymphocytes. The other β2 integrins (i.e. αxβ2 (CD11c/CD18) and αdβ2 (CD11d/CD18) are not expressed on lymphocytes.
An important member of the α4 family of integrins for leukocyte adhesion and trafficking is very late antigen-4 (VLA-4, CD49d/CD29, α4β1)150, 151. VLA-4 binds to its ligand VCAM-1, and is chiefly responsible for lymphocyte and monocyte adhesion to vascular endothelium. Studies in autoimmune encephalitis (EAE) have demonstrated a critical role for VLA-4 for lymphocyte homing into the CNS. VCAM-1, which is expressed at very low levels in CNS microvessels, is highly induced by proinflammatory stimuli in EAE and in patients with MS 152.
α4β7 is a pivotal integrin involved in gut-homing. This integrin binds to mucosal addressin cell adhesion molecule-1 (MAdCAM-1) 128, 129 and plays a critical role in homing of lymphocytes to Peyer’s patches by supporting their binding to high endothelial venules. Both α4β1 and α4β7 can be activated to a high-avidity state 153. The β7 integrin null mouse shows reduced homing of nai–ve lymphocytes to Peyer’s patches 48, which is even further reduced in double mutant mice lacking both β7 integrin and L-selectin 31,154. Homozygous null mutations for integrins α4 and β1 lead to embryonic lethality 155. α4-null chimeric mice have normal thymus cellularity and circulating T cells at birth, yet after a few weeks the thymus becomes atrophic 156, 157. This data supports an important role for α4 integrins for the maintenance of thymocyte populations after birth. As integrin β7 deficiency does not have a comparable effect it is then likely that the deficiency for α4β1 integrin is responsible for this finding.
Integrin αEβ7 (CD103) was first identified by a monoclonal antibody that recognizes lymphocytes in intestinal tissue (i.e. human mucosal lymphocyte antigen-1, HML-1). The 175kDa alpha E chain pairs exclusively with the β7 chain also present in α4β7. αEβ7 is expressed by over 90% of intraepithelial lymphocytes and subsets of dendritic cells (DC). Its role in lymphocyte traffic remains unclear, however recently it has emerged as an important marker of CD4+ and CD8+ regulatory T cells as well as for DC subsets responsible for imprinting gut tropism to T cells 158–160.
Members of the immunoglobulin superfamily share structural and genetic features with immunoglobulin molecules, and each contains at least one immunoglobulin domain (Figure 4). An immunoglobulin domain consists of two β-pleated sheets held together by a disulfide bond. Some molecules of the immunoglobulin superfamily are expressed on the vascular endothelium, where they act as counter-receptors for leukocyte integrins. Three of these immunoglobulins are of particular importance for leukocyte adhesion in inflammatory bowel disease. Intercellular adhesion molecule-1 (ICAM-1) or CD54, Vascular Cell Adhesion Molecule-1 (VCAM-1) or CD106, and Mucosal Addressin Cell Adhesion Molecule-1 (MAdCAM-1) all play important roles in the dysregulated trafficking of leukocytes that occurs during IBD.
ICAM-1 is a homodimeric molecule constitutively expressed on a variety of cell types including resting endothelial cells. Its expression on endothelial cells is up-regulated by stimulation with cytokines 161. ICAM-1 is one of the principal ligands for the β2-integrins LFA-1 (CD11a/CD18) and Mac-1 (CD11b/CD18) 162, although in the context of transmigration it seems that CD11a predominantly binds to ICAM-1, whereas CD11b binds to multiple ligands 163. Under resting conditions, the affinities of LFA-1 and Mac-1 for ICAM-1 are low 139, but activation of leukocytes causes conformational changes in LFA-1 and Mac-1 that greatly increases their affinities for ICAM-1. Although ICAM-1 cannot support rolling on its own, isolated LFA-1 I domain supports rolling on ICAM-1164. On neutrophils, E-selectin engagement can induce partially activated LFA-1 to mediate LFA-1-dependent rolling on ICAM-1 69. An antisense oligonucleotide that blocks the translation of ICAM-1 mRNA (i.e. Alicaforsen) has been evaluated in IBD, yet its efficacy has been limited 165.
ICAM-2, like ICAM-1 is a homodimeric molecule 146 Different from ICAM-1 which has five Ig-like domains, ICAM-2 has only two (figure 4). It is expressed constitutively on endothelial cells, including HEV, on certain leukocytes and on megakaryocytes and platelets 166. The expression of ICAM-2, different from ICAM-1 does not appear to be upregulated by inflammatory stimuli, thus it is thought to be involved mostly during constitutive trafficking. ICAM-2-deficient mice have normal leukocyte homing and maturation and do not suffer from spontaneous inflammatory conditions 167.
VCAM-1(CD106) is a type-1 transmembrane protein that contains either six or seven immunoglobulin domains 168, 169 and is expressed on both large and small vessels after stimulation by cytokines 170–172. Expression on resting endothelial cells is very low or absent. On cultured endothelial cells, sustained expression of VCAM-1 lasts at least 24 hours. VCAM-1 may also be found on epithelial cells, dendritic cells, Kupffer cells and on smooth muscle cells within atherosclerotic lesions 172–174. VCAM-1 is primarily an endothelial ligand for very late antigen-4 (VLA-4 or α4β1) and promotes the adhesion of lymphocytes, monocytes, eosinophils, and basophils. VCAM-1-deficiency results in embryonic lethality. A conditional VCAM-1-mutant mouse that lacks VCAM-1 on endothelial cells showed a lymphocyte homing defect to bone marrow and impaired humoral responses 175, 176. Mice hypomorphic for VCAM-1 expression were protected from atherosclerosis
MAdCAM-1 contains two immunoglobulin domains and a mucin domain 128, and is the main ligand for α4β7 integrin 129. It is expressed at high levels by HEV of Peyer’s patches and in lamina propria venules128. The frequency of MAdCAM-1 positive vessels is increased in human IBD 177 and in animal models of colitis 178. It is also present in the inflamed pancreas 179 and may be inducible in other locations by proinflammatory cytokines such as TNF-α and IL-1 181.
In addition to binding to α4β7 integrin, MAdCAM-1 isolated from young mice also supports rolling through L-selectin 130. This double role of MAdCAM-1 may underlie the synergistic effects of L-selectin and α4β7 integrin in lymphocyte homing to gut-associated lymphoid tissue31, 44. MAdCAM-1 is up-regulated in the IL-10 knockout mouse, which develops spontaneous colitis 178, and in intestinal lesions of patients with IBD 177 supporting an important role in chronic inflammatory conditions.
As there are excellent reviews in the topic 182 we will only briefly introduce models in which the anti-adhesion strategy has been evaluated with the goal of attenuating inflammation. Three broad categories of colitic models have been used to explore this strategy. A spontaneous model (i.e., the cotton top tamarin model), four chemically induced models (i.e. acetic acid-, 2,4,6-trinitrobenzenesulfonic acid (TNBS), dextran sulfate sodium (DSS) and formalin-induced colitic models) and two others that develop colitis due to deficient regulatory responses (i.e. CD45Rbhigh CD4 lymphocyte transfer, IL10-deficient mice).
A wasting syndrome was reported in cotton-top tamarins, an endangered South American marmoset, in 1976. An association between colitis and colonic adenocarcinoma in CTT was reported in 1981 183 and their chronic colitis formally characterized by Madara in 1985 184. Similar to patients with UC, CTT develop recurrent spontaneous flares of colitis with distortion of the mucosal architecture, suggesting chronic dysregulated immune responses. No identifiable pathogens to which the histologic findings may be attributed were identified. Like human IBD, colitis in CTT responds to treatment with the 5-ASA compound sulfasalazine 184.
The administration of diverse chemical agents such as: sulfated polysaccharides (e.g. carrageenan, amylopectin sulfate, dextran sulfate (DSS)), rectal instillation of diluted acetic acid, TNBS, and intravenous injection of immune complexes followed by chemical irritation of the colon, all result in the development of colitis. A common shared mechanism for all of these models is the disruption of the epithelial barrier with increased exposure of intestinal microflora to the intestinal immune system. The observation that administration of such distinct chemical irritants result in similar histologic findings suggests that the colitis is rather an unspecific, stereotypic response to injury (125). The original disturbance does not determine the specific nature of the resulting lesions but likely serves as a trigger of common immunologic responses. The inflammatory response in all of these models is self-limited and resolves spontaneously upon removal of the injurious agent. Therefore these models do not accurately replicate the specific chronic pathogenetic mechanisms that mediate disease maintenance in human ulcerative colitis or colonic CD.
Intrarectal administration of acetic acid (AA) results in barrier dysfunction and induces diffuse colitis in a dose-dependent manner. At the tissue level, AA induces diffuse ulcers of the distal colon with alterations in crypt architecture with occasional transmural inflammatory infiltrates that vary according to the severity of the process 185. As an acute colitis, it is better suited to address the early pathogenetic mechanisms of IBD lesions.
Intracolonic co-administration of 50% ethanol, which breaches the mucosal barrier along with the hapten TNBS induces a sustained colitis that persists up to 8 weeks. The inflammatory infiltrate is mixed, composed of neutrophils, macrophages and lymphocytes. Granulomas are present in over 50% of animals. Disruption of the barrier by ethanol and TNBS are both required for the induction of colitis and administration of one or the other fails to induce disease 186. As the inflammatory process last for up to 2 months additional information may be obtained beyond the very early developmental stages of colitis.
DSS administered orally is a reproducible and technically simple way to induce colitis in rodents. Differential susceptibility to DSS has been reported in several mouse strains, yet the agent induces some degree of colitis in all mice. DSS disrupts the mucosal barrier and induces a colitic process that is more dependent on innate immune responses, as it develops even in the absence of T, B and NK cells 187. Yet in mice with an intact immune system there is eventual activation of T cells responses, which is superimposed on the primary innate responses induced by the chemical agent.
This model is part of an early family of IBD animal models induced by deposition of immune-complexes in the colon, triggered by the mucosal injury that results after administration of dilute formalin. The lesion is mostly localized to the mucosa with damage to the epithelium, loss of the goblet cells and a predominant acute inflammatory infiltrate 188. Although some of these features are shared with human IBD, the resultant inflammatory process is transient and resolves spontaneously.
Powrie et al described in 1994 that transfer of CD45RBhigh cells into immunodeficient recombinase activating gene (RAG)-deficient or severe combined immunodeficient (SCID) mice results in the development of colitis 189. Most nai–ve T cells are CD45RBhigh whereas regulatory T cells are not within this subset. Co-transfer of the CD45RBlow cells, which contain regulatory T cell subsets, prevents and cures colitis. The bacterial flora remains an important factor in this model, as transfer of CD45RBhigh cells into gnotobiotic recipients results in attenuated disease 190. Different from previously discussed models, adaptive immune responses are fundamental in CD45RBhigh colitis, enabling the assessment of the role of adhesion molecules expressed by pathogenic T cells and the contribution of specific T cells subsets to the disease process.
IL-10-deficient mice develop chronic transmural colitis 191. The disease is characterized by T-helper (Th)1 responses during early stages and by Th2 responses during late stages. Lymphocyte development and antibody responses are normal. Histological findings in the colon include mucosal hyperplasia, mixed inflammatory infiltrates and aberrant expression of major histocompatibility complex class II molecules on epithelia. The bacterial flora remains an important player, as mice kept under specific pathogen-free conditions develop only local inflammation within the proximal colon. Thus IBD in this model results from dysregulated immune responses likely triggered by bacterial antigens. IL-10 is probably an essential regulatory cytokine for the maintenance of intestinal homeostasis. The inflammatory process is chronic and therefore suited to address mechanisms related to the chronic stages of IBD.
Interference with lymphocyte recirculation to the intestine by targeting specific adhesion molecules and chemokine receptors is an attractive strategy for drug development in IBD 19. This approach has already resulted in the development of several agents that have reached clinical use, e.g., natalizumab (Biogen/Elan Pharmaceuticals, www.tysabri.com) which targets the α4 integrin, shared by the α4β1 and α4β7 integrins. MLN02 (Millenium Pharmaceuticals, www.mlnm.com/rd/inflammation/candidates/mln02.asp) targets specifically the α4β7 integrin. Most recently Traficet EN (Chemocentryx Corporation), which targets the small-intestinal-specific chemokine receptor CCR9 has been tested in IBD. The therapeutic success of natalizumab has been overshadowed by the appearance of serious complications 21. Numerous studies have targeted adhesion molecules in animal models of IBD (Table 1), yet much more remains to be learned. A greater knowledge of the basic mechanisms that underlie dysregulated chronic inflammatory trafficking, as well as the differences between trafficking to the small or large intestine will be required for the further development of the anti-adhesion strategy in IBD.
Given the critical role of the selectins in mediating the initial steps of leukocyte recruitment to inflammatory sites, they would appear to be obvious potential therapeutic targets in IBD. Yet few reports attempted to target these molecules in animal models of IBD or reported disease attenuation by targeting a selectin. In the earliest report by Podolsky et al., colitic Cotton-top tamarins, which suffer from a spontaneous relapsing colitis reminiscent of UC, were administered two different antibodies against E-selectin. Despite close attention to dosing, half-life and tissue levels, these antibodies had no significant attenuating effect on the severity of the colitis 192. In a subsequent report, colitis was induced in P-selectin-deficient, and P- and E-selectin double-deficient mice using acetic acid or TNBS. Damage scores were not different in the P-selectin-deficient mice when compared to wild-type mice, whereas the double-deficient mice showed enhanced leukocyte recruitment and more severe disease 193. More recently, Sans and colleagues reported a series of experiments that explored the role of the selectins in the TNBS-induced rat model of colitis 194. Antibodies against E-, L- and P-selectin decreased rolling fluxes at certain time points, while the anti-P-selectin monoclonal antibody also decreased the number of adherent leukocytes. Yet no attenuating effect on the severity of the colitis was observed, even with the anti-P-selectin antibody. By contrast, the following year the same group assessed the effect of an anti-P-selectin antibody and P-selectin-deficiency in dextran sulfate sodium (DSS)-induced colitis in mice, and observed an attenuating effect of the severity of the colitis for both 195. These authors propose that this discrepancy may result from inter-species variability, different experimental approaches or distinct pathophysiological mechanisms related to the different ways of inducing colitis in the different models. In agreement with this study, anti-P-selectin antibodies interfered with leukocyte rolling and adhesion in DSS-treated mice 196 and attenuated TNBS-colitis in rats 197.
We are aware of only one study that targeted selectin ligands in a mouse model of colitis. Rijcken and colleagues treated mice with an antibody against PSGL-1 (2PH1), and studied its effects on rolling and adhesion through intravital microscopy. In addition they assessed its effect on colonic inflammation by means of myeloperoxidase assays, a surrogate marker for neutropil infiltration and histological assessment of the severity of colitis 198. Blockade of PSGL-1 with 2PH1 reduced the number of rolling leukocytes, increased rolling velocities and attenuated colitis, supporting an important role for PSGL-1 for leukocyte recruitment in this model. Recently PSGL-1 deficient CD4+ T cells were shown to induce colitis in RAG−/− mice indistinguishable to that of WT cells 199, whereas the expression of PSGL-1 in colonic endothelium was induced by DSS administration 92. The functional role of PSGL-1 in colitis remains unclear.
The first report of an attenuating effect by targeting an integrin in an animal model of IBD appeared in 1992 when Wallace and colleagues administered an anti-CD18 mAb to rabbits before and after induction of colitis by TNBS. Both neutrophilic infiltration and epithelial injury improved with this strategy, regardless of whether the antibody was administered prior to or 12h after TNBS administration 200. This result was confirmed in TNBS colitis in rats using 2 antibodies against CD11b/CD18, administered 2 hrs prior to and 3 days after induction of colitis 201. Immune complex colitis induced by formalin administration in rabbits also responded to an antagonist of CD11b/CD18 202. Recently, Grisham and coworkers have reported attenuated colitis induced by CD4+ T cells isolated from LFA-1 (CD11a/CD18)-deficient (Itgal−/−) mice, and attenuated colitis when DSS was administered to CD18 (Itgb2−/−) and CD11a null mice 203,204. An antibody against complement receptor 3, which serves as a ligand for αMβ2 (Mac-1) and αxβ2 integrins ameliorated TNBS- and CD45RBhigh-induced colitides 205.
Targeted blockade of the α4 integrins in an animal model of IBD was first described by Podolsky and colleagues in cotton-top tamarins. An anti-human α4 integrin (HP1/2) known to interfere with the interaction of VLA-4 with VCAM-1 or a saline placebo control were administered daily for 7 days to colitic animals. Colonic biopsies obtained before and after treatment demonstrated that HP1/2 attenuated acute activity but had no effect on the chronic inflammatory inflitrates 192. An antibody that targeted a combinatorial epitope on the α4β7 integrin was then successfully used to attenuate colitis in cotton-top tamarins. Following a similar protocol to that of Podolsky et al, cotton-top tamarins were randomized to receive daily injections for 7 days of anti-α4β7 or an isotype control, with acquisition of colonic biopsies at three time points and close monitoring of a clinical effect on stool consistency. Significant histological improvement was observed. Clinically cotton-top tamarins showed improvement in stool consistency during the first 24h after administration of the antibody. This therapeutic response persisted for over 2 weeks after completion of the therapeutic regimen 206. Monoclonal antibodies against α4β7 integrin and MAdCAM-1 blocked lymphocyte recruitment and reduced the severity of colonic inflammation in the CD45Rbhigh transfer model of colitis, one of the best studied IBD murine models 207. Similarly, anti- MAdCAM-1 antibodies significantly reduced colonic injury and decreased the number of β7 integrin-positive cells in the colonic mucosa in mice with DSS-induced colitis 208, 209.
ICAM-1 and VCAM-1 mRNA levels are increased in inflamed intestine of both TNBS colitic mice 194, 210 and IL-10-deficient mice 211; in the latter mRNA levels for MAdCAM-1 were also increased. Similarly, in the CD45Rbhigh colitic model, mRNA levels for the three immunoglobulins increased 212. MAdCAM-1 is up-regulated in chronically inflamed small and large intestine of patients with UC and CD 177, 179. Antibodies against MAdCAM-1 ameliorate DSS-induced colitis 213.
In TNBS-induced colitis, ICAM-1 was slightly upregulated and VCAM-1 was upregulated at least 8-fold, as measured by an intravenous radiolabeled antibody strategy useful to assess adhesion molecule expression on the endothelial surface 214. Intravital microscopy of colonic venules revealed increased leukocyte rolling and adhesion. An antibody to VCAM-1 was able to relieve some of the symptoms of disease and also reduced leukocyte adhesion. This is probably the strongest mechanistic evidence suggesting an involvement of VCAM-1 in intestinal inflammation, although TNBS-induced colitis differs significantly from human IBD. An anti-ICAM-1 antibody also attenuated TNBS colitis in rats 197. Colonization by Lactobacillus casei attenuated leukocyte recruitment to the colon, potentially by preventing the upregulation of ICAM-1 induced by TNBS 215. Evidence for increased adhesion molecule expression in humans comes from a study based on cultured human intestinal endothelial cells isolated from IBD patients, which support increased adhesion of monocytic cells (U937) and neutrophils compared with controls 216, 217, although a molecular mechanism was not identified. In therapeutic studies, antibodies to ICAM-1 protected rats from acute colitis induced by acetic acid and DSS 218 ,219, 220. In addition, an ICAM-1 antisense oligonucleotide both prevented and reversed DSS colitis in mice 221 and ICAM-1-deficient mice showed decreased mortality and attenuated colitis induced by DSS 222. The endothelial molecule appears to be more important than leukocyte ICAM-1, as CD4+ T cells isolated from ICAM-1-deficient mice induced colitis in Rag1−/− or Rag2−/− mice indistinctly from that induced by WT mice 199.
Although there is a body of literature describing the physiology of lymphocyte trafficking, our knowledge of leukocyte recruitment during conditions of chronic inflammation as seen in patients with IBD is limited. This is especially true in inflammatory cell trafficking to the small intestine, in part due to the lack of animal models that develop disease in the terminal ileum. However, two murine models that develop chronic ileitis similar to that characteristic of CD have now been described, i.e., SAMP1/Yit and TNFΔARE mice 223, 224. This has facilitated investigation of the adhesion pathways that are involved in lymphocyte trafficking specifically to the small intestine under conditions of chronic inflammation. Both the SAMP1/YitFc and TNFΔARE models of chronic ileitis show many of the features that are characteristic of CD 182, 225, including the chronic and unrelenting course, the critical role of T lymphocytes and the specific localization of the inflammatory process to the small intestine 226.
In 1998, Matsumoto et al. described a mouse strain that develops chronic inflammation predominantly in their terminal ilea and truly recapitulated many of the histopathological findings of CD in humans 224. Unlike all the previously described models of IBD, intestinal inflammation develops spontaneously in the small intestine, in the absence of chemical, immunologic or genetic manipulation. These mice were derived from the SAMP1 strain of senescent-accelerated mice (SAM), which originated from AKR inbred mice 227. SAMP1/Yit mice represent a sub-line of the SAMP1 strain, uniquely identified by the development of spontaneous skin lesions that correlate with intestinal inflammation 224. After selective inbreeding for the presence of skin lesions, followed by transfer from conventional to specific pathogen-free conditions, the early senescence phenotype was lost and the presence of discontinuous chronic ileitis was first described 224.
After over 30 generations of brother-sister matings at the University of Virginia, and the appearance of several phenotypic differences not present in the donor colony, a sub-strain was designated SAMP1/YitFc 228, 229. Similarly to the original strain, these mice have 100% disease penetrance, and exhibit patchy, transmural inflammation that is primarily localized to the terminal ileum, with formation of granulomata, as well as dense infiltration of the terminal ileal lamina propria by lymphocytes, neutrophils and macrophages. In addition these mice develop extraintestinal manifestations that target many of the organs also affected by human IBD. These include pyoderma-like skin lesions, ocular involvement and perianal disease, which is similar to the perianal manifestations observed in humans with CD, including fissures, rectal prolapse, fistulae and abscess formation 229.
Immunologically, the ileitis is characterized by increased production of IFN-γ and TNFα from lymphocytes by 4 weeks, predating any histological evidence of inflammation 228, 229. The role of Th17 cells in the pathogenesis of the disease remains to be tested. The disease is mediated by lymphocytes that infiltrate the lamina propria and mesenteric lymph nodes, display an activated phenotype, and adoptively transfer disease to SCID mice 228. The adoptive transfer of SAMP1/YitFc CD4+T cells predominantly induces ileitis, different from that transferred by CD45Rbhigh T cells that cause colitis, which is thought to reflect the inherent ability of SAMP1/YitFc T cells to recirculate preferentially to the small intestine. The disease in these mice also responds to therapeutics currently used in the treatment of CD, including corticosteroids, and monoclonal antibodies against TNFα230 and α4 integrins, all of which ameliorate disease severity 231.
A second model of chronic ileitis was described by Kontoyiannis et al. in 1998 223. The TNFΔARE model develops both a predominantly Crohn’s-like chronic ileitis similar to that of SAMP/Yit mice, and deforming rheumatoid arthritis-like joint disease in both heterozygous and homozygous mice. Like SAMP1/Yit, the disease in TNFΔARE mice responds to immunoblockade of TNF-α, supporting the relevance of this model to the human diseases. TNFΔARE mice were developed by targeted deletion of 69 base pairs of AU-rich elements in the 3′ UTR region of the gene that encodes TNF-α. This deletion results in increased stability of the mRNA for TNF-α, increased synthesis of TNF-α protein and elevated levels of systemic TNF-α 223. Interestingly, the increased levels of TNF-α do not result in uniform pan-enteritis, but predominantly in ileitis. Homozygous TNFΔARE+/+ mice develop severe ileitis and arthritis, and die at an early age. Heterozygous mice (+/−) develop disease by 8 weeks-of-age that worsens progressively, yet they reach sexual maturity and procreate. A related strain (i.e. B6.129S-Tnftm2Gkl/Jarn, MGI: 3720980) on the BL6 background has been generated by over 20 generations of continuous backcrosses between TNFΔARE mice on mixed genetic background (i.e. C57BL6 and 129S6), generated as previously described 223 to C57BL6/J mice. The C57BL6/J genetic background did not alter the localization, time course or severity of the intestinal inflammation. The defined C57BL/6 genetic background greatly facilitates immunologic studies that would be difficult or impossible on the original strain with a mixed genetic background.
Dissecting the lymphocyte-trafficking pathways in chronic murine ileitis may help optimize the current therapeutics in clinical use in IBD, as well as to identify new potential therapeutic targets. To that effect we have targeted many of the adhesion molecules and chemokines using these two models of ileitis.
The role played by L-selectin in nai–ve T-cell trafficking has been relatively well characterized 232, whereas the role it plays during induction or maintenance of chronic inflammation is incompletely understood.
Expression of L-selectin is often considered to correlate with antigenic inexperience and to be indicative of the nai–ve state. Although nai–ve T cells indeed express L-selectin, it is also expressed in central memory T cells 50 at other times. L-selectin is re-expressed after lymphocyte activation in in vitro systems, and subpopulations of memory and effector lymphocytes express L-selectin 33, 50, 233. Therefore, L-selectin may continue to play a role in the recruitment of antigen-experienced subpopulations of lymphocytes important for the pathogenesis of chronic inflammatory diseases, where L-selectin ligands (expressed in PNAd+ and MAdCAM-1+ vessels) have been demonstrated within effector sites. PNAd+ vessels are consistently found within inducible lymphoid aggregates or follicles (ILF), also known as tertiary lymphoid organs (TLO) 124, 127. Thus, recruitment of L-selectin-expressing T cells likely occurs at these sites, and interference with these pathways may be of therapeutic value 122, 123.
The models of chronic ileitis have allowed us to assess for the first time whether the trafficking pathways to the small and large intestine were shared or distinct from those to other intestinal segments under conditions of dysregulated inflammation. Systematic targeting of the different families of adhesion molecules with neutralizing antibodies against single molecules or in combination was performed in an adoptive transfer model in which CD4+ T cells from SAMP1/YitFc mice are adoptively transferred into MHC-matched SCID mice. The predominant disease in the recipients is ileitis, not colitis. Single (data not shown) or combined blockade of P-, E- and/or L-selectin failed to attenuate the ileitis in SCID mice or in SAMP1/YitFc mice with established ileitis 91 (Figure 5). However, administration of the anti-L-selectin antibody MEL-14 prior to the development of inflammation had a protective effect, suggesting a requirement for L-selectin during the initiation stages of the disease (Rivera-Nieves, J, unpublished results).
The selectins possess a C-type lectin domain at their amino terminus that interacts with carbohydrate determinants expressed by ligands expressed on high endothelial venules (HEV) and PSGL-1 expressed by adherent leukocytes 26. Indeed, all of the L-selectin ligands identified to date (i.e., CD34, GlyCAM-1, sgp200, PSGL-1, endomucin, MAdCAM-1) contain carbohydrate moieties 134. The carbohydrate post-translational modifications found on HEVs that are necessary for L-selectin recognition include sialylation, fucosylation and carbohydrate sulfation 124. Chronic inflammatory processes induce the appearance of HEV-like vessels in numerous human diseases and animal models of chronic inflammation 234. A likely role of these inflammation-induced structures is to support recruitment of L-selectin-expressinglymphocytes directly into effector sites, where during chronic inflammation numerous tertiary lymphoid organs appear 127.
The MECA-79 monoclonal antibody stains HEVs and blocks L-selectin-dependent lymphocyte adhesion 235. Glycoproteins carrying the PNAd epitope are recognized as relevant L-selectin ligands, as no additive effects on lymphocyte homing was observed after simultaneous blockade of PNAd and L-selectin. The MECA-79 epitope is found in PP, MLN, PLN but not within the intestinal LP, except in chronic inflammation 236. Yet this pattern is not restricted to the intestine and PNAd reactivity has been reported in over 20 immune-mediated animal models and human conditions, including asthma, diabetes, psoriasis, RA, Hashimoto’s thyroiditis, Grave’s disease, cutaneous lymphoma and allograft rejection 124.
Immunohistochemical studies in SCID mice before and after CD4+ T cell adoptive transfer from donor SAMP1/YitFc mice demonstrated that the PNAd epitope is expressed after the development of inflammation in these mice. However, single or combined blockade of MAdCAM-1 (MECA-367) and PNAd (MECA-79) provided no therapeutic benefit compared to mice treated with isotype-matched control antibodies 91.
Inoue and colleagues described for the first time the therapeutic effect of neutralization of PSGL-1 with the monoclonal antibody 2PH1 in SAMP1/Yit mice 237. Administration of 2PH1 to mice with established disease significantly reduced infiltration of the LP by CD4, CD8 and CD68+ subsets, and the villus height was partially restored, although no effect on the hypertrophy of the muscularis (a chronic structural change) was noted. Their emphasis was on the role of PSGL-1 on monocyte recruitment to the intestine; however when bone marrow chimeric SAMP1/Yit mice that lacked PSGL-1 in their hematopoietic compartment were generated, their disease was equal in severity to that of PSGL-1-sufficient SAMP1/Yit mice suggesting that PSGL-1 on monocytes was not the critical molecule. The small intestinal microvasculature expressed PSGL-1 and a therapeutic effect of blockade of PSGL-1 was demonstrated in SAMP mice using a different antibody (i.e. 4RA10) 91. It is likely that the therapeutic effect of this strategy is due to its interference with multiple recruitment pathways (Figure 6). Interfering with PSGL-1 would block rolling on P-selectin expressing endothelial cells, rolling on platelets, platelet-leukocyte interactions and endothelial-platelet interactions.
The importance of the α4 integrins in lymphocyte trafficking is well established in the literature 4. α4β7 and α4β1 integrins are crucial for the dichotomy in migration to gut or to non-intestinal sites in physiological recruitment. In this setting,α4β7-expressing cells migrate preferentially to the gut, whereas lymphocytes expressing the α4β1 integrin (VLA-4) preferentially traffic to non-intestinal sites235, 238. Expression of these integrins under physiological conditions is reciprocal, and gut-homing α4β7+ cells tend to be α4β1neg or low and vice versa 235, 238. Recent clinical trials with a humanized mAb to α4 integrins in patients with Crohn’s disease have suggested that α4 integrins may be relevant therapeutic targets 18, 20. Yet while targeting both α4 integrins (α4β7, α4β1) through its shared α4 moiety is effective in CD, a more specific mAb that targeted α4β7 specifically was effective in UC but not in CD 239. This data supports divergent trafficking pathways between small and large bowels. Further understanding of potential mechanisms of induction of small intestinal specific homing have recently emerged with the discovery that retinoic acid is essential for the imprinting of T cells with a small intestinal homing phenotype 240,241, 242.
Expansion of the β7 integrin-expressing CD4+T cell population and of MAdCAM-1 after T cell transfer suggests a role for these molecules in the pathogenesis of chronic ileitis. Yet, neutralizing monoclonal antibodies against the β7 integrin subunit (FIB-504) orα4β7 integrin (DATK-32) were ineffective in reducing the severity of acute or chronic infiltrates, compared to isotype antibody-treated SCID mice with chronic ileitis88. In agreement a β7-deficient SAMP1/Yit substrain recently generated develops ileitis indistinguishable from that of β7-sufficient SAMP1/Yit mice (Gorfu, G. Rivera-Nieves, J, McDuffie, M., Pizarro TT, Cominelli, F and Ley K, unpublished observations). To attenuate ileitis, blockade of at least two integrins might be required, for example using an anti-α4 integrin antibody, which interferes with both α4β7 andα4β1 integrins 88.
The expression of ICAM-1 and VCAM-1 was increased in intestinal microvessels of SAMP1/YitFc mice and the related SCID adoptive transfer model mice in the presence of inflammation, consistent with their induction under conditions of chronic inflammation. In addition, combined blockade of ICAM-1 and VCAM-1 was able to attenuate ileitis in adoptively transferred SCID mice, whereas blockade of either alone was not sufficient 231. This data supports the hypothesis that the trafficking pathways under conditions of chronic inflammation are highly redundant and resistant to therapeutic intervention. This is further supported by the complete lack of efficacy by the same combined intervention in SAMP1/Yit mice, in which the inflammatory process develops over many months with the likely acquisition of highly redundant trafficking pathways 231.
In a separate set of experiments, neutralization of MAdCAM-1 was found to be ineffective at attenuating ileitis in SCID mice, despite clear induction of this molecule with the development of chronic inflammation. It is likely that T cells continue to home to the intestine using VCAM-1/α4β1 and ICAM-1/LFA-1 after the MAdCAM-1/α4β7 pathway has been neutralized. This is supported by the efficacy of the simultaneous blockade of α4 integrins (clone PS2) and ICAM-1 231, which targets multiple mechanisms of arrest 88.
During the past decade, the success of anti-TNF-α strategies has revolutionized the treatment of IBD 243. Yet, only about 70% of patients respond to this therapy, driving the continued search for other anti-inflammatory strategies for the treatment of the remaining 30% of patients who do not respond and for those that respond initially but eventually fail. Therefore, alternative biological therapies that target other pathways of the chronic inflammatory cascade must be evaluated in CD 14, 243. Interference with leukocyte recirculation to the intestine by targeting specific adhesion molecules has resulted in the development of agents that have advanced into clinical use, e.g., Natalizumab (Biogen/Elan Pharmaceuticals, www.tysabri.com) and MLN02 (Millenium Pharmaceuticals, www.mlnm.com/rd/inflammation/candidates/mln02.asp), which has been evaluated in CD and UC 239.
A pivotal role for α4β7 integrin/MAdCAM-1 in lymphocyte trafficking to the gut has become dogma under physiological conditions 4, 235, fueling the evolution of these anti-α4 strategies in IBD. The evidence for a similar non-redundant role in chronic gut inflammation is less clear. Antibody-blocking studies in animals and humans suggest efficacy 192, 207–209, but specificity is unclear. Despite the rapid progression from the bench to bedside of the anti-α4 therapeutic strategy18, 192, much remains to be learned regarding its fundamental mechanism of action. Our data in the ileitis models support thehypothesis that compensatory mechanisms allow continued trafficking after neutralization of theα4β7 integrin/MAdCAM-1 pathway that is pivotal for physiologic and pathologic traffic to the colon. Further dissection of the adhesive pathways in chronic ileitis may help to optimize current therapies already in clinical use for CD 18, minimize risks and potentially to further understand the mechanisms of inflammatory lymphocyte recruitment to other organs as well.
Recent studies have shown that early blockade of CCL25 using a small molecule antagonist (CCX282, Traficet-EN, Chemocentryx Corp.) effectively prevented ileitis in TNFΔARE mice, confirming a critical role for this molecule during the induction phase of chronic ileitis (Zheng Wei, Chemocentryx Corp., personal communication). However when we targeted this molecule in a different model of ileitis (SAMP1/Yit), the strategy was only efficacious during early disease 244. Targeting this and other chemokines alone or in combination with anti-adhesion strategies may be promising in subsets of patients with IBD. Small molecule approaches may be particularly usefulas adjunctive therapies.
Physiological leukocyte recruitment is a highly regulated process with a limited number of decision points along every step of the adhesion cascade 4, 23, 24. Expression of specific combinations of adhesion molecules on lymphocyte subpopulations determines their capacity to reach specific tissues, where cognate endothelial ligands are often restrictively expressed (e.g. MAdCAM-1: gut vs. VCAM-1: elsewhere). This results in orderly recruitment that begins and terminates physiological inflammatory responses. Under conditions of chronic inflammation, continuous dysregulated production of pro-inflammatory cytokines results in inappropriately increased adhesion molecule expression (e.g. MAdCAM-1), as well as aberrant expression of molecules not normally expressed by specific lymphocyte subpopulations and tissues, thus increasing the chances of lymphocytes contacting appropriate endothelial ligands. Therefore, when one adhesion molecule is blocked, others compensate for its deficiency. These redundant adhesive pathways favor adhesiveness over tissue specificity, resulting in failure of termination of the inflammatory response and perpetuation of dysregulated leukocyte recruitment. Further dissection of redundant adhesive pathways in chronic ileitis may help to optimize therapies already in clinical use and potentially develop new strategies for the treatment of CD that in two thirds of patients involves the small intestine.