Transepithelial Neutrophil Migration Is CXCR1 Dependent In Vitro and Is Defective in IL-8 Receptor Knockout Mice¹

Most infections start at mucosal surfaces, where the pathogens make contact with the epithelial barrier and trigger inflammation. In response to chemotactic signals, neutrophils leave the blood vessels, traverse the lamina propria to the epithelial barrier, and cross the polarized epithelial cell layer into the lumen. This process is strictly regulated at the molecular level, through the sequential elaboration of chemokines and chemokine receptors, both by the neutrophils themselves and by the resident tissue cells. Although there is extensive information on the extravasation process and on interactions of neutrophils with endothelial cells, less is known about the molecular events that guide neutrophils through peripheral tissues to these infected mucosal sites.

Neutrophils are critical effector cells of the antibacterial defense of the urinary tract. Previous studies have shown that the genetic deficiency of the C3H/HeJ mouse causes a dysfunctional neutrophil response with decreased resistance to infection. In normal mice, neutrophil depletion was shown to impair bacterial clearance (1, 2). These observations illustrate the importance of understanding how infections initiate neutrophil recruitment and the molecular interactions of neutrophils with epithelial cells. In the urinary tract infection (UTI)³ model, we have studied the response of human kidney epithelial cells to infection with uropathogenic Escherichia coli (3, 4). The epithelial cells were shown to secrete chemokines and IL-8 was identified as the main chemokine involved in transepithelial neutrophil migration (4, 5). In vivo studies in the murine UTI model identified macrophage-inflammatory protein-2 as an epithelial IL-8 equivalent in the mouse urinary tract (2).

IL-8 and other neutrophil-directed chemokines mediate their biological responses by binding to cell surface chemokine receptors. Human neutrophils express two high affinity IL-8 receptors, CXCR1 (IL-8RA) and CXCR2 (IL-8RB), which have been cloned and extensively characterized (6, 7). They are members of the large family of serpentine receptors with seven transmembrane domains that couple to Bordetella pertussis toxin-sensitive heterotrimeric G-proteins for signal transduction (8). CXCR1 displays greater ligand specificity than CXCR2 and binds with high affinity to IL-8 and granulocyte chemotactic protein (9). CXCR2 binds multiple CXC chemokines in addition to IL-8, including epithelial neutrophil-activating peptide 78 (ENA-78), neutrophil-activating peptide-2, growth-related oncogene (GRO)-α, -β, -γ, and LPS-induced CXC chemokine. In contrast to humans, mice express one main CXC chemokine receptor that binds several IL-8-like CXC chemokines, including macrophage inflammatory protein-2, KC, ENA-78, and granulocyte chemotactic protein-2. The murine gene was recently deleted from the mouse genome by homologous recombination in embryonic stem cells, resulting in an IL-8R homologous knockout mouse (mIL-8Rh KO) (10). Neutrophils from these mice fail to migrate in response to the CXC chemokines, but have intact sensitivity for other activation pathways.

This study investigated the involvement of chemokine receptors in neutrophil interactions with mucosal barriers. The chemokine receptor expression by human urinary tract epithelium was examined, as was the role of these receptors for neutrophil migration across E. coli-infected epithelial cell layers in vitro. The in vivo relevance was confirmed in IL-8R KO mice. The results demonstrate that urinary tract epithelial cells express both CXCR1 and CXCR2, that receptor expression is increased by infection, and that chemokine receptors are critical for neutrophil migration across infected epithelial layers in vitro and in vivo.

Tryptic soy agar (TSA) and FCS were from Difco (Detroit, MI). Monoclonal mouse anti-human IL-8 (subclone 4G9/A5/A7) and polyclonal rabbit anti-human IL-8 Ab were kindly supplied by Sandoz (Vienna, Austria), and the anti-mIL-8Rh Abs were from Santa Cruz Biotechnology (Santa Cruz, CA). RB6-8C5, a rat IgG2b mAb specific for murine neutrophils was a kind gift from Dr. A. Sjöstedt (Umeå University, Sweden), Dr. W. Conlan (Trudeau Institute, Saranac Lake, NY), and Dr. R. Coffman (DNAX Research Institute, Palo Alto, CA). Polyclonal rabbit anti-rat Ab, monoclonal mouse IgG, FITC-conjugated goat anti-mouse IgG F(ab′)₂ Ab, and FITC-streptavidin were from Dakopatts (Stockholm, Sweden). FITC-labeled IL-8 was kindly provided by Glaxo Wellcome (Research Triangle Park, NC). Human monoclonal anti-CXCR1 Ab (clone 42705.111; MAB330 and FAB330F), human monoclonal anti-CXCR2 Ab (clone 48311.211; MAB331 and FAB331F), recombinant GRO-α, and ENA-78 were purchased from R&D Systems (Oxon, U.K.). Rabbit anti-mouse Igs, mouse APAAP complexes, Fast-Red substrate, and swine anti-rabbit Igs/FITC Igs were purchased from Dako (Copenhagen, Denmark). Mayer hematoxylin was from KEBO Laboratory (Stockholm, Sweden) and Mount-quick “AQUEOUS” was from Daido Sangyo (Tokyo, Japan). Buffered glycerol was purchased from ACO (Stockholm, Sweden) and contained 2% 1,4-diazabicyclo-(2, 2, 2)octane from Sigma (St. Louis, MO). RPMI 1640 and gentamicin were from Flow (Gothenburg, Sweden). Polymorphprep was from Nycomed Pharma AS (Oslo, Norway). Transwells (24.5 mm diameter, polycarbonate membrane with 3-μm pores) were from Costar (Cambridge, MA), and the culture flasks were from Falcon (Gothenburg, Sweden). Polyethylene catheters (outer diameter 0.61 mm) were purchased from Clay Adams (Parsipanny, NJ). Purified LPS was a kind gift from Dr. Alf Lindberg (Karolinska Institute and Hospital, Stockholm, Sweden).

E. coli Hu734 of serotype O75:K5:H-, hly⁺, ColV⁺ is a lac⁻ mutant of the wild-type pyelonephritis strain GR12 (11). E. coli 1177 of serotype O1:K1:H7 was isolated from a child with acute pyelonephritis. The strains were maintained by passage on TSA plates. For experiments, Luria broth was inoculated with colonies from the TSA plates and incubated overnight at 37°C. Under these conditions, E. coli Hu734 and E. coli 1177 both express P and type 1 fimbriae. Bacteria were harvested by centrifugation at 3000 rpm for 10 min (RP centrifuge; Hettich Rotanta, Malmö, Sweden). The pellet was resuspended in 0.01 M of PBS (pH 7.2) to a concentration of 1–2 × 10⁹ CFU/ml. The bacterial concentration was confirmed by viable counts.

The human A498 kidney epithelial cell line (HTB44; American Type Culture Collection, Manassas, VA) was grown in culture medium RPMI 1640 supplemented with 0.05 mg/ml gentamicin and 5% FCS. Confluent A498 epithelial cell layers were prepared on Transwell inserts as previously described (3). Briefly, 1 ml of cells (∼3 × 10⁵ cells/ml) were seeded onto inverted Transwell inserts and allowed to settle for 6 h at 37°C in a 5% CO₂ atmosphere. The remaining medium was then removed, and the Transwell inserts were moved to cluster plates containing fresh medium. Plates were incubated at 37°C in a 5% CO₂ atmosphere until the cell layers were fully confluent (14 days). When grown under these conditions, A498 cells formed confluent nonpolarized layers on both sides of the transwell filter (3).

The cell layers were infected with E. coli Hu734 for 24 h before the addition of neutrophils. At time 0 h, fresh culture medium (1.5 ml) was added to the top well, and fresh medium or medium containing stimulant (E. coli, 10⁸ bacteria/ml) (2.7 ml) was added to the bottom well of the transwell inserts. The cluster plates were then incubated at 37°C in a 5% CO₂ atmosphere for 24 h, a time previously shown to induce maximal neutrophil migration across A498 cell layers (3).

Cell layers were also exposed to the LPS preparations, at concentrations ranging from 10 ng/ml to 100 mg/ml.

Neutrophils were isolated from peripheral blood of healthy human volunteers as described (3). The cell suspension contained >97% neutrophils as determined by Wright’s Giemsa stain and was 98% viable as determined by trypan blue exclusion. Neutrophils (1.5–3 × 10⁶) were added to the infected cell layers after the spent medium had been removed from the top well, and replaced with medium containing neutrophils. Neutrophil migration was quantified by counting of the neutrophils that had passed from the upper to the lower well using a hemocytometer chamber. Migration was expressed in percent of the neutrophils added to the top well.

To evaluate the role of IL-8 receptors on epithelial cells, anti-CXCR1 or anti-CXCR2 Abs were added to the upper and lower compartments of infected epithelial cell layers. After 0.5 h, the cell layers were washed two times. Neutrophils were added to the upper compartment, and neutrophil transmigration was quantified as described above.

IL-8 receptor expression in the human urinary tract was examined using immunohistochemistry. Surgical specimens were obtained from patients with bladder and kidney cancer, and tissue sections were selected from the tumor-free zone, and the normal morphology was confirmed by the pathologist. Tissues were frozen and maintained as described (12).

Samples for immunostaining were fixed in 50% and 100% acetone for 5 min each, incubated with a 1:50 dilution of mouse anti-human CXCR1 and CXCR2 Abs for 3 h at room temperature, washed in PBS (three times), and incubated with rabbit anti-mouse Igs at a 1:50 dilution for 60 min at room temperature in a moist chamber. The sections were washed again in PBS and incubated with mouse APAAP complexes diluted 1:50 in PBS-Sap for 60 min as above. After washes in PBS, the Fast-Red substrate containing levamisole (prepared according to the manufacturer’s instructions) was added to the sections and left to incubate for 20 min. The sections were thereafter washed in TBS, pH 7.6, counterstained with Mayer hematoxylin for a few seconds, and mounted with Mount-quick “AQUEOUS”. The samples were investigated under a light microscope (Microphot, Nikon, LRI instrument AB). Mouse IgG1 at a final concentration of 5–10 μg/ml and a normal rabbit Ig fraction were used as negative controls.

The specificity of the anti-CXCR1 or CXCR2 Abs for the functional chemokine receptors was controlled by preincubation of the tissue sections with IL-8, GRO-α, or ENA-78 (10 μg/ml).

IL-8R expression by A498 epithelial cells was examined by confocal microscopy. Cell layers were fixed with 4% paraformaldehyde and stained with FITC-labeled monoclonal anti-CXCR1 or anti-CXCR2 (10 μl/ml) Abs. The cell nuclei were costained with propidium iodide. Finally, filters were washed, air dried on glass slides, and examined with an MRC-1024 confocal equipment (Bio-Rad Laboratories, Hemel-Hampstead, U.K.) attached to a Nikon Eclipse E800 upright microscope (Nikon, Tokyo, Japan).

For FACS analysis, the A498 cells were grown to confluency in 25-ml culture flasks at 37°C in a 5% CO₂ atmosphere. The cells were stimulated with E. coli Hu734 (10⁸ bacteria/ml) with medium as a negative control, detached from the flasks by incubation with PBS/EDTA (0.2 g EDTA, 5 g NaCl, 0.25 g KCl, 1.78 g Na₂HPO₄ × H₂O, 0.25 g KH₂PO₄, 1 liter H₂O) for 20 min at 37°C in a 5% CO₂ atmosphere and harvested by centrifugation at 1000 rpm (Hettich Rotanta RP centrifuge, Malmö, Sweden) for 10 min. The cell pellet was suspended in 20 μl RPMI containing 0.3% BSA (RPMI-BSA) and FITC-labeled anti-CXCR1 or anti-CXCR2 (10 μl/ml) mAbs and incubated on ice for 30 min. Cells were washed twice in RPMI-BSA and analyzed by flow cytometry in an EPICS Profile II instrument (Coulter, Miami, FL). A total of 5000 cells were counted in each sample.

Functional receptors for IL-8 were detected using FITC-labeled IL-8 (10 μg/ml) with FITC-streptavidin (10 μg/ml) as a negative control. Cells and FITC-labeled IL-8 were incubated on ice for 90 min, washed twice in RPMI-BSA, and analyzed by flow cytometry as described.

Breeding pairs of IL-8 receptor-deficient BALB/c mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Homozygous mIL-8Rh KO mice and the homozygous IL-8R proficient BALB/c mice were bred in the animal facilities at the Department of MIG, University of Lund. Female BALB/c and IL-8R KO mice were used at 9–16 wk of age.

E. coli Hu1177 were injected into the mouse urinary tract as previously described (13). Following ether anesthesia, 0.1 ml of the bacterial suspension was installed into the bladder through a soft polyethylene catheter. The catheter was removed immediately after inoculation, and no further manipulations were conducted. Infection was quantified by viable counts on tissue homogenates from mice sacrified by cervical dislocation under anesthesia at different times after inoculation. Kidneys and bladders were removed aseptically and homogenized in 5 ml of PBS in sterile disposable plastic bags with a Lab Stomacher 80 homogenizer (Seward Medical UAC House, London, U.K.). Serial dilutions, 50 μl, in sterile PBS were plated on TSA. After overnight culture, plates with 30–300 colonies were selected and the bacterial numbers were counted. The number of CFU/entire tissue was determined after adjustment for the dilution factor.

Urine samples were collected before infection, and 2, 6, 24 h, and 7 days after infection. Urination was induced by gentle compression of the mouse abdomen. Polymorphonuclear neutrophil numbers in fresh urine were counted microscopically using a hemocytometer chamber. The samples were centrifuged and the supernatants were stored at −20°C for subsequent chemokine analysis. Mice with more than 50 × 10⁴ leukocytes/ml in a preinoculation sample or with a positive urine culture were excluded.

IL-8 receptor expression in the mouse urinary tract was examined on kidney and bladder sections as described for human tissues. After freezing in nitrogen, the samples were fixed for immunostaining and incubated with anti-mIL-8R Abs in a 1:40 dilution for 3 h. After washing in PBS three times, the sections were incubated with swine anti-rabbit Igs/FITC Igs at a 1:30 dilution for 2 h at room temperature in a moist chamber. Samples for RB6 fluorescence immonostaining were incubated with a 1:200 dilution of rat anti-mouse granulocyte mAb (RB6-8C5) or rat IgG for 3 h at room temperature. After washing in PBS three times, FITC labeled goat-anti-rat secondary Ab was added at a 1:50 dilution in PBS and incubated for 1 h at room temperature. After three washes in PBS, the samples were examined in a fluorescence microscope (Nikon) equipped with a 100 W mercury lamp (Osram, Berlin, Germany) and Ploem-pac with the filter set for FITC. To prevent UV light quenching, buffered glycerol containing 2% 1,4-diazabicyclo-(2, 2, 2)octane was used as a mounting medium were fixed as described above.

The Mann Whitney U test was used to compare experimental groups.

Confluent A498 epithelial cell layers on the underside of Transwell polycarbonate filters were preinfected with E. coli Hu734. Neutrophils were added to the upper well, and migration to the lower well was quantified after 3 h (Fig. 1,A). Infection stimulated the epithelial cells to secrete IL-8 (17 ng/ml), and neutrophil migration across the E. coli-stimulated cell layers was IL-8 dependent, as shown in inhibition experiments with anti-IL-8 Abs. Addition of monoclonal anti-IL-8 Abs to the upper and lower wells reduced E. coli-induced neutrophil migration down to the level of the uninfected control (Fig. 1 A).

These results demonstrate that IL-8 is the major chemokine involved in neutrophil migration across the E. coli-infected epithelial cell layers. Washing of the cell layers and transfer to new cluster plates did not reduce neutrophil migration, suggesting that the active form of IL-8 was bound to cell surface receptors.

IL-8 has been shown to bind proteoglycans and specific chemokine receptors on endothelial cells, but receptor expression by epithelial cells has not been investigated. To address this question, biopsies were obtained from patients undergoing surgery of the urinary tract. Tissue sections from the renal pelvis and the bladder were stained with monoclonal anti-CXCR1 or anti-CXCR2 Abs (Fig. 2). Both receptors were detected in the bladder epithelium, with CXCR2 being more abundant than CXCR1. The renal pelvic epithelium expressed both receptors, with more intense staining for CXCR1 than for CXCR2. Neutrophils in the subepithelial space expressed more CXCR1 than CXCR2.

The Abs were specific as shown by inhibition with IL-8 for CXCR1 and GRO-α or ENA-78 for CXCR2 (Fig. 3, G and H). GRO-α or ENA-78 pretreatment of the epithelial cells caused a reduction in anti-CXCR2 Ab binding down to the level of the medium control, but had no effect on the binding of CXCR1 (Fig. 4).

The effect of infection on epithelial chemokine receptor expression was examined using the A498 human kidney epithelial cell line. By confocal microscopy, uninfected cells were found to be receptor negative, but a marked increase in the surface expression of both CXCR1 and CXCR2 was detected in infected cells (Fig. 3, A–D). This increase was quantified by flow cytometry showing fluorescence intensities that were 89% and 96% higher than the uninfected control with virtually complete separation of the peaks (Fig. 3, E and F).

Purified LPS was tested for effects on chemokine expression by the A498 cells. At concentrations <100 μg/ml, no stimulatory effect was observed (data not shown). The cells have previously been shown to lack surface CD14 expression (14).

These results demonstrate that human uroepithelial cells express CXCR1 and CXCR2 and that in vitro infection increases the expression of both receptors.

The increase in CXCR1 and CXCR2 receptor expression following infection was accompanied by an increase in IL-8 binding to the cells. By flow cytometry, a higher binding of FITC-labeled IL-8 to the infected A498 cells was detected (56% above the uninfected control) (Fig. 3, G and F). Pretreatment of the epithelial cells with anti-CXCR1 Ab inhibited this increase (Fig. 3,G), but anti-CXCR2 Abs had no inhibitory effect (Fig. 3 F).

These results demonstrate that CXCR1 was the functional receptor for IL-8, and that the increase in CXCR1 receptor expression accounted for the increase in IL-8 binding to the epithelial cells following infection.

The involvement of epithelial chemokine receptors in transepithelial neutrophil migration was examined in the Transwell assay across E. coli-infected epithelial cell layers (Fig. 1 B). Pretreatment of the infected epithelial cells with anti-CXCR1 Abs reduced neutrophil migration to 39.3% (p < 0.004) of the infected control. The CXCR2 Abs alone had no significant effect (77.2%; NS), but maximal inhibition was obtained using both CXCR1 and CXCR2 Abs (18.8%; p < 0.0001). Neutrophil pretreatment with CXCR1 Ab reduced neutrophil migration to 44.8% (p < 0.004), but the CXCR2 Abs had no significant effect (95.2%; NS).

These results demonstrated that CXCR1 is essential for the E. coli-stimulated neutrophil migration across human uroepithelial cell layers.

UTI was induced in BALB/c and mIL-8Rh KO mice by intravesical inoculation with E. coli 1177. Receptor expression in vivo was examined by staining of kidney and bladder sections with anti-IL-8R Abs (Fig. 5,A). Infection of BALB/c mice caused a rapid increase in epithelial IL-8 receptor expression (Fig. 5 A), but receptor expression was not detected in the mIL-8Rh KO mice either before or after infection.

The in vitro experiments predicted that disruption of the IL-8 receptor should abrogate neutrophil migration across epithelial cell layers also in vivo. This hypothesis was confirmed using mIL-8Rh KO mice (Fig. 6). Intravesical inoculation of control mice with E. coli 1177 caused a rapid neutrophil response as evaluated by histology and immunohistochemistry (Fig. 6, A and B). Bladder (Fig. 6,A) tissue sections showed neutrophils migrating through the submucosa to the epithelial lining and crossing the epithelial barrier into the lumen. Maximal tissue densities of neutrophils were recorded around 6 h after infection (Fig. 5 B). Neutrophils were also detected in urine starting at 2 h after infection and with a peak after 6 h.

Neutrophil recruitment was delayed in the mIL-8R knockout mice compared with the controls. Tissue sections showed the presence of neutrophils only after 24 h. At later times, neutrophils accumulated under the epithelial barrier. After 7 days, the tissues from mIL-8R KO mice were full of neutrophils that reached as far as the lamina propria, but were unable to cross the epithelium into the urine (Fig. 6,B). As a consequence, neutrophil numbers excreted into the urine remained low at all times (Fig. 5 B).

These observations confirmed the in vitro studies and demonstrated that IL-8 receptors and their chemokine ligands are a limiting factor for neutrophil migration across the infected epithelial barrier of the urinary tract.

Tissues were harvested from the infected mIL-8Rh KO and control mice at different times after intravesical inoculation with E. coli 1177. Bacterial persistence was determined by viable counts on tissue homogenates. The control mice had no bacteria in the tissues after 7 days. The mIL-8Rh KO mice, in contrast, were unable to clear infection and had high bacterial counts in the tissues after 7 days. Most of the mice developed bacteremia and symptoms and had to be sacrificed.

Mucosal infections elicit a chemokine-dependent neutrophil response, causing the neutrophils to migrate to and to cross the epithelial barrier into the mucosal lumen (2). Although chemokine receptors may be predicted to be important for this process, their expression in the mucosa and involvement in neutrophil-epithelial cell interactions has not been examined. Bacteria causing UTI have been shown to elicit epithelial IL-8 responses, and neutrophil migration across infected uroepithelial cell layers in vitro was shown to be IL-8 dependent (4). This study investigated chemokine receptors on epithelial cells using the UTI model. The results demonstrate that CXCR1 and CXCR2 are present on the human urinary tract epithelium, that expression by uroepithelial cells is increased by infection, and that CXCR1 plays a crucial role in IL-8-dependent neutrophil migration across infected human uroepithelial cell layers in vitro. Studies of IL-8R KO mice confirmed these observations. Transepithelial migration was abrogated in the KO mice and the neutrophils were trapped under the epithelium (15). The mice became highly susceptible to infection and eventually developed tissue pathology (16). The results illustrate the importance of epithelial CXC receptors in directing neutrophils at mucosal infection sites.

CXC chemokines mediate their biological responses by binding to chemokine receptors belonging to the large serpentine receptor family with seven transmembrane loops, linked to a G-protein for signal transduction (6, 7). CXCR1 and CXCR2 share 77% amino acid sequence homology, and their genes are localized on human chromosome 2q35 (17, 18). The two receptor subtypes differ notably in their N-terminal extracellular domains, and in their C-terminal intracellular domains, and differ in ligand specificity. Several cell types have been shown to express CXCR1 and CXCR2, including endothelial cells (19), basophils (20), dendritic cells (21), mast cells (22), type 1 helper cells (23), and eosinophils (24), but these receptors have been most extensively studied on neutrophils (25, 26, 27, 28). Here, we show for the first time that human epithelial cells express both CXCR1 and CXCR2. The receptors were present in the apical portion of bladder epithelial cells and in the entire cytoplasm of kidney epithelial cells, as shown by staining of human tissue sections. Thus, the molecular basis for CXCR-dependent neutrophil-epithelial cell interactions is in place, along the mucosal lining of the human urinary tract. The difference in relative amount of these receptors expressed in kidneys or bladder provides a mechanism for differential cell recruitment to these sites.

In earlier studies, infection was shown to up-regulate CXCR2 expression in epidermoid carcinoma cell lines (29) and minor vestibular glands (30), but no CXCR1 response was detected. Chemokine receptor expression by infected epithelial cells has not previously been investigated. Dwinell et al. (31) examined the constitutive mRNA levels in uninfected intestinal epithelial cells and found little if any message. This was consistent with the low CXCR1 and CXCR2 expression before infection, but after infection of the A498 cells there was an increase in both CXCR1 and CXCR2 in vitro. Experimental infection caused a rapid increase in expression of the murine IL-8 receptor by epithelial cells in kidney and bladders in vivo. CXCR1, but not CXCR2, was shown to account for the increased binding of IL-8 to infected cells. This was expected because CXCR1 was suggested to mediate IL-8-induced chemotaxis at the site of inflammation, where the concentration of IL-8 is high, whereas CXCR2 has an active role in the initiation phase of polymorphonuclear neutrophil migration in more distant sites from the inflammation, where the concentration of IL-8 is at the picomolar level (32, 33, 34, 35, 36).

The increased neutrophil migration across infected cell layers in vitro was mainly due to the CXCR1-bound IL-8, as Abs to CXCR2 alone did not significantly reduce migration. There was, however, an additive effect when both Abs were used, suggesting cooperation between the two receptors. The results thus suggested that CXCR1 on epithelial cells was required to present IL-8 to incoming neutrophils and to lead them through the cell layer into the lumen. CXCR2 on epithelial cells may aid in this process, but is not sufficient to support neutrophil migration. As CXCR2 is more promiscuous than CXCR1 and several chemokines in addition to IL-8 can be produced by uroepithelial cells (our unpublished observations), their function in the urinary tract mucosa may relate to these chemokines more than to IL-8.

The vessel wall and mucosal barrier present quite different challenges to the migrating neutrophil. The occurrence of CXCR1 and CXCR2 on endothelial cells is still controversial, and their function in neutrophil extravasation is unknown. However, there is some evidence that the IL-8 dimer formation is required for receptor activation (37), and that it is critical in the binding of heparan sulfates and Duffy Ag receptors on endothelial cells (38), both participating in the chemokine presentation (39, 40, 41, 42, 43, 44). This study demonstrates that CXC receptors are present in the mucosa and are involved in neutrophil migration across epithelial cell layers, but the molecular details of this interaction are not known. In our model, it is possible that the IL-8 dimer forms a bridge between the neutrophil and the epithelial cells, but these interactions need further study. In neutrophils, CXCR1 seems to be important for chemotaxis, and receptor recycling has been postulated as an important mechanism of chemotaxis, maintaining the sensitivity of the neutrophils to the chemotactic gradient (25, 45). Epithelial cells, in contrast, are stationary cells with an entirely different function. They are the last cells to encounter the neutrophils on their way through the tissues, and likely to have a different function as compared with endothelial cells and connective tissue compartments. In providing a stable source of chemokines they orchestrate the exit of activated neutrophils from tissues that might otherwise be destroyed by neutrophil degranulation. We propose that epithelial cells are programmed to maintain a more stable and high expression of chemokines and chemokine receptors than other cell types.

The in vivo relevance of epithelial CXCR expression was confirmed using IL-8Rh KO mice. Although no single rodent homologue for human IL-8 has been identified (46, 47), several CXC and CC chemokines are produced in response to UTI in the mucosa in vivo (1, 2). These chemokines converge on the single murine IL-8 receptor homologue, which was deleted by Cacalano et al. (10) to construct the mIL-8Rh KO mouse. This study showed a dysfunctional neutrophil response in the IL-8R KO mice compared with the BALB/c controls. By immunohistochemistry, neutrophils were shown to accumulate in bladder and kidney tissue. They were unable to cross the epithelium into the lumen resulting in low urine neutrophil numbers at all times. The influx of neutrophils showed that the chemotactic gradient was intact despite the lack of IL-8 receptors but that the epithelium formed an impermeable barrier. The chemotactic gradient might be due to IL-8 and established through CXCR-independent mechanisms like the Duffy Ags or heparane sulfates. Alternatively, chemotaxis may be achieved through other mediators like C5a or bacterial products like fMLP (48). These observations illustrate the compartmentalization of the signals and effector molecules involved in neutrophil migration and the highly regulated interaction of this event.

The infection related recruitment of neutrophils to the urinary tract results in so-called “pyuria.” For decades, pyuria has been recognized as a sign of UTI and used as a diagnostic tool, but the molecular mechanisms by which bacteria induce this neutrophil response have not been understood. The following scenario may now be proposed. Neutrophil recruitment starts when epithelial cells in the urinary tract mucosa secrete chemokines and ends when the neutrophils leave the tissues by crossing the epithelial barrier into the lumen. Expression of functional IL-8 receptors is crucial in order for neutrophils to cross epithelial barriers. The IL-8Rh KO mouse showed several different phenotypes relevant to UTI. They became highly susceptible to infection, and developed systemic disease resembling acute pyelonephritis in humans (B. Frendéus, unpublished observations). The dysfunctional receptors caused neutrophil accumulation in the tissues and the animals developed tissue pathology (L. Hang, unpublished observation). Thus, epithelial cell chemokine receptors are critical both for the neutrophil dependant mucosal defense and for tissue integrity. This explains why there is a need to regulate neutrophil migration across epithelial barriers and why cells resident have developed mechanisms for differential expression of chemokines and their receptors.

We thank Dr. A. E. I. Proudfoot (Geneva Biomedical Research Institute, Glaxo Wellcome Research and Development SA, Geneva, Switzerland) for supplying the FITC-labeled IL-8.