A characteristic of many enteropathies is increased epithelial permeability, a potentially pathophysiological event that can be evoked by T helper (Th)-1 (i.e., IFN-γ) and Th2 (i.e., IL-4) cytokines and bacterial infection [e.g., enteropathogenic Escherichia coli (EPEC)]. The green tea polyphenol (−)-epigallocatechin gallate (EGCG) has immunosuppressive properties, and we hypothesized that it would ameliorate the increased epithelial permeability induced by IFN-γ, IL-4, and/or EPEC. EGCG, but not the related epigallocatechin, completely prevented the increase in epithelial (i.e., T84 cell monolayer) permeability caused by IFN-γ exposure as gauged by transepithelial resistance and horseradish peroxidase flux; EGCG did not alleviate the barrier disruption induced by IL-4 or EPEC. IFN-γ-treated T84 and THP-1 (monocytic cell line) cells displayed STAT1 activation (tyrosine phosphorylation on Western blot analysis, DNA binding on EMSA) and upregulation of interferon response factor-1 mRNA, a STAT1-dependent gene. All three events were inhibited by EGCG pretreatment. Aurintricarboxylic acid also blocked IFN-γ-induced STAT1 activation, but it did not prevent the increase in epithelial permeability. Additionally, pharmacological blockade of MAPK signaling did not affect IFN-γ-induced epithelial barrier dysfunction. Thus, as a potential adjunct anti-inflammatory agent, EGCG can block STAT1-dependent events in gut epithelia and monocytes and prevent IFN-γ-induced increased epithelial permeability. The latter event is both a STAT1- and MAPK-independent event.
- intestinal barrier
- signal transducer and activator of transcription-1
the intestinal epithelium delineates the gut lumen from the body proper–the contiguous single-cell layer functioning as a barrier to antigens and microorganisms derived from the diet and colonic microbiota (30). Many human gut disorders have in common a disrupted intestinal epithelial barrier. For example, increases in both paracellular and transcellular permeability have been shown in patients with inflammatory bowel disease (IBD) or in resected tissues from these individuals examined ex vivo (35). Certainly, increased gut permeability contributes to the pathophysiology of IBD, and it has been suggested that a “leaky” gut may be the primary defect in a cohort of patients (27). Clearly, maintenance of the epithelial barrier is a key component of innate immunity.
Treatment of human intestinal epithelial cell (IEC) monolayers with cytokines representative of T helper (Th)-1 (e.g., IFN-γ, TNF-α) and Th2 dominated responses (e.g., IL-4, IL-13) or with pathogenic Escherichia coli results in decreased barrier function as measured by decreased transepithelial resistance (TER) and increased permeability to low (e.g., mannitol, inulin) and high [e.g., horseradish peroxidase (HRP), ovalbumin] molecular weight marker molecules (3, 6, 11, 23, 24, 33). These in vitro analyses have been corroborated to some degree by a lesser number of in vivo and ex vivo investigations (16). Moreover, numerous studies have demonstrated derangement or perturbation of the epithelial tight junction and tight junction-associated and cytoskeletal proteins accompanying cytokine or pathogen-induced increases in epithelial permeability, implicating the tight junction as a functional endpoint of the physiological stimulus (33, 42). In contrast to the awareness of factors that increase epithelial permeability, comparatively little is known of mechanisms/signals that enhance epithelial barrier function (19)—a regrettable situation given its' importance in gut homeostasis (30).
Recently, the putative anti-cancer properties of the green tea polyphenol (−)-epigallocatechin gallate (EGCG) have been complemented by the demonstration of a protective effect for green tea extract in the IL-2-deficient model of murine colitis (40). Given that EGCG has been found to inhibit activation of IFN-γ-induced signal transduction events (28, 37) and that IFN-γ has been implicated in several animal models of gut inflammation (5), we hypothesized that a therapeutic benefit of EGCG could be via inhibition of IFN-γ signaling and maintenance of epithelial barrier function. We sought to determine whether EGCG could inhibit the increase in epithelial monolayer permeability evoked by IFN-γ, IL-4, and enteropathogenic E. coli (EPEC) and then to assess epithelial and monocyte intracellular signals that might be affected by EGCG, particularly the STAT proteins, as ubiquitous signal transducers of cytokine receptors (32).
EGCG significantly prevented the epithelial barrier dysfunction evoked by IFN-γ but not that caused by IL-4 or EPEC. Similarly, IFN-γ-induced STAT1 activation in epithelial cells and monocytes, but not IL-4-induced STAT-6 activation, was inhibited by EGCG. However, another inhibitor of STAT1 activation, aurintricarboxylic acid (ATA), also blocked IFN-γ-induced STAT1 phosphorylation but did not ameliorate the permeability defect, suggesting that IFN-γ modulation of epithelial permeability is STAT1 independent. Thus EGCG is capable of blocking IFN-γ signaling mediated via STAT1 in immune (i.e., monocytes) and immune accessory cells (i.e., epithelial), and can also preserve the barrier function in epithelia exposed to IFN-γ, an event that is likely STAT1 independent.
Cell culture and reagents.
T84 human colonic epithelial cells were cultured at 37°C with 5% CO2 in a 1:1 mixture of DMEM and Ham's F-12 medium supplemented with 2% (vol/vol) penicillin-streptomycin and 1.5% (vol/vol) HEPES (all from Invitrogen, Burlington, ON) and 10% fetal calf serum (CanSera, Toronto, ON, Canada) (26). The THP-1 human monocytic cell line was cultured as above except that RPMI 1640 replaced the DMEM/F-12 mixture. Before experimental use, THP-1 cells were treated with 10 nM phorbol 12-myristate 13-acetate (Sigma, St. Louis, MO) for 48 h to induce differentiation into a macrophage-like phenotype (38). For molecular assays, 2 × 106 cells/well were seeded in six-well plates and cultured until confluent.
Human recombinant IFN-γ was from R&D Systems (Minneapolis, MN) and human recombinant IL-4, EGCG, epigallocatechin (EGC), ATA, 2-methylthio-ATP (MTA), tyrphostin AG 490 (AG 490), cyclohexamide (CHX), and fludarabine were from Sigma. The MEK/ERK inhibitors, PD-98059 and U-O126, and the p38 inhibitor SB-203580, were from Calbiochem (San Diego, CA). Enteropathogenic E. coli (EPEC) was provided by Dr. P. M. Sherman (Hospital for Sick Children, Toronto, ON, Canada) and was cultured in Luria Bertani broth as previously described (33). Cytokines, bacteria, and pharmacological agents were used at the concentrations specified in the figure legends and are based on salient publications.
Assessment of epithelial permeability.
One million T84 cells were seeded onto 1-cm2 semipermeable filter supports (pore size 400 nm; Costar, Corning, Cornell, NY), cultured under the above conditions, and used when TER (an index of paracellular permeability) of the monolayer was ≥1,000 Ω·cm2 as measured by a voltometer and electrodes (Millipore, Bedford, MA) (8). Cytokines and inhibitors were added to the basal compartment, and EPEC was added to the apical compartment of the culture well. TER is expressed as the percentage of pretreatment resistance to normalize for variation in absolute values between individual monolayers (8). HRP flux assays, typically a measure of transcellular transport, were performed 48 h postcytokine treatment. The amount of intact, apical-to-basal translocated HRP (type VI, Sigma) over 2 h was determined as previously described (3) and is expressed as percent recovery of total HRP added (i.e., 10 μM).
Western blot analysis.
Whole cell lysates were prepared by rocking cells in ice-cold RIPA buffer containing protease (complete protease inhibitor cocktail; Roche, Indianapolis, IN) and phosphatase inhibitors (100 mM NaF, 100 mM NaVO3) for 30 min at 4°C. Lysates were clarified by centrifugation, and the supernatant was collected and stored at −70°C. Protein concentration was determined by using the Bradford microplate assay (Bio-Rad, Hercules, CA). Samples (20–40 μg protein) in reducing-loading buffer were boiled and electrophoresed through 4–10% (29:1 acrylamide/bisacrylamide) SDS gels. Separated proteins were electroblotted to Immobilon nitrocellulose membrane (Millipore, Bedford, MA) and blocked in 5% nonfat Carnation powdered milk/Tris-buffered saline/Tween 20 (TBST) for 1 h. Primary antibodies used were anti-pY701-STAT1, 1:2,000 (Upstate Biotechnology, Lake Placid, NY); anti-pY641-STAT6, 1:1,000; anti-caspase-3, 1:2,000 (Cell Signaling, Beverly, MA); anti-STAT6, 1:1,000 (Zymed Laboratory, San Francisco, CA); anti-STAT1, 1:4,000, anti-interferon response factor (IRF)-1, 1:2,000; and anti-actin, 1:500 (Santa Cruz Biotechnology, Santa Cruz, CA). Blots were washed and incubated with secondary antibody-HRP conjugates for 1 h [goat anti-rabbit or rabbit anti-mouse (both at 1:4,000; Santa Cruz Biotechnology)], washed extensively, and immunoreactive proteins were visualized by using enhanced chemiluminescence (Amersham Pharmacia, Piscataway, NJ) and by exposing the membrane to Kodak XB-1 film.
Nuclear extracts were prepared after the method of Andrews and Faller (2) with minor modifications (8). Protein concentration was determined as above. EMSAs were conducted according to a previously published protocol (4). Briefly, nuclear extracts (5–10 μg of protein) in binding buffer were incubated for 30 min with [32P]dCTP (New England Nuclear Life Science Products, Boston, MA)-labeled oligonucleotide probe (hSIE) containing a high-affinity STAT1 binding site (5′-GTCGACATTTCCCGTAAATC-3′ and 5′-TCGACGATTTACGGGAAATG-3′; Ref. 41) or the STAT6 binding site (8). Samples were electrophoresed through a nondenaturing 6% (40:1 bis/acrylamide) polyacrylamide gel for 2.5 h at 120 V, dried under vacuum at 80°C, and visualized by autoradiography after overnight exposure (−70°C) to Kodak XAR film.
Total RNA was isolated by using TRIzol (Invitrogen) following the manufacturer's protocol and quantified by UV-spectrophotometry. cDNA was generated (RevertAid RNAse H−, MBI Fermentas, Hanover, MD) from polyA+ mRNA following the manufacturer's protocol using 2 μg of total RNA and oligo(dT) primers. After 30 cycles of PCR (Platinum Taq, Invitrogen) with 2 μl cDNA, amplified fragments were electrophoresed through 2% agarose/Tris-boric acid-EDTA (TBE) gels in 1× TBE buffer and visualized under UV light. Primers were obtained by using Primer3 software (http://www.genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi/) with published mRNA sequences (GenBank). Primers used were IRF-1 forward: 5′-CGA TAC AAA GCA GGG GAA AA-3′ (10 pmol), IRF-1 reverse: 5′-TAG CTG CTG TGG TCA TCA GG-3′ (10 pmol); β-actin forward: 5′-CCA CAG CAA GAG AGG TAT CC-3′ (3 pmol), reverse: 5′-CTG TGG TGG TGA AGC TGT AG-3′ (3 pmol).
TER and HRP flux data were compared by ANOVA followed by Newman-Keuls statistical comparisons, where P ≤ 0.05 was set as the level of statistically significant difference. Data are presented as means ± SE, and n = number of epithelial preparations.
EGCG inhibits IFN-γ but not IL-4 or EPEC evoked increases in epithelial permeability.
Exposure of T84 monolayers to IFN-γ evoked a significant loss of barrier function that was both time and dose dependent, as assessed by TER. The addition of 20 ng/ml IFN-γ to the basal side of T84 monolayers had negligible effects on TER over a 24-h period (data not shown) but resulted in a significant (∼50%) decrease in TER by 48 h posttreatment (Fig. 1A). Furthermore, a 1- to 3-h exposure to IFN-γ, followed by washout, did not result in diminished barrier function when monolayers were assessed 48 h later. Rather, a minimum of 6 h of continuous exposure to IFN-γ was required to evoke a drop in TER 48 h later [control = 104 ± 6%; IFN-γ (20 ng/ml; 3 h) = 95 ± 5%; IFN-γ (6 h) = 65 ± 9%* of pretreatment TER; n = 3, *P ≤ 0.05 compared with control]. The IFN-γ-induced drop in TER required protein synthesis, being completely blocked by CHX (1 μg/ml) cotreatment [control = 87 ± 5, IFN-γ (20 ng/ml; 48 h) = 15 ± 2%; IFN-γ + CHX = 142 ± 2% of pretreatment TER; n = 3, P ≤ 0.05 across all groups] (36).
Monolayers incubated with EGCG and IFN-γ maintained their pretreatment TER (Fig. 1A), and this was not due to EGCG-mediated increases in TER, because EGCG only did not alter TER compared with control values. Furthermore, cotreatment of T84 monolayers with the related polyphenol, EGC (100 μM) did not prevent the IFN-γ-induced drop in TER [control = 104 ± 6, IFN-γ (20 ng/ml; 48 h) = 36 ± 5%, IFN-γ + EGC = 52 ± 9% of pretreatment TER; n = 3, P ≤ 0.05 compared with control]. Thus the barrier-preserving property of EGCG is not common to all green tea constituents. Extending the consideration of epithelial barrier function, we tested the ability of EGCG to prevent IFN-γ increases in the transepithelial flux of HRP, which crosses epithelia via transcellular and paracellular pathways (3). Over the 2-h incubation period, HRP translocation markedly increased across IFN-γ-treated monolayers compared with control and EGCG-only-treated epithelia. Complementing the TER data, concomitant treatment of epithelia with EGCG abolished the increased HRP translocation evoked by IFN-γ exposure (Fig. 1B).
To assess the specificity of the EGCG barrier-preserving effect, we examined T84 monolayers treated with other agents known to increase epithelial permeability, specifically IL-4 and EPEC (3, 8, 33). In contrast to the marked inhibition of the IFN-γ-induced TER decrease, treatment of T84 monolayers with EGCG did not attenuate the reduction in TER induced by IL-4 or EPEC (Fig. 1C).
IFN-γ-induced increases in epithelial permeability are not accompanied by evidence of cell death.
The reduced TER and increased HPR flux induced by IFN-γ could be a consequence of cytokine cytotoxicity. Western blot analysis of whole cell protein extracts and assessment of caspase-3 levels (whole and cleaved) as an early marker of apoptosis provided no data in support of IFN-γ-induction of apoptosis in this study (data not shown).
IFN-γ activation of STAT1, but not IL-4 activation of STAT6, is inhibited by EGCG.
Our interest in the signaling events that regulate epithelial barrier function in response to immune mediators (8, 25) coupled with the selective ability of EGCG to inhibit IFN-γ, but not IL-4, permeability increases led to an examination of STAT activation in IFN-γ- or IL-4-treated T84 monolayers. Incubation of T84 cells with IFN-γ induced obvious, early (30 min) STAT1 phosphorylation and DNA-binding activity (Figs. 2A and 3A). Assessment of the kinetics of the IFN-γ-induced STAT1 tyrosine phosphorylation showed that continued incubation with IFN-γ over 24 h resulted in a peak of STAT1 phosphorylation (pSTAT1) at 4 h, followed by a progressive diminution until 24 h posttreatment when pSTAT1 levels were identical to control levels (Fig. 2B). When T84 cells were exposed to IFN-γ for 30 min only, increased pSTAT1 was detectable for ≤3 h (Fig. 2C), whereas a 4-h IFN-γ exposure resulted in elevated pSTAT1 ≤6–8 h after withdrawal of the cytokine (Fig. 2D). Decreases in the level of activated STAT1 paralleled increases in nonphosphorylated STAT1 levels, consistent with the STAT1 gene being autoregulated by STAT1 transcriptional activity (Fig. 2, B–D). We have previously examined IL-4-induced STAT6 activation in T84 cells by using EMSA in which there was a dose-dependent increase in STAT6 DNA-binding activity that peaked 30 min post-IL-4 stimulation (8).
To explore any effect of EGCG on IFN-γ- and IL-4-treated epithelia, we examined the effect of EGCG treatment on cytokine-induced STAT phosphorylation. Due to the rapid induction of STATs after stimulation, we adopted a pretreatment regimen in which EGCG was added to T84 cells 30 min before incubation with cytokine. IFN-γ-evoked pSTAT1 was markedly reduced in T84s pretreated with EGCG (Fig. 3A), whereas EGCG alone had no effect on pSTAT1 immunoreactivity. Similarly, the STAT1 activation in THP-1 cells treated with IFN-γ was virtually abolished (Fig. 3A). In marked contrast, EGCG did not affect IL-4-induced pSTAT6 levels in T84 epithelial or THP-1 macrophages (Fig. 3B). Notably, and consistent with the TER data, the related polyphenol EGC did not affect IFN-γ-induced pSTAT1 (Fig. 3C).
To further characterize the EGCG-mediated inhibition of STAT1 activity, we examined DNA-binding by EMSA and mRNA and protein levels of IRF-1, a known STAT1-regulated gene (34). EGCG pretreatment blocked IFN-γ-induced STAT1 DNA binding activity, and, as expected from this result, upregulation of IRF-1 mRNA transcript and protein (Fig. 4, A–C).
IFN-γ-induced disruption of epithelial barrier function is STAT1 independent.
The correlation of EGCG inhibition of IFN-γ-evoked STAT1 activation and preservation of epithelial barrier functions suggested that IFN-γ-induced increases in epithelial permeability could be STAT1 dependent. To test this supposition, we employed a series of pharmacological inhibitors in TER and molecular studies. The reputed STAT1 inhibitor fludarabine (Flu) (17) did not abrogate the IFN-γ-induced drop in TER [control = 104 ± 3, IFN-γ (20 ng/ml; 48 h) = 44 ± 6%, IFN-γ + Flu (50 μM) = 42 ± 4% of pretreatment TER ± SE; n = 8, P ≤ 0.05 across all groups] but also failed to block STAT1 DNA-binding on EMSA (data not shown). Somewhat surprisingly, the Jak2 inhibitor AG 490 was unable to prevent the IFN-γ-induced drop in TER [control = 113 ± 7, IFN-γ (20 ng/ml; 48 h) = 62 ± 4%, IFN-γ + AG 490 (10 μM) = 71 ± 5% of pretreatment TER; means ± SE; n = 3, P ≤ 0.05]. ATA has recently been shown to inhibit IFN-γ-induced STAT1 phosphorylation in murine macrophages (9). Pretreatment of T84 cells with ATA strongly inhibited IFN-γ-induced pSTAT1 (Fig. 5A, inset) but did not block the increase in epithelial permeability (Fig. 5A). Additionally, the methylase inhibitor, MTA, was recently shown to inhibit STAT1 DNA binding after IFN treatment (29), but again this agent could not prevent the IFN-γ-induced hyperpermeability in T84 monolayers (Fig. 5B).
IFN-γ-induced disruption of epithelial barrier function is Erk and p38 MAPK independent.
Finally, in a further attempt to define the signaling pathways important for IFN-γ barrier disruption, we examined MAPK pathways. IFN-γ has been demonstrated to activate ERK and p38 MAPKs (18, 22), ERK activity is required for IL-17-induced barrier increases (21), and EGCG has been shown to block MAPK signaling (10). Thus we sought to assess the role of these kinases in IFN-γ-induced barrier disruption. T84 monolayers were treated with the MAPK inhibitors PD-98059, U-O126 (both block ERK activation), or SB-203580 (inhibits p38) and concomitantly exposed to IFN-γ. Forty-eight hours posttreatment, measurement of TER indicated that use of these pharmacological inhibitors of MEK/ERK and p38 MAPK did not affect the IFN-γ-induced barrier dysfunction (Fig. 6).
Cytokines directly modulate many epithelial functions. For example, in vitro studies have shown that, among others, IFN-γ and IL-4 increase IEC monolayer permeability (11, 23, 24). The present findings confirm earlier studies showing decreased TER in IFN-γ- or IL-4-treated epithelia and demonstrate IFN-γ-induced increased transepithelial movement of the large protein HRP, indicative of increased transcellular movement of lumen-derived substances. Similar in vivo cytokine effects could have a major impact on gut homeostasis, such that increased access of lumenal material could precipitate or exaggerate pathophysiological processes (16). Increased epithelial permeability and a concomitant increase in tissue levels of IFN-γ or IL-4 is common in inflammatory enteropathies (14, 20), and so, strategies to neutralize this cytokine-induced hyperpermeability at the level of the epithelium would be of therapeutic benefit.
Current anti-inflammatory therapies rely heavily on steroids and broad-spectrum immunosuppressives, an unacceptable position that is increasingly leading to the characterization and use of biologicals (e.g., anti-TNF-α) and neutraceuticals. EGCG falls within the latter category and is receiving considerable attention as an anti-inflammatory agent (7). Indeed, because EGCG is the predominant polyphenol in green tea, it seems likely that it was largely responsible for the reduced inflammation observed in spontaneously colitic IL-2-deficient mice drinking green tea (40). However, this study addressed neither the target cell nor the mechanism of EGCG action. We posited a role for EGCG as an epithelial barrier-enhancing agent and found that it did, in fact, completely prevent increases in T84 monolayer permeability evoked by IFN-γ. However, the increased epithelial permeability caused by IL-4 or EPEC was unaffected by EGCG cotreatment. Thus the ability of EGCG to limit IFN-γ-induced increases in epithelial permeability is likely a component of the anti-inflammatory nature of this polyphenol. We add the caveat that if EGCG, or green tea, is to complement current anti-inflammatory therapies, a real benefit will probably be restricted to disorders involving specific cytokines (e.g., IFN-γ).
IFN-γ disruption of epithelial barrier function has been correlated with redistribution of perijunctional F-actin and reduced zonula occludens-1 (ZO-1) and occludin (42). Yet paradoxically, reduced TER can precede altered tight junction (TJ) structure, and dramatic changes in TER can occur with doses of IFN-γ that have only subtle effects on TJ structure or composition (6). Our preliminary observations also suggest that IFN-γ-induced changes in TER are associated with subtle changes in ZO-1 and occludin (personal observation). We consistently find a lack of obvious apoptosis (gauged by caspase-3 cleavage) 48 h posttreatment with IFN-γ, which contrasts with a recent report in which incubation of T84 monolayers with IFN-γ for a longer time period (72 h) resulted in increased caspase-3 activity but only modest increases in additional apoptotic readouts (6). However, apoptosis was discounted as mediating the IFN-γ effect, because pharmacological inhibition of caspases did not block increased permeability (6). Direct cytotoxic effects of IFN-γ incubation have been reported, but these data relate to the HT-29 colonic epithelial cell line grown on plastic and thus are not directly comparable to our studies (13). In addition, oxidant damage has been implicated in IFN-γ-induced increases in epithelial permeability (39), but this is unlikely to be the case here, because EGC, which has similar antioxidant properties to EGCG, did not affect IFN-γ disruption of T84 barrier function. These data provide a structural basis for the IFN-γ modulation of TER. However, the intracellular signaling pathway(s) that mediate IFN-γ-induced epithelial barrier dysfunction are unclear.
STAT1 is the principle intracellular signaling molecule activated by IFN-γ. Menegazzi et al. (28) showed that IFN-γ-induced STAT1 activation in carcinoma-derived cell lines of nongut origin was blocked by EGCG. We found that EGCG significantly reduced STAT1 activation in T84 epithelia and THP-1 monocytes/macrophages. IFN-γ-induced increases in IRF-1 (a STAT1-regulated gene) were diminished in both cell types treated with EGCG, confirming functional inhibition of STAT1 activity. EGCG reduces the activation of other STAT proteins, such as STAT3 (28), but it is not a pan-STAT inhibitor, because it did not affect IL-4-evoked STAT6 activation in enterocytes or monocytes. A number of mechanisms could account for EGCG inhibition of IFN-γ-induced STAT1 activation: 1) interaction with the IFN-γ receptor; 2) inhibition of Jak2 activity (Jak1 is less likely a target, because this would predict inhibition of IL-4 STAT6 activation); or 3) blockade of STAT1 phosphorylation sites. Regardless of the precise mechanism of action, the clear inhibition of IFN-γ-driven STAT1 signaling in immune (i.e., macrophages) and immune accessory cells (i.e., epithelia) underscores the anti-inflammatory potential of EGCG, a naturally occurring polyphenol.
With respect to epithelial barrier function, concomitant EGCG treatment prevented IFN-γ-induced increases in epithelial permeability, as gauged by TER and HRP flux measurements. The facts that IFN-γ-induced decreased barrier function occurred 48 h posttreatment, the longevity of the IFN-γ-induced STAT1 activation and the dramatic inhibition of STAT1 activation by EGCG all suggest that IFN-γ regulation of epithelial permeability was a STAT1-dependent phenomenon. Validated, highly specific, pharmacological inhibitors of STAT1 activity are not available. Flu, previously reported to block STAT1 activation in human peripheral blood mononuclear cells (hPBMC) (17), inhibited neither IFN-γ-induced epithelial STAT1 DNA-binding activity nor decreases in TER. More perplexing was the inability of a Jak2 inhibitor to block IFN-γ-induced increases in T84 monolayer permeability. Moreover, experiments with ATA, previously shown to block STAT1 activation (9), revealed that this agent was as effective as EGCG in preventing IFN-γ-induced STAT1 tyrosine phosphorylation but did not ameliorate the epithelial barrier defect. Recent findings documenting a requirement for STAT1 arginine methylation for nuclear translocation (29) led us to the methylase inhibitor MTA. The upregulation of IRF-1 expression by IFN-γ was reduced significantly by MTA, whereas the barrier dysfunction remained unaffected. Thus, whereas EGCG would certainly exert an anti-inflammatory effect by reducing IFN-γ-driven STAT1-mediated events in enterocytes, the summation of the present data indicate that IFN-γ-induced epithelial barrier disruption is a STAT1-independent event. The data support earlier studies in which partial inhibition of STAT1 nuclear localization did not diminish the drop in TER caused by exposure to conditioned medium from hPBMC activated by bacterial superantigens (25).
The question of how EGCG prevents the IFN-γ-induced epithelial barrier defect remains. Exposure to IL-4 increases epithelial permeability via phosphatidylinositol 3-kinase (PI3-kinase) (8), and IFN-γ may activate PI3-kinase in immune cells (31). Because EGCG did not affect IL-4-evoked T84 hyperpermeability, it seems unlikely that EGCG interference with PI3-kinase is the means by which the alleviation of IFN-γ-induced epithelial permeability defect occurred. IFN-γ treatment of nonepithelial cell lines can elicit ERK and p38 MAPK activation (18, 22), ERK has been implicated in epithelial TJ function/formation (21), and EGCG has been shown to interfere with MAPK signaling (10). Employing pharmacological inhibitors of MEK/ERK and p38, at maximal doses (1, 12, 15), we found that neither kinase was required for IFN-γ-induced barrier dysfunction. Finally, exposure to EPEC reduces TER via activation of myosin light-chain kinase and contraction of the TJ-associated actinomyosin cytoskeleton (33, 43). EGCG did not affect EPEC-induced barrier dysfunction, further highlighting its action as a putatively specific IFN-γ signaling antagonist, at least with respect to the regulation of enteric epithelial barrier function.
In summary, three key findings are presented. First, EGCG prevents increased paracellular (and transcellular) enteric epithelial permeability induced by IFN-γ but not that evoked by IL-4 treatment or EPEC infection. Second, IFN-γ-induced STAT1, but not IL4-induced STAT6, activation and transcriptional activity in epithelial cells and monocytes is inhibited by EGCG. Third, physiological, molecular, and pharmacological analyses indicate that the EGCG preservation of epithelial barrier integrity in the face of IFN-γ challenge is independent of its ability to prevent STAT1 activation and that the IFN-γ effect is not via ERK or p38 MAPKs. Thus, as an anti-inflammatory agent, EGCG blocks IFN-γ-STAT1-driven events in immune and nonimmune cells and specifically antagonizes IFN-γ-induced increases in epithelial permeability, an event that is not directly related to inhibition of STAT1 activation.
Financial support was provided by Canadian Institutes for Health Research (CIHR) Operating Grant MT-13421 (to D. M. McKay). J. L. Watson and H. Cameron are recipients of Canadian Digestive Disease Foundation (CDDF)/CIHR doctoral student scholarships.
Technical support from Jun Lu is gratefully acknowledged.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 the American Physiological Society