Multidrug Resistance-Associated Transporter 2 Regulates Mucosal Inflammation by Facilitating the Synthesis of Hepoxilin A₃¹

Transmigration of polymorphonuclear neutrophils (PMNs)³ across epithelial surfaces represents a shared phenomenon among a diverse array of inflammatory mucosal conditions. For example, salmonellosis, shigellosis, and pneumonia, as well as autoimmune/idiopathic states such as Crohn’s disease, ulcerative colitis, bronchitis, chronic obstructive pulmonary disease, and oral diseases such as periodontitis all culminate in the destructive breach of the protective outer epithelium by activated PMNs. Furthermore, the severity and clinical outcome of these inflammatory diseases correlate with the extent of PMN infiltration (see Refs. 8 and 12). An emerging concept over the past decade suggests an orchestrated movement of PMNs from the bloodstream to luminal sites of inflammatory stimuli. For example, in response to inflammatory stimuli within the intestinal lumen, IL-8, as well as other CXC chemokines, are secreted from the basolateral surface of enterocytes prompting PMN escape from the bloodstream and eventual accumulation within the submucosal space (1). In concert, the potent PMN chemoattractant hepoxilin A₃ (HXA₃) is secreted from the apical surface of the epithelial barrier where it establishes a chemotactic gradient to guide PMNs from the submucosa across the epithelium, a process constituting the final step in PMN recruitment to the mucosal lumen (2). HXA₃ is a hydroxyepoxide metabolite formed from the intermediate 12S-hydroperoxyeicosa-5Z,8Z,10E,14Z-tetraenoic acid (12S-HpETE), the main product of 12S-lipoxygenase (LOX) activity (3, 4, 5). Hepoxilin synthase activity, responsible for the conversion of 12S-hydroxy-5Z, 8Z, 10Z, 14Z-eicosatetraenoic acid to HXA₃ has been confirmed for human platelet 12-LOX, human epidermal LOX type 3, and murine and porcine 12/15-LOX (5, 6). Recently, a critical role for HXA₃ in PMN-mediated events in both intestinal and pulmonary inflammation has been demonstrated (2, 7, 8, 9).

To perform its specific function of drawing PMNs into a mucosal lumen, HXA₃ must be secreted from the apical surface of epithelial cells, thereby establishing a concentration gradient across epithelial tight junctions to incite paracellular PMN movement. HXA₃ performs this function as a pure chemoattractant that does not induce neutrophil activation (10), leaving that event to occur in the presence of pathogens and other stimuli following arrival at the epithelial luminal surface (2). Despite the critical nature of PMN transmigration in pathogen elimination and disease pathophysiology, many of the details regarding the regulation of PMN migration across mucosal surfaces remain undefined. Herein, we describe a mechanism for the vectored secretion of HXA₃ from the apical surface of epithelial cells involving the ATP-binding cassette (ABC) transporter multidrug resistance-associated protein 2 (MRP2). In general, the physiological role of ABC transporters covers a wide spectrum of functions varying from the transport of excretory compounds and toxins to the elimination of xenobiotics (11). Our findings reveal a previously unappreciated role of MRP2 in meditating the inflammatory response as we demonstrate that inflammatory signals up-regulated MRP2 expression in association with 12-LOX activity, a critical element in the synthesis of HXA₃. Furthermore, our studies not only provide vital insight into the contribution of ABC transporters in the pathogenesis of inflammatory disorders of the gastrointestinal tract with respect to mechanisms underlying PMN recruitment but also identify novel targets for the treatment of epithelial-associated inflammatory conditions such as inflammatory bowel disease (IBD).

T84 intestinal epithelial cells (passages 40–58) were grown as previously described (12). For transport experiments, T84 cells were seeded at 250,000/cm² on inverted polycarbonate membrane Transwell inserts (4.5-mm diameter) and cultured at 37°C, 90% relative humidity, and 5% CO₂. HCT-8 (at passages 30–40) were seeded on inverted Transwell inserts at a density of 40,000/cm² and cultured to days 7–10 postseeding for efflux experiments. The formation of restrictive monolayers (at least 800 ohm · cm²) was monitored for all cell types by measurement of transepithelial electrical resistance using an EVOM Voltometer (World Precision Instruments).

Nonagitated microaerophilic bacterial cultures were prepared as previously described (12). Salmonella enterica serotype Typhimurium (S. typhimurium) SL1344 (wild type), EE633 (sipA⁻), AvrA⁻, VV341 (hilA⁻) as well as Escherichia coli F-18 were used in this study and have previously been described (12, 13).

The PMN transmigration assay using inverted cell culture monolayers of polarized cells has been described previously (12).

For experiments involving ABC transport inhibitors, epithelial cells were pretreated for 1 h with either cyclosporin A (10–1000 nM), verapamil (0.1–10000 nM), probenecid (10–100 μM), or MK571 (50–100 μM). Following pretreatment, the drug was removed by washing with HBSS⁺ containing Mg²⁺, Ca²⁺, and 10 mM HEPES (HBSS; pH 7.4) before infecting polarized T84 monolayers at the apical surface with wild-type S. typhimurium SL1344.

The apical surface of polarized T84 cell monolayers was infected with S. typhimurium SL1344 or EE633 as previously described (2, 10). Subsequent HXA3 identification was conducted according to the protocol described by Mrsny et al. (2). This method was used to obtain the data (see Fig. 3C).

T84 or HCT-8 monolayers grown on 4.5-cm² Transwells were drained of medium and washed with warm HBSS⁺ before infection with S. typhimurium (multiplicity of infection, 100:1) for 1 h at 37°C. The bacteria were washed away with warm HBSS⁺ and 750 μl of HBSS⁺ was added to the apical surface, with 2 ml of HBSS⁺ in the basolateral chamber. The Transwells were incubated at 37°C and the buffer in the apical compartment was collected every 2 h and replenished with the same volume of fresh HBSS⁺. Collected buffer was pooled and stored on ice in the dark. After 6 h (three collections), all pooled samples received 0.5 ng of PGB₂ (Cayman Chemical) as an internal standard and was brought to a pH of 3–4 with HCl before extraction. Samples were extracted by solid phase as previously described (14). Samples were eluted in 5 ml of methanol, concentrated under nitrogen and resuspended in 100 μl of methanol. Samples were stored at −80°C until analyzed.

HXA₃ (8-hydroxy-11,12-epoxy-eicosatetraenoic acid) was identified by liquid chromatography/mass spectrometry/mass spectrometry (LC/MS/MS)-based lipidomic analyses (9, 15). Extracted samples suspended in methanol were analyzed by a triple quadruple linear ion trap LC/MS/MS system (MDS SCIEX 3200 QTRAP; Applied Biosystems) equipped with a LUNA C18–2 minibore column (Phenomenex) using a mobile phase (methanol:water:acetate, 65:35:0.03, v/v/v) with a 0.35 ml/flow rate. MS/MS analyses were conducted in negative ion mode and fatty acids were identified and quantified by the multiple reaction monitoring mode using specific transitions for HXA₃ (335→27 m/z) and 12-HETE (319→179 m/z). Specific retention time for HXA₃ and 12-HETE was established with a synthetic standard (BIOMOL) and transition ions established by MS/MS analyses using enhanced product ion mode with appropriate selection of the parent ion in quadruple 1. Calibration curves (1–1000 pg) for HXA₃, 12-HETE, and PGB₂ were established with synthetic standards.

T84- or HCT-8-polarized monolayers were apically infected with either wild-type S. typhimurium (SL1344), EE633, or the bacterial negative control commensal isolate E. coli F-18 (12). Immunoprecipitation was performed using the Seize X Mammalian Immunoprecipitation Kit (Pierce) according to the manufacturer’s instructions. The bead column was cross-linked with 70 μg of goat anti-MRP2 polyclonal or goat anti-multidrug resistance (MDR) polyclonal Abs (Santa Cruz Biotechnology). Whole cell lysates added to each column were normalized to 0.5 mg/column. Eluted protein was electrophoresed on a nondenaturing gradient SDS-PAGE gel (8–16% Tris-glycine; Cambrex Biosciences). Proteins were transferred using a standard semidry transfer method onto nitrocellulose and probed with MRP2 or MDR polyclonal Abs (1/500) overnight at 4°C followed by washes and incubation with HRP-labeled donkey anti-goat Abs (1/1000; Santa Cruz Biotechnology) for 1 h. Bands were detected using SuperSignal West Pico Chemiluminescent Substrate (Pierce).

Based on a modification of Barthel et al. (16), female C57BL/6 mice, 6–8 wk old (The Jackson Laboratory) were orally treated with 75 μl of streptomycin sulfate (stock 50 mg/ml; Sigma-Aldrich) in sterile distilled water following a 4-h starvation period of both food and water. After 24 h, the mice were starved for 4 h and orally gavaged with 5 × 10⁸ CFU of the appropriate bacterial strain in 500 μl of sterile HBSS, monitored for 48 h, and sacrificed.

The diagnosis of IBD was based on clinical, endoscopic, and histological criteria, and infectious colitis was ruled out by stool cultures. The collection of samples was approved by the Institutional Review Board at the University of Chicago. Intestinal biopsy specimens were obtained from four ulcerative colitis and four Crohn’s disease patients during routine endoscopy upon informed consent. Control colonic biopsy specimens were obtained from age- and sex-matched individuals undergoing surgery for cancer resection.

Murine intestinal sections were harvested from the proximal colon of mice at necropsy. Human tissue was obtained from inflamed or noninflamed biopsy samples of matched patient samples acquired through the Department of Pathology at the University of Chicago as a generous gift from Dr. J. R. Turner. All tissues were imbedded in Tissue-Tek OCT compound (Sakura Finetek USA), flash frozen in liquid nitrogen, and sectioned (5-μm thickness) on a 2800 Frigocut cryostat (Reichert-Jung). Hematoxylin and Eosin Y staining and scoring were performed as previously documented (17).

Murine and human colon sections were fixed in cold acetone (−20°C) for 10 min and air dried for 5 min. Blocked sections were incubated at 4°C overnight in goat anti-MRP2 polyclonal Ab (Santa Cruz Biotechnology) diluted 1/750 or 1/250 in TBS for mouse and human tissue, respectively. Sections were washed in TBS, followed by a 1-h incubation with FITC-conjugated donkey anti-goat Ab (Santa Cruz Biotechnology) diluted 1/1000 and 1/250 in TBS for mouse and human sections, respectively. Sections were mounted with1 M Tris-HCl (pH 8.0)/Vectashield (Vector Laboratories) mixture (1:1) and analyzed by immunofluorescent microscopy at ×20 oil magnification (Nikon Eclipse TE-2000S).

This assay was performed according to the manufacturer’s instructions using membrane product SB-MRP2-SF9-ATPase (Solvo Biotechnology). ATPase activity of the substrates probenecid (1 mM; BIOMOL) and HXA₃ (1, 10, 100 μM; BIOMOL) was measured with respect to baseline activity.

Cells were treated with 10 μM 5-chloromethylfluorescein diacetate (CMFDA; Molecular Probes) for 1 h to allow uptake and intracellular conversion to the fluorescent MRP substrate 5-chloromethylfluorescein (CMF) (18). CMF, actively transported by MRPs as the glutathione conjugate glutathione-methylfluorescein (GS-MF), was detected by measuring the appearance of fluorescence in the bathing solutions (excitation 495 nm and emission 535 nm; LS-5 spectrofluorometer; PerkinElmer Life and Analytical Sciences).

Permeability studies were conducted with T84 monolayers incubated in HBSS and used [³H]mannitol (0.088 μM, 0.147 MBq) as a control probe for low permeability paracellular transport. Radioactivity was determined by liquid scintillation counting, and coefficients of permeability for probe molecules were similar to those found in the literature for viable monolayers (19). Transport experiments in the apical to basal (A→B) direction were initiated by adding 200 μl of the drug solution (CMFDA) to the donor chamber. Transport experiments in the basal to apical (B→A) direction were initiated by adding 1 ml of CMFDA to the basal chamber. At predetermined times over the course of the experiment, 100-μl samples were taken from the respective chamber and replaced with fresh HBSS. Transport studies conducted in the presence of the MRP inhibitor MK571 (100 μM with 0.2% DMSO) were conducted as above. The cumulative mass of the drug transported from donor to receiver chambers was assessed over the time course of the experiment (dM/dt) using linear regression analysis. The following equation was used to fit the data: dM/dt = p·A·C₀, where p is the permeability coefficient (cm/s), A is the surface area of the Transwell membrane (0.333 cm²), and C₀ the initial drug concentration in the donor chamber (nmol/cm³). In all experiments, the amount of drug transported from donor to receiver was ≤5% of the donor drug mass, ensuring the validity of C₀ as a constant in the equation.

Plasmids used to generate siRNAs were constructed with the pSUPER vector (Oligoengine) using the method described by Brummelkamp et al. (20). Briefly, two oligonucleotides were designed incorporating a 19-nt sequence from each target transcript (shown in italics) or its reverse compliment separated by a short spacer region and a portion of the BglII or HindIII restriction sites. Oligonucleotides specific for the human gene MDR1 encoding P-gp and a random control sequence were described previously (21). Oligonucleotides for the human gene encoding MRP2 (GenBank accession no. NM_000392) were: 5′-GATCCCCGTATCAGGTTTGCCAGTTATTCAAGAGATAACTGGCAAACCTGATACTTTTTGGAAA-3′ and 5′-AGCTTTTCCAAAAAGTATCAGGTTTGCCAGTTATCTCTTGAATAACTGGCAAACCTGATACGGG-3′. After annealing, dsDNAs had overhanging restriction sites and were ligated into digested pSUPER and transformed into competent E. coli DH5α by standard methods and plated on Luria-Bertani with ampicillin (50 μg/ml). Plasmids were extracted, purified (QIAprep Spin Miniprep Kit; Qiagen) and sequenced for confirmation. Once confirmed, bulk plasmid was prepared for transfection using a Qiagen Plasmid Midi Kit. Plasmids were transfected into HCT-8 intestinal epithelial cells using Lipofectamine 2000 (Invitrogen) as described previously (21).

BALB/c mice and C.B-17scid/scid (SCID) mice (6–8 wk old) were purchased from The Jackson Laboratory (BALB/c) and Massachusetts General Hospital (SCID). Colitis was induced in SCID mice by adoptive transfer of CD4⁺CD45RB^high T cells isolated from splenic cells of BALB/c mice as described previously (22, 23, 24). Briefly, CD4⁺ T cells were enriched from the spleens of BALB/c mice using anti-CD4 (L3T4) a MACS magnetic separation system (Miltenyi Biotec) according to the manufacturer’s instruction. Enriched CD4⁺ T cells were then labeled with CyChrome- conjugated anti-mouse CD4 mAb (BD Pharmingen) and PE-conjugated anti-CD45RB mAb (BD Pharmingen) and then sorted into a CD45RB^high (highest staining 30%) fraction; sorting was performed with Mo-Flo (DakoCytomation). Each SCID mouse was injected i.v. with 200 μl of PBS containing 5 × 10⁵ CD4⁺CD45RB^high T cells. Mice were weighed and monitored for appearance and signs of soft stool and diarrhea daily during the experimental period.

For 12/15-LOX inhibition, T84 cell monolayers were incubated in the presence of baicalein (stock concentration at 1 mM in DMSO) in medium for 48 h at 37°C. Subsequently, S. typhimurium SL1344 was added to the apical surface of the T84 monolayers for 1 h at 37°C and the cells were then washed free of nonadherent bacteria and processed for the PMN transmigration assay. For 5-LOX inhibition, the T84 cells were incubated for 24 h in the presence of caffeic acid (stock concentration at 22 mM in DMSO). Both inhibitors were purchased form BIOMOL.

PMN isolation was limited to repetitive donations by 10 different donors over the course of these experiments. Due to variation in both PMNs and transepithelial resistance between monolayers (baseline resistance of ∼1500 ohm · cm²), data were analyzed within an individual experiment and not between experiments. All results are expressed as the mean ± SD of an individual experiment performed in triplicate. Values of p values were calculated according to Student’s t test and values <0.05 were considered statistically significant.

Elucidation of mechanisms governing HXA₃ synthesis and apical secretion could provide novel therapeutic strategies to impede and/or limit neutrophil involvement in deleterious events at epithelia associated with a number of chronic inflammatory conditions. HXA₃ must be secreted from the apical surface of epithelial cells to establish a chemotactic gradient across tight junction complexes to incite the paracellular movement of PMNs (2). Given the structural characteristics of HXA₃ and its strict apical secretion requirement, we considered the involvement of ABC transporters as potential substrates for HXA₃. P-gp and MRP2 were promising candidates due to their localization on apical membranes of enterocytes (25, 26, 27). In vitro treatment of polarized human intestinal epithelial cell monolayers (T84) with inhibitors of either P-gp (cyclosporin A and verapamil) or MRP2 (probenecid and MK571) demonstrated a progressive, dose-dependent decrease in the ability of pathogenic S. enterica Typhimurium (S. typhimurium) to induce HXA₃-mediated PMN transepithelial migration (Fig. 1). Moreover, at the maximal doses tested, none of these drugs had an effect on PMN migration induced by imposed gradients of fMLP nor did they affect S. typhimurium internalization by T84 cell monolayers (data not shown). These results suggest that inhibitors of ABC transporters known to localize to the apical surface of intestinal epithelial cells block the recruitment of PMNs across epithelial cell monolayers induced by an inflammatory stimulus driven by HXA₃; in this case, apical infection with the enteric pathogenic S. typhimurium.

ABC transporter inhibitors are known to be poorly selective (28). To better define the relative roles played by P-gp and MRP2 in PMN transmigration, polarized epithelial cell monolayers were transfected with plasmids generating siRNA for one of each of these two transporters. Decreased expression of MRP2 significantly attenuated (>80% inhibition) S. typhimurium-induced transepithelial PMN migration while monolayers generating siRNA specific for MDR1 (encoding P-gp) showed PMN transmigration in response to S. typhimurium infection equivalent to control monolayers (Fig. 2,a). Importantly, neither siRNA indirectly affected the ability of PMN to transmigrate as PMN migration in response to an imposed fMLP gradient was similar under all conditions (Fig. 2,a). Modulation of MRP2 and P-gp expression by siRNA was verified by immunoblot analysis (Fig. 2,b). Densitometric analysis showed a 75% decrease in MRP2 with siRNA in S. typhimurium-infected cells and a complete loss for P-gp (data not shown). Inhibition of transepithelial PMN migration induced by S. typhimurium was consistent with the extent of protein reduction for MRP2 but not for P-gp (Fig. 2, a and b), suggesting that MRP2, and not P-gp, is involved in PMN transepithelial migration.

In polarized epithelial cells, the majority of MRP transporters localize to the basolateral membrane (18, 27). To examine functional MRP2 transporter expression related to inflammatory events, directional efflux of the MRP substrate CMFDA from polarized epithelial cell monolayers was studied (Fig. 2 c). CMFDA passively diffuses into cells where upon hydrolysis of the acetate moieties by cytosolic esterases it becomes the fluorescent MRP substrate CMF; CMF is actively transported by MRPs as the glutathione conjugate GS-MF. In the baseline state, T84 cell monolayers have an ∼25-fold higher permeability coefficient in the apical to basolateral (A→B) direction relative to the B→A direction (20.3 ± 1.04, and 0.8 ± 0.03 × 10⁻⁶ cm/s, respectively). This relationship favors an overall basolateral secretion, or A→B transport, of GS-MF. Interestingly, T84 monolayers infected at their apical surface with wild-type S. typhimurium decreased the A→B:B→A permeability coefficient ratio to 1.8, with an 89% reduction of probe movement in the A→B direction, suggesting activation of an apical efflux pathway. Addition of the MRP2 inhibitor MK571 to T84 monolayers infected with wild-type S. typhimurium restored the A→B transport of GS-MF to 43% compared with controls with no effect on transport in the opposite direction. As a negative control, nonpathogenic S. typhimurium (VV341) only modestly reduced GS-MF transport (15%; p < 0.05). In contrast, P-gp functional activity was not increased during S. typhimurium infection (21).

The above results indicate that inflammatory events can enhance apical MRP2-specific efflux activity. We next sought to determine whether HXA₃ serves as a substrate for MRP2. Interestingly, MRP2 exhibits preference for negatively charged, hydrophobic molecules such as the eicosanoid cysteinyl leukotriene (29, 30, 31, 32). HXA₃ shares these characteristics at neutral pH, suggesting it fits the profile of an MRP2 substrate. Since ABC transporters hydrolyze ATP during active substrate transport, the ATPase activity of purified inside-out membrane vesicles isolated from Spodoptera frugiperda (Sf9) cells expressing the MRP2 transporter were examined following HXA₃ addition. MRP2-containing membrane vesicles exhibited a dose-dependent increase in ATPase activity in the presence of HXA₃ (Fig. 2 d). Addition of the MRP2 substrate inhibitor probenecid was used as a positive control to depict maximal ATPase activity. Such data provide direct evidence that HXA₃ can be a substrate for MRP2.

Previous studies have shown that the S. typhimurium effector protein SipA is required for the ability of S. typhimurium to induce PMN transepithelial migration (13, 16, 33). We examined the role of virulence factors, such as SipA, to incite functional expression of MRP2 by comparing MRP2 expression levels following infection with either wild-type S. typhimurium (SL1344) or the isogenic SipA mutant strain (EE633). Our results show that S. typhimurium stimulated MRP2 expression in a SipA-dependent manner (Fig. 3, a and b); P-gp levels were not affected by S. typhimurium infection during a 1-h infection (Fig. 3,a). To control for specificity of SipA, we performed this assay using a commensal E. coli strain lacking SipA (F-18) as well as an isogenic strain of S. typhimurium mutated for an alternative effector protein, AvrA, that retains SipA expression. As anticipated, the AvrA mutant induces MRP2 expression (Fig. 3,b), whereas E. coli F-18 fails to do so. Moreover, complementing with a sipA-expressing plasmid (AJK63) restores the ability of the SipA mutant (EE633) to induce MRP2 protein expression (Fig. 3 b). Thus, under these conditions S. typhimurium infection of T84 cell monolayers modulates MRP2 expression with no effect on P-gp expression.

Using SipA as a prototypical virulence element, we could also show its involvement in the release of HXA₃ since wild-type S. typhimurium elicited significantly more HXA₃ release from T84 cell monolayers compared with the SipA mutant (EE633), and this was correlated, respectively, with the ability to facilitate PMN migration (Materials and Methods: Method 1) (Fig. 3 c). These results suggest that induction of intestinal inflammatory events, in this case SipA-mediated induction of HXA₃ release, stimulates MRP2 expression to provide a potential apical secretion mechanism for HXA₃. Furthermore, since 12-HETE is a biomarker for the HXA₃ biosynthetic pathway and is easily detected by LC/MS/MS, we next measured the amount of 12-HETE released from intestinal epithelial cells displaying reduced levels of MRP2 (via siRNA) during an infection with S. typhimurium. We found that the inhibition of 12-HETE from infected epithelial cells with reduced levels of MRP2 was at least 20% (n = 2) compared with infected siRNA vector control monolayers. Similarly, 12-HETE release by S. typhimurium-infected was also profoundly inhibited by pharmacological treatment with a MRP2 inhibitor (MK571 100 μM; data not shown). To further substantiate a link between MRP2 and the hepoxilin biosynthetic pathway, we found that HXA₃ release (measured by Method 2 in Methods and Materials) by infected T84 cells or HCT-8 cells was inhibited 55–90% (n = 2) by the MRP2 inhibitor probenecid (100 μM).

We next determined whether MRP2 expression is modulated during an infection with S. typhimurium in vivo (16). Mice infected with wild-type S. typhimurium exhibited severe intestinal pathology characterized by a predominant PMN cellular infiltrate and extensive tissue damage, whereas mice infected with the SipA mutant demonstrated only a slight level of intestinal inflammation (Fig. 4, a–c). Apical localization of the ABC transporter MRP2 was uniformly increased at the villus tips during the acute phase of intestinal inflammation (Fig. 4, d–f), confirming MRP2 up-regulation by pathogenic S. typhimurium. In contrast, P-gp was not up-regulated (data not shown and Ref. 21). Collectively, these data are consistent with our in vitro findings suggesting MRP2 may play a role in acute salmonellosis.

Although chronic inflammatory diseases have been correlated with excessive PMN recruitment and destruction of protective epithelial barriers, many aspects of events that initiate and perpetuate PMN transmigration are poorly understood. However, using a mouse leukocyte-type 12/15- LOX Ab, an earlier study by Shannon et al. (34) recognized a fundamental change in the pattern of 12/15-LOX expression in colonic tissue during the onset of IBD; inflamed colonic tissue expressed increased levels of 12-LOX with a corresponding increase in 12-LOX activity relative to healthy colonic tissue. However, there are four enzymes known to possess significant 12-LOX activity in humans: platelet-type 12-LOX (ALOX12), 12(R)-LOX (ALOX12B), epidermis-type LOX (ALOXE3), and 15-LOX type 1 (ALOX15) (35, 36, 37, 38). 15-LOX type 1 is the human homolog of murine leukocyte-type 12/15-LOX (39), and we have previously shown that PMN transepithelial migration in response to infection with S. typhimurium is dependent on 12/15-LOX activity (2, 9). Formation of 12-HpETE by these LOXs is the first step for the formation of HXA₃ (40).

To delineate the in vivo relevance of the above findings during chronic states of inflammation, we sought to determine whether 12-LOX inhibition blocks inflammatory outcomes associated with an IBD model induced by transfer of CD4⁺CD45RB^high T cells (41, 42). The CD4⁺CD45RB^high T cell transfer model of IBD is well characterized, has a requirement for intestinal microorganisms, and involves the full spectrum of the immunoinflammatory cascade, including the infiltration of PMN to the intestinal mucosa. Following disease induction, inhibition of the 12/15-LOX activity (baicalein administration by i.p. injection every other day for 17 days) resulted in complete resolution of gross morphological changes, including rectal prolapse (Fig. 4, g–i). This improvement was consistent with histopathology results (Fig. 4, j–l). As shown in Table I, quantification of these findings revealed that colons from PBS-treated (control) mice exhibited mucosal erosion, mucosal wall thickening, and moderate crypt abcessation with an accompanying inflammatory infiltrate. In contrast, the degree of colonic inflammation in the baicalein-treated cohort was significantly decreased and the neutrophilic infiltrate returned to control levels following baicalein treatment (Fig. 4 m). These results demonstrate that the severity of mucosal inflammation induced in the CD4⁺CD45RB^high T cell transfer model of IBD was significantly attenuated by the 12/15-LOX inhibitor baicalein, which acts to prevent the synthesis of HXA₃. We were not able to perform similar experiments using either one of the MRP2-selective inhibitors since injection or oral delivery did not result in topical delivery to the colon, the required site of action.

We next posed the question of whether the 12/15-LOX activity and MRP2 expression might be coordinated as a regulated element of response to an inflammatory signal. In the presence of the 12/15-LOX inhibitor baicalein, S. typhimurium-infected monolayers in vitro showed significantly lower levels of MRP2 than those induced by this pathogen in the absence of the inhibitor (Fig. 5,a). Similar treatment with caffeic acid, a 5-LOX inhibitor, failed to affect enhanced MRP2 expression induced by S. typhimurium infection (data not shown). Examination of intestinal tissues from CD45RB^high T cell transfer mice revealed that therapeutic intervention with baicalein reduced apical membrane MRP2 expression (Fig. 5 b), suggesting a link between the regulation of 12/15-LOX activity and MRP2 expression during both acute and chronic inflammatory events. Taken together, these results imply that the 12/15-LOX pathway is required for maximal induction of MRP2 expression associated with epithelial inflammation.

To determine the in vivo relevancy of the above findings with respect to nonpathogen-induced human disease, we assessed whether MRP2 is up-regulated in patients with IBD. Indeed, as shown in Fig. 5,c, we provide direct evidence demonstrating that intestinal biopsy specimens obtained from IBD patients with either active Crohn’s disease or ulcerative colitis exhibit marked MRP2 up-regulation at the intestinal epithelial apical surface compared with healthy intestinal segments. One hundred percent of the ulcerative colitis patients (four of four) and 50% of the Crohn’s disease patients (two of four) examined in this study demonstrated an up-regulation of MRP2. None of the age- and sex-matched control biopsies showed any increase above baseline levels (Fig. 5 c).

Given that migration of PMNs across mucosal surfaces contributes to epithelial cell dysfunction in a host of mucosal diseases (2, 7, 8, 12), the data provided here not only identify a new mechanism underlying the vectored release of HXA₃ but also reveal interference with HXA₃ synthesis or apical secretion from epithelial cells as novel therapeutic strategies for the treatment of mucosal inflammatory disorders such as IBD. We used S. typhimurium as a tool to induce an acute intestinal inflammatory response and found that during active states of intestinal inflammation apical expression of the efflux transporter MRP2 is profoundly up-regulated, functionally active, and capable of using HXA₃ as a substrate. The molecular basis of this observation is consistent with the role of the S. typhimurium type III-secreted product SipA, as this effector protein has been found to be both necessary and sufficient for induction of PMN transepithelial migration across model intestinal epithelia (13). In keeping with its role as a virulence factor, our data now reveal that SipA mediates PMN transepithelial migration by directly affecting the release of HXA₃ via modulation of MRP2 expression. Thus, identification of the molecular mechanism by which SipA activates PMN transepithelial migration through MRP2 defines a novel proinflammatory signaling cascade essential for the pathogenesis of S. typhimurium that likely represents a paradigm of mucosal pathogen-elicited events.

Insights into pathogen-elicited active inflammation of the intestine may provide important information relating to mechanisms of disorders underlying IBD, which appear to be unrelated to pathogen colonization (43). Therefore, we hypothesize that although evolution of such responses most likely targeted pathogens such as S. typhimurium, aberrant activation of such pathways may lead to induction of mucosal inflammation associated with some chronic diseases of the intestine and airway. Consistent with this notion, we found a profound up-regulation of MRP2 at the apical surface of the colonic epithelium in murine models of chronic intestinal inflammation as well as in intestinal biopsies from patients presenting with active Crohn’s disease and ulcerative colitis. Under normal healthy conditions, the most abundant constitutive expression of human and rat MRP2/MRP2 mRNA is found in the renal proximal tubule brush border membrane and the hepatocyte canalicular membrane (44). Significantly lower levels of MRP2 have been found in the small intestine, with exclusive localization at the apical brush border membrane of villi (44). Furthermore, MRP2 expression decreases in intensity from the villus tip to the crypts, where no expression has been observed (44). At this location, MRP2 is thought to play an important role in drug disposition. We now describe an unanticipated function of MRP2 as it is uniformly up-regulated at the villus tips of the apical surface of epithelial cells during active states of inflammation and plays a pivotal role in the inflammatory response.

Secreted HXA₃ then establishes a gradient through the tight junction complex to provide a chemotactic gradient used by PMNs to target the lumen of mucosal tissues at sites of inflammation (2). The observation that HXA₃ can serve as a substrate for the apical efflux transporter MRP2 fits well with the appreciation that other eicosanoids such as the cysteinyl leukotriene C₄, a metabolite of the 5-LOX pathway, is one of the highest affinity MRP2 substrates characterized to date (29, 30, 31, 32). Detailed studies, however, are required to fully characterize and provide direct evidence for MRP2 transport of hepoxilin.

There are four enzymes that possess 12-LOX activity in humans and the most abundant LOX expressed in epithelial cells is 12/15-LOX. In humans, unlike mice, 12HpETE is not the major product of 12/15-LOX (ALOX15, 4:1 ratio of 15-HpETE:12-HpETE) (45, 46). However, 12/15-LOX is highly expressed in epithelial cells, including intestinal epithelial cells (45, 46), and the synthesis of 12-HpETE through this pathway is likely significant. At present it is unclear which 12-LOX enzymes specifically contribute to HXA₃ production in the human mucosa and thus the potential contribution of ALOX12, ALOX12B, or ALOXE3 cannot be excluded. Our studies, however, determined that pharmacological inhibition of 12/15-LOX activity (both in vitro and in vivo) was found to not only block HXA₃ synthesis but also the enhanced apical expression of MRP2 typically observed as part of the inflammatory process.

Our observation that 12/15-LOX activity is coupled to MRP2 up-regulation and apical localization implies that either a metabolite of the 12/15-LOX pathway or an active form of the enzyme itself participates in MRP2-related cellular changes associated with inflammation. However, it is unclear at this point in our investigation at what level or through what mechanism this association is achieved. Nevertheless, previous studies have demonstrated that cyclooxygenase 2 expression enhances the functional activity of P-gp, providing a precedent of a causal link between arachidonic acid metabolism and expression of an ABC transporter (47). Furthermore, given that epithelial cells appear to regulate 12/15-LOX and MRP2 expression in a coupled fashion, several novel therapeutic strategies are potentially available to treat inflammatory conditions, such as IBD, based upon regulation of 12/15-LOX and MRP2. Underscoring such important clinical implications, we show that treatment with baicalein, a 12-LOX inhibitor, which inhibits HXA₃ synthesis as well as the enhanced apical expression of MRP2, resolved PMN transmigration/intestinal inflammation in a murine model of IBD.

Although our studies provide the first description that MRP2 is involved in the mechanisms that promote active states of intestinal inflammation (i.e., PMN recruitment), other ABC transporters, namely, P-gp (MDR1), have been found to play a role in the development of colitis (48), albeit by a completely different mechanism than MRP2. Consistent with this observation, we have recently determined that prolonged apical colonization of intestinal epithelial cells by wild-type S. typhimurium (∼4 h) leads not only to a profound functional decrease in P-gp but that the presence of P-gp adversely influences the ability of S. typhimurium to invade host cells (21). These results demonstrate that MRP2 and P-gp can be differentially regulated within the same tissue during a disease process. This observation supports the finding that during the progression of chronic renal failure MRP2 is up-regulated in both the kidney and liver, whereas P-gp remains unaffected (49).

In summary, we are beginning to understand how the PMN chemoattractant HXA₃ is secreted in a vectoral fashion through a unique efflux transport system located at the apical surface of the intestinal epithelium that involves the ABC transporter MRP2. This study demonstrates a critical link between apically expressed MRP2 and the HXA₃ biosynthetic pathway. Surprisingly, inhibition of an enzyme critical for the synthesis of HXA₃, 12/15-LOX, was found to also suppress the apical expression of MRP2 induced by inflammatory signals. This unanticipated coupling of 12/15-LOX and MRP2 identified in an in vitro acute inflammatory model initiated by a pathogen was recapitulated in an in vivo mouse model of IBD and was consistent with observations made in biopsy samples obtained from IBD patients. Since HXA₃ has also been shown to be involved in pathogen-induced inflammation of both intestinal (2, 9) and pulmonary (7) epithelial cells, it is likely that our results describe a generalized mechanism governing HXA₃ synthesis and its subsequent apical release that could provide novel therapeutic strategies to impede and/or limit PMN involvement in deleterious events associated with a number of acute and chronic inflammatory conditions.

We thank Dr. Bryan Hurley for critical reading of this manuscript. We also thank Dr. Jerrold R. Turner at the University of Chicago for supplying the IBD specimens.

One of the authors (R.J.M.) is affiliated with a company focusing on the commercial development of therapeutics derived from this research.