Interferon-γ Alters Downstream Signaling Originating from Epidermal Growth Factor Receptor in Intestinal Epithelial Cells

Background: Inflammation alters epithelial signaling. We studied how IFN-γ regulates EGFr signaling. Results: IFN-γ regulated EGFr expression, membrane localization, tyrosine residue phosphorylation, and linkage to MAPK. EGF no longer inhibited ion transport in IFN-γ-treated cells. Conclusion: EGFr signaling outcomes are altered by inflammation. Significance: These effects may contribute to diarrheal and other symptoms of inflammatory bowel diseases. The epidermal growth factor receptor (EGFr) regulates many cellular functions, such as proliferation, apoptosis, and ion transport. Our aim was to investigate whether long term treatment with interferon-γ (IFN-γ) modulates EGF activation of downstream signaling pathways in intestinal epithelial cells and if this contributes to dysregulation of epithelial ion transport in inflammation. Polarized monolayers of T84 and HT29/cl.19A colonocytes were preincubated with IFN-γ prior to stimulation with EGF. Basolateral potassium transport was studied in Ussing chambers. We also studied inflamed colonic mucosae from C57BL/6 mice treated with dextran sulfate sodium or mdr1a knock-out mice and controls. IFN-γ increased intestinal epithelial EGFr expression without increasing its phosphorylation. Conversely, IFN-γ caused a significant decrease in EGF-stimulated phosphorylation of specific EGFr tyrosine residues and activation of ERK but not Akt-1. In IFNγ-pretreated cells, the inhibitory effect of EGF on carbachol-stimulated K+ channel activity was lost. In inflamed colonic tissues, EGFr expression was significantly increased, whereas ERK phosphorylation was reduced. Thus, although it up-regulates EGFr expression, IFN-γ causes defective EGFr activation in colonic epithelial cells via reduced phosphorylation of specific EGFr tyrosine residues. This probably accounts for altered downstream signaling consequences. These observations were corroborated in the setting of colitis. IFN-γ also abrogates the ability of EGF to inhibit carbachol-stimulated basolateral K+ currents. Our data suggest that, in the setting of inflammation, the biological effect of EGF, including the inhibitory effect of EGF on Ca2+-dependent ion transport, is altered, perhaps contributing to diarrheal and other symptoms in vivo.

The precise etiology of the two entities of chronic inflammatory bowel disease, Crohn disease and ulcerative colitis, remains unknown despite a vast body of research. In Crohn disease, there is a chronic T helper 1 (Th1)-driven immune response with overproduction of inflammatory cytokines, such as IFN-␥ and interleukin (IL)-12 (1). Moreover, increased serum levels and mucosal production of IFN-␥ in Crohn disease have also been demonstrated (2). In mice, the frequently used acute dextran sulfate sodium (DSS) 2 colitis model is mainly a Th1 cytokine-driven inflammation with up-regulation of IFN-␥, IL-12, IL-1, and TNF-␣ (3). A key role for IFN-␥ in the pathogenesis of DSS colitis has been shown in both C57BL/6 and Balb/c mice (4). Indeed, IFN-␥ Ϫ/Ϫ mice are resistant to DSS-induced colitis (5). In another model of spontaneous colitis, the mdr1a Ϫ/Ϫ mouse, IFN-␥ mRNA and protein levels are elevated more than 50-fold in mice with established colitis compared with controls, and production of IFN-␥ is also elevated in stimulated mesenteric lymph nodes (6,7). Interestingly, the human MDR gene is located on chromosome 7, a susceptibility locus for inflammatory bowel diseases (8).
Proinflammatory cytokines, such as IFN-␥ and TNF-␣, can induce tight junction reorganization and/or epithelial damage that causes a leaky mucosal barrier (9,10). In health, a polarized monolayer of columnar intestinal epithelial cells normally forms a tight barrier to the access of toxic agents to the mucosal immune system (11). However, IFN-␥ disrupts tight junctions and thereby disturbs paracellular transport mechanisms. The barrier function of the epithelium is also vital to maintain the electrical gradients necessary for controlled regulation of absorption versus secretion of electrolytes and water. Dysregulation of the balance between absorption and secretion by the epithelium can lead to secretory diarrhea, a major symptom of inflammatory bowel diseases.
The epidermal growth factor receptor (EGFr) has been shown to play a role in repair and cell growth, but it also exerts important influences on epithelial transport in the gastrointestinal tract. The most important downstream signaling cascades are the Ras-Raf-MEK-ERK as well as the phosphatidylinositol 3-kinase (PI3K)-AKT-mTOR pathways. Ligand binding leads to EGFr dimerization and autophosphorylation and to tyrosine phosphorylation of other proteins. The mitogen-activated protein kinase (MAPK) and PI3K pathways contribute to proliferation and play different roles in epithelial ion transport. We showed previously that transactivation of EGFr by G proteincoupled receptor agonists could limit excessive chloride secretion via the MAPK pathway (12). Another study documented the role of the PI3K pathway in calcium transport (13). EGF inhibits carbachol-induced ion transport in intestinal epithelial cells by a mechanism that involves activation of PI3K as well as MAPK pathways (14,15). EGFr signaling also plays an important role in overall intestinal homeostasis. This is perhaps best demonstrated in studies using transgenic mice with defective EGFr (waved-2) or lacking the EGFr ligand, TGF-␣ (waved-1). Both models display an increased susceptibility to colitis induced by the administration of DSS (16,17). Conversely, overexpression of TGF-␣ reduces susceptibility to DSS colitis (18). These data point to a predominant role for EGFr in protecting against the development of chronic intestinal disease.
Our group showed previously that ion transport responses are diminished in tissues from colitic mice and that EGF could restore normal ion transport responses to calcium-and cyclic nucleotide-dependent agonists (19). Thus, despite its inhibitory effect in normal tissue, EGF normalized ion transport in the setting of inflammation by a mechanism that involved ERK-MAPK and PI3K. However, the reason for the differential effect of EGF in normal and inflamed tissues was unknown. Nevertheless, these observations have clinical relevance because EGF enemas have been shown to be beneficial in reducing diarrhea and other symptoms in patients with left sided ulcerative colitis (20). We hypothesized that factors in inflamed tissues might modulate the precise consequences of EGFr activation in the intestinal epithelium. Taking IFN-␥ as an exemplar of an inflammatory mediator relevant to colitis, this study was designed to test this hypothesis in vitro and in vivo.
Physiological Solutions-The composition of the Ringer's solution used for Western blot studies was as follows: 140 mM Na ϩ , 120 mM Cl Ϫ , 5. Tissue Culture-Human T 84 colonic epithelial cells were cultured as described previously (21). In brief, cells (passages 20 -40) were grown in 1:1 Dulbecco's modified Eagle's/Ham's F-12 medium with L-glutamine and 15 mM HEPES (Mediatech Inc., Manassas, VA), supplemented with 5% newborn calf serum (Invitrogen) at 37°C in 5% CO 2 . Cells were fed twice a week and then seeded onto 30-mm Millicell transwell polycarbonate filters (10 6 cells) for Western blot analysis or 12-mm filters (5 ϫ 10 5 cells) for Ussing chamber experiments. Cells were cultured for 10 -15 days on filters prior to use. When grown on polycarbonate filters (Millipore), T 84 cells acquire the polarized phenotype of native colonic epithelia.
Cells grown on polycarbonate filters were stimulated basolaterally with IFN-␥ (1000 units/ml) for up to 72 h. After preincubation with IFN-␥, cells were stimulated with EGF (100 ng/ml) from the basolateral side for up to 90 min.
Immunoprecipitation and Western Blot Analysis-T 84 monolayers were preincubated with IFN-␥ or medium for the indicated time periods and then washed three times with warm Ringer's solution and allowed to equilibrate at 37°C for 30 min. Afterward, cell monolayers were stimulated with EGF (100 ng/ml) for up to 90 min. The reaction was stopped by washing with ice-cold Ringer's solution. This was followed by lysing the cells in lysis buffer (1% Triton X-100, 1 g/ml leupeptin, 1 g/ml pepstatin, 1 g/ml antipain, 100 g/ml phenylmethylsulfonyl fluoride, 1 mM Na ϩ -vanadate, 1 mM sodium fluoride, and 1 mM EDTA in phosphate-buffered saline) for 45 min. Cells were scraped off the filters, and the protein content of the lysates was assayed using a Bio-Rad protein assay kit. Samples were adjusted for an equal amount of protein and incubated with 5 g of monoclonal anti-EGFr antibody for 1 h at 4°C on a rotating platform. This was followed by incubation with protein A-agarose beads (Upstate Chemicon, Temecula, CA) for 1 h at 4°C on a rotating platform. The protein A-agarose-antibodyantigen complex was pelleted by centrifugation for 3 min at 1200 rpm and washed three times with ice-cold Ringer's solution. The beads were then resuspended in 2ϫ gel-loading buffer (50 mM Tris, pH 6.8, 2% SDS, 200 mM dithiothreitol, 40% glycerol, 0.2 bromphenol blue) and boiled for 4 min before loading onto a 7.5% polyacrylamide gel to resolve proteins. For whole cell lysate immunoblotting, an aliquot of lysate was diluted in 2ϫ gel-loading buffer (50 mM Tris, pH 6.8, 2% SDS, 200 mM dithiothreitol, 40% glycerol, 0.2 bromphenol blue) and boiled for 4 min before the samples were loaded on a 4 -15% polyacrylamide gel to resolve proteins.
Resolved proteins were transferred onto polyvinylidene membranes (Millipore). The membrane was then blocked with 1 or 5% (for anti-phosphoantibodies) skim milk in Tris-buffered saline (TBS) for at least 30 min. This was followed by incubation with the primary antibody in 1% skim milk or 5% BSA (for anti-phospho-antibodies) overnight. After five washes with Tris-buffered saline containing 1 or 0.1% (for anti-phosphoantibodies) Tween, the membranes were incubated with an HRPconjugated secondary antibody (1:3000) for 30 min. After additional washing with wash buffer, proteins were detected using an enhanced chemiluminescence detection kit (GE Healthcare). Densitometric analysis of Western blots was performed using NIH Image software.
Phospho-Akt1 ELISA-Whole cell lysates from T 84 cells grown on filters were analyzed for phospho-Akt1 using a commercially available enzyme-linked immunsorbant assay (Path-Scan Phospho-Akt1(Ser 473 ) Sandwich ELISA kit, Cell Signaling Technology) according to the manufacturer's instructions.
RNA Isolation and Real-time PCR-RNA isolation from T 84 cells was performed using the RNeasy Plus minikit (Qiagen, Valencia, CA) according to the manufacturer's instructions. DNA was removed using the TURBO DNA-free kit (Ambion, Austin, TX) according to the manufacturer's recommendations. RNA concentration was determined by absorbance at 260 and 280 nm. cDNA was synthesized using a high capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA). Real-time PCR was performed using MESA GREEN qPCR MasterMix Plus for SYBR assays (Eurogentec, San Diego, CA) on a StepOnePlus real-time PCR system using Step One software, version 2.0 (Applied Biosystems, Foster City, CA). The conditions for real-time PCR were as follows: 5 min at 95°C for enzyme activation followed by 40 cycles of 15 s of 95°C for denaturation and 1 min of 60°C for annealing and extension. The melt curve was performed at 50°C for 1 h. Human GAPDH was used as an endogenous control (sense, CATGTTCGTCATGGGTGTGAACCA; antisense, AGT-GATGGCATGGACTGTGGTCAT). The primer sequences for human EGFr were as follows: sense, TTTGCCAAGGCAC-GACTAACAAG; antisense, ATTCCCAAGGACCACCTCA-CAGTT. All measurements were made in triplicate, and results were analyzed by the ⌬⌬CT method.
Confocal Microscopy-5 ϫ 10 5 T 84 cells were seeded onto 12-mm Millicell-HA filters for 6 -8 days before stimulation with IFN-␥. 48 h after stimulation, cells were fixed with 10% formalin in PBS for 10 min and permeabilized in 0.3% Triton X-100 in PBS for 10 min. Between steps, cells were washed three times with PBS. Blocking was performed with 20% donkey serum in PBS for 1 h, followed by a single washing step. As a primary antibody, rabbit anti-EGFr antibody was diluted in 2% BSA. Cells were incubated with primary antibody overnight at 4°C in a humidified chamber. After washing (three times), secondary Alexa-488-conjugated donkey anti-rabbit antibody (excitation/emission maxima at 495/ 519 nm; Molecular Probes, Eugene, OR) was added at a 1:500 dilution in 20 mg/ml BSA in PBS for 30 min at room temperature. After washing (three times with PBS), cells were incubated with Hoechst 33258 (excitation/emission maxima at 352/461 nm; Molecular Probes) in PBS (1:500) for 20 min at room temperature. The final washing stage consisted of four washes with PBS. Then the cells on the filter membrane were transferred onto glass slides and mounted in ProLong Gold Antifade Reagent (Molecular Probes). Confocal microscopy was performed using a Zeiss LSM 510 laser-scanning confocal system on a Zeiss Axioscope 2 upright microscope (Zeiss, Jena, Germany). Analyses of data were performed using the Zeiss LSM 5 Image Examiner software (Zeiss).
Electrophysiological Studies-T 84 cells seeded on permeable 12-mm Transwells were grown until confluence. Prior to Ussing chamber experiments, monolayers were pretreated with IFN-␥ (1000 units/ml) for 24 h. The duration of pretreatment was chosen to avoid effects on transepithelial resistance that occur with longer exposure to the cytokine. The concentration was chosen based on previous ion transport studies with IFN-␥ from our group (23). Cells were mounted in Ussing chambers (window area ϭ 0.6 cm 2 ) and bathed in oxygenated (95% O 2 and 5% CO 2 ) Ringer's solution (basolateral salt solution) at 37°C. After equilibration, monolayers were voltage-clamped to zero potential difference by application of short circuit current (I sc ). Amphotericin B (100 M) was then added to the apical side of the chamber to permeabilize the apical membrane, and after 30 min, an apical-to-basolateral K ϩ gradient (40 mM to 5 mM) was established. Ouabain (100 M) was added basolaterally to inhibit Na ϩ -K ϩ -ATPase. Because of the applied potassium gradient, changes in I sc (⌬I sc ) are wholly reflective of electrogenic potassium currents through the basolateral membrane. Changes in I sc were recorded in response to carbachol (100 M) with or without prior stimulation with EGF (100 ng/ml) for 20 min.
Animals-8-Week-old C57BL/6 mice, 15-week-old FVB, and 12-week-old mdr1a Ϫ/Ϫ (on an FVB background) mice were housed under conventional conditions and provided with food and water ad libitum. Animal studies were approved by the University of California, San Diego, Committee on Investigations Involving Animal Subjects.
Induction of Colitis-Acute colitis was induced in C57BL/6 mice by giving 3% DSS (molecular weight 36,000 -56,000, MP Biomedicals, LLC, Solon, OH) orally in drinking water for 7 days as first described by Okayasu et al. (24). Mice were sacrificed on day 8 after 7 days of DSS treatment or water as control. FVB mice were killed at 12 weeks of age; mdr1a Ϫ/Ϫ mice were observed for signs of spontaneously developing colitis, such as diarrhea, blood in stool, and ruffled fur, and sacrificed after such signs developed, typically at 12-15 weeks of age. All mice were sacrificed by cervical dislocation. Colons were removed, and the mucosal layer was stripped of serosal layers. Mucosal tissues were transferred in ice-cold Ringer's solution, dissected with clean tools, and incubated in ice-cold lysis buffer. Protein lysis and Western blot analysis were performed as described for cell culture experiments.
Statistical Analysis-All data are expressed as means Ϯ S.E. for a series of n experiments. Student's t test or analysis of variance with Student-Newman-Keuls post-test was used to com-pare mean values as appropriate using GraphPad Instat 3 software (Graph Pad Software, La Jolla, CA). Values of p Ͻ 0.05 were considered significant.

IFN-␥ pretreatment does not alter EGF-induced EGFr phosphorylation but increases EGFr expression at both mRNA and protein level in T 84 cells.
a, EGFr was immunoprecipitated, and phosphotyrosine or total EGFr was sequentially blotted in immunoprecipitates from T 84 cells in response to EGF stimulation at the indicated time points. Cells were pretreated for 48 h with IFN-␥ or medium. A representative Western blot is shown. b, densitometric analysis (n ϭ 8) showed a significant increase of phosphorylated EGFr in controls at 15 min of EGF stimulation. IFN-␥ did not affect EGF-induced EGFr phosphorylation in T 84 cells. Data are expressed as percentage of control without EGF stimulation. Asterisks denote significant differences from control (*, p Յ 0.05). c, EGFr mRNA expression in response to IFN-␥ treatment over 48 h, normalized to GAPDH (n ϭ 6, performed in triplicate). d, representative Western blot of whole cell lysates, sequentially probed for total EGFr or lamin A/C as loading control. e, densitometric analysis reveals significantly increased EGFr expression in IFN-␥-treated cells at the indicated time points with EGF stimulation. Data are expressed as -fold control. Asterisks denote significant differences from control (*, p Յ 0.05; **, p Յ 0.01). Error bars, S.E.

IFN-␥ Does Not Alter Overall EGF-induced EGFr
Phosphorylation-It has been shown that IFN-␥ alone is capable of transactivating the EGFr in T 84 cells (23,25). Based on these observations, we sought to determine if preincubation of T 84 cells with IFN-␥ alters EGF-induced phosphorylation of its receptor. We treated T 84 cells with IFN-␥ (1000 units/ ml) for 48 h prior to EGF stimulation (100 ng/ml) for up to 60 min. As expected, EGF significantly increased EGFr phosphorylation over time in control cells (Fig. 1, a and b). However, whereas basal EGFr phosphorylation was increased by IFN-␥, the cytokine did not alter EGF-induced phosphorylation of its receptor.
IFN-␥ Increases Epithelial EGFr Expression-We hypothesized that the apparent inability of IFN-␥ to modulate EGFinduced phosphorylation of its receptor might be due to changes in expression of the receptor after IFN-␥ treatment. To investigate this, we first performed real-time PCR on T 84 cell lysates pretreated for 48 h with IFN-␥. We found significantly higher expression of EGFr mRNA in treated compared with control cells (Fig. 1c). We next investigated whether this was accompanied by increased expression of the receptor at the protein level. As shown in Fig. 1, d and e, treatment with IFN-␥ for 48 h, followed by stimulation with EGF for 0 -90 min, revealed significantly higher levels of EGFr after pre-treatment with IFN-␥ compared with controls (p Ͻ 0.05 versus control, n ϭ 3). This increase was sustained at each time point of EGF stimulation. To confirm that this phenomenon was not limited to T 84 cells, we also treated HT29/cl.19A cells with IFN-␥ prior to challenging the cells with EGF for up to 90 min. IFN-␥ also increased EGFr expression in this cell line (Fig. 2, a and b).

IFN-␥ Causes Relocalization of EGFr in T 84
Cells-To verify the effect of IFN-␥ (1000 units/ml, 72 h) on EGFr levels, we analyzed EGFr expression by confocal imaging (Fig. 3). As predicted, IFN-␥ markedly increased staining for EGFr (green). However, whereas EGFr in control cells appeared to localize distinctly to the membrane, as demonstrated by close apposition to the submembrane actin (red), the receptor relocalized, in part, to the cytoplasm in cells treated with IFN-␥ and showed a more diffuse staining pattern (Fig. 3a). After 72 h of IFN-␥ exposure, a certain disruption of the cytoskeleton is evident. This might be mediated by AMP-activated protein kinase, as shown previously by our group (26). In the current study, we concentrated on the change in localization of the receptor. This change was reflected by a diminished intensity of EGFr co-localized with submembrane actin in IFN-␥-treated cells. Changes in EGFr localization were verified by scans along randomly assigned lines on representative confocal images in Fig. 3a using the LSM 510 colocalization software to con- firm overlap of fluorescence intensities (arbitrary units) (Fig.  3b). The peaks associated with each marker (EGFr (green) and submembrane actin (red)) showed diminished overlap in cells treated with IFN-␥ versus untreated cells. This shift of EGFr enrichment was further substantiated by quantification of total actin pixels (red) colocalized with total EGFr pixels (green). The extent of total actin red pixels that were colocalized with total EGFr green pixels in untreated cells was 19.5 Ϯ 1.0%. This value decreased by one-third in IFN-␥-treated cells (13.0 Ϯ 2.1%) (p Ͻ 0.05, n ϭ 3, 3-5 visual fields). Thus, the failure of IFN-␥ to potentiate EGF-induced EGFr phosphorylation, despite increasing levels of the receptor, may reflect the fact that the cytokine relocalizes the receptor to a site where it may be resistant to ligand binding and/or activation.

IFN-␥ Treatment Selectively Abrogates Phosphorylation of Different EGFr Tyrosine Residues-
The signaling consequences of EGFr activation depend on the precise census of receptor tyrosine residues that are phosphorylated. Thus, it was of interest to assess whether IFN-␥ also altered this pattern of tyrosine phosphorylation. We assessed phosphorylation of five EGFr tyrosine residues that are known to be autophosphorylation substrates (Tyr 992 , Tyr 1068 , Tyr 1086 , Tyr 1148 , and Tyr 1173 ; Fig. 4, a-e). Pretreatment with IFN-␥ significantly decreased the ability of EGF to induce phosphorylation of Tyr 1068 and Tyr 1086 (Fig. 4, a and b). On the other hand, phosphorylation of Tyr 992 was delayed but ultimately not reduced in IFN-␥-treated cells (Fig. 4c), whereas that of Tyr 1148 and Tyr 1173 was not significantly affected (Fig. 4, d and e). IFN-␥ Modifies ERK but Not PI3K Activation-To explore further the downstream signaling outcomes originating at EGFr in IFN-␥-treated cells, we assessed activation of two major signaling pathways involved in EGFr regulation of ion transport, the ERK MAPK and PI3K pathways. In IFN-␥treated cells, we detected a significant increase in ERK phosphorylation at baseline and following EGF stimulation compared with controls (Fig. 5, a and b). However, when compared with base-line levels in cells not exposed to EGF, the relative ability of EGF to increase ERK phosphorylation was actually reduced by IFN-␥ treatment (Fig. 5c). Therefore, EGF had a diminished incremental effect on ERK activation when this pathway had previously been activated by IFN-␥. In contrast, when we investigated the effect of IFN-␥ on EGF-stimulated PI3K signaling, as measured by phosphorylation of the downstream target, Akt-1, neither base-line nor EGF-stimulated phosphorylation of Akt was significantly influenced by pretreatment with IFN-␥ (Fig. 5, d and f). The findings were confirmed when PI3K activation was alternatively assessed using an ELISA for phosphorylated Akt-1 (Fig. 5f). Overall, we conclude that IFN-␥ emphasized the relative ability of EGFr to signal via PI3K at the expense of ERK.

Experimental Colitis Induces Opposing Effects on EGFr
Expression and ERK Activation-To explore whether our cell culture results could be translated into the setting of disease, we investigated EGFr and its downstream signaling targets, ERK and Akt-1, in two murine models of experimental colitis: acute colitis induced by DSS and spontaneous colitis in mdr1a Ϫ/Ϫ

Ion Transport Modulation by Interferon-␥ in Epithelial Cells
mice. For both models of experimental colitis, it has been shown that levels of IFN-␥ are elevated (6,7). Colonic tissues from C57BL/6 mice subjected to acute DSS colitis showed a significant increase in EGFr expression compared with tissues from control C57/BL6 mice (Fig. 6, a and b). Similarly, in mdr1a Ϫ/Ϫ mice that had developed spontaneous colitis, colonic tissues also showed significantly greater EGFr protein levels than seen in their FVB littermates (data not shown). This is in line with the results in T 84 cells treated with IFN-␥, showing increased levels of the receptor. Next, we sought to investi-gate if the downstream signaling targets of EGFr are modulated in inflamed colonic tissue from mice.
In DSS colitis, mucosal tissues exhibited similar levels of total ERK as seen in normal mice, but ERK phosphorylation was essentially absent (Fig. 6, c and d). Moreover, total Akt-1 expression was similar and Akt-1 phosphorylation was not significantly changed compared with controls (data not shown). Similar trends were observed in mdr1a Ϫ/Ϫ mice (data not shown). Therefore, although we cannot necessarily ascribe our findings specifically to an action of IFN-␥, we conclude that FIGURE 5. IFN-␥ increases ERK activation and reduces the capacity of EGF to activate ERK but not Akt-1. a, representative Western blot of whole cell lysates of T 84 cells, probed for phosphorylated or total ERK. b, in IFN-␥-treated cells, densitometric analysis revealed a significant increase in ERK phosphorylation at baseline (IFN-␥ versus control; *, p Յ 0.05; n ϭ 8). c, following EGF stimulation, increased ERK phosphorylation was detected in both groups. However, the relative increase in phosphorylation compared with time 0 revealed a reduced capacity of EGF to induce ERK activation in IFN-␥-treated cells. d, representative Western blot of whole cell lysates of T 84 cells, probed for phosphorylated (Ser 473 ) or total Akt-1. e, the relative increase in phosphorylation over time with EGF stimulation showed no influence of prior IFN-␥ treatment on EGF-induced Akt phosphorylation. f, phosphorylation of Akt-1, measured by ELISA, was significantly increased by EGF over time, but this was not influenced by the presence or absence of IFN-␥ treatment (*, p Յ 0.05 after 30 min of EGF stimulation; n ϭ 3 in duplicate). Error bars, S.E. murine colitis is associated with an increase in EGFr expression and a biasing of downstream signaling targets away from ERK, as seen in cell culture.

IFN-␥ Modifies Ability of EGF to Reduce Basolateral Potassium Currents in T 84
Cells-Because IFN-␥ pretreatment altered intracellular signaling pathways downstream of EGFr in intestinal epithelial cells, it was of interest to examine whether this also had functional consequences. We decided to investigate specifically the basolateral potassium current because EGF had previously been shown to inhibit this conductance in cells treated with calcium-dependent agonists, such as carbachol (CCh) (27)(28)(29)(30). Confluent monolayers of T 84 cells were treated with IFN-␥ (24 h) and then mounted in Ussing chambers and studied under conditions where differences in I sc are the result of potassium currents across the basolateral membrane (31). A 24-h period of IFN-␥ preincubation was chosen to avoid the changes in barrier function that occur after 48 -72 h of IFN-␥ exposure and are probably mediated by AMP-activated protein kinase (26). As expected, IFN-␥ reduced the basal level of potassium conductance (Fig. 7a). Moreover, EGF inhibited the CChstimulated ⌬I sc in control cells, as expected, but this effect was lost in cells that had been treated with IFN-␥ (Fig. 7a). These findings imply that functional consequences of EGF treatment may be altered in epithelial cells exposed to IFN-␥.

DISCUSSION
In this study, IFN-␥ was shown to alter EGFr expression, as well as signaling pathways originating from the receptor, in FIGURE 6. In experimental colitis, mucosal EGFr expression is increased, whereas ERK phosphorylation is reduced. 8-Week-old C57BL/6 mice were treated with 3% DSS in drinking water to induce an acute colitis or tap water as control. a, representative Western blot of lysates from mucosal tissue from distal colon, probed for total EGFr or ␤-actin as a loading control. b, elevated EGFr expression in mice treated with DSS for 7 days in drinking water (*, p Ͻ 0.05). c, representative Western blot of lysates from mucosal tissue from distal colon, probed for phosphorylated or total ERK. d, reduced ERK expression in mice treated with DSS (****, p Ͻ 0.0001). Error bars, S.E. FIGURE 7. IFN-␥ leads to loss of the inhibitory effect of EGF on basolateral K ؉ channel activity in T 84 cells. T 84 cell monolayers were pretreated with or without IFN-␥ for 24 h before mounting in Ussing chambers. Short circuit current (I sc ) measurements, reflective of basolateral K ϩ channel activity, were recorded after stimulation with CCh in the presence or absence of EGF. a, in control cells, EGF significantly inhibited CCh-stimulated K ϩ channel activity (p Ͻ 0.05, n ϭ 7). However, in IFN-␥ pretreated cells, which showed a reduced CCh-stimulated I sc , the inhibitory effect of EGF was lost (n ϭ 7, not significant (n.s.)). b, hypothetical model for the results of this study. In the non-inflamed state (left half), EGF binds to its receptor and activates both MAPK and PI3K pathways, which lead to an inhibitory effect on basolateral CCh-induced potassium currents. In contrast, during inflammation (right half), IFN-␥ levels are elevated locally. This leads to a shift of the downstream signaling pathways in favor of EGF-stimulated PI3K. Furthermore, the inhibitory effect of ERK on basolateral CCh-induced potassium currents is lost. Error bars, S.E. colonic epithelial cells. Modulation of signaling pathways downstream of EGFr by IFN-␥ probably accounts for functional changes in ion transport responses in our cell model and may also explain previous observations we have made in mouse models of experimental colitis, where EGF could partially restore impaired sodium absorptive responses in disease (19). IFN-␥ plays a pivotal role in experimental colitis as well as in human inflammatory bowel disease. We therefore used a prolonged IFN-␥ pretreatment to mimic a chronic state of inflammation.
In the current studies, IFN-␥ tended to increase base-line EGFr phosphorylation, consistent with our earlier results showing that IFN-␥ transactivates EGFr (23) but had no effect on overall EGF-induced EGFr phosphorylation. IFN-␥ reportedly had different effects on EGFr phosphorylation in two other cell lines tested. In A431 cells, IFN-␥ caused rapid and reversible independent tyrosine phosphorylation of EGFr, whereas in HepG2 cells, the presence of EGF was necessary for IFN-␥ to transactivate the receptor (25,32). This suggests various activation patterns and effects of inflammatory cytokines in different contexts. Even so, our studies imply that the differential effects of EGF in normal and inflamed settings that we have previously reported (19) probably cannot simply be attributed to a global change in EGFr activation.
Although we did not detect overall changes in EGF-induced phosphorylation of EGFr under the conditions studied, we observed increased expression of the receptor at both the mRNA and protein level. This effect was reproduced in the HT-29cl19a human colonic epithelial cell line and is consistent with a report of increased EGFr in the human breast cancer cell line MDA 468 (33) after incubation with IFN-␥. We also detected a change in receptor localization measured by immunofluorescent staining. We conclude that long term stimulation with IFN-␥ actually decreases receptor abundance on the cell surface, presumably by stimulating receptor internalization (34). This might partly account for the fact that IFN-␥ did not increase EGFr activation upon subsequent EGF stimulation, despite up-regulation of receptor expression. Our group demonstrated previously that barrier disruption in cells exposed to IFN-␥ for a prolonged time period was mediated by AMP-activated protein kinase (26). In the current studies, we did not examine tight junction proteins, but we detected a disruption of the actin cytoskeleton in the immunofluorescence studies. Whether AMP-activated protein kinase activation leads to internalization of EGFr was not tested here but deserves further study.
To seek an alternative explanation for the ability of inflammation to modify EGFr signaling, the degree of phosphorylation of specific EGFr tyrosine residues was studied. In fact, of the five tyrosine residues tested, we detected significantly lower phosphorylation of two residues, whereas the other residues were not significantly affected. We therefore suggest, based on these and other data from our group (35)(36)(37)(38), that IFN-␥ may activate protein-tyrosine phosphatases directed against specific EGFr residues. This would be concordant with our observations that, in vitro, IFN-␥ increases the expression and activity of one of the protein-tyrosine phosphatases, namely PTPN2 (38). We could also demonstrate increased PTPN2 expression in intestinal biopsy specimens from patients with Crohn disease (38). Other inflammatory stimuli, such as reactive oxygen species, can influence protein-tyrosine phosphatase activity as well (39,40). Moreover, differences in the phosphorylation pattern of tyrosine residues in EGFr can lead to distinct cellular outcomes in intestinal epithelial cells (37). Upon activation of EGFr, distinct downstream signaling pathways, such as ERK-MAPK, PI3K/Akt-1, or the JAK/STAT pathway, are activated. We did not directly test the functional effects of the specific EGFr tyrosine residues affected by IFN-␥, but the shift toward the PI3K pathway (see below) probably indicates a functional consequence of the observed dephosphorylation of a specific constellation of tyrosine residues.
We found a significant increase in ERK but not Akt phosphorylation at baseline after incubation with IFN-␥. Conversely, IFN-␥ reduced the relative ability of EGF to activate ERK but did not alter its ability to activate Akt-1. This implies that the consequences of EGFr activation may be redirected in inflammation. Previously, our group has shown that ERK accounts for termination of CCh-induced chloride secretory responses (15,41,42). Conversely, EGF favors the PI3K pathway over the ERK-MAPK pathway in modulating colonic ion transport responses in colitic mice (19). Our data shown here are concordant with these previous studies. We also investigated mucosal protein levels of ERK and Akt, as well as EGFr, in two models of experimental colitis. EGFr levels were elevated in inflamed tissues in agreement with previous studies (43,44). Because EGFr plays a pivotal role in mucosal restitution, the elevated EGFr levels could be explained as an adaptation to promote mucosal healing (45)(46)(47)(48) and PI3K activation. This was also proposed in previous studies of experimental colitis in mice with EGFr dysfunction or EGFr ligand deficiency (16,17). ErbB4, a EGFr family member, is also increased in tissues from mice with colitis as well as in biopsies from patients suffering from Crohn disease (49).
To further investigate the consequences for ion transport, we used T 84 cells to demonstrate that IFN-␥ abolishes the inhibitory effect of EGF on CCh-induced basolateral potassium currents. Treatment of polarized intestinal epithelial cells with IFN-␥ leads to down-regulation of many transporters and ion pumps, such as basolateral potassium channels, the Na ϩ -K ϩ -2Cl Ϫ cotransporter, and Na ϩ -K ϩ -ATPase (50 -52). However, an ability of IFN-␥ to selectively modulate the effect of EGF on ion transport responses had not been tested. Because basolateral potassium efflux is a driving force for apical calcium-dependent chloride secretion (53), the finding that IFN-␥ reverses the inhibitory effect of EGF on a Ca 2ϩ -dependent potassium current could partly explain diarrheal symptoms in disorders such as chronic inflammatory bowel diseases. Ion transport is reduced overall if IFN-␥ is present. In our experimental setting, IFN-␥ produced significantly reduced basolateral potassium current, as expected. Nevertheless, small changes in K ϩ current could influence additional chloride secretion in inflammation. EGF was no longer able to inhibit the basolateral potassium current following IFN␥ treatment, implying a subtle change in mucosal transport homeostasis. Earlier studies from our group have shown that direct activation of EGFr by EGF leads to inhibition of basolateral potassium channel activity via activation of

Ion Transport Modulation by Interferon-␥ in Epithelial Cells
the PI3K pathway rather than the ERK-MAPK pathway (27). On the other hand, if EGFr is transactivated by G protein-coupled receptors, shedding of TGF-␣ leads to activation of ERK, limiting chloride secretion by inhibiting apical chloride channels (12,15,54,55). Further, McCole et al. (19) demonstrated that EGF also influences ENaC activity in colitis via PI3K. Our findings may also imply, therefore, that other effects of EGF may be altered in epithelial cells exposed to IFN-␥. Fig. 7b summarizes our hypothetical model whereby in the non-inflamed state (left half), ERK predominantly limits basolateral CCh-induced potassium currents when EGF activates its receptor. This inhibitory effect is apparently diminished during inflammation (right half) due to a shift to the PI3K pathway and loss of ERK signaling. It is likely that the basolateral potassium channel targeted in our study is the Ca 2ϩ -activated KCNN4 channel (56,57). However, this will require clarification in future studies.
In summary, in Crohn disease and ulcerative colitis, one of the most common and disabling symptoms is diarrhea. Diarrheal disorders are ultimately the result of decreased absorption and/or increased secretion of fluid and electrolytes (58). It has been shown that EGFr plays a protective role in inflammatory conditions and that ERK negatively regulates chloride secretion. In this work, we demonstrate that IFN-␥ causes defective EGFr activation in intestinal epithelial cells, with a shift toward the PI3K pathway in inflammation. These alterations in downstream signaling pathways originating at EGFr lead to functional consequences for ion transport responses in disease. Preincubation of intestinal epithelial cells with IFN-␥ reverses the inhibitory effect of EGF on carbachol-induced potassium currents, which are required for anion secretion. Insights into perturbed activation of growth factor receptors and their functional consequences may lead to novel treatment strategies for chronic inflammatory bowel disease. We must, however, be mindful of the caveat that administration of EGFr ligands by themselves should be approached with caution due to the increased risk of colorectal cancer in ulcerative colitis patients, although such an approach has already been demonstrated to have clinical merit for the treatment of left-sided ulcerative colitis (20).