|Home | About | Journals | Submit | Contact Us | Français|
The primary function of neutrophils in host defense is to contain and eradicate invading microbial pathogens. This is achieved through a series of swift and highly coordinated responses culminating in ingestion (phagocytosis) and killing of invading microbes. While these tasks are usually performed without injury to host tissues, in pathologic circumstances such as sepsis, potent antimicrobial compounds can be released extracellularly, inducing a spectrum of responses in host cells ranging from activation to injury and death. In the lung, such inflammatory damage is believed to contribute to the pathogenesis of diverse lung diseases, including acute lung injury and the acute respiratory distress syndrome, chronic obstructive lung disease, and cystic fibrosis. In these disorders, epithelial cells are targets of leukocyte-derived antimicrobial products, including proteinases and oxidants. Herein, we review the mechanisms involved in the physiologic process of neutrophil transepithelial migration, including the role of specific adhesion molecules on the leukocyte and epithelial cells. We examine the responses of the epithelial cells to the itinerant leukocytes and their cytotoxic products and the consequences of this for lung injury and repair. This paradigm has important clinical implications because of the potential for selective blockade of these pathways to prevent or attenuate lung injury.
This article reviews the mechanisms involved in transmigration of neutrophils through lung epithelial cells during inflammation. Identification of the signaling pathways involved will help identify novel targets to prevent or attenuate lung injury.
In their primary function in host defense, neutrophils have been likened to a night watchman because they continuously patrol the distant reaches of the lung and other organs, searching for invading microbial pathogens that they hunt down, ingest (phagocytose), and destroy (1, 2). To achieve this purpose, neutrophils possess a potent antimicrobial arsenal that includes oxidants, proteinases, and cationic peptides (3). Oxidants such as O2− and H2O2 are produced by the phagocyte NADPH oxidase and are potently microbicidal (4). Granules within the cytoplasm of neutrophils contain powerful proteolytic enzymes and cationic proteins that can digest a variety of microbial substrates. When neutrophils internalize microbial pathogens, these cytotoxic compounds are typically released directly into the phagosome, compartmentalizing both the pathogen and the cytotoxic products. However, under pathologic circumstances, these compounds are released into the extracellular space and can damage host tissues. Importantly, neutrophils and their responses in the context of an inflammatory response are inherently beneficial; only when their responses become excessive or unregulated does injury to host tissues ensue (1).
Integral to the effective functioning of neutrophils in host defense in the lung and other organs is their ability to egress from the vasculature and migrate through the tissues to the site of infection. During this journey into the lung, neutrophils pass through the endothelium, interstitial tissues, and epithelium before ending up in the airspaces. It is also during this time that unrestrained activation of neutrophils in response to microbial or host-derived stimuli may result in release of cytotoxic compounds that can injure vicinal host cells. While it is clear that neutrophils can emigrate from the vasculature into the airspaces without causing injury (5–7), there is compelling evidence from observations in humans and in experimental models that in pathologic circumstances, neutrophils are primary perpetrators of inflammatory injury to the lung and other organs. For example, neutrophil influx into the alveolar space correlates with lung injury as manifest by an increase in permeability of the alveolo-capillary membrane (8). Further, in some (but not all) animal models of acute lung injury, neutrophil depletion is protective (9, 10). It is believed that during the translocation of neutrophils from the vasculature to the airspaces, activation may be excessive and/or prolonged, leading to extracellular release of cytotoxic compounds that can induce a spectrum of responses in neighboring cells ranging from activation to injury and death. To understand what goes awry in pathologic conditions, it is necessary to review the physiologic processes involved in neutrophil transmigration into the lung.
Neutrophils exit the circulation by a well-characterized series of events involving adhesion to and transmigration across the vascular endothelium into the interstitial space (11). A detailed review of these events is beyond the scope of this manuscript, but several key points that are relevant to transepithelial migration will be briefly summarized. The first step in neutrophil emigration from the circulation involves adhesion to the vascular endothelial cells. Traditionally, these events have been viewed as involving three discrete phases: rolling, activation, and firm adhesion (11, 12). However, recent studies have added additional events to this sequence including tethering, slow rolling, modulation of adhesion strength, intraluminal crawling, and finally transcellular and paracellular migration (11). The initial step in leukocyte adhesion is capture (or tethering) mediated by interactions between L, E, and P-selectins and P-selectin glycoprotein ligand (PSGL1) and α4β1 (VLA4) integrin. L-selectin is expressed by leukocytes, P-selectin is expressed by inflamed endothelium and platelets, E-selectin is expressed by inflamed endothelium, and PSGL1 is expressed by endothelium and some leukocytes. The subsequent rolling step is mediated by interactions between selectins and PSGL-1 and other glycosylated ligands. It is noteworthy that L-selectin and P-selectin require shear stress to support adhesion (13), and this may explain in part why selectins are not involved in neutrophil–epithelial adhesion and transmigration. Subsequent to rolling is slow rolling (mediated by selectin-triggered signaling), followed by arrest of the neutrophils on the endothelial surface involving β1 and β2 integrins and their cognate binding partners. Integrins are transmembrane glycoproteins expressed on leukocytes that are composed of α and β chains. In the case of β2 integrins, there is a common β chain (CD18) and a variable α chain (CD11a, b, c, or d). The phase of leukocyte arrest is mediated by interactions between the β2 integrin CD11a/CD18 (LFA-1) and intercellular adhesion molecule (ICAM)1, α4β1 (VLA-4) and vascular cell adhesion molecule (VCAM)1, and α4β7 and MADCAM1. The next phase involves adhesion strengthening and spreading that is mediated by activation of outside-in and inside-out intracellular signaling pathways. The phase of intravascular crawling that follows is mediated by interactions between CD11b/CD18 (αmβ2 or Mac-1) and ICAM1. Leukocytes then transmigrate across the endothelium, taking either a paracellular route involving platelet endothelial cell adhesion molecule (PECAM)1, CD99, junctional adhesion molecules (JAMs), and ESAM or a transcellular route involving ICAM1, PECAM1, and caveolins.
Analogous to the endothelial events, neutrophil migration across epithelia can also be considered in three sequential stages: adhesion, migration, and postmigration events (14) (Figure 1). The initial stage of neutrophil transepithelial migration is characterized by adhesion of the neutrophils to the epithelial membrane. However, distinct from interactions with the endothelium, neutrophil adhesion to the epithelium occurs on the basolateral as opposed to the apical surface. Further, the neutrophils that arrive at the epithelium have a “prior history” inasmuch as they have just crossed the endothelium and migrated through interstitial tissues before arriving at the base of the epithelium. These prior adhesive and migratory events, in conjunction with exposure to the chemoattractants that have spurred this movement, undoubtedly result in an alteration in the state of activation (“priming”) of the migrating neutrophils. The neutrophils subsequently travel between adjacent epithelial cells (the interepithelial “migration tunnel”), crossing interepithelial junctions. Of importance, the epithelium and thus the interepithelial migration tunnel through which the neutrophils crawl is significantly longer (20 μm in length or more) than the equivalent structure between endothelial cells (a few micrometers at most) (15). In addition, there is no evidence that neutrophils take a transcellular route through epithelial cells unlike transendothelial migration. Finally, after traversing the epithelial barrier, neutrophils emerge on the apical surface of the epithelium and adhere to this surface for varying periods of time. During these events, multiple, sequential adhesive interactions must occur in order for the neutrophils to completely traverse the epithelial monolayer.
Our understanding of transepithelial neutrophil migration is less comprehensive than that of transendothelial migration. Further, much of our knowledge about transepithelial neutrophil migration derives from in vitro studies with intestinal epithelium, with relatively fewer studies using lung and other types of epithelial cells. In this review, we will discuss in detail the current knowledge of the adhesive interactions during each of these three stages of transepithelial migration. As mentioned above, we will compare and contrast the steps involved in transepithelial neutrophil migration in different types of epithelia and with those involved in transendothelial migration (Tables 1 and and22).
Much of our knowledge of the mechanisms of transepithelial neutrophil migration derives from in vitro studies of neutrophil transmigration across cultured epithelial cells, although a few studies have examined these events in vivo (16). While earlier studies employed epithelial cells cultured on the upper surface of tissue culture inserts (17, 18) with neutrophil transmigration proceeding in a (nonphysiologic) apical to basolateral direction, more recent studies have employed a more physiologic arrangement in which epithelial cells are cultured on the underside of inverted tissue culture (e.g., Transwell) chambers (19–21). After allowing epithelial cell attachment, the chambers are then restored to the upright position and allowed to grow and attain confluence. Neutrophils are subsequently added to the upper chamber (basolateral surface of the epithelia) with a chemoattractant added to the lower chamber (apical surface of the epithelia) such that neutrophils are induced to migrate in the physiologic basolateral-to-apical direction (Figure 2). Some investigators have developed even more elaborate systems in which endothelial and epithelial cells are grown on opposite sides of a semipermeable membrane and neutrophils are induced to migrate sequentially across the endothelial and then the epithelial layer (22, 23).
In the first stage of transepithelial migration in vivo, neutrophils adhere to the basolateral epithelial surface (24) via β2 integrins (19). In most epithelial cell types, CD11b/CD18 is the critical molecule involved in the initial adhesion of neutrophils to the basolateral surface (14, 19, 25). This is in contrast to neutrophil–endothelial adhesion, which is mediated by the combined effects of CD11a, b, and c (26, 27). In intestinal and urinary epithelium, neutrophil adhesion is mediated almost exclusively by CD11b/CD18 while CD11a and CD11c appear to have little or no role (19, 28). In the bronchial and alveolar epithelium, β2 integrins are similarly critical for neutrophil adhesion (29, 30), with CD11b/CD18 playing the predominant role (31, 32), although CD11a and CD11c appear to participate under certain circumstances (21, 33).
In the presence of inflammatory stimuli, neutrophil–epithelial adhesion is increased in a β2 integrin–dependent manner (34, 35). Moreover, β2 integrin binding is a prerequisite to neutrophil migration across both the intestinal and airway epithelium, as blocking antibodies against either subunit almost completely prevent neutrophil migration (19–21). Further confirming the critical role of CD11b/CD18 in transepithelial neutrophil migration, neutrophils from patients with leukocyte adherence deficiency (LAD), a genetic disorder characterized by the absence of CD11/CD18 expression on the neutrophil cell surface, fail to migrate across a cultured GI epithelial monolayer (19). Because of this defect in neutrophil migration, patients with LAD suffer from recurrent mucosal infections (36).
Despite the overwhelming dependence of neutrophil migration on the β2 integrins, CD18-independent transmigration can occur in some situations (37–39). In the intestine, β1 and β3 integrins play no role in neutrophil transepithelial migration (15). Further, there are reports of CD18-independent neutrophil migration in the lung (36, 40–43). Because neutrophils emigrate primarily through capillaries in the lung (44) (in contrast to postcapillary venules in the systemic circulation), and because the pulmonary circulation is characterized by lower pressures and close proximity to the epithelium, it is perhaps not surprising that neutrophil migration into the lung involves distinct mechanisms compared with neutrophil transmigration in other organs such as the intestine and kidney.
CD29 (β1 integrin) appears to be involved in neutrophil transmigration in the lung, although its mechanisms of action may involve adhesion to fibroblasts or interstitial matrix rather than epithelial cells (39). Neutrophil CD44, a glycosylated membrane receptor implicated in a variety of cell adhesion events including neutrophil adhesion to endothelium (45), also plays a role in transepithelial migration, where its activation negatively regulates migration (46, 47). While selectins mediate neutrophil adhesion to the endothelium (11), they do not appear to be involved in adhesion to or transmigration through the epithelium (15, 48, 49).
The differences in cell surface molecules involved in initial adhesion of neutrophils to the epithelium versus endothelium are perhaps to be expected, given that transepithelial migration proceeds in the basolateral-to-apical direction, while transendothelial migration proceeds in the apical-to-basolateral direction (15). Furthermore, because neutrophil adhesion to the endothelium but not the epithelium occurs in an environment of shear force due to blood flow, distinct mechanisms are required (15, 50).
It has long been known that neutrophil CD11b/CD18 binding to epithelial surfaces is critical to neutrophil–epithelial adhesion and subsequent transmigration, and epithelia clearly express CD11b/CD18 ligands on their basolateral surface (51). However, the search for the epithelial counter-receptor for CD11b/CD18 has proven elusive, in part because there may be multiple CD11b/CD18 ligands so that standard inhibitory studies are unrevealing (15, 49). One obvious candidate is ICAM-1, a known CD11b/CD18 ligand (52) that is critical for neutrophil adhesion to and migration across the endothelium (53, 54) as well as transepithelial T cell (55) and possibly eosinophil (56) migration. In fact, inflammatory stimuli up-regulate ICAM-1 on the surface of intestinal (25, 57, 58), conjunctival (57), renal, and bladder (59–61), as well as tracheal (32, 33, 62), bronchial (31, 63–65), and alveolar (66, 67) epithelial cells. Furthermore, neutrophil adherence to the apical surface of cultured tracheal (33), bronchial (31, 68), and intestinal (58) epithelial cells is ICAM-1 dependent. Finally, ICAM-1 inhibition may have a protective effect in lung injury (69, 70) and sepsis (71), although there are conflicting reports (70) and these results may be due to effects on endothelial rather than epithelial ICAM-1 (72).
Despite the evidence suggesting that ICAM-1 mediates neutrophil–epithelial adhesion, ICAM-1 is expressed exclusively on the apical epithelial surface of both intestinal (25, 58) and alveolar (66, 67) epithelial cells, thus precluding a role in initial adhesion to the basolateral surface during transepithelial migration. In fact, studies using blocking antibodies have definitively established that ICAM-1 binding is not critical to neutrophil migration across intestinal epithelium (19, 35). Still, ICAM-1 may play a role in neutrophil transmigration across other types of epithelia. In the lung, although ICAM-1 is unlikely to mediate neutrophil transmigration across the alveolar epithelium (67), there is some controversy in the literature regarding the role of ICAM-1 in neutrophil transmigration across the airway epithelium, where ICAM-1 is expressed on the basolateral surface (55). Whereas Liu and colleagues found only a very minimal reduction in neutrophil transmigration across a bronchial epithelial monolayer (20), there are two reports of substantial inhibition of neutrophil transmigration by ICAM-1 antibodies (21, 73). Finally, neutrophil transepithelial migration across uroepithelial cells (28) and possibly the gingival epithelium (74) appears to be mediated by ICAM-1.
VCAM-1 expression in bronchial (65) and renal tubular (59) epithelial cells is induced by exposure to inflammatory stimuli in some reports but not in others (31), and there are reports that neutrophil–epithelial adhesion is VCAM-1 dependent (75). However, this interaction has not been shown to be important in neutrophil transepithelial migration.
Other candidate epithelial ligands for β2 integrins include the JAMs, transmembrane junctional proteins that are members of the Ig superfamily. JAM-A, a ligand for CD11a/CD18 (76), and JAM-C, a ligand for CD11b/CD18 (77), mediate transendothelial neutrophil migration (76, 78, 79) although there are conflicting reports (80, 81). Since JAM-A (82) and JAM-C (83) are expressed on epithelial cells, it is plausible that they serve as ligands for CD11b/CD18. However, JAM-A–blocking antibodies do not impede neutrophil transmigration across the intestinal epithelium (84), and no role for JAM-A in neutrophil transmigration across the alveolar epithelium has been demonstrated (85). Still, JAM-A is also expressed by neutrophils (84), and while epithelial JAM-A is apparently not involved in transepithelial migration, neutrophil JAM-A may play a role in the final stages of transepithelial migration by mediating neutrophil detachment from the epithelium (86).
By contrast, JAM-C plays a critical role in neutrophil transmigration across epithelial cells (83). Furthermore, epithelial JAM-C binds specifically to neutrophil CD11b/CD18 and mediates neutrophil adhesion to and migration across epithelia via direct ligation of CD11b/CD18 (83). However, JAM-C is expressed at the level of the desmosome and its binding to CD11b/CD18 occurs at sites distal to the initial adhesive interactions with the basolateral membrane of the epithelium (87). Furthermore, JAM-C blockade only partially (~ 50%) decreases neutrophil transepithelial migration and does not inhibit initial adhesion to the epithelial monolayer, suggesting the existence of other, uncharacterized epithelial ligands that bind CD11b/CD18 before JAM-C. Nonetheless, JAM-C is an important ligand for CD11b/CD18 and is critical to subsequent steps in transepithelial neutrophil migration. Although attributed to its role in transendothelial neutrophil migration, the importance of JAM-C in an in vivo murine model of acute pulmonary inflammation (88) may also be related to its role as a CD11b/CD18 ligand during transepithelial neutrophil migration.
Since CD11b/CD18 can also bind carbohydrates (89), which are known to be involved in neutrophil adhesion to the endothelium (12), some have postulated that epithelial cell surface carbohydrates may mediate neutrophil transmigration via binding to CD11b/CD18. Indeed, Colgan and colleagues demonstrated that carbohydrate interactions are involved in neutrophil transepithelial migration, as pretreatment with various polysaccharides inhibited neutrophil transmigration (48). Moreover, several carbohydrates similarly inhibit epithelial cell adhesion to purified CD11b/CD18, suggesting that the mechanism by which soluble carbohydrates inhibit neutrophil transepithelial migration is by interfering with binding of neutrophil CD11b/CD18 to cognate ligands on the epithelial cell surface (90). In addition, of the various carbohydrates tested, fucoidin, which binds relatively specifically to CD11b/CD18, was the most potent inhibitor of epithelial binding to CD11b/CD18 and fucosidase treatment or inhibition of proteoglycan synthesis of the epithelial cells reduced neutrophil adhesion, suggesting that fucosylated proteoglycans are the epithelial ligands for neutrophil CD11b/CD18. Several fucosylated proteoglycans are expressed on the epithelial cell surface (91), but the identity of the specific molecule(s) that binds to CD11b/CD18 is unknown, although several potential candidates have been identified but not fully characterized (90). Notably, despite their role in transmigration via CD11b/CD18 binding, soluble carbohydrates did not prevent neutrophil adhesion to the epithelium (48). This suggests that the role of carbohydrates in neutrophil transmigration occurs during later stages, after firm adhesion has been established, and that other CD11b/CD18 ligands mediate the initial binding phase.
Nectins are transmembrane proteins involved in epithelial cell adhesion, but whether they serve as ligands for CD18/CD11b or are otherwise involved in epithelial–neutrophil adhesion has not been determined (14).
In summary, CD18/CD11b is clearly the most important neutrophil surface adhesion molecule mediating the initial adhesion of neutrophils to the basolateral surface of the epithelium. There are multiple epithelial ligands for CD18/CD11b, including fucosylated proteoglycans and JAM-C. The role of ICAM-1 in initial adhesive events remains uncertain. Finally, other known CD18/CD11b ligands, including ICAM-2, IC3b, heparin, and fibrinogen, do not appear to be involved in this process (15, 50).
Once neutrophils have firmly adhered to the basolateral surface of the epithelium, they begin to migrate across the epithelial monolayer through the lateral paracellular space (transmigration tunnel). It is well established that neutrophils transmigrate across the epithelium exclusively via a paracellular rather than transcellular route (24, 50, 92–94). This is in contrast to transendothelial neutrophil migration, which can occur via either paracellular or transcellular routes (95). Interestingly, neutrophils tend to migrate in clusters (19, 85, 96) and preferentially at the tricellular corners, which are favorable for transmigration given the discontinuous nature of the tight junctions at these sites (85). In the lung, neutrophils migrate through tricellular corners located at the junctions of two alveolar type I cells and one alveolar type II cell (97) and may be guided by pulmonary fibroblasts (16, 98).
The process of neutrophil transmigration also depends on sequential adhesive interactions as the neutrophils crawl through the “tunnel” between epithelial cells. One of the cell surface molecules involved in this process is CD47, which has a well-characterized role in transendothelial migration (99). CD47 is expressed both on the basolateral surface of epithelia and on neutrophils (100), and its expression in epithelium is up-regulated experimentally in response to inflammatory stimuli (101) and in inflammatory bowel disease (IBD) (102). Further implicating its role in transmigration, CD47-deficient mice demonstrate defective neutrophil accumulation in a murine model of Escherichia coli peritonitis, resulting in enhanced bacterial growth and increased mortality (103). In vitro studies initially showed that blocking antibodies against CD47 inhibited neutrophil transmigration across an intestinal epithelial monolayer (100), but it subsequently became clear that transmigration was delayed rather than fully prevented (101). Interestingly, both epithelial and neutrophil CD47 are involved in the process of transmigration across the intestinal epithelium (100, 101). In the distal lung (alveolar) epithelium, CD47 plays a role in monocyte transmigration (104), but its role in neutrophil transmigration has not yet been demonstrated.
The mechanism by which CD47 mediates transepithelial neutrophil migration has not been fully defined, although CD47 does not appear to be a ligand for CD11b/CD18. Furthermore, blocking antibodies do not interfere with neutrophil adhesion to the epithelial surface (100), and epithelial cells do not adhere to purified CD47 (15), suggesting that the role of CD47 in transepithelial migration occurs subsequent to the initial adhesive events. In fact, blocking antibodies result in accumulation of neutrophils within the epithelial monolayer (in the inter-epithelial tunnel) despite decreased transmigration to the apical surface (100). Therefore, some have suggested that CD47 might function to trigger de-adhesion from CD11b/CD18 (49).
Neutrophil CD47 plays a critical role in transepithelial migration, as CD47 is redistributed to the neutrophil cell membrane during transmigration and preincubation of neutrophils with blocking antibodies inhibits transmigration across either epithelial monolayers or cell-free filters (101). It appears that in neutrophils, CD47 ligation triggers downstream signaling cascades including tyrosine kinase-mediated events, leading to enhanced neutrophil migration, presumably via regulation of cytoskeletal reorganization (101). In addition, CD47 on the neutrophil cell surface binds SIRPα in cis (i.e., also on the neutrophil cell surface) and regulates neutrophil migration in part via SIRPα-dependent mechanisms (105, 106). Since SIRPα interacts with SHP-1 and SHP-2, it may be that the signal transduction–modulating effects of CD47 are mediated through SIRPα and SHP-1/SHP-2. The role of SIRPα in regulation of neutrophil migration may extend beyond its role as a ligand for CD47, as it appears to regulate transmigration by PI3K-dependent, tyrosine kinase–independent mechanisms, whereas tyrosine kinases but not PI3K appear to play a role in CD47-mediated signaling pathways. Furthermore, antagonism of SIRPα results in inhibition of neutrophil transmigration rather than the delay in transmigration attributable to CD47 (105). In addition, the related protein SIRPβ1, which is not a CD47 ligand, also regulates neutrophil transmigration, probably in an inhibitory fashion via signaling through an unidentified adaptor protein (107). This evidence further suggests that SIRP family members control neutrophil transmigration via CD47-independent mechanisms.
The role of epithelial CD47 in facilitating neutrophil transmigration has not been elucidated, although it is clear that epithelial CD47 is critical, as preincubation of epithelial cells with a CD47-blocking antibody inhibits transmigration (100, 101) and neutrophil transmigration across CD47-deficient epithelia is increased after transfection with CD47 (101). Further, neutrophil migration into the alveolar space and the attendant increase in lung permeability in response to LPS on Gram-negative bacteria is attenuated in CD47-deficient mice, an effect that is attributable to neutrophil (as opposed to endothelial or epithelial) CD47 (280). However, little is known about the mechanisms by which epithelial CD47 mediates neutrophil transmigration. Some hypothesize that neutrophil SIRPα binds epithelial CD47 in trans, suggesting that this interaction may initiate functional responses in the epithelium during leukocyte transepithelial migration (14).
In addition to CD47 and SIRPα, neutrophil migration through the epithelial monolayer is mediated by another member of the JAM family, JAM-like protein (JAML), known to be expressed on neutrophils and to mediate neutrophil adhesion to the endothelium (108). It has recently been shown that neutrophil JAML binds to the coxsackie and adenovirus receptor (CAR), an Ig superfamily receptor expressed at epithelial tight junctions, and that this interaction is important for neutrophil transmigration (109).
In summary, after initial adhesion to the basolateral epithelial surface, neutrophils migrate across the epithelium via the interepithelial tunnel by mechanisms using the cell surface molecules CD47, SIRPα, and SIRPβ, followed by binding of neutrophil JAML to CAR.
Once neutrophils have completely traversed the epithelial monolayer, they participate in adhesive interactions with the apical epithelial surface such that they are retained on the lumenal side despite fluid flow (pulmonary edema in the lung and diarrhea in the gut) and mechanical forces (cough in the lung and intestinal peristalsis in the gut). In this location, neutrophils can constitute a defense barrier and perform their essential function of eradicating invading microorganisms (14). In fact, in patients with inflammatory bowel disease, high numbers of neutrophils are present in the lumen bound to the apical surface of the epithelium (102). Further, neutrophil adhesion to the luminal surface may be one reason why bronchoalveolar lavage (BAL) recovers only a small percentage of alveolar neutrophils (44). Binding of neutrophils to the apical epithelium is also mediated by specific adhesion molecules, which have begun to be elucidated. As discussed above, ICAM-1, an adhesion molecule expressed predominantly on the apical surface of epithelial cells (25, 58, 67), is up-regulated in response to inflammatory stimuli (25, 58, 61, 63, 64). Importantly, ICAM-1 is a known ligand of neutrophil CD18/CD11b (52) and mediates neutrophil–epithelial adhesion (58). While ICAM-1 is aptly suited to serve in tethering neutrophils at mucosal surfaces, this remains to be proven (14, 15, 49, 58, 85, 102).
In addition to ICAM-1, neutrophils may be retained at the apical epithelial surface via Fc interactions. Specifically, it is hypothesized that neutrophil Fc receptors bind to soluble antibodies that are in turn bound to specific ligands on the apical epithelial surface. Indeed, the intestines of patients with inflammatory bowel disease are characterized by auto-antibodies bound to the apical epithelial membrane (110). Furthermore, exogenous antibodies directed against antigens expressed on the apical surface of intestinal epithelial cells prevent neutrophil detachment from the epithelium after transmigration in experimental models (111). Whether prolonged retention of neutrophils via binding to the Fc domain of autoantibodies against lung epitopes contributes autoimmune pulmonary disease in such disorders as systemic lupus erythematosis is not known.
It is recently appreciated that mechanisms are necessary to clear neutrophils from the epithelial surface after successful transmigration. As neutrophils traverse the epithelium and arrive at the luminal (apical) aspect of the tissue, we presume that it is disadvantageous for prolonged exposure to activated neutrophils. A monoclonal antibody screen of apical neutrophil ligands revealed the existence of a mechanism to actively promote the clearance of neutrophils from the luminal surface of mucosal epithelia (112). This antibody (clone OE-1) was found to recognize decay accelerating factor (DAF, also known as CD55). DAF is classically known as a complement regulatory protein, which inhibits complement-mediated cell lysis and is highly expressed on the apical aspect of mucosal (e.g., lung and intestine) epithelial cells. While not fully understood at present, mapping of the binding site for the OE-1 antibody revealed that DAF may compete with ICAM-1 to dispel neutrophils from the epithelial surface after transmigration. CD97 expressed on neutrophils has been shown to bind and interact with DAF (113), although molecular details of this response are not well understood.
One of the principal functions of epithelia is to form a selective barrier between various body compartments of different composition or between the body and the environment, a function that is achieved by inter-epithelial junctions termed tight and adherens junctions (114). During leukocyte transmigration, the epithelial barrier must open (at least transiently) to allow the passage of leukocytes. During immune surveillance or even during a normal immune response to an invading pathogen, leukocytes migrate across epithelia without damaging the epithelia (6, 94, 115, 116). To achieve paracellular transmigration without damage to the epithelium, there are close cell–cell contacts and highly regulated mechanisms responsible for signaling the opening and closing of the tight junctions without compromising barrier function (24). Some evidence exists that neutrophils may actively “reseal” epithelial junctions after transmigration. For example, upon arrival at the apical membrane surface, neutrophils release adenine nucleotides (ATP and AMP), which are subsequently metabolized to adenosine (117). Adenosine liberated in this manner is then available to bind apically expressed adenosine receptors, wherein one of the functional endpoints is the reestablishment of epithelial tight junction complexes (118).
In pathologic states, the passage of large numbers of activated neutrophils can result in damage to the epithelium (119, 120), both due to the release of toxic substances and by the mechanical force exerted by the neutrophil pseudopod resulting in microscopic wounds in the epithelium (94, 114). This neutrophil-mediated damage to the epithelial barrier results in the paracellular permeability, which in turn leads to the leakage of fluids that characterize acute lung injury (ALI)/acute respiratory distress syndrome (ARDS) and diarrheal illnesses and facilitates the entry of bacterial toxins and microorganism into the tissues. We propose that there are three distinct mechanisms involved in opening the epithelium: (1) highly regulated disassembly and resassembly of TJ, (2) mechanical force resulting in epithelial wounds, and (3) degradative effects of neutrophil-derived mediators. We will discuss each of these mechanisms below. Furthermore, we propose that the distinct mechanisms by which neutrophils can migrate across the epithelial barrier may unify apparently conflicting reports of the effects of neutrophil transmigration on epithelial permeability (6).
Although neutrophils transmigrate across the epithelium via the paracellular space (50, 92–94), the transmigration does not inevitably result in increased epithelial permeability, as demonstrated both in vitro (94, 115) and in vivo (6, 121). Small numbers of neutrophils can transmigrate without any change in permeability, as measured by electrical resistance and macromolecular flux (6, 15, 94, 120), and although larger numbers of transmigrating neutrophils (especially if activated) induce increased epithelial permeability, this can be transient and reversible (24, 93, 120, 122). This suggests that there are mechanisms by which tight junctions are opened and closed to allow for paracellular neutrophil transmigration without marked and permanent increased permeability or damage to the epithelium. There is a developing literature that begins to define the mechanisms by which neutrophil adhesion to the epithelium triggers intracellular signaling pathways that culminate in a rapid, organized opening and closing of tight junctions to allow for the passage of neutrophils with only a minimal transient or no increase in permeability.
Early ultrastructural studies revealed that transmigrating neutrophils make close contacts with other neutrophils and with adjacent epithelial cells, such that an barrier impermeable to macromolecules is maintained (24, 93, 94). Notably, the ultrastructural appearance of these intimate contacts is similar to those seen during neutrophil transendothelial migration (24), and it has been demonstrated that these points of intimate contacts are comprised by specific endothelial cell surface proteins, notably JAM A, which forms a transient ring around the transmigrating neutrophil (123). In addition, there are functional data to suggest that adhesion of neutrophils to the basolateral epithelial surface may be the critical event that triggers the opening and resealing of tight junctions sufficiently to allow for transmigration (115, 124). Adhesion of neutrophils to the basolateral epithelial surface results in epithelial permeability that is independent of neutrophil transmigration or soluble factors released by neutrophils (19, 24, 115, 124, 125). Probably because of this adhesion-mediated increased permeability, neutrophil transmigration is more efficient in the physiologic basolateral-to-apical direction (19). The increased permeability induced by early adhesive events suggests that neutrophil adhesion to the surface of epithelial cells triggers intracellular signaling events within the epithelial cell leading to opening of the tight junctions. Neutrophil adhesion to the epithelial surface triggers phosphorylation of myosin light chain (MLC) (124), which in turn regulates contraction of the actomyosin ring, leading to opening of the TJ (126). In addition, neutrophil adhesion stimulates tyrosine phosphorylation of TJ proteins (124), which is known to regulate permeability (124, 127). The specific kinases responsible for phosphorylation of MLC or TJ proteins are beginning to be identified (128, 129). Interestingly, immunofluorescence studies have revealed that adhesion-dependent epithelial paracellular permeability is not due to loss or redistribution of junctional proteins away from the junction, which occurs only during neutrophil transmigration (124). The regulated opening of the junctions is quickly followed by rapid closure of the junction, in which JAM-A plays a critical role (84). Consistent with this putative role, JAM-A–deficient mice display increased permeability in a model of intestinal inflammation (130). The closure of the junctions leads to a rapid resealing of the epithelium and only transient alterations in permeability (84), which is critical to maintain barrier function during immune surveillance under physiologic conditions.
While the mere adhesion of neutrophils to the epithelial cell surface can trigger rapid opening and closing of the tight junctions, resulting in only transiently increased permeability, neutrophil transmigration itself may result in a higher degree of permeability that may or may not be reversible. These effects are attributable to both the mechanical forces of transmigration with the creation of microscopic defects and to soluble factors released by activated neutrophils as they migrate. The magnitude of these effects and the rapidity of their repair determine whether gross changes in macromolecular permeability and disease progression result from neutrophil transmigration. We will discuss each of these in turn.
During transepithelial neutrophil migration, “scout” neutrophils are the initial cells to interact with and cross the epithelium individually. As transmigration progresses, neutrophils tend to migrate in clusters following in the “tracks” of the leading leukocytes (19, 96, 120). The migration of large numbers of activated neutrophils (“high-density migration”) results in the formation of large epithelial wounds due to the mechanical separation of epithelial cells at the site of the interepithelial junction (94, 96, 120). These “wounds” are thought to represent precursors of the macroscopic areas of denuded epithelium (ulcerated lesions) that characterize inflammatory mucosal diseases such as IBD and ALI/ARDS (15). Although there are conflicting reports regarding the contribution of neutrophil-derived soluble mediators to the creation of these wounds (24, 96), the preponderance of evidence suggests that these wounds are at least in part created by mechanical forces (24, 116, 120). The repair of these wounds, which is achieved initially by epithelial cell flattening, extension of lamellipodia, and contraction of actin/myosin rings (120), may determine the extent to which permeability persists, resulting in leakage of fluid in disease states such as diarrheal illnesses and ALI.
In addition to the effect of neutrophil adhesion on opening of TJ and the mechanical effects of neutrophil transmigration on epithelial permeability, soluble mediators released by transmigrating neutrophils may influence epithelial permeability and function. These soluble mediators include proteases, cationic peptides, and reactive oxygen species (ROS). The importance of soluble mediators in neutrophil–epithelial interactions is underscored by the fact that neutrophils do not release toxic substances while in circulation, but only once they are adherent to the endothelium, interstitium, or epithelium (131). These soluble mediators and their effects on the epithelium will be discussed below.
Of the neutrophil proteinases, elastase, a serine proteinase stored in azurophilic granules and released upon neutrophil activation, is the most completely characterized. The role of elastase in the pathogenesis of neutrophil-mediated inflammatory diseases, including ALI/ARDS and inflammatory bowel disease, is strongly supported in the literature. The BAL fluid (132–135) and plasma (136) of patients with ALI/ARDS are characterized by high levels of elastase, and these levels correlate with the severity of lung injury (134, 136). In animal models, elastase administration causes lung injury (137–139) and elastase inhibition has a protective effect on lung injury (140–144), as measured by several variables including permeability of the alveolocapillary membrane. There is also strong evidence to support the pathogenic role of elastase in the pathogenesis of inflammatory bowel disease (145–148) as well as cystic fibrosis (138, 149, 150) and chronic obstructive pulmonary disease (151, 152). Although the use of proteinase (elastase) inhibitors has had variable effects in ALI (153, 154), this may be because of limited study design (151), because patients are administered the inhibitor after the disease process has initiated (155), or because concentrated elastase activity may exist in a “protected space” between neutrophils and epithelia from which antiproteinases are excluded (156).
Yet, while elastase likely plays a pathogenic role in tissue destruction and permeability in some inflammatory diseases, it remains unclear whether this is due to the destructive effects of elastase on the epithelium per se in addition to its well-established degradative effects on the basement membrane matrix (151, 157) and endothelium (158, 159). The effects of elastase on the epithelium are controversial. We have used an in vitro model to demonstrate that elastase released by neutrophils traversing an epithelial monolayer induces increased epithelial permeability via reorganization of the actin cytoskeleton and the intercellular junctions of epithelial cells adjacent to transmigrating neutrophils (96). Importantly, these effects of elastase on epithelial permeability facilitate further neutrophil transmigration, resulting ultimately in the creation of circular defects (wounds) in the monolayer, as has also been observed by others (120, 160). Although there may be several distinct mechanisms by which elastase disrupts the apical junctions and increases epithelial permeability, we have described proteolytic cleavage of E-cadherin by elastase (96), analogous to the effect of elastase on endothelial VE-cadherin (158), as one potential mechanism. Thus, degradation or down-regulation of junctional proteins may enable migrating neutrophils to “loosen” the interepithelial junctions, thereby inducing an increase in epithelial permeability and promoting transmigration of trailing neutrophils (161, 162).
There are conflicting reports in the literature regarding the role of elastase in mediating the permeability induced by neutrophil transmigration. While elastase clearly induces epithelial permeability in several studies (96, 163), in other studies protease inhibitors failed to prevent the fall in resistance induced by neutrophil transmigration and therefore had no effect on the rate of transmigration (24, 115). The conflicting results of the various studies may be attributable to the specificity of the various in inhibitors used, the type of purified elastase used, the directionality of the transmigration experiments, the epithelial cell line used, the culture conditions (such as coating of inserts), the density of neutrophil transmigration, or the methods for measuring permeability (125) or transmigration. Another explanation for these discrepant results is that proteinases are released into a “protected space” between the epithelium and the transmigrating neutrophil, in which they can achieve high concentrations and from which larger molecules such as endogenous antiproteinases are excluded (40, 164). The extent to which experimental conditions compensated for this phenomenon may explain variable results.
In addition, increased epithelial permeability mediated by the opening of interepithelial junctions, diseases in which the pathogenesis involves transepithelial neutrophil migration, such as ALI/ARDS and inflammatory bowel disease (165), are often characterized by loss of epithelial cells resulting in not only permeability (166) but also denuded epithelium that can appear microscopically as ulcerations. In both ALI (167–173) and inflammatory bowel disease (174, 175), the loss of epithelial cells can be due to apoptosis, which is in part triggered by neutrophil transmigration (176). More specifically, recent studies from our laboratory have demonstrated that elastase secreted by transmigrating neutrophil is responsible not only for degradative effects on junctional proteins, leading to epithelial permeability (96), but also induces epithelial cell apoptosis. Importantly, elastase induces epithelial cell apoptosis via nondegradative mechanisms, by cleaving proteinase activated receptors (PARs), leading to the activation of specific intracellular signaling pathways that culminate in apoptosis (177, 178). In addition, PAR activation has also been shown to result in apoptosis and increased permeability in the intestinal epithelium (179). Further support for the role of elastase in epithelial cell apoptosis derives from in vivo models in which intratracheal elastase administration induces lung epithelial cell apoptosis (180). Interestingly, elastase has similar pro-apoptotic effects on the endothelium (181). In addition, elastase may also have direct cytotoxic effects on the epithelium (182, 183), although this is controversial (24, 160), and overall the epithelium seems to be less vulnerable than the endothelium to the cytotoxic effects of elastase (184). Finally, unopposed elastase activity may prevent epithelial healing (185).
Whether or not these effects of elastase facilitate or are required for neutrophil transmigration across the basement membrane (138, 186, 187), endothelium (155, 188–190), or epithelium (24, 96, 115) is controversial—although elastase is necessary for neutrophil migration into the lung in some circumstances (142, 151, 158, 191) but not in others (192, 193). Finally, in addition to its effects on epithelial permeability, elastase may also facilitate neutrophil transmigration by modulating de-adhesion from the epithelial surface (194, 195). Ongoing studies in our laboratory aim to elucidate the effects of elastase on epithelial intercellular junctions and apoptosis, the intracellular signaling mechanisms by which these effects occur, and the extent to which these effects facilitate neutrophil transmigration.
In addition to serine proteases, neutrophils produce matrix metalloproteinases (MMPs), which are also implicated in tissue injury in the acute inflammatory process (196, 197). The BAL fluid (198–200) and plasma (201, 202) of patients with ALI/ARDS have elevated concentrations of MMPs, which correlate with clinical severity (203), and MMP inhibition is protective in experimental models of lung injury (191, 196, 204, 205), although there is some conflicting data in the literature (206, 207). Importantly, the implicated MMPs are derived from neutrophils (197) as well as alveolar macrophages (208). Similarly, several MMPs, including neutrophil-derived MMPs, are up-regulated in both patients with inflammatory bowel disease (209, 210) and animal models of inflammatory bowel disease (211), and MMP expression correlates with disease severity (209, 210). Moreover, MMP deficiency or inhibition attenuates tissue injury in animal models of inflammatory bowel disease (211–214), although it is unclear whether MMPs derived from neutrophils (197, 209) or other cell types (211) are the most critical for disease pathogenesis.
While MMPs clearly degrade the extracellular matrix (ECM), with nearly every ECM component a potential substrate of MMPs (215), the effects of MMPs on the endothelia and epithelia are less clear. Certain MMPs do appear to play a role in maintaining epithelial integrity (211), in part via proteolytic cleavage of E-cadherin and occludin, leading to tight and adherens junction disassembly (215–217). By analogy, endothelial permeability is regulated by MMP degradation of occludin (218) and E-cadherin (219). The identity of the specific MMPs that target specific junctional proteins, resulting in increased permeability of various types of epithelium, remains to be determined.
In addition to mediating tissue injury, MMPs may also pave the way for neutrophil influx, as with elastase. MMPs have the potential to influence neutrophil migration to the inflammatory site through multiple mechanisms, including assembly of the actin cytokeleton, modulation of cell surface adhesion molecules, and proteolysis of the ECM (220). In some experimental models of ALI (204, 221, 222) but not in others (223, 224), MMP inhibition decreases neutrophil influx. This suggests that tissue injury induced by MMPs may indeed facilitate neutrophil transepithelial migration, as is the case with the transepithelial migration of other leukocytes (225) but not with transendothelial neutrophil migration (189). The diverse effects of the various MMPs on tissue injury and neutrophil migration have only begun to be elucidated (187, 211, 223, 226).
In addition to proteases, cationic peptides called defensins are a major component of azurophilic granules. Defensins are released principally into the phagolysosome but also into the extracellular space upon neutrophil stimulation (227). Defensins are antimicrobial against both gram-positive and gram-negative bacteria as well as fungi and enveloped viruses via permeabilization of their cell membranes (227). As with other antimicrobial mediators, defensins induce permeability in cultured epithelia via both cytotoxic (228–231) and noncytotoxic (232) mechanisms, and have been shown to cause endothelial injury as well (233). Moreover, when administered intratracheally to mice, defensins induce lung injury, as assessed by lung permeability, oxygen saturation, and mitochondrial damage, in a dose-dependent manner (234). Defensin concentrations have been found to be greatly elevated in the BAL fluid of patients with inflammatory lung diseases, including ARDS (235) and cystic fibrosis (230), and plasma defensin concentrations correlate with severity of lung injury (236). Taken together, this suggests that defensins are likely to play a pathogenic role in diseases characterized by neutrophil-mediated epithelial injury.
In addition to proteinases and defensins, neutrophil-derived oxidants are thought to play a major role in epithelial injury in neutrophil-mediated diseases, including ALI/ARDS and inflammatory bowel disease. Tissue injury due to oxidants is thought to be a key factor in the pathogenesis of ALI/ARDS (237), as levels of plasma (238) and lung (239–241) oxidants, likely of neutrophil origin (242), are increased in patients with ALI/ARDS and correlate with mortality (238). There are also data from clinical trials to suggest that antioxidant therapy may attenuate lung injury (243), although this data is inconclusive (244). Moreover, reactive oxygen and nitrogen species have been shown to induce lung injury, as assessed by permeability and histologic examination, in animal models of ALI (245–250), and the injurious oxidants have been shown to be neutrophil derived (251, 252). In one study, mice whose neutrophils are deficient in NADPH oxidase, the key enzyme necessary to generate a respiratory burst, were protected from increased lung permeability in a sepsis model (253). In inflammatory bowel disease, oxidant levels are elevated in inflamed intestinal mucosa (254, 255), and oxidant-induced injury likely contributes to mucosal injury and clinical exacerbations (256).
In addition to their indirect proinflammatory effects, including the activation of redox-sensitive transcription factors that up-regulate expression of various cytokines and chemokines (244, 257), oxidants have a direct effect on epithelial injury. Oxidants have been shown to induce epithelial cell death, either apoptotic (258–265) or necrotic (258–260), depending on the dose and duration of exposure (266), in both animal models of lung injury and in in vitro studies. Oxidants also increase epithelial permeability via disruption of tight junctions and redistribution of junctional proteins (267–270). The effects of oxidants on cell death, tight junction integrity, and permeability have been shown in various types of epithelial cells (115, 262, 267, 271, 272). Still, although there is extensive evidence that oxidants induce epithelial permeability, experiments with neutrophils from patients with chronic granulomatous disease suggest that oxidants are not the only factor inducing epithelial permeability during neutrophil transmigration (24) (see other mechanisms above), and neutrophil transepithelial migration may proceed independently of oxidant-induced junctional disruption (115). By comparison, although oxidants induce endothelial permeability via cell death (273–275) and disruption of tight junctions (276), neutrophil transmigration can induce endothelial permeability via mechanisms independent of oxidants (184, 277).
Much recent attention has been paid to understanding the innate mechanisms involved in the resolution of inflammation at mucosal sites. Of particular interest are series of lipid mediators termed the lipoxins and resolvins (278). Lipoxins are bioactive eicosanoids derived from membrane arachidonic acid by the combined action of 5-lipoxygenase (LO) and 12-LO or 15-LO (i.e., transcellualr biosynthesis). A number of in vitro and in vivo studies have revealed that lipoxins, and specifically lipoxin A4 (LXA4), serve as an innate “stop signal,” functioning to control local inflammatory processes (278). LXA4 has been demonstrated to inhibit neutrophil transmigration across both endothelia and epithelia both in vitro and in vivo. Synthetic lipoxin analogs exhibit greater potency for these actions than the native compound, likely due to a longer half-life consequent to decreased metabolism to inactive compounds.
An additional aspect of lipid metabolism was recently defined: that of inflammatory resolution via a switch from proinflammatory (e.g., leukotrienes and prostaglandins) to anti-inflammatory (e.g., lipoxins) lipid mediators. Such a switch occurred through temporal induction of 15-LO pathways via cyclic AMP responsive elements in the 15-LO gene and revealed that these functionally distinct lipid profiles drive neutrophils toward a program of inflammatory resolution (278). Extensions of these findings identified the “resolvins” as key control points for initiation of inflammatory resolution. The discovery of resolvins was based on a plethora of previous reports indicating that omega-3 polyunsaturated fatty acids (omega-3 PUFA) are beneficial to a number of cardiovascular and immunoregulatory responses. Ensuing studies revealed the existence of novel series of lipid mediators, derived from either eicosapentanoic acid (C20:5, 18-series resolvins) or docosahexaenoic acid (C22:6, 17-series resolvins), which potently initiate the resolution phase of acute inflammation. One such resolvin, termed RvE1, was recently shown to potently attenuate allergic airway inflammation in a murine model (279).
In summary, the pathogenesis of inflammatory diseases affecting mucosal surfaces in the lung and GI tract involves the disruption of endothelial and epithelial barriers. There is considerable evidence that a balance must exist between pro- and anti-inflammatory mechanisms during active states of disease. A tipping of this balance in either direction results in either establishment of chronic inflammation or the active resolution of acute inflammation. Herein, we have delineated the mechanisms by which neutrophils migrate from the intravascular space across the epithelium, including the roles of adhesion molecules such as β2 integrins, intercellular adhesion molecules, junctional adhesion molecules, carbohydrate ligands, and others. We have included evidence from models of both alveolar and bronchial epithelium as well as intestinal and other types of epithelium, and address the apparent similarities and differences between the interactions of neutrophils with the different types of epithelial cells as well as endothelial cells. We have discussed the mechanisms by which neutrophils injure the epithelium, including regulated disassembly of tight junctions, mechanical force, and degradative effects of soluble mediators including elastase, MMPs, defensins, and oxidants. This injury leads to epithelial cell apoptosis and sloughing, resulting in enhanced permeability, which in the lung allows for extravasation of edema into the alveolar spaces, manifesting clinically as bilateral infiltrates, compromised gas exchange, and diminished lung compliance. We hope that an improved understanding of the mechanisms by which neutrophil migrate across and injure epithelia will ultimately lead to the development of therapeutic endeavors with the goal to prevent or mitigate lung injury and hasten repair of the damaged lung.
This work was supported by grant #HL090669 from the National Institutes of Health to G.P.D. and HL60569, DK50189, and DE13499 to S.P.C.
Originally Published in Press as DOI: 10.1165/rcmb.2008-0348TR on October 31, 2008
Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript.