Roles of integrin activation in eosinophil function and the eosinophilic inflammation of asthma

Abstract

Eosinophilic inflammation is a characteristic feature of asthma. Integrins are highly versatile cellular receptors that regulate extravasation of eosinophils from the postcapillary segment of the bronchial circulation to the airway wall and airspace. Such movement into the asthmatic lung is described as a sequential, multistep paradigm, whereby integrins on circulating eosinophils become activated, eosinophils tether in flow and roll on bronchial endothelial cells, integrins on rolling eosinophils become further activated as a result of exposure to cytokines, eosinophils arrest firmly to adhesive ligands on activated endothelium, and eosinophils transmigrate to the airway in response to chemoattractants. Eosinophils express seven integrin heterodimeric adhesion molecules: alpha4beta1 (CD49d/29), alpha6beta1 (CD49f/29), alphaMbeta2 (CD11b/18), alphaLbeta2 (CD11a/18), alphaXbeta2 (CD11c/18), alphaDbeta2 (CD11d/18), and alpha4beta7 (CD49d/beta7). The role of these integrins in eosinophil recruitment has been elucidated by major advances in the understanding of integrin structure, integrin function, and modulators of integrins. Such findings have been facilitated by cellular experiments of eosinophils in vitro, studies of allergic asthma in humans and animal models in vivo, and crystal structures of integrins. Here, we elaborate on how integrins cooperate to mediate eosinophil movement to the asthmatic airway. Antagonists that target integrins or the effectors that regulate integrins of eosinophils represent potentially promising therapies in the treatment of asthma.

INTRODUCTION

The eosinophil is a highly motile white blood cell containing distinctive cytoplasmic granules [1, 2]. First described in 1879 by Paul Ehrlich, eosinophils have been studied extensively as potent defenders against parasitic helminths. The cuticles of helminths are destroyed by granule proteins and superoxide radicals released or generated by eosinophils [3, 4]. The view of eosinophils as benign immune sentinels and effectors has become complicated over the past few decades by recognition of the possible deleterious role of eosinophils in immune hypersensitivity syndromes, including asthma, dermatitis, and rhinitis [2, 5–9].

The prevalence of asthma worldwide is increasing for reasons that are unclear [10–12]. Recruitment of eosinophils to the airway is believed to exacerbate asthma and contribute to the chronic character of asthma [13, 14]. Thus, the study of how eosinophils traffic from blood to the airway is of considerable importance. Integrins, which are among the most versatile of known cellular receptors [15, 16], have generated particular interest as likely determinants of how eosinophils arrest in the postcapillary venules of the bronchi and mediate extravasation of eosinophils from blood to the airway wall and airspace [17]. Specifically, integrins have been studied in relationship to rolling and arrest of eosinophils on endothelium, migration through endothelium and the underlying basement membrane, and traversing of bronchial epithelium into the airway lumen [18, 19]. It is important to note that this multi-step paradigm of eosinophil and granulocyte diapedesis applies to postcapillary vessels of the bronchial circulation; that movement to the parenchymal tissue of the lung is little understood [20].

As depicted in Figure 1, purified human blood eosinophils express seven integrin heterodimers: α4β1 (CD49d/29), α6β1 (CD49f/29), αMβ2 (CD11b/18), αLβ2 (CD11a/18), αXβ2 (CD11c/18), αDβ2 (CD11d/18), and α4β7 (CD49d/β7) [21–23]. Each type of heterodimer interacts with its own set of ligands, which may be deposited in extracellular matrix or a counter-receptor on other cells. Understanding functions of the integrin complement on a given cell type is complicated by the fact that each integrin may be present in several conformational states, have varying levels of expression, and cluster in the plasma membrane [24, 25]. The movement of eosinophils into the asthmatic lung, therefore, can be expected to involve a complex interplay of integrin receptors in different states of activation interacting with a diverse set of ligands on bronchial endothelium and cells within tissue.

Figure 1. Integrins of eosinophils.

Schematic of the seven heterodimeric integrin adhesion receptors expressed by eosinophils. Functions and ligands assigned to integrins have been deduced in various assays employing eosinophils. Asterisks represent subunits that contain the I (Insert)-domain. Subunits that are underlined contain the I-like domain. ICAM, intercellular adhesion molecule; MAdCAM-1, mucosal addressin cell adhesion molecule-1; FG, fibrinogen; LN, laminin; VCAM, vascular cell adhesion molecule; VN, vitronectin.

Although less is known about integrins of the eosinophil in comparison to integrins of lymphocytes, neutrophils, or platelets, considerable information is available about the surface phenotype of the eosinophil, how integrins and integrin effectors dynamically regulate interactions of eosinophils with ligands, and how these interactions lead to properties of eosinophil adhesion, migration, survival, and recruitment in asthma. Here, we summarize recent advances in the comprehension of integrin structure, integrin function, and integrin modulation of eosinophils. Therapeutic implications of such advances are discussed.

INTEGRINS OF EOSINOPHILS

α4β1 (CD49d/29, VLA-4)

Roles of α4β1 of eosinophils

α4β1 has been studied more extensively than other eosinophil integrins because of its role in mediation of rolling at physiological shear rates [26, 27] and adhesion [21, 28–31] of purified blood or airway eosinophils on vascular cell adhesion molecule 1 (VCAM-1, CD106). VCAM-1 is an integrin counter-receptor upregulated in the asthmatic lung [32]. Figure 2 depicts the arrangement of immunoglobulin (Ig)-like modules of VCAM-1 and the loop in Ig-like module 1 that contains the Ile-Asp-Ser-Pro-Leu (IDSPL) recognition sequence that interacts with α4β1. Expression of endothelial VCAM-1 is induced by T helper cell type 2 (Th2) mediators, strongly implicating VCAM-1 in α4β1-mediated eosinophil recruitment from blood to the airway of human asthmatics [33, 34]. Endothelial VCAM-1 may also be expressed in response to activators of NF-κB or of activator protein (AP)-1, including thrombin, homocysteine, angiotensin II, reactive oxygen species, or migration inhibitory factor [35–39]. Eosinophils may even upregulate local amounts of VCAM-1 on endothelial cells through production of hypothiocyanous acid (HOSCN), a membrane permeable product that is generated in response to oxidation of thiocyanate (SCN⁻) by eosinophil peroxidase [40]. There is little [41, 42] or no [34, 43] expression of α4β1 on purified human neutrophils, suggesting that eosinophil recognition of VCAM-1 mediated by α4β1 is an important mechanism by which selective infiltration of eosinophils over neutrophils is achieved in asthma [44, 45]. Although it has been reported that α9β1 on neutrophils can mediate migration on VCAM-1 [42], in our hands neutrophils do not adhere to immobilized recombinant soluble VCAM-1 (our unpublished results) under conditions in which eosinophils adhere readily [30, 31].

Figure 2. Structure and schematic of VCAM-1 splice forms.

Crystal structure of the first two N-terminal immunoglobulin (Ig)-like modules of 6d- and 7d-VCAM-1 [75]. The crystal structure is color coded to match the schematic of VCAM-1 modules depicted to the right. In the crystal structure, module 1 contains an IDSPL loop (highlighted in black) that is recognized by α4β1 of eosinophils. A second IDSPL motif recognized by α4β1 of eosinophils is present in module 4 of 7d-VCAM-1, which is also recognized by αMβ2. Module 4 is absent from 6d-VCAM-1, the likely result of exon skipping during mRNA splicing [78]. Two intra-domain disulfide bonds are present in modules 1 and 4 while modules 2, 3, 6, and 7 are predicted to contain only one such disulfide. The disulfide is missing from module 5. Individual modules have been color coded based on amino acid sequence identity: modules 1 and 4 = 73%, modules 2 and 5 = 60%, modules 3 and 6 = 60%. The internal homology and similarity of intron sizes between modules 1–3 and 4–6 suggests that modules 4–6 may have arisen in evolution by gene duplication of ancestral modules 1–3 [163]. The figure was created with RasMol.

In addition to mediating rolling and firm adhesion, α4β1 supports eosinophil migration on VCAM-1. Thus, blocking antibodies recognizing the α4 subunit inhibit eosinophil migration through Transwell pores coated with recombinant VCAM-1 in response to Regulated on Activation, Normal T Cell Expressed and Secreted (RANTES) and across VCAM-1-expressing endothelial cells induced by supernatants of peripheral blood mononuclear cells obtained from antigen-sensitized atopic asthmatics [46, 47]. Migration of eosinophils across endothelial cell monolayers in response to eotaxin is inhibited by anti-α4 [48]. Ligation of α4β1 by VCAM-1 has been shown to modulate functions beyond eosinophil rolling, adhesion, and migration. That is, adhesion of eosinophils to VCAM-1 potentiates superoxide anion generation [49–51]. The respiratory burst is inhibited following preincubation with anti-α4 antibody or the WAY103 small molecule α4 antagonist [50]. In contrast, anti-α4 or WAY103 enhances eosinophil-derived neurotoxin release from granules of eosinophils adherent to VCAM-1 [50]. Thus, ligation of α4β1 by these two antagonists has different activities – stimulation of granule release and inhibition of respiratory burst.

The β1 subunit localizes to podosomes of blood eosinophils adherent to VCAM-1 [52] or of airway eosinophils adherent to VCAM-1, intercellular adhesion molecule (ICAM)-1 (CD54), fibrinogen, vitronectin, or albumin [31]. The podosome is a transient structure that mediates interaction with and proteolysis of adhesive ligands in extracellular matrix [53]. Podosomes are enhanced in both size and quantity of airway eosinophils adherent to VCAM-1 in comparison to blood eosinophils [31], and the increase can be replicated by treatment of blood eosinophils with interleukin (IL)-5 or tumor necrosis factor (TNF)-α [52]. It is noteworthy that eosinophils do not form the more stable focal adhesions found in fibroblasts adherent to VCAM-1 [52]. Airway eosinophils are more migratory than blood eosinophils [54]. The increased size and/or number of podosomes of airway eosinophils may, therefore, contribute to the increased mobility.

As described above, the conformational states of integrins determine whether an integrin is functional for cell adhesion and migration. Activation of integrins is accomplished by “inside-out” signaling whereby chemokines and cytokines interact with cell-surface receptors and initiate signaling pathways that target the cytoplasmic domain of integrins. A comparison of α4β1 activity on eosinophils and the Jurkat T lymphocytic cell line suggests that α4β1 on eosinophils is tightly controlled. That is, the 15/7, HUTS-21, or 9EG7 activation-sensitive antibodies do not react with β1 of purified blood eosinophils using identical protocols in which the antibodies react robustly with β1 of Jurkat cells [28, 30], indicating that β1 of Jurkat cells is in a higher activation state in comparison to eosinophils. Thus, purified blood eosinophils do not adhere or migrate on the α4β1 ligand, fibronectin [28, 55], whereas Jurkat cells constitutively adhere to fibronectin [28] and migrate on surfaces coated with fibronectin in response to serum (our unpublished results). Coating of Transwell membranes with fibronectin inhibits formyl-Met-Leu-Phe (fMLP)-stimulated migration of eosinophils in comparison to uncoated membranes, indicating that fibronectin may even be inhibitory to eosinophil migration [55]. Adhesion and migration on fibronectin, therefore, requires a highly active form of α4β1 that is not present on purified blood eosinophils. “Forcing” β1 into a high activation state by incubation of blood eosinophils with the 8A2 activating antibody to β1 results in adherent eosinophils that roll less well on VCAM-1 [27], stimulates eosinophil adhesion to fibronectin [28], and decreases migration of eosinophils across monolayers of human umbilical vein endothelial cells [56]. Incubation of purified eosinophils from blood of allergic subjects with Mn²⁺ exposes the activation-sensitive epitope in the β1 hybrid domain recognized by mAb 15/7 [57] and enhances adhesion to VCAM-1 [58]. RANTES, monocyte chemotactic protein (MCP)-3, and eotaxin transiently increase eosinophil interaction with VCAM-1 and of a Leu-Asp-Val (LDV)-containing peptide [59, 60]. Thus, steps in eosinophil recruitment mediated by α4β1 are dynamically regulated by the allosteric structure of α4β1 present on the eosinophil surface.

Conformations of β1 of eosinophils

We have employed conformation-sensitive antibodies to probe β1 conformation on purified or nonpurified blood or airway eosinophils in segmental antigen challenge or steroid withdrawal clinical models of allergic human asthma. Figure 3 depicts a model in which several states of β1 activation are queried with three conformation-sensitive antibodies: N29, HUTS-21, and 9EG7. N29 recognizes an activation-induced epitope in the N-terminal region of the plexin, semaphorin, and integrin (PSI) domain [61], HUTS-21 recognizes an activation-induced epitope in the hybrid domain [62] that is also recognized by mAb 15/7 [57, 63], and 9EG7 recognizes an activation-induced epitope in the epidermal growth factor (EGF) domains of the “leg” [64, 65]. The locations of these epitopes in various structures adopted by β1, based on structures of αVβ3 and αIIbβ3 deduced from X-ray crystallographic and electron microscopic studies [24, 66], suggest that the antibodies recognize increasingly activated forms in the order N29<HUTS-21<9EG7. We have found that N29 but not HUTS-21 or 9EG7 recognize purified blood and airway eosinophils [31]. Such eosinophils adhere to immobilized VCAM-1 and not fibronectin [31], indicating that the N29⁺/HUTS-21⁻/9EG7⁻ putative form of α4β1 depicted to the middle in Figure 3 recognizes VCAM-1 but not fibronectin. Jurkat cells, which express the N29⁺/HUTS-21⁺/9EG7⁺ putative form of α4β1 depicted to the right in Figure 3, react highly with all three antibodies [30], adhere to fibronectin [67], and adhere even better to VCAM-1 compared to eosinophils [30]. Thus, both N29⁺/HUTS-21⁻/9EG7⁻ and N29⁺/HUTS-21⁺/9EG7⁺ forms of α4β1 prefer VCAM-1 over fibronectin. Eosinophils that are purified from blood are more reactive to N29 compared to nonpurified eosinophils in whole blood from the same donor, indicating that β1 is susceptible to activation during the purification procedure [68]. In asthmatics with a dual response phenotype, N29 reactivity of unfractionated blood and bronchoalveolar lavage (BAL) eosinophils 48 hours after segmental antigen challenge is increased compared to blood eosinophils before challenge (our unpublished results). Moreover, N29 reactivity of BAL eosinophils correlates with the percentage of eosinophils in BAL fluid. These results indicate that the N29 immunoreactive form of β1 is likely important in movement of primed blood eosinophils out of the circulation and into the airway in response to segmental antigen challenge. Indeed, we speculate that VCAM-1 may not be recognized by the N29⁻/HUTS-21⁻/9EG7⁻ putative form of α4β1 illustrated to the left in Figure 3.

Figure 3. Antibody probes of α4β1 conformation and affinity.

Schematic of three putative conformations of α4β1 that might be expressed by eosinophils as modeled on the crystal structures of αVβ3 and αIIbβ3 and adapted from Luo BH et al. [164]. The bent, closed form of α4β1 is presumably not recognized by any of the activation-sensitive antibodies (left). Extension of α4β1 reveals the epitope in the β1 PSI domain recognized by N29 - this conformation is presumed to most closely depict the conformation of α4β1 of blood or airway eosinophils (middle). Further activation results in swing-out of the hybrid domain and exposure of the epitope recognized by HUTS-21, along with separation of the integrin legs and exposure of the epitope recognized by 9EG7 (right). The latter form of α4β1 is presumably present on Jurkat, EoL-3, or Mn²⁺- or PMA-activated AML14.3D10 eosinophilic cells [30].

Based on the utility of N29 as a measure for activation of α4β1 in the antigen-challenge model, N29 positivity of blood eosinophils was assayed in the inhaled corticosteroid (ICS) withdrawal model of controlled elicitation of asthma. In the ICS model, mild asthmatics undergo true or sham ICS withdrawal in a randomized, placebo-controlled, two-period crossover study. ICS withdrawal results in a decrease in the forced expiratory volume in one second (FEV₁) [69], increase in circulating and sputum eosinophils [69], and a higher percentage of VCAM-1-positive vessels in comparison to subjects before withdrawal [32]. We found that N29 epitope expression, which was quantified by flow cytometry on eosinophils in whole blood, correlated inversely with FEV₁ both after ICS withdrawal and throughout the ICS study [68]. In response to ICS withdrawal, N29 epitope expression correlated with FENO (fraction of exhaled nitric oxide) [68], an established asthma marker and indicator of airway inflammation [70, 71]. Indeed, receiver-operator characteristic curve analysis [72] indicates that N29 reactivity is a better predictor of FEV₁ < 95% of baseline than FENO or sputum eosinophils [68]. The ICS study may be the first demonstrated example in which clinical measurements of a disease are predicted by and/or correlate with the activation state of a particular integrin subunit expressed on a circulating cell type.

α4β1 recognition of VCAM-1 modules

The allosteric structural forms assumed by α4β1 of eosinophils in antigen challenge and ICS withdrawal models of asthma are important in the recognition of VCAM-1 modules and alternatively spliced VCAM-1 forms. Shown in Figure 2, the loop between β strands C and D in the first immunoglobulin (Ig)-like module of VCAM-1 is accessible on the protein surface and contains the core integrin-recognition sequence IDSPL [73–77]. A second IDSPL site is present in the CD loop in module 4 [74, 76]. Alternative splicing of mRNA encoding VCAM-1 generates two protein forms in humans: a variant consisting of seven Ig-like modules, 7d-VCAM-1, containing both putative integrin-binding sites in modules 1 and 4, and a variant containing six Ig-like modules, 6d-VCAM-1, that is missing the putative integrin-binding site in module 4 and contains only the site in module 1 [78]. Endothelial cells appear to express more 7d-VCAM-1 compared to 6d-VCAM-1 based on quantitation of mRNA from HUVEC cells treated with TNF-α [78]. The distance of module 1 from the endothelial surface may be expected to facilitate α4β1-mediated eosinophil capture from the circulation. That is to say, module 1 of 7d-VCAM-1 may extend 3.7 nanometers further from the endothelial surface than module 1 of 6d-VCAM-1 or 11.1 nanometers further than module 4 based on the length of individual Ig-like modules calculated from rotary shadowing electron microscopic images of recombinant soluble 7d-VCAM-1 adsorbed to mica [77]. The first two N-terminal modules of 7d-VCAM-1 can mediate both rolling and firm adhesion of blood eosinophils via α4β1 [27], indicating that this extension allows ready interaction with eosinophil α4β1. By studying a set of recombinant forms of VCAM-1, we have shown that α4β1 mediates robust adhesion of purified blood eosinophils to constructs containing module 1, including 6d-VCAM-1, 7d-VCAM-1, and the construct containing only modules 1–3, 1–3VCAM-1, whereas there is less adhesion mediated by α4β1 to the construct containing only modules 4–7, 4–7VCAM-1 [30]. Thus, purified blood eosinophils more readily recognize module 1 in comparison to module 4 of VCAM-1. While module 1 may promote the initial capture of eosinophils, the close proximity of module 4 to the endothelial wall may help strengthen adhesion and facilitate movement of activated blood eosinophils out of the circulation. Static adhesion to module 4 of eosinophils or eosinophilic cell lines can be enhanced following incubation with IL-5, Mn²⁺, or PMA [30]. That 7d-VCAM-1 is divalent may be of additional consequence for α4β1-mediated eosinophil diapedesis. In other words, it may be that module 1 and 4 of a single VCAM-1 molecule can be ligated simultaneously by two different α4β1 integrin molecules co-expressed on an individual eosinophil cell. The head of α4β1, if modeled after the crystal structure of αVβ3, has dimensions between 4.5–9.0 nanometers [66], smaller than the distance that bridges modules 1 and 4 [77]. Structural variations between modules 2 and 5 may differentially orient modules 1 and 4, respectively, for recognition by α4β1 of eosinophils. That is, while module 2 of VCAM-1 is disulfide-bonded in the crystal structure shown in Figure 2 [75], module 5 is unusual for Ig-like modules in lacking such a bond based on proteolysis studies with endoproteinase Glu-C [79]. Of note, EoL-3 and AML14.3D10 eosinophilic leukemic cell lines do not recognize module 4 of VCAM-1 even though these cells express more surface α4β1 and display a more conformationally active form of β1 that is better recognized by N29, HUTS-21, and 9EG7 [30]. Recognition of modules 1 and 4 by eosinophils is, therefore, controlled by more than α4β1 expression level and activation state alone. The β1 subunit of eosinophils but not of eosinophilic cell lines is in podosomes [52], which may partly explain the differences between the cell lines and eosinophils. Experiments described below indicate a role for αMβ2 as well in eosinophil recognition of VCAM-1 module 4.

αMβ2 (CD11b/18, Mac-1)

αMβ2 is present on purified blood eosinophils in a conformational state that is constitutively less active than α4β1. αMβ2 of unstimulated purified blood eosinophils mediates low levels of static adhesion to module 4 of 7d-VCAM-1 [30]. Baseline interaction is regulated by phosphoinositide-3 kinase (PI3K), in as much as adhesion to module 4 and not module 1 of 7d-VCAM-1 is blocked completely by wortmannin or LY294002 [30], both known to block αMβ2 integrin-mediated adhesion of eosinophils to ICAM-1 or albumin in response to IL-5 [80, 81]. That inhibitors specific for αMβ2 completely block adhesion to module 4 even though the recognition is partly α4β1-dependent [30], raises the possibility of integrin crosstalk and/or cooperativity between αMβ2 and α4β1. Further evidence that recognition of module 4 by αMβ2 is significant comes from the AML14.3D10 and EoL-3 eosinophilic cell lines, which do not express αMβ2 and do not adhere to module 4 [30].

Whereas α4β1 of blood eosinophils constitutively ligates VCAM-1 even in the absence of stimulation by cytokines and irrespective of “inside-out” signaling, αMβ2 recognition is influenced greatly by activation. Incubation with IL-5 enhances αMβ2-mediated static adhesion of blood eosinophils to ICAM-1 or modules 1 or 4 of VCAM-1 [30, 80, 81]. The αMβ2-mediated adhesion of IL-5-stimulated eosinophils to modules 1 or 4 of VCAM-1 is mostly refractory to inhibition by wortmannin or LY294002 [30]. Adhesion of IL-5-stimulated eosinophils to ICAM-1 or albumin is, however, blocked by these inhibitors [80, 81]. Thus, recognition of diverse ligands by αMβ2 of eosinophils is differentially regulated. Indeed, under assay conditions that mimic blood flow, IL-5 incubation has the paradoxical effect of decreasing, rather than increasing, adhesion of blood eosinophils to 7d-VCAM-1 in the parallel plate flow chamber within minutes [30]. The decrease in adhesion is dependent on αMβ2 since it can be reversed by anti-αM antibody [30]. Such a finding is consistent with the observation that IL-5 promotes the release in vivo of eosinophils from bone marrow via a mechanism that is dependent on β2 integrins and PI3K and that can be inhibited by wortmannin [82]. The release of eosinophils may be facilitated further by L-selectin shedding from the surface of eosinophils in response to IL-5 stimulation [83]. These results suggest that the recognition of VCAM-1 by αMβ2 may be tightly controlled by IL-5. Namely, the IL-5-induced conformation of αMβ2 may transiently enhance and then inhibit eosinophil adhesion in a process that may depend on αMβ2 activation and be associated with L-selectin shedding. The consequence of such release would be to promote the subsequent movement of eosinophils through the vascular wall into the airway lumen. In support of this conjecture, we have shown that airway eosinophils purified following segmental antigen challenge express a conformationally active form of αMβ2 recognized by the CBRM1/5 anti-αM conformation-sensitive antibody [31]. Figure 4 depicts the epitope in the I (insert)-domain of the αM subunit recognized by CBRM1/5 [84]. Two different structures believed to represent the inactive or active form of the αM I-domain have been crystallized in buffers containing manganese or magnesium, respectively [85, 86]. Even though the CBRM1/5 epitope is exposed in the presumed inactive crystal form, the epitope is not well recognized by CBRM1/5 [84]. Thus, the I-domain of αM likely undergoes a change in shape or conformation upon activation that facilitates better recognition of the I-domain by CBRM1/5 [84, 87]. As such, the enhanced reactivity of CBRM1/5 of airway compared to blood eosinophils indicates that the αM I-domain of airway eosinophils is structurally different. Such airway eosinophils recognized by CBRM1/5 adhere to diverse ligands, including VCAM-1, albumin, ICAM-1, fibrinogen, and vitronectin via αMβ2 [31]. That is to say, a prominent characteristic of αMβ2 is its ability to recognize a plethora of structurally diverse ligands and extracellular matrix proteins through the Lys²⁴⁵-Arg²⁶¹ segment of the αM I-domain [88]. Blood eosinophils purified before or after antigen challenge are not well recognized by CBRM1/5 and exhibit little or no adhesion to such ligands [31]. Intranasal administration of IL-5, a regulator of αMβ2, causes eosinophil migration into the airway lumen of mice [89]. This movement is blocked by intraperitoneal administration of the p85a dominant negative form of the PI3K regulatory subunit fused to human immunodeficiency virus-transactivator of transcription (HIV-TAT) [89]. αMβ2 expression is upregulated on human or mouse airway eosinophils following antigen challenge in comparison to blood eosinophils [90–93]. Elevation of αMβ2 has also been observed on migratory human blood eosinophils [48]. The upregulation of αMβ2 on eosinophils can be achieved following incubation of blood eosinophils with IL-5, granulocyte/macrophage colony-stimulating factor (GM-CSF), fMLP, or platelet activating factor (PAF) [91, 94]. Expression of αM on blood eosinophils is increased following incubation with calcium ionophore or after only minutes of phorbol 12-myristate 13-acetate (PMA) stimulation, indicating that blood eosinophils have preformed stores of αMβ2 that can be rapidly mobilized to the cell surface [21].

Figure 4. I-domain of αMβ2.

Schematic (left) of the I-domain of αMβ2 color coded to match the crystal structure (right) [86]. α-helices-1, 3, and -7 of the I-domain (shown in green) undergo conformational changes in response to αMβ2 activation [84]. Residues recognized by CBRM1/5 are labeled and highlighted in black. The figure was created with PyMol.

Transendothelial migration of eosinophils triggered by eotaxin is inhibited by anti-αM [48]. The chemokines RANTES, MCP-3, and complement (C)5a expose the CBRM1/5 activation epitope of αMβ2 [59, 95] and promote migration of blood eosinophils [59, 95–98]. Treatment with IL-5 exposes the CBRM1/5 epitope of blood eosinophils and mimics the phenotype of airway eosinophils by inducing αMβ2-mediated adhesion of blood eosinophils to ICAM-1, albumin, vitronectin, and fibrinogen [31, 99]. IL-5 is upregulated in the blood and airway of human asthmatics [100–103], and the IL-5 receptor is down-regulated on airway eosinophils compared to blood eosinophils, an indication that the IL-5 receptor on airway eosinophils has been occupied [104]. Degranulation of eosinophils stimulated by GM-CSF and PAF is mediated by αMβ2 [105]. Therefore, upregulation and activation of αMβ2 appears to represent a major and long-lasting consequence of exposure of eosinophils to mediators of eosinophilic inflammation.

αDβ2 (CD11d/18)

The αD and αM I-domains share 60% amino acid sequence identity [106]. As such, αDβ2 on eosinophils may be expected to resemble αMβ2 in the promiscuous recognition of several ligands. In αDβ2-transfected HEK293 human embryonic kidney or IC-21 macrophage cells, αDβ2 mediates adhesion to fibrinogen, vitronectin, VCAM-1, or fibronectin and migration on vitronectin [106]. In unstimulated or IL-5-treated purified blood eosinophils, αDβ2 has been reported to contribute to adhesion on VCAM-1 in static or flow conditions [21, 107]. Like αM, airway eosinophils express higher levels of αD in comparison to blood eosinophils [21, 31], and expression of αD on blood eosinophils is increased to a level similar to the expression on airway eosinophils following minutes of treatment with PMA, calcium ionophore, or three days of culture in IL-5 [21, 107]. Eosinophils that are purified from blood express elevated surface levels of αD compared to eosinophils in whole blood from the same donor indicating that upregulation of αDβ2, like activation of β1 integrins [68], is a consequence of eosinophil activation during purification [68]. Thus, as with αMβ2, eosinophils appear to have preformed stores of αDβ2 that can be rapidly mobilized. We have found that αD stains diffusely in unstimulated blood eosinophils adherent to VCAM-1 but is present in structures in the substrate plane in airway eosinophils or PMA-activated blood eosinophils adherent to VCAM-1 [52]. These αD-positive structures do not co-localize with podosomes, raising the possibility that αDβ2 recognizes VCAM-1 independent of recognition by α4β1. We have not observed αD-dependent adhesion of unstimulated or IL-5-stimulated purified blood eosinophils to VCAM-1 using anti-αD antibodies but do not exclude the possibility that αD, like αM, contributes to eosinophil recognition of VCAM-1 [30].

α4β7 (CD49d/β7)

α4β7 of blood eosinophils supports static adhesion on mucosal addressin cell adhesion molecule-1 (MAdCAM-1) and mediates rolling on MAdCAM-1 and VCAM-1 under flow [27]. Incubation with PAF induces elevated static adhesion of eosinophils to MAdCAM-1 via α4β7 [29]. Crosslinking of β7 on blood eosinophils by soluble VCAM-1 or Fib 30 antibody increases GM-CSF mRNA expression and survival of blood eosinophils, indicating that α4β7 is a regulator of eosinophil survival [108]. The increased survival is blocked by anti-GM-CSF, but not by antibodies to IL-3 or IL-5, pointing to GM-CSF synthesis as the likely target by which ligation of α4β7 increases viability [108].

α6β1 (CD49f/29)

α6β1 is a well-known receptor for laminins. α6β1 supports adhesion of eosinophils purified from allergic donors to human laminin obtained from pepsin digestion of placenta, while eosinophils purified from nonallergic donors adhere less well [23]. We did not find specific adhesion of blood eosinophils purified from normal, allergic, or asthmatic subjects or of airway eosinophils to mouse laminin-1, raising the question of if, when, and/or how eosinophils are able to recognize laminin [31]. Eosinophil migration through Matrigel, a basement membrane mix that contains laminin-1, is reduced in response to anti-β1 [109].

αLβ2 (CD11a) and αXβ2 (CD11c/18)

αLβ2 is mostly selective for ICAMs and has a narrower ligand-binding specificity than αMβ2 or αDβ2 [110]. In vitro transendothelial chemotaxis of eosinophils to eotaxin is partly inhibited by anti-αL [48]. αXβ2 shares ligands with αMβ2 [111]. There is little, if any, known about the role of αXβ2 on eosinophils, making αXβ2 the most enigmatic of integrin heterodimers expressed by eosinophils. We have found no inhibition of blood or airway eosinophil adhesion to ICAM-1, albumin, or fibrinogen by antibodies to αLβ2 or αXβ2, indicating that these two integrins are inactive on eosinophils for these particular ligands compared to αMβ2 [31].

SUMMARY OF EOSINOPHIL INTEGRINS

The unique set of integrin heterodimers expressed by eosinophils mediates diverse functions of eosinophil rolling, firm adhesion, migration, respiratory burst, degranulation, and viability. In asthma, these functions are important in orchestrating interactions of eosinophils with ligands expressed by airway cells. α4β1 and αMβ2 are likely the two most important integrins that mediate eosinophil adhesion and movement. These two integrin heterodimers differ in several respects. α4β1 of unstimulated blood eosinophils mediates rolling along VCAM-1-expressing airway endothelium and constitutively ligates modules 1 and 4 of VCAM-1, whereas αMβ2 is constitutively less active and primarily recognizes only module 4 of VCAM-1. α4β1 in the blood and αMβ2 en route to or within the airway can undergo activation. We propose that α4β1 and αMβ2 may cooperate to mediate adhesion and release from luminal ligands like VCAM-1, followed by adhesion to and movement on diverse pericellular endothelial ligands mediated by activated αMβ2. Specifically, α4β1 expressed on circulating eosinophils may undergo activation, i.e., in response to allergen, allowing α4β1 to better recognize module 1 of 7d-VCAM-1 or 6d-VCAM-1. Following eosinophil capture on endothelium, both α4β1 and αMβ2 co-expressed on the eosinophil may compete for binding of module 4 of the aforementioned 7d-VCAM-1 molecule to facilitate firm adhesion. Subsequent activation of αMβ2 by IL-5 or chemokines may cause αMβ2 to assume a dominant role in recognition of module 4 and a significant role in recognition of module 1. Such recognition may involve a transient increase in adhesion to VCAM-1 mediated by αMβ2, followed by release from VCAM-1 of αMβ2 within minutes under flow in the presence of IL-5 and release of α4β1 due to possible proteolysis in podosomes [52]. The consequence of such release may be to promote αMβ2-mediated movement of eosinophils on diverse endothelial ligands in transit to the airway. In essence, movement through airway endothelium may, therefore, involve a “hand-off” whereby αMβ2 replaces α4β1 as the principal adhesive and migratory integrin. The “hand-off” may be facilitated by IL-5 and/or chemokines including members of the eotaxin family, which can shift integrin usage away from α4β1 to β2 integrins [112]. In fact, IL-5 can even inhibit binding of the 15/7 anti-β1 conformation-sensitive antibody to eosinophils from allergic donors [58] and, as noted, induce L-selectin shedding and release from endothelial selectins. 6d-VCAM-1 with one integrin recognition module, and 7d-VCAM-1 with two, may regulate differentially such recruitment by virtue of valency of the splice form, distance of the integrin binding site from the endothelial surface, and/or role of module 4. Activation of αMβ2 and diapedesis to the pulmonary vasculature may be facilitated further by degradation of matrix proteins by metalloproteinases in podosomes, transient adhesive structures of IL-5- and TNFα-activated blood eosinophils or airway eosinophils.

The remaining five integrins of eosinophils – αDβ2, α4β7, α6β1, αLβ2, and αXβ2 – likely play supportive roles to α4β1 and αMβ2 in eosinophil recruitment. αDβ2 and αXβ2 of activated eosinophils potentially recognize VCAM-1 and additional ligands similar to those recognized by αMβ2. The enhanced survival of eosinophils in the airway lumen may involve ligation of α4β7 to soluble adhesive ligands such as fibronectin. α6β1 of migrating eosinophils may interact with laminin in transit through vascular basement membrane, while αLβ2 potentially ligates ICAMs. The functions of αDβ2, α4β7, α6β1, αLβ2, and αXβ2 of eosinophils, however, warrant further exploration.

TARGETING EOSINOPHIL INTEGRINS IN VIVO

Several competitive inhibitors that target integrin heterodimers have been devised to interfere with the recruitment or number of eosinophils in the asthmatic airway, and thus, diminish airway eosinophilic inflammation and particular aspects of asthma, to which eosinophils are thought to contribute [113]. These therapies have been evaluated in clinical human trials and diverse animal models including primate, guinea pig, sheep, rabbit, mouse, and rat. Enigmatically, as elaborated below, the studies in humans and in animal models have reached different conclusions.

Most therapeutic strategies involving integrins of eosinophils have targeted α4, the common subunit of α4β1, the most important counter-receptor of eosinophils recognizing VCAM-1, and α4β7, potentially important in recognition of MAdCAM-1. The utility of α4 subunit antagonists as therapeutic agents in clinical disease was first realized with the development of Antegren (Tysabri, Natalizumab) a humanized anti-α4 antibody that has shown effectiveness in clinical trials for multiple sclerosis and Crohn’s disease [114]. In parallel, a number of small molecule inhibitors including BIO-1211 (compound 28) have been developed. BIO-1211 has a 200-fold greater selectivity for the active compared to inactive form of α4β1 and is based on the LDV sequence from the alternatively spliced connecting segment-1 (CS-1) peptide of cellular fibronectin [115, 116]. Despite promising results in animal models of asthma, development of BIO-1211 was discontinued due to lack of efficacy in phase II clinical trials of asthma conducted by Merck and Biogen [117]. In preliminary studies, a second LDV mimetic modeled after BIO-1211, IVL745, caused only a modest reduction in sputum eosinophils in human patients with mild-to-moderate atopic asthma following inhalation and had no effect on the early or late asthmatic response to inhaled allergen or markers of airway inflammation [118]. The futures of IVL745 and another α4β1 antagonist, compound HMR 1031, under investigation by Sanofi-Aventis in phase II trials of asthma, are unclear [117, 119]. HMR 1031 is a potent α4β1 small molecule antagonist that selectively blocks binding of α4β1 to VCAM-1 and fibronectin [120]. When administered by aerosol for 8 days to patients with mild-to-moderate persistent asthma, HMR 1031 failed to relieve house dust mite- or methacholine- induced airway eosinophilia, exhaled nitric oxide production, or soluble markers of airway hyperresponsiveness [121]. Two orally active dual α4β1/α4β7 antagonists, TR14035 and R411, were examined in clinical trials by Tanabe/GlaxoSmithKline and Roche, respectively [117]. TR14035 is no longer listed in GlaxoSmithKline’s therapeutic pipeline [117], R411 was discontinued as noted on Roche’s 2006 annual investor report [122], and several other α4β1 antagonists, including GW-559090 from GlaxoSmithKline [123, 124] or RBx-7796 (Clafrinast) from Ranbaxy [125], are no longer listed in either company’s pipeline.

The relative failure of α4 antagonists in humans is puzzling given the effectiveness of these compounds in animal models of asthma. Thus, while described above, BIO-1211, HMR 1031, IVL745, TR14035, or GW559090X α4β1 antagonists have negligible benefit in human asthma clinical trials, these compounds have had significant effects in sheep [118, 121, 126], mouse [121, 127], rat [118, 123, 128], or guinea pig [123] models of asthma. The HP1/2, PS2/3, PS/2, TA-2, or Max-68P anti-α4 blocking antibodies, CS-1 peptide ligand anti-α4β1 mimic, or α4β1 small molecule inhibitors reduce either numbers of eosinophils in the airway or ameliorate inflammatory histopathology or airway allergic responses in the guinea pig [129–133], sheep [134–137], mouse [138–140], rat [141, 142], or rabbit [143]. A complication of such studies is the failure to differentiate between effects resulting from eosinophils, roles of integrins of eosinophils, or roles of other cell types [144–147]. Of additional consequence, there are reports of cellular movement independent of β1 or β2-integrins in a model of T cell migration within a 3D-collagen gel [148] or by granulocytes null for β2 in response to fMLP [149], suggesting that movement of eosinophils or leukocytes in disease may involve processes other than, or in addition to, integrins. In any case, more informative models of inflammation that are better able to define the specific role of integrins of eosinophils in asthma and disease may result in improved therapeutic outcome.

Approaches that target integrins in eosinophil-associated diseases including asthma likely can be significantly improved. One problem in the development of such inhibitors may be the functional overlap and redundancy of integrins. In other words, α4β1, αLβ2, and αMβ2 may be able to compensate for one another to some extent in vivo [150–152]. Thus, simultaneous targeting of several integrins may be a potentially more advantageous strategy than antagonizing only one integrin molecule. Indeed, the success of corticosteroids or β2-adrenergic receptor agonists (bronchodilators) in dampening airway inflammation could be said to result from such broad activity on several inflammatory targets at once [119]. Nonetheless, morbidity and mortality continue to increase in eosinophil-related diseases including asthma despite the availability of corticosteroids formulated for delivery to the lung [153]. Clearly, there is a need for additional therapies that inhibit eosinophilic inflammation. Therapies that broadly target integrins would offer the possibility of potently suppressing eosinophil-related pathologies with greater specificity and with lesser side effects compared to current treatments. Such therapy could target recruitment mechanisms involving both α4 and β2 integrins. This strategy may prove efficacious against neutrophils as well as eosinophils and be of use in subjects with severe or persistent asthma with predominantly neutrophilic inflammation [154–156]. As noted previously, β2 integrins of human eosinophils recognize VCAM-1 and numerous matrix proteins of the pulmonary vasculature, mediate adhesion and movement of eosinophils, and are regulated by PI3K, a target of therapeutic value in the murine asthma model [89]. Movement of human eosinophils is antagonized by anti-β2 [54, 157–160]. In fact, such migration is inhibited even more effectively with a combination of anti-α4 and anti-β2 than either alone [160]. Several models of allergic inflammation in animals indicate that β2 integrins may be promising therapeutic targets in future treatment regimens of asthma [142, 143]. A major worry is immunocompetency arising from antagonism of β2 integrins. An example is efalizumab (Raptiva^®, Genentech), a humanized anti-αL monoclonal antibody that has been approved for the treatment of psoriasis [161]. Efalizumab has shown moderate efficacy in reducing accumulation of eosinophils in the airway and in attenuating the late asthmatic response of human subjects with atopic asthma [162]. Efalizumab is immunosuppressive and may, therefore, be contra-indicated in patients with infection, who are already immunocompromised, or who are pregnant [161]. One strategy to improve integrin antagonists like efalizumab may be to develop inhibitors that bind only certain integrin activation states. Targeting specific structural conformers of integrins, in turn, may block only certain integrin-ligand interactions and minimize side effects.