Receptor for Advanced Glycation Endproducts Mediates Neutrophil Migration across Intestinal Epithelium¹

Polymorphonuclear leukocyte (PMN,³ neutrophil) transmigration across mucosal surface epithelial lines plays a key role in many inflammatory processes such as inflammatory bowel diseases (IBD), cholangitis, cholecystitis, bronchial pneumonia, bronchitis, pyelonephritis, and cystitis. During inflammation, local activated PMN emigrate from microvascular endothelial blood vessels through interstitial tissues and migrate across mucosal epithelial monolayers in a continuous manner. This process requires sequential bidirectional interactions between heterodimeric β₂ integrins on PMN with proteins and carbohydrates on the surface of endothelial and epithelial cells (1, 2, 3). As demonstrated by transendothelial migration studies, the PMN cell surface β₂ integrins CD11a/CD18 (LFA-1) and CD11b/CD18 (Mac-1, CR3) interact with endothelial cell surface adhesion molecules such as ICAM-1, PECAM-1, and selectins (4, 5, 6) to regulate PMN rolling, adhesion, and subsequent transmigration across vascular endothelial monolayers. A key role for β₂ integrin in PMN transmigration across mucosal epithelium has also been demonstrated. Compared with the fact that PMN transendothelial migration uses both CD11a/CD18 and CD11b/CD18, PMN migration across epithelial monolayers requires only CD11b/CD18 (7, 8). Studies have shown that blocking CD11b/CD18 function completely abrogates PMN transepithelial migration (7, 9).

Similar to that of PMN migration across endothelium, multiple steps of PMN-epithelial adhesion mediated by interactions of PMN β₂ integrin with series heterogeneous epithelial ligands are proposed during the process of migration across epithelial monolayers (6, 10). It is convincing that known endothelial ligands such as cell adhesion molecules and selectins are not involved in PMN transepithelial migration (6, 11). Instead, several specific carbohydrate and protein molecules have been identified on epithelial cell surfaces that serve as potential CD11b/CD18 ligands and modulate PMN transepithelial migration. These include recently identified epithelial-fucosylated glycoproteins (1) and junction adhesion molecule (JAM) family protein (JAM-C) (12, 13, 14, 15). However, inhibitory Abs directed against these molecules only partially block PMN transepithelial migration, suggesting the existence of additional ligands for CD11b/CD18 along the epithelial paracellular pathway. Chavakis et al. (16) recently reported that the receptor for advanced glycation endproducts (RAGE) on activated endothelial monolayers serves as an adhesive receptor for β₂ integrins on PMN. This leaves the intriguing question of whether RAGE is also expressed in epithelia and serves as a counterreceptor for CD11b/CD18 during PMN transepithelial migration in the setting of mucosal inflammation.

RAGE is an Ig superfamily cell surface protein originally identified in endothelial cells that plays a crucial role in developing diabetic complications (17, 18, 19). Subsequent studies further defined RAGE as a pattern recognition receptor that binds to multiple other molecules such as amyloid components, proinflammatory cytokine-like mediators of the S100/calgranulin family, and amphoterin (high-mobility group box 1 protein) by recognizing specific three-dimensional structures on these proteins (20, 21, 22). Studies have found that RAGE expression is up-regulated in the diabetic vasculature and also in other tissues/cells under pathophysiological conditions (23, 24, 25). Cell surface ligation of RAGE results in rapid and sustained cellular activation through multiple intracellular signaling pathways that lead to the propagation of inflammatory responses (5, 26, 27). A recent finding that RAGE serves as a counterreceptor for β₂ integrin further suggests its involvement in inflammation, possibly, via additional mechanisms related to leukocyte recruitment and accumulation in the tissue (16).

Although RAGE is regarded as an important propagator of inflammation during diabetes (21, 28), its role in other settings, especially in other inflammatory conditions that are nondiabetic related, has not been fully explored. Recent work by Foell et al. (29) found increased expression of human PMN-derived S100A12 (EN-RAGE), a ligand of RAGE, in inflamed intestinal lumen from patients with IBD. Although this provides indirect evidence that RAGE may be involved in the disease activity of IBD, questions still remain regarding the expression of RAGE in normal intestinal epithelia and whether this expression is up-regulated under colonic inflammatory conditions such as IBD. In this study, we examined the expression and localization of RAGE in intestinal epithelia and investigated its involvement in intestinal inflammation. Our results demonstrate that normal intestinal epithelial cells express low-level RAGE that is localized at the cell-cell contacts close to the epithelial apical intercellular cell junctions. However, in inflammatory states, significant increases of RAGE were found in intestinal epithelia and human colonic tissues of patients with active IBD, including Crohn’s disease and ulcerative colitis (UC). Our observation of elevated RAGE expression in inflammatory epithelial cells supports the hypothesis that RAGE facilitates PMN transepithelial migration via interactions with the leukocyte β₂ integrin CD11b/CD18 during an inflammatory episode.

Human intestinal epithelial T84 cells were grown in DMEM/F-12 supplemented with 6% newborn calf serum or FBS. Two other lines of human intestinal epithelial cells, HT29 and CaCO2-BBE (9), and human kidney 293 cells (American Type Culture Collection; ATCC), were cultured in DMEM supplemented with 10% FBS. Human pulmonary epithelial cell, A549, was maintained in Ham’s F-12K medium with 2 mM l-glutamine, 1.5 g/L NaHCO₃, and 10% FBS. For transmigration experiments, epithelial cells were grown on collagen-coated, permeable polycarbonate filters with 5-μm pores (Corning; Fisher Scientific) as described previously (9). PMN were isolated from the whole blood of normal human volunteers (approved by the institutional review board committee) by Dextran and Ficoll sedimentation as previously described in detail (9). Isolated PMN were resuspended in HBSS devoid of calcium or magnesium (HBSS−) (4°C) at a concentration of 5 × 10⁷cells/ml and were used within 4 h after isolation. Human large intestine/colon tissue specimens from normal volunteer donors and active IBD patients were obtained from the National Disease Research Interchange of the National Resource Center (Philadelphia, PA). These tissue specimens, which were fresh-frozen or snap-frozen shortly after surgical removal, were sectioned at 5- to 10-μm thickness using a Leica CM3050S cryostat cryosectioning system (Leica Microsystems) and mounted on coverslides followed by air-drying before immunostaining was performed. Normal human colon sections were also purchased from Biochain. A mouse monoclonal (IgG2b) and a goat polyclonal Ab against human RAGE were obtained from R&D Systems. A high-titer polyclonal mouse anti-RAGE Ab was also generated in our laboratory by immunization with soluble RAGE extracellular domain (exRAGE) and glutathione S-transferase (GST) fusion protein. This Ab was used to confirm the expression of RAGE expression in epithelial cells and in tissue sections. The hybridoma of the inhibitory monoclonal anti-CD11b Ab 44aacb (also termed 44a, IgG2a) was obtained from ATCC, and the produced Ab was purified using protein A-conjugated Sepharose. The noninhibitory monoclonal anti-CD11b Abs, VIM12 and LM2/1 (30), and anti-CD11c Ab CBR-p150/4G1 were obtained from Abcam and Cell Sciences, respectively. Anti-E-cadherin (E-cad) and zona occludence protein-1 (ZO-1) Abs were purchased from Serotec. The monoclonal anti-JAM-A Ab J10.4 (IgG1), which binds to leukocyte cell surface but has no inhibitory effect (31, 32), and mouse isotype IgGs (IgG2a and IgG2b) (BD Biosciences) were used as controls in the protein binding, cell adhesion and cell migration assays. The goat anti-human CD55/DAF polyclonal Ab was obtained from R&D Systems. HRP-conjugated or Alexa Fluor 488 (495/519) and Alexa Fluor 568 (578/603)-conjugated secondary Abs and control mouse IgG were obtained from Molecular Probes. Recombinant human TNF-α and human IFN-γ were purchased from Upstate Biotechnology and used at 20 and 10 ng/ml, respectively. Purified human fibrinogen (FBG) was purchased from Sigma-Aldrich.

Soluble recombinant proteins consisting of the exRAGE and GST or His-tag were prepared. In brief, cDNA encoding the extracellular domain of human RAGE was amplified by PCR from a human colon Marathon cDNA library (Clontech Laboratories) or RT-PCR amplified from T84 epithelial cells using the following primers: 5′-atatcttaagaccctggaaggaagcaggatg and 5′-atatagatctttcgatgatgctgatgctgac. To produce exRAGE-GST fusion, a DNA fragment corresponding to the exRAGE devoid of the signal peptide was amplified using an additional sense primer 5′-atatgaattcgctcaaaacatcacagcccggat and the same antisense primer and inserted downstream of the GST moiety in the pGEX-4T1 vector (Amersham Biosciences) through EcoRI and XhoI sites. After expressing in bacteria (BL21), the exRAGE-GST was purified on glutathione-agarose (Sigma-Aldrich) and eluted with 10 mM reduced glutathione. Purity was verified by SDS-PAGE and Western blot analysis. Soluble exRAGE-His tag chimeras were also prepared as described previously (33). This recombinant exRAGE was previously tested for binding to several RAGE ligands, including AGE-BSA, amyloid fibrils, and glycosaminoglycans (33).

Cells (1–2 × 10⁶) were solubilized in 1× SDS-containing sample buffer and boiled under nonreducing or reducing conditions. Equal amounts of proteins from different cell types were analyzed by SDS-PAGE with 10% acrylamide gels. For Western blotting experiments, proteins on acrylamide gel were transferred onto nitrocellulose. After blocking with 5% nonfat milk, the membranes were blotted with monoclonal (2–5 μg/ml) or polyclonal anti-RAGE Ab (1/1000 dilution) for 1 h (25°C) or overnight (4°C). After washing, membranes were incubated with HRP-conjugated goat anti-mouse or donkey anti-goat secondary Ab, followed by detection using ECL reagents (Amersham Biosciences).

Immunofluorescence labeling of epithelial cell monolayers cultured on permeable supports was performed as previously described (12, 22). Briefly, monolayers were fixed and permeabilized using cold ethanol (−20°C, 20 min). The monolayers were subsequently blocked with 5% normal goat serum or BSA in HBSS then incubated with primary Abs followed by Alexa Fluor-conjugated secondary Abs. Monolayers were then mounted in the ProLong antifading embedding solution (Molecular Probes) and analyzed using a confocal microscope (Zeiss). Images shown were representative of at least three experiments with multiple images taken per slide. For Z-series, optical sections were recorded at 0.5-μm intervals. As a control for background labeling, monolayers were incubated with comparable concentrations of irrelevant IgG and secondary Abs. Frozen human large intestine/colon tissue sections were fixed with 100% ethanol, or fixed with 3.7% paraformaldehyde (5 min, 20°C) and permeabilized with 0.03% Triton X-100 (5 min, 4°C), followed by immunofluorescence staining and analysis by confocal microscopy.

To assay RAGE expression in T84 cells after inflammatory cytokine treatment, we incubated T84 monolayers with a culture medium containing TNF-α (20 ng/ml) or IFN-γ (10 ng/ml), or a combination of the two, for 24, 48, and 72 h, respectively (34, 35). After washing, the monolayers were fixed, blocked with 5% normal goat serum or BSA, followed by incubation with anti-RAGE Abs and labeling with Alexa Fluor-conjugated secondary Abs. In parallel experiments, anti-RAGE Ab binding was also detected by a peroxidase (HRP)-conjugated secondary Ab followed by color development using ABTS and OD measurement at 405 nm. The OD reading from control IgG binding served as the background and was subtracted from the values for the experimental group.

Functionally active CD11b/CD18 and CD11c/CD18 were purified to homogeneity from large quantities of human PMN (∼10¹⁰ cells) by immunoaffinity chromatography (1, 8, 12, 31, 36). To coat on microtiter plate well surfaces (96-well plate, flat-bottom; ICN Biomedical), purified β₂ integrins (∼100 μg/ml) in 1% N-octyl-β-d-glucopyranoside-containing buffer (150 mM NaCl, 2 mM MgCl₂, 2 mM CaCl₂, and 100 mM Tris (pH 7.4)) were diluted 20-fold with HBSS and immediately added to the wells (50 μl/well) followed by incubation overnight (4°C) to allow protein binding. After blocking the nonspecific binding by 1% BSA, the wells were then incubated with exRAGE-GST or His-tag chimeras (5 μg/ml) in the presence or absence of inhibitory reagents for 1 h at 37°C. Wells were then washed with HBSS, and the bound exRAGE chimeras were detected by an anti-GST Ab (for exRAGE-GST) or an anti-His tag Ab (for exRAGE-His) followed by HRP-conjugated secondary Abs and color development using ABTS. Nonspecific binding to BSA-coated wells was used as blanks and subtracted from the experimental wells to calculate specific binding.

Adhesion of T84 cells to immobilized CD11b/CD18 was performed as described previously (1). Briefly, 96-well plates coated with CD11b/CD18 were blocked by 1% BSA in HBSS for 1 h at 20°C. In the meantime, T84 cells were elicited by trypsin/EDTA and incubated with 5 μg/ml 2′,7′-bis-(2-carboxyethyl)-(and-6)-carboxyfluo-rescein (BCECF), acetoxymeyhyl ester (Molecular Probes) in HBSS for 15 min at 37°C. After removing free dye by washing, fluorescence-loaded T84 cells were incubated in CD11b/CD18-coated wells (∼2.5 × 10⁵ cells/well in 150 μl) with or without Abs or reagents, in a stationary place at 37°C for 1 h to allow for cell adhesion. To quantify cell adhesion, wells were gently washed three times, and the fluorescent intensity of each well before and after washing was determined using a fluorescent plate reader at excitation/emission wavelengths of 485/535 nm (Biotek). Cell adhesion to BSA-coated wells served as the blank control.

PMN transepithelial migration experiments were performed using confluent T84 cell monolayers cultured in an inverted fashion on collagen-coated Transwells as described previously (9, 37). With this setup, PMN (10⁶ cells) were added to the upper chamber of T84 monolayers in 150 μl of HBSS with and without Abs or recombinant proteins. PMN transmigration across cell monolayers in a physiologically relevant basolateral-to-apical direction was induced by adding the chemoattractant fMLP (1 μM) into the lower chamber (9, 37). After incubation with the Transwell setups at 37°C for 0.5-, 1-, and 2-h time periods, PMN migration into the lower chambers were quantified by myeloperoxidase assays (9, 37).

Recent studies demonstrated that in endothelial cells, RAGE serves as an extracellular ligand for the β₂ integrins CD11b/CD18 and CD11c/CD18 during leukocyte transendothelial migration (16). Given the fact that CD11b/CD18 plays a key role in PMN migration across intestinal epithelium, we evaluated the expression of RAGE in intestinal epithelial cells and its ability to serve as a ligand for CD11b/CD18 during PMN transepithelial migration. To examine RAGE expression in human intestinal cells, we performed PCR using oligonucleotide primers corresponding to multiple regions of RAGE and cDNAs derived from human colon tissues and human intestinal epithelial cells. As shown in Fig. 1,A, PCR amplified two major DNA fragments (arrows) with sizes of 1.2 and 0.9 kb from the human colonic cDNA (Clontech). DNA sequencing revealed that the amplified 1.2-kb fragment matched the full-length RAGE (GenBank accession no. AB036432) and the shorter sequence corresponded to a N-terminal-truncated RAGE form that had been reported previously (GenBank accession nos. AB061669 and AY755628). We then performed immunoblotting to assess the expression of RAGE protein in intestinal epithelial cells. As shown in Fig. 1,B, immunoblotting using a RAGE-specific Ab detected a prominent RAGE protein of ∼35 kDa from detergent lysates of two lines in human colonic epithelial cells, T84 and HT29. The observed RAGE-banding patterns from these colonic cells were virtually identical with that detected from lung epithelial cell A549 (Fig. 1), in which RAGE expression has been reported previously (38). Expression of RAGE in another colonic epithelial cell, Caco2, was also detected in the experiments (data not shown) (39). No RAGE band was detected from the kidney-derived cell 293 (Fig. 1,B), which agreed with previous reports by others (40, 41). Confocal fluorescence imaging of T84 monolayers after labeling with anti-RAGE Ab revealed that RAGE is a membrane component of epithelial lateral membranes. As shown in the en face and X-Z images (Fig. 1 C), significant amounts of RAGE were localized at the lateral membranes of T84 cells.

We further performed immunofluorescence labeling using Abs against RAGE and other membrane marker proteins to determine the localization of RAGE in intestinal epithelial (T84) monolayers. As shown in Fig. 2, double labeling using an anti-RAGE Ab and an Ab against the apical adherent junction marker, E-cad, revealed significant colocalization of these proteins, suggesting that RAGE is located at intestinal epithelial lateral membranes close to the apical cell junction complexes. A portion of RAGE seemed very close to the apical site of T84 monolayers with the pattern of staining similar, but not identical, to the staining of the tight junction marker ZO-1 (42) (Fig. 2), indicating that RAGE is not a tight junction-associated protein. Colocalization of RAGE with E-cad, but not with ZO-1, was further indicated in the X-Z images showing double staining patterns of RAGE/E-cad and RAGE/ZO-1 (Fig. 2, lower panel), respectively. RAGE labeling was also clearly different from staining with the apical membrane marker decay-accelerating factor (DAF) (4), confirming that RAGE is a not epithelial apical component.

As an inflammation amplification factor, RAGE expression was found increasing during inflammation (23, 43, 44, 45, 46). To test whether RAGE plays a role in intestinal inflammation, we examined RAGE expression in intestinal epithelial monolayers treated with or without proinflammatory cytokines. In these experiments, T84 monolayers were cultured in the presence or absence of TNF-α and/or IFN-γ for various time intervals (34, 35), and the expression of RAGE in T84 cells was analyzed by immunofluorescence staining. We observed that RAGE expression along the lateral membranes of T84 cells was significantly increased after treatment with either TNF-α or/and IFN-γ. Fig. 3,A shows the enhanced RAGE labeling in T84 after a 72-h treatment with a combination of TNF-α and IFN-γ. To quantitatively measure RAGE expression in TNF-α or IFN-γ-treated monolayers, we labeled the cytokine-treated and nontreated T84 cells with anti-RAGE Ab and detected the Ab labeling with a peroxidase (HRP)-conjugated secondary Ab followed by enzymatic-based color assays (detailed in Materials and Methods). As shown in Fig. 3 B, RAGE expression was clearly enhanced in T84 cells during the time course of TNF-α or IFN-γ treatment. With a combination treatment of the two cytokines, RAGE expression was increased near 3-fold after 72 h, suggesting a synergistic effect of IFN-α and IFN-γ. These data demonstrate that RAGE can be up-regulated in colonic epithelial monolayers under certain inflammatory conditions, implying that it may play a role in modulating intestinal inflammation and cellular dysfunctions.

The expression and localization of RAGE in intestinal epithelia was further examined by studying normal human colon tissue samples from two donors and inflammatory colon tissue samples from three IBD patients, including two patients with active UC and one patient with Crohn’s disease. As shown in Fig. 4, immunofluorescence staining of the normal human intestine tissue sections confirmed that RAGE is expressed at intestinal epithelial cell lateral membranes near the luminal side. The results of human colon tissue studies are thus in accordance with the RAGE labeling observed in cultured colonic epithelial monolayers (Figs. 1 and 2). Studies of inflammatory colon tissues revealed a significant increase in RAGE expression along the epithelial linings, which is in agreement with the observation that RAGE is up-regulated in proinflammatory cytokine-treated epithelial monolayers (Fig. 3). As shown in Fig. 4, in tissue sections from patients with UC (UC1 and UC2), RAGE staining was remarkably brighter and much more diffused compared with that in normal intestinal tissue sections. In the inflamed colon tissue sections, anti-RAGE labeled not only the lateral membranes of epithelial cells near the luminal side (Fig. 4), but also the basolateral membranes of the epithelial cells. A similar bright but diffused labeling pattern was observed in colon tissue sections from the patient with Crohn’s disease (Fig. 4).

Given the crucial role of β₂ integrin CD11b/CD18 in PMN transmigration across the intestinal epithelium (47) and that RAGE is a ligand for integrins during PMN transendothelial migration (16), we next investigated whether RAGE can serve as an adhesive ligand for CD11b/CD18 during PMN transmigration across the colonic epithelium. In vitro binding assays were first performed to confirm that CD11b/CD18 from PMN is capable of binding to the exRAGE. In the experiments, CD11b/CD18 purified from PMN (1, 12, 31) was immobilized in microtiter wells followed by incubation with soluble exRAGE fusion proteins. As shown in Fig. 5,A, soluble exRAGE (exRAGE-GST) directly bound to CD11b/CD18 in a dose-dependent manner. Compared with binding to CD11b/CD18, exRAGE binding to immobilized CD11c/CD18, another leukocyte β₂ integrin (36), was significantly less (Fig. 5,A). No binding of RAGE to immobilized BSA (control) was detected under the same conditions. Because β₂ integrin-mediated ligand binding is generally through the extracellular “I-domain” of CD11b (30, 48), we thus further tested the involved binding region of CD11b/CD18 and the binding specificity using Abs against different CD11b domains. As shown in Fig. 5,B, 44a, an inhibitory Ab against the I-domain of CD11b (49), completely blocked RAGE binding to CD11b/CD18. Conversely, VIM12, a noninhibitory Ab against the C terminus of CD11b (30), did not block the interaction, suggesting that RAGE specifically binds to the I-domain of CD11b. Similar inhibition were also observed using another I-domain-targeted inhibitory Ab CBMR1/29 (1, 12, 30, 49) (results not shown). In concordance with this observation, we also found that the binding of RAGE to CD11b/CD18 was blocked by EGTA (Fig. 5,B), which depleted the divalent cation (Mg²⁺) required for CD11b binding, and competitively inhibited by FBG, another known CD11b/CD18 ligand that binds to the CD11b I-domain (Fig. 5 B).

The physiological significance of RAGE-CD11b/CD18 interactions was characterized using cell adhesion assays. We evaluated the role of epithelial RAGE in mediating T84 cell adhesion to immobilized CD11b/CD18. In these experiments, T84 cells elicited by trypsin/EDTA were preloaded with fluorescent dye BCECF (Molecule Probes). Labeled T84 cells were added to 96-well microtiter wells coated with CD11b/CD18 or BSA (served as blank) and incubated for 1 h at 37°C. As shown in Fig. 6, T84 cells (no inhibitor) strongly adhered to the wells coated with CD11b/CD18 (47.1 ± 5.9% of total applied cells), whereas <6% of cells (5.9 ± 0.4% of total applied) bound to BSA-coated wells (blank). In addition, adhesion of T84 cells to the CD11b/CD18-coated surface was significantly inhibited by the anti-RAGE Ab and soluble exRAGE fusion proteins. As shown in Fig. 6,B, compared with the isotype-matched control IgGs or the control protein (GST only), cell adhesion was reduced to 25.5 ± 2.7% in the presence of the anti-RAGE mAb (10 μg/ml) and ∼20–28% in the presence of exRAGE chimeras (20 μg/ml each) (19.1 ± 2.3 and 27.9 ± 4.1% for exRAGE-GST and exRAGE-His, respectively). In the same experiment, the inhibitory anti-CD11b mAb 44a completely abolished cell adhesion (Fig. 6 B), indicating that the T84 cell adhesion was exclusively mediated by CD11b/CD18. Compared with the strong inhibition by anti-CD11b Ab, the less inhibition of T84 cell adhesion by anti-RAGE Ab and soluble exRAGE proteins is not surprising, given that other epithelial counterreceptors are also reported in colonic epithelial cells for CD11b/CD18 (1, 11, 12).

To answer whether RAGE-mediated adhesion is engaged in PMN transepithelial migration, we assayed the effects of anti-RAGE Abs and RAGE extracellular fusion protein in PMN migration across epithelial (T84) monolayers in a physiologically relevant basolateral-to-apical direction toward the chemoattractant fMLP (9, 37). As shown in Fig. 7, in the absence of inhibition (no inhibition), freshly isolated PMN showed a robust migration across T84 monolayers, and nearly 40% of total applied PMN transmigrated into the fMLP-containing lower chambers after 1 h (Fig. 7). As also seen, PMN transmigration was completely blocked by the inhibitory Ab 44a (Fig. 7), indicating that PMN migration across T84 monolayers is CD11b/CD18-dependent. In addition, both the anti-RAGE Ab and the exRAGE-GST fusion protein significantly inhibited the migration of PMN across T84 monolayers (Fig. 7). Serving as controls, both mouse isotype controls, IgG2a and IgG2b, and a mAb against human JAM-A, J10.4 (32, 50), showed no inhibition on PMN transepithelial migration. Given the fact that RAGE can serve as an adhesive ligand for CD11b/CD18, these results suggest that RAGE expressing on the epithelial cell lateral membranes plays an important role in facilitating PMN transepithelial migration, presumably through interactions with β₂ integrin CD11b/CD18 on migrating PMN.

It has been well established that RAGE serves as an amplification factor in various inflammatory cellular dysfunctions and is associated with the development of diabetic vascular complications, neurodegenerative disorders, rheumatoid arthritis, and other inflammatory diseases (23, 51, 52, 53, 54). As a member of the Ig superfamily proteins, RAGE is capable of binding to multiple ligands thus mediating diverse functions. Due to the critical role of RAGE in regulating diabetes-associated chronic inflammations, most research on RAGE has been focused on diabetes-related organs and cells. Although several studies have reported RAGE expression in some epithelial cells, including eye epithelial cells (55, 56), lung epithelial cells (38), and skin cells (57), so far there is no direct evidence of RAGE in the intestinal epithelium, in particular under inflammatory conditions. In this study, we demonstrated that RAGE is expressed in both cultured human intestinal epithelial cells and normal and inflammatory intestinal tissues. By immunofluorescence staining of epithelial monolayers and human colon tissue sections, we observed that RAGE is localized at the lateral membranes of epithelial monolayers and concentrated juxtapose to the epithelial apical junction complexes. We have also demonstrated that RAGE plays a significant role in regulating PMN transepithelial migration, presumably by serving as a counterreceptor for the leukocyte integrin CD11b/CD18.

Several lines of evidence derived from this study indicate that epithelial RAGE mediates PMN transepithelial migration via interaction with the PMN cell surface CD11b/CD18. First, in vitro direct binding assays showed a specific interaction between RAGE and CD11b/CD18 (Fig. 5). Our data further defined the major binding region on CD11b for RAGE is in the I-domain of CD11b (Fig. 5,B). Second, in vitro cell adhesion experiments clearly indicated that RAGE is capable of mediating epithelial adhesion to CD11b/CD18 (Fig. 6). Third, we observed that both anti-RAGE Ab and soluble exRAGE fusion protein significantly inhibited PMN migration across epithelial monolayers (Fig. 7). To exclude the possibility that inhibition of anti-RAGE Ab on PMN transmigration is due to the effect of Fc, we used mouse isotype-matched IgGs and noninhibitory mouse mAb (J10.4) as negative controls. Additional experiments using the Abs against PMN Fc receptor also indicated that PMN Fc receptor might not be involved in anti-RAGE-mediated PMN transmigration (data not shown). In addition, we also observed that RAGE expression in epithelial monolayers was significantly up-regulated after treatment with the inflammatory cytokines (Fig. 3). Enhanced RAGE expression was also found in colon tissues from patients with active IBD (Fig. 4). Given the fact that RAGE-mediated function is involved in propagating inflammatory responses in many other inflammatory diseases, increasing epithelial RAGE found in active intestinal inflammatory diseases suggests that it may play a role in augmenting intestinal inflammation. From our results, it is possible that RAGE on epithelial cells enhances inflammatory responses by facilitating leukocyte migration across the monolayers through interactions with CD11b/CD18 on PMN. Additional evidence supports a crucial role of RAGE in intestinal inflammation including a recent report by Foell et al. (29), indicating that expression of another RAGE ligand, S100A12, is also significantly increased in IBD patients.

Given the paracellular localization of RAGE on epithelial monolayers, the binding of RAGE with the PMN cell surface CD11b/CD18 and its role in modulating PMN transepithelial migration is particularly interesting. Like RAGE molecule, CD11b/CD18 can bind to a wide variety of ligands, including ICAMs (58), JAM-C (12, 13), FBG (59), factor H and X (60), heparin (2), leukocyte elastase (61), and others. Of all the ligands investigated so far, ICAM-1 is the best characterized. It has been well established that ICAM-1 plays an essential role in regulating leukocyte transendothelial migration via interactions with leukocyte β₂ integrin. Although ICAM-1 has been found in the epithelium, this molecule is expressed at the apical surface of epithelial monolayers, thus it is not suitable to serve as an adhesive ligand for PMN migration across epithelial monolayers in a physiologically relevant basolateral-to-apical direction (11). The cellular localization of RAGE that is physically close to epithelial apical junctions suggests that RAGE may not serve as the adhesive ligand that mediates the initial adhesion of migrating PMN to the epithelia. Instead, its function may resemble another CD11b/CD18 ligand, JAM-C. As shown in the previous study, JAM-C is a lateral membrane component that is concentrated at desmosomes of epithelial monolayers and serves as an adhesive ligand for CD11b/CD18 during PMN transepithelial migration (12).

Although this study focused on characterizing RAGE as an adhesive ligand that mediates PMN-epithelial cell adhesion via binding to CD11b/CD18, RAGE may also serve as an important signal transduction receptor during PMN transepithelial migration. Unlike PMN migration across endothelium, which may follow both paracellular and intercellular pathways (62), PMN transepithelial migration is generally believed to occur through the paracellular space and through the epithelial cell-cell junction complexes. Previous studies suggested that PMN-epithelial interactions during transepithelial migration trigger a transient opening of epithelial tight junctions as well as a temporary disruption of epithelial barriers to allow PMN to pass through (10, 63). However, the involved signaling molecules and their mediated signaling pathways are not clear. In endothelial cells, studies found that ligation of RAGE results in a disruption of the monolayer barrier function (26), leading to increasing endothelial permeability through activation of p38 MAPK (64) and protein kinase C (65). Because in epithelium, RAGE is localized juxtapose to the apical cell junctions, it would be interesting to know whether interactions of RAGE with CD11b/CD18 during PMN transmigration would initiate similar downstream signals that interfere the epithelial junction/barrier functions and thereby open the way for PMN to migrate across.

In summary, the present study provides evidence that RAGE is expressed at the lateral membranes of intestinal epithelial monolayers that is up-regulated under inflammatory conditions, and that RAGE serves as a novel adhesive counterreceptor for leukocyte CD11b/CD18 and facilitates PMN transepithelial migration. These findings demonstrate a new mechanism for modulating PMN transepithelial migration, and thus may provide an additional target for potential anti-inflammatory therapy for IBD.

The authors have no financial conflict of interest.