Compartmentalized Response of IL-6/STAT3 Signaling in the Colonic Mucosa Mediates Colitis Development

Diarrhea, rectal bleeding, abdominal pain, and body weight loss are hallmarks of inflammatory bowel diseases (IBD) (1). IBD encompasses two main pathologies: Crohn disease and ulcerative colitis. The pathogenesis of both disorders is not well understood, but increasing evidence suggests that locally produced cytokines (2, 3) and chemokines (4, 5) play an important role not only in the establishment, but also in the progression of the pathologies.

STAT3 is a transcription factor that regulates genes involved in programed cell death, proliferation, differentiation, migration, and survival downstream of a variety of cytokines and growth factors (6–8). In IBD, STAT3 is constitutively active in intestinal epithelial cells (IECs) and in immune cells located at the intestinal mucosa (9–11). Furthermore, spontaneous colitis has been observed in mice lacking STAT3 in macrophages (12, 13), but T cell–specific deletion of STAT3 has a protective effect in autoimmune disease models (14, 15). In IECs at the colonic mucosa, STAT3 displays a proliferative and prosurvival function important for tumor initiation in colitis-associated cancer, a major complication associated with IBD (10). It is clear, therefore, that tissue-specific functions of STAT3 are necessary for maintaining intestinal homeostasis and understanding the mechanisms controlling the activation of STAT3 in the colonic mucosa during the course of inflammation is highly relevant.

In this report, by using two models of colonic inflammation, we describe that IEC polarization plays an important role in the regulation of IL-6/STAT3 signaling in the intestinal epithelium. We observed that several cell types secrete IL-6 in the colonic mucosa, including macrophages and IECs. However, in the context in which inflammation is triggered and an intact epithelium remains, only IECs are able to release IL-6 intraluminally to induce the activation of STAT3 in neighbored IECs. Activation of STAT3 in such conditions occurs mainly in the epithelial cells lining the apical surface of the crypts and happens in response to high levels of basolateral IFN-γ. We also observed that basal levels of IFN-γ produced by CD4⁺ cells play an important role in the apical polarization of the IL-6Rα. In the context of epithelial damage, such as that observed in ulcerated areas during colitis, STAT3 activation occurs in epithelial cells along the whole crypt axis, maybe because of the disruption in the compartmentalization of the IL-6 signal established by the epithelial barrier. Thus, epithelial barrier loss during colitis is an important factor that directly influences the activation of STAT3 via IL-6 in epithelial cells. In conclusion, polarization of the intestinal epithelium by IFN-γ could be essential for priming/conditioning the inflammatory response mediated by STAT3 in mucosal tissue.

All procedures were reviewed and approved by the Cinvestav Institutional Committee for Care and Use of Laboratory Animals. Six- to eight-week-old C57BL/6J and CRTAM-deficient (kindly donated by Genentech, San Francisco, CA) mice were bred at the Center for Research and Advanced Studies of Cinvestav. Animals were housed in a standard day and night cycle with free access to food and water.

Recombinant human and murine IFN-γ and IL-6 were obtained from PeproTech (Rocky Hill, NJ) and used at 100 U/ml and 20 ng/ml for in vitro treatments and 2 and 10 μg/kg for in vivo treatments. Abs were used as manufacturer suggested. p-Hist3 (sc-56745), STAT1 (sc-346), STAT3 (sc-482), GAPDH (sc-32233), Muc-2 (sc-15334), E-cadh (sc-59778), and IL-6R (BE0047) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA) and Bio X Cell. Claudin-1 (ab15098) and F4/80 (ab6640) were obtained from Abcam. p-STAT1 (9167S), p-STAT3 (9134P), AKT1 (2938), p-AKT473 (4060), PARP (9541), PCNA (13110), Ki67 (9129), and IL-6 (CST 12912) were purchased from Cell Signaling and Claudin-2 (516100) from Invitrogen. HRP-conjugated secondary Abs were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA).

Colonic epithelial cell lines SW480 and Caco-2 were obtained from American Type Culture Collection and were grown in Transwells with DMEM and high glucose medium with 10% FCS and antibiotics. Cells were maintained in a humidified incubator with 5% of CO₂.

Proximal colon was identified, and a 1–2-cm segment was removed. The viscus was opened and then rinsed in cold PBS 1×. The colon was incubated in a Ca²⁺ chelation solution for 30 min at room temperature (RT) (966 mM NaCl, 1.5 mM KCl, 10 mM HEPES, 10 mM Tris, 27 mM Na EDTA or Na EGTA, 45 mM sorbitol, 28 mM sucrose, and 0.1% BSA). After chelation, the tissue was manually shaken to liberate the crypts. The tissue was removed, and the solution was centrifuged at 200 rpm for 2 min. The supernatant was removed, and the pellet was lysed in radioimmunoprecipitation assay buffer for Western blot or F12 medium for cytokine treatment (16, 17).

Colonic epithelial cell lines SW480 and Caco-2 were stimulated with recombinant human IFN-γ (no. 300-02) or IL-6 (no. 200-06). Control cells were treated with BSA. For colon explants and colonic isolated crypt treatment, a colon segment or crypts were incubated at RT for 1 h with recombinant murine cytokines (IFN-γ no. 315-05 or IL-6 no. 216-16) dissolved in F12 medium. The tissue or crypts were collected for analysis.

For cytokine administration, mice were randomly divided, and recombinant murine IFN-γ (no. 315-05) or IL-6 (no. 216-16) was administered i.p. Intraluminal administration of recombinant murine IFN-γ and IL-6 dissolved in 100 μl of sterile PBS was carried out using a 20GX1.5′′ probe (Cadence Science, Ref. 9920). Thirty minutes after intrarectal administration, the animals were euthanized, and the colon was collected and processed for analysis. Control mice were administrated with mouse seric albumin (MSA).

Cytokine inhibition in SW480 and Caco-2 cells was performed with human anti–IL-6R Ab (AF-227-NA) or human anti–IL-6 Ab (ab6672). IECs were treated with IFN-γ or IL-6 in presence of neutralizing Abs or isotype controls. The Abs were used according to the manufacturer’s instructions.

Mice received 2.5% (w/v) of dextran sulphate sodium (DSS) (molecular mass 40 kDa, Carbosynth, CA) dissolved in tap water. Mice were sacrificed, and colons were assessed for weight and length and processed for histology. Colon was processed as previously reported by us for immunofluorescence and Western blot analysis (18).

Cytokines were quantified from supernatants of culture media of stimulated cells by using the OptEiA Human IL-6 ELISA Set (RUO-555220; BD Biosciences), according to the manufacturer’s instructions. Briefly, well plates were coated overnight with the capture Ab for cytokine using carbonate buffer (pH 9.5) at 4°C, blocked with 1× PBS supplemented with 10% FBS (1 h). Supernatant dilutions were made using blocking buffer, added to the wells of the plate, and incubated for 2 h at RT. Detection Abs and streptavidin/HRP were added together and incubated for 1 h at RT. The o-phenylenediamine substrate solution was added, and after 15-min incubation at 37°C, reaction was halted with 2 N of H₂SO₄. Absorbance at 450 nm was read using an absorbance microplate reader (Tecan Sunrise, Salzburg, Austria).

Colon samples in radioimmunoprecipitation assay buffer (150 mM NaCl, 1% NP-40, 0.5% deoxycholic acid, 0.1% SDS, and 50 mM Tris; pH 8) were homogenized using Tissue Master using Omni International (Kennesaw, GA) and sonicated for 10-s maximal output. Protein concentration was determined using the BCA Protein Assay (Pierce, Thermo Fisher Scientific) and 20–25 μg of protein were loaded in denaturing SDS-PAGE. GAPDH and actin were used as loading controls.

Colon specimens were cryopreserved in O.C.T. (Sakura Finetek, Torrance, CA) and sectioned in 20 μm using a cryostat (Leica, Nussloch, Germany). Then, they were fixed with paraformaldehyde (20 min; RT) and permeabilized with 100% methanol (20 min; −20°C). Nonspecific binding interactions were prevented by blocking the tissue with 2% BSA (BSA, Sigma-Aldrich, St. Louis, MO) for 1 h at RT, and primary Abs were incubated at 4°C overnight. Alexa Fluor–labeled secondary Abs were incubated then for 1 h, and nuclei were counterstained with DAPI. Representative images were taken on an LSM 510 Confocal Microscope (Zeiss, Jena, Germany) with 20×, 40×, and 60×.

Statistical significance was assumed at p < 0.05, p < 0.01, or p < 0.001. All results are displayed as mean ± SEM.

STAT3 regulates several biological processes in the gut, including colorectal cancer development, IBD establishment, and epithelial regeneration (6, 7, 19). We therefore investigated the presence of active STAT3 in the colonic mucosa of wild-type (WT) mice treated with DSS from 1–6 d to mimic different degrees of injury in the colonic mucosa. As shown in Fig. 1A, increased levels of STAT3 phosphorylated at Y705, from now on referred to as p-STAT3, were observed starting 4 d after DSS treatment. In colitic animals, p-STAT3 was enriched at nucleus of colonic IECs located along the whole crypt axis (Fig. 1B, 1C, arrows). No nuclear p-STAT3 staining was detected in IECs of healthy control animals. However, p-STAT3 was detected at nuclei of nonepithelial cells that were bordering the epithelial crypts in both control and colitic animals (Fig. 1C, asterisk). Activation of STAT3 in SW480 cells and in isolated crypts exposed to IL-6 strongly suggested a direct implication of the cytokine in the generation of p-STAT3 in IECs during colitis (Fig. 1D, 1E). Therefore, we analyzed IL-6 secretion in the mucosa of colitic mice exposed to DSS for 1–4 d. As shown in Fig. 1F, IL-6 secretion augments in mucosal explants of colitic mice starting at day 3 of treatment, a process that strongly increased on day 4, when STAT3 activation was clearly noticed (Fig. 1A). Thus, our findings suggest that IL-6 could be inducing activation of STAT3 in IECs of colitic mice.

IL-6 is produced mainly by immune cells underneath the intestinal epithelium, such cells face the basolateral side of epithelial cells [Fig. 2A, arrow and (7, 20)]. We therefore investigated whether IL-6 administered basolaterally was able to activate STAT3 in IECs in vivo. To this end, C57BL/6J mice were injected i.p. with IL-6, and STAT3 activation was analyzed in the colon 2 h postinjection. As shown in Fig. 2B, increased levels of p-STAT3 were detected in the colonic mucosa of IL-6–treated animals. No changes in p-STAT1, total STAT1, or total STAT3 were observed. However, p-STAT3 analysis by Western blotting isolated colonic crypts or by performing immunofluorescence staining revealed that basolateral administration of IL-6 induces nuclear accumulation of STAT3 in nonepithelial cells. Those cells were always located in close proximity with the IECs, forming the crypts (Fig. 2B, 2C, asterisk). Nonepithelial cells displaying nuclear p-STAT3 staining were F4/80 positive, suggesting that basolateral administration of IL-6 activates STAT3 in macrophages (Fig. 2D, arrowhead; Supplemental Fig. 1A, asterisk). Interestingly, i.p. administration of IL-6 induces sporadic accumulation of p-STAT1 at nuclei of IECs located at the crypt surface (Supplemental Fig. 1B, arrow).

Next, we sought to identify the alternative mechanism involved in the activation of STAT3 in IECs. In fibroblasts, STAT3 activation is mediated by IFN-γ (21), a proinflammatory cytokine increased at the basolateral side of IECs during colitis (4, 22, 23). Thus, we analyzed p-STAT3 levels in colonic cell lysates and colonic isolated crypts of mice administered i.p. with IFN-γ for 2 h. Similar to the findings with IL-6, basolateral administration of IFN-γ strongly increased p-STAT3 in the colonic mucosa of C57BL/6J animals, and, as expected, IFN-γ also induced STAT1 activation. No changes in total STAT1 or total STAT3 were perceived (Fig. 2B). Interestingly, we observed STAT3 activation after basolateral stimulation with IFN-γ and nuclear accumulation of p-STAT3 in both epithelial (Fig. 2B, 2C, arrow) and nonepithelial cells (Fig. 2D, Supplemental Fig. 1A, asterisk). p-STAT3–positive IECs were scattered along the whole crypt axis; however, they were clearly enriched at the apical surface of the crypts (Fig. 2C, 2D, white arrows). Furthermore, p-STAT3–positive nonepithelial cells that were induced after basolateral administration of IFN-γ were positive for F4/80 and were observed in close apposition to the IECs, as shown in Fig. 2D and Supplemental Fig. 1A (white asterisk). Contrary to p-STAT3, a more uniform distribution of p-STAT1 in IECs along the crypt axis was induced by IFN-γ (Supplemental Fig. 1B). Furthermore, a more detailed analysis for STAT1 and STAT3 activation revealed that IFN-γ induced a fast activation of STAT1 in IECs, such as the process started at the crypt surface. However, over time, nuclear staining of p-STAT3 was detected in the same cells following a similar pattern of activation as the one displayed by p-STAT1 (Supplemental Fig. 1C). Thus, taken together, these results strongly suggest that IFN-γ actively contributes to the activation of STAT3 in differentiated IECs.

To understand the mechanism by which IFN-γ activates STAT3 in IECs, we stimulated SW480 with IFN-γ for 0.5–6 h. As shown in Fig. 3A, STAT3 phosphorylation was strongly induced starting 30 min after cytokine treatment, but its presence declined rapidly. Furthermore, p-STAT3 was not detected in cell lysates of SW480 cells treated with IFN-γ for 24 h, but it was present in cells exposed to IL-6 or the mixture IL-6/IFN-γ for the same period of time (Supplemental Fig. 2A). Conversely, IFN-γ presence strongly induced phosphorylation of STAT1 for up to 24 h even in the presence of IL-6 (Fig. 3A, Supplemental Fig. 2A). No changes in STAT1 or STAT3 were observed during the course of the experiment. Thus, these results suggest that IFN-γ transiently activates STAT3 in IECs, but in contrast, STAT3 activation by IL-6 or IL-6/IFN-γ results in a long-term effect.

Indirect activation of STAT3 signaling by secreted factors could be triggered in an autocrine or paracrine way (24). Thus, we investigated whether STAT3 activation downstream of IFN-γR was mediated by a secreted factor. We therefore analyzed the phosphorylation of STAT3 as a function of time in SW480 cells exposed to IFN-γ for 30 min, followed by washout (Fig. 3B, Supplemental Fig. 2B). Phosphorylation of STAT3 after IFN-γ removal lasted for several hours (Fig. 3B), and the conditioned medium obtained from IFN-γ–treated cells induced STAT3 phosphorylation in freshly plated SW480 cells (Fig. 3C). No changes for STAT3 were detected during the course of the experiment (Fig. 3B, 3C). Following a similar protocol, no changes in p-STAT3, p-STAT1, STAT1, or STAT3 were detected in SW480 stimulated with BSA (carrier) alone (data not shown). Next, we analyzed STAT3 activation in IEC pretreated with brefeldin A (BfA), a potent transport inhibitor used to prevent cytokine release (25). As shown in Fig. 3D, the activation of STAT3 induced by IFN-γ was strongly reduced in the presence of BfA, but no changes in p-STAT1 were noticed. In agreement with previous published results, STAT3 activation induced by IL-6 was also suppressed by BfA. These results suggest that in IECs, long-term activation of STAT3 by proinflammatory cytokines requires the activation of the secretory pathway.

Because of our results in Fig. 1, we next investigated whether IFN-γ induces the secretion of IL-6 in IECs. As shown in Fig. 3E, IL-6 was enriched in the supernatant of nonpolarized SW480 cells exposed to IFN-γ. Furthermore, as reported previously, IL-6 treatment induced a strong release of IL-6 by IECs. No synergistic effect was noticed when SW480 cells were treated with both cytokine mixtures, IL-6/IFN-γ (data not show). Furthermore, treating SW480 cells with BfA before IFN-γ or IL-6 stimulation prevented the release of IL-6 into the media (Fig. 3E). Also, IFN-γ and IL-6 treatment induced IL-6 secretion in freshly isolated colonic crypts obtained from C57BL/6J mice (Fig. 3F). Interestingly, IFN-γ induced IL-6 release in several cell lines, including IECs, fibroblasts, and macrophages (Fig. 3E, 3F, Supplemental Fig. 2C, 2D). However, the amount of IL-6 secreted by those cells after IFN-γ stimulation was considerably lower than the one induced by IL-6 (fibroblasts) or LPS (macrophages). In fibroblasts, IL-6 release induced by TGF-β treatment was observed at similar extent as the one induced by IFN-γ, corroborating the viability of our assay (26). Interestingly, IFN-γ treatment induced only a transient secretion of IL-6 in IECs, suggesting that it does not affect IL-6 expression (Supplemental Fig. 2E). To analyze if IFN-γ was inducing STAT3 activation via autocrine stimulation of IL-6/STAT3 signaling, we evaluated p-STAT3 in SW480 and Caco-2 cells treated with IFN-γ in the presence of neutralizing Abs against IL-6 signaling. As expected, anti–IL-6R or anti–IL-6 Ab strongly reduced the activation of STAT3 IFN-γ−mediated in comparison with cells treated with the cytokine plus an isotype control (Fig. 3G, Supplemental Fig. 2F). Thus, taken together, these results strongly suggest that in IECs, IFN-γ induces STAT3 activation by stimulating IL-6 secretion.

Next, we evaluated the secretion of IL-6 by the IECs lining the colonic mucosa of colitic mice. Thus, we investigated the presence of IL-6 in the mucosa of C57BL/6J animals in basal conditions. As shown by immunofluorescence studies, IL-6 staining was detected in epithelial cells and nonepithelial cells at the colonic mucosa (Fig. 2A). IL-6–positive IECs were observed along the crypt axis, including the surface (arrows). Nonepithelial, IL-6–positive cells expressed the macrophage marker F4/80 and were detected surrounding the crypt base (Fig. 4A, white asterisk; Fig. 4B). However, increased staining for IL-6 was clearly noticed in epithelial cells present at the mucosa of colitic mice when compared with their nontreated counterparts (Fig. 4A, arrows). Furthermore, the number of double-positive cells expressing F4/80 and IL-6 was also augmented (Fig. 4B). It is noteworthy to mention that patchy areas enriched with double-positive cells (F4/80/IL-6) were frequently observed in the inflamed mucosa of colitic mice. In agreement with our findings, the induction in IL-6 secretion was enhanced in both mucosal explants and isolated crypts obtained from colitic mice (Fig. 4C). However, as expected, IL-6 secretion was more pronounced in the colonic explants when compared with isolated crypts. Thus, all these results strongly suggest that IEC present in the colon constantly produce and secrete IL-6, and this process is enhanced during colitis.

IECs elicit a complex response to cytokines depending on several factors such as cytokine concentration, receptor availability, and stage of differentiation (27–30). Therefore, we examined STAT3 activation and IL-6 secretion in fully differentiated Caco-2 cells that were exposed in a polarized way to proinflammatory stimuli. To this end, Caco-2 cells were plated in Transwell filters, and the stimulus was added at the apical or basolateral side. As shown in Fig. 5A, no activation of STAT3 was observed in Caco-2 cells stimulated apically with IL-6 or IFN-γ. However, as expected, STAT3 activation was induced once IL-6 or IFN-γ were administered basolaterally. No synergistic effect in the activation of STAT3 was obtained in monolayers stimulated with IL-6 and IFN-γ at the basolateral side. Importantly, TGF-β and LPS did not affect STAT3 phosphorylation status when administered apically or basolaterally. No significant changes in STAT3 were noticed with any of the treatments. Surprisingly, IL-6 was detected at the apical compartment of Caco-2 monolayers that were treated with IFN-γ or IL-6 basolaterally, and the administration of the mixture IL-6/IFN-γ potentiated such effect (Fig. 5B), but this process was not observed in Caco-2 cells stimulated with IFN-γ at the apical side. Also, we detected a strong release of IL-6 at the basolateral after stimulating Caco-2 cells with IFN-γ at the basolateral side. The basolateral administration of TGF-β or LPS has no influence in the presence of IL-6 either at the apical or basolateral side. However, the apical stimulation of Caco-2 cells with IL-6 triggered the basolateral release of IL-6, an effect that was not observed in monolayers stimulated with IFN-γ at the apical side. Thus, these results strongly suggest that activation of STAT3 in IECs stimulated with IFN-γ is mediated by the endogenous IL-6 that is secreted at the apical side.

IL-6R and IFN-γR are located at the basolateral side of epithelial cells (31, 32); however, STAT3 activation in Caco-2 cells stimulated with cytokines was only observed in monolayers secreting IL-6 at the apical side, suggesting that other components were involved in the process. Thus, we speculated that IL-6R expression and location in IECs could also be affected by IFN-γ, similar to the observed with IL-10R (33). Therefore, we next evaluated the presence of IL-6Rα and the accessory coreceptor, the gp130, in IECs exposed to IFN-γ. Interestingly, IFN-γ treatment rapidly increased IL-6Rα and gp130 protein levels in Caco-2 cells (Fig. 5C). Furthermore, immunofluorescence staining demonstrated that IL-6Rα was also relocated to the apical surface in Caco-2 cells treated with the cytokine (Fig. 5D). Next, we evaluated the distribution of IL-6Rα in the colonic mucosa of C57BL/6J mice. IL-6Rα was highly enriched in nonepithelial cells surrounding the colonic crypt in both control and DSS-treated animals (Supplemental Fig. 2G). Also, as reported previously, positive staining for IL-6Rα at the basolateral side of IECs located at the crypt base was observed [Supplemental Fig. 2H and (31)]. Those cells were in close contact with nonepithelial, IL-6Rα–positive cells (white asterisk). However, and unexpectedly, we noticed the presence of IL-6Rα at the apical cell surface of IECs located in the bottom of the crypt (Supplemental Fig. 2H, arrowheads). The apical staining for IL-6Rα was more noticeable in IECs present at the surface of the crypt, and the presence of the receptor in those cells was strongly enhanced after colitis induction (Fig. 5E). To investigate the functionality of the IL-6Rα detected at the apical membrane in IECs of the colonic mucosa, we next administered C57BL/6J mice intraluminally with carrier alone (MSA), IL-6, or IFN-γ for 30 min, and STAT3 and STAT1 activation was evaluated by Western blot (Fig. 5F). As expected, activation of STAT3 was only observed in the animals that received IL-6, but no changes were detected after IFN-γ or MSA administration. Immunofluorescence staining demonstrated activation of STAT3 in epithelial cells located at the apical surface of the colonic crypt (Fig. 5G). Thus, we conclude that during inflammation, IFN-γ regulates the activation of STAT3 in IECs by polarizing both the receptor and the secretion of the cytokine.

Apical administration of IL-6 failed to induce STAT3 activation in Caco-2 cells but evoked a strong phosphorylation of the molecule in colonocytes in vivo (Fig. 5A, 5F). Thus, the discrepancy between our in vitro and in vivo results prompted us to speculate that basal release of IFN-γ induced by colonic microbiota (34) was essential for IL-6Rα polarization. Therefore, IL-6/STAT3 signaling in the mucosa of Crtam-deficient mice was analyzed. As shown in Supplemental Fig. 3A, in the absence of Crtam, the production of IFN-γ induced by the resident microbiota is greatly reduced because of functional impairment of CD4⁺ T cells (35). However, no changes in IL-23, an unrelated cytokine, were noticed. Low levels of IFN-γ were likely a consequence in the reduction of colonic CD4⁺ T cells because no differences in the number of IFN-γ–producing macrophages were detected between Crtam knockout (KO) and WT animals (Supplemental Fig. 3B). Similar to that observed in WT animals, i.p. administration of IFN-γ induced activation of STAT1 and STAT3 in the colonic mucosa of Crtam KO mice without affecting STAT1 or STAT3 levels (Fig. 6A). However, STAT3 activation was greatly reduced in Crtam-deficient mice when compared with WT animals (Supplemental Fig. 3C). Activation of STAT1 arose in IECs along the whole-crypt axis (Fig. 6B, white arrows) and in nonepithelial cells present in the mucosa (Fig. 6B, white asterisk). However, p-STAT3 staining was noticed preferentially in nonepithelial cells (Fig. 6C, white asterisk) and was almost absent in epithelial cells. Nonetheless, some epithelial cells at the bottom of the crypt displayed nuclear staining for p-STAT3 (white arrows). Furthermore, the presence of IL-6Rα at the luminal side of the enterocytes was greatly reduced in the animals lacking Crtam (Fig. 6D). Interestingly, despite in the differential response observed in STAT3 activation, no morphological changes were noticed in the colonic mucosa of Crtam-deficient mice when compared with WT animals (Supplemental Fig. 3D). However, the amount of Claudin-1, an epithelial protein induced by STAT3 (36) to strength barrier function in the gut, was strongly decreased in the colon in the absence of Crtam (Fig. 6E). Furthermore, Claudin-1 and -2 were present in the cellular junctions in IECs. However, the distribution of both molecules along the crypt axis displayed a slightly different pattern in the mucosa of Crtam-deficient mice when compared with WT animals (Fig. 6F, Supplemental Fig. 3E). In addition, mucin-2, another molecule that is regulated by STAT3 in epithelial cells, was strongly reduced in the absence of Crtam (Fig. 6F), and the protein was mostly accumulated in the cytosol of goblet cells (Supplemental Fig. 3F). The i.p. administration of Crtam KO mice with IFN-γ for 72 h partially rescued those phenotypes (data not shown).

Given our results, we then analyzed the role of STAT3 activation in the mucosa of Crtam-deficient mice that were induced to colitis. Disease Activity Index analysis showed a similar evolution in the development of the disease in Crtam KO and WT mice (Fig. 7A). Also, analysis of weight loss and colon length in colitic WT and Crtam-deficient mice reported similar changes in both strains (Supplemental Fig. 4A, 4B). Furthermore, we also noticed that in Crtam-deficient mice, the activation and response of STAT3 in the colonic mucosa of colitic mice followed a similar pattern to the one observed in WT animals (Fig. 7B, Supplemental Fig. 4C). However, the secretion of IL-6 was greatly reduced in the mucosa of Crtam-deficient mice (Fig. 7C), and p-STAT3 staining was observed mainly at the nuclei of nonepithelial cells (Fig. 7D, asterisk). Furthermore, analysis of colonic isolated crypts showed a decrease in p-STAT3 in Crtam-deficient mice when compared with WT animals (Fig. 7E). Microscopic inspection of the colonic mucosa of colitic mice demonstrated less infiltration in the colon of Crtam KO. However, it was noticeable that in the absence of Crtam, a more widespread damage occurs in the epithelium (Fig. 7F, arrow). Consequently, histopathological evaluation of colonic samples according to the parameters established by Rayudu et al., demonstrated no differences between both strains (Supplemental Fig. 4D). Thus, these results suggest that IL-6/STAT3 signaling is altered in the mucosa of Crtam-deficient mice.

IBD are chronic pathologies of unknown etiology (37). Our work in the last years has unveiled an essential role of the intestinal epithelial barrier in the development and establishment of IBD. However, the apparatus by which IECs contribute to IBD progress is not fully understood. In this study, we analyzed the mechanism triggering STAT3 signaling in the colonic mucosa during inflammation, and we demonstrated the importance of IECs polarization and barrier function in this process. Our findings uncovered an unknown role for IFN-γ in the activation of STAT3 in differentiated IECs located at the apical region of the crypts during colitis. Also, these results highlight an active role of the epithelial barrier in the compartmentalization of IL-6/STAT3 signaling. Finally, we demonstrated that impaired production of IFN-γ by CD4 T cells largely attenuated IL-6/STAT3 signaling in IECs, but not in immune cells.

Understanding the pathophysiology of IBD has been a challenge that limited the development of more efficient therapies. The use of IBD-like models has been broadly implemented as an approach to investigate the role of factors known to directly influence the outcome of the disease. However, contradictory results have been obtained using similar models (e.g., DSS) and specific factors (e.g., IL-6), mainly because we have obviated the complexity of the tissue (several cell types with specific intrinsic characteristics are affected) and the broad variety of mediators that are present in the inflamed mucosa (cytokines, growth factors, chemokines, etc). One essential component in the maintenance of the colonic mucosa that has not been fully considered in this process is the polarity of the epithelium. The epithelial barrier in the colonic mucosa is formed by a single layer of fully polarized epithelial cells with two well-defined domains, apical and basolateral. Functionally and structurally, both domains are different, and the role of such polarization in the development of IBD is not totally understood. By using a model of colitis based in epithelial injury (DSS-mediated colitis), we now demonstrate that in addition to the epithelial barrier function, the polarization of IEC is important for the development of IBD-like symptoms. Our results highlight the role of the exclusion of the cytokine receptors at apical and basolateral domains in the compartmentalization of the signaling pathways that activated in the colonic mucosa during inflammation. Such process could play an important role for the maintenance of the intestinal epithelial barrier and for the restraining of the inflammatory responses such as those described for TLR (38). Therefore, these findings could be important to explain why the same signaling (STAT3) displays protective or detrimental effects in the colonic mucosa during colitis (31, 39). In fact, according to our results, the mechanisms underlying the activation of STAT3 in the colonic mucosa imply the establishment of feedback loops between apical and basolateral receptors as the observed in the tolerizing response mediated by TLR (38).

The biological relevance of the polarized response triggered by IFN-γ in IECs could be envisioned as part of two different scenarios (Supplemental Fig. 4E). First, in the context of an intact epithelial barrier, basolateral secretion of IFN-γ at low levels will drive the apical enrichment of IL-6Rα in IECs. If a challenge induces the release of high levels of IFN-γ, mimicked in this study by the i.p. injection of the cytokine, apical secretion of IL-6 will stimulate the activation of STAT3 in crypt surface epithelial cells to strength epithelial barrier functions (36, 40). Lack of activation of STAT3 in IECs after i.p. administration of IL-6 strongly supports this theory. The context in which high amounts of IFN-γ are released in a mucosa with an intact epithelial barrier needs to be investigated but could be ranging from helminth to bacterial or even viral infections (41). Thus, activation of STAT3 in such a context could be an important component of the immune response orchestrated by the organism to fight or clear infections (40). The second scenario involves a situation in which the epithelial barrier is compromised, such as that observed during DSS-mediated colitis. In those circumstances, IL-6 produced by immune cells (e.g., macrophages) will diffuse from the interstitial compartment to the luminal region to interact with the apical receptor in differentiated epithelial cells. Activation of STAT3 in IECs in the latter scenario could be also mediated by IL-10 given that IFN-γ triggers the expression of the cytokine receptor in the epithelium (33). However, it is noteworthy to mention that activation of STAT3 in colonic mucosa of colitic mice has been linked to the presence of several other cytokines, including IL-22 and EGF (24, 42). Therefore, the feedback loops established between all those cytokine receptors could dictate the final outcome. However, giving that polarization and expression of cytokine receptors in the intestinal epithelium is regulated by IFN-γ, the compartmentalization of several signaling pathways in the gut could be subject to the amount of IFN-γ delivered by immune cells. Thus, the mechanisms and machinery responsible for controlling the polarization of the receptors needs to be investigated because such process could be indispensable for the proper development of the innate and adaptive immune responses of the colonic mucosa.

The biological function for the polarization of IL-6/STAT3 in IECs present at the crypt surface needs to be investigated; nevertheless, our results provide some clues. For example, IFN-γ induces a sequential activation of STAT1 and STAT3 in the same epithelial cell, something similar to the observed during priming of immune cells by IFN-γ (43, 44). In macrophages, for example, transient accumulation of STAT1 in the nucleus has been described during priming and commitment into specific cell lineages (45). Therefore, we can speculate that a similar function could be performed by STAT proteins in epithelial cells. Furthermore, because p-STAT3, but not p-STAT1, was detected at the nuclei of IECs of colitic mice, we believe that STAT3 plays a more prominent role as a transcription factor, and mucosal p-STAT1 could be rather implicated in the inhibition of cell death (46). Activation of STAT3 transcriptional activity in IECs after cytokine stimulation could be important for the maintenance of epithelial barrier function in the intestinal epithelium by modulating tight junction proteins or reducing cell damage (10, 40, 47). However, an immunomodulatory role for such signaling cannot be ruled out based in our findings as proposed in other contexts (48).

The mechanism by which mucosal CD4⁺ T cells maintain basal levels of IFN-γ necessary for inducing polarization of IL-6Rα in the IECs remains elusive. However, the tolerogenic CD4⁺ T cell response triggered by bacterial components could play an active role in such process. In fact, it has been shown that gut microbiota stimulates not only cell maturation, but also commitment of the immune system cells toward specific lineages or functions (49, 50). The impaired secretion of IFN-γ in the colonic mucosa of Crtam-deficient mice strongly supports this theory. Basal levels of IFN-γ in mucosal tissue during such tolerogenic responses could be essential for inducing polarization of cytokine receptors by enhancing IEC differentiation (30). Thus, a mechanism of STAT3 activation like the one described in this review could therefore be influenced/regulated by the microbiota and might be an important immunomodulatory mechanism in the tolerogenic or proinflammatory immune response.

IFN-γ is a pleiotropic cytokine that regulates polarization, commitment, and differentiation of several cell types (27, 30, 51, 52). The mechanisms behind these processes remain highly unknown. In this study, we observed that IFN-γ induced polarized secretion of IL-6 in IECs in a similar fashion to the observed in endometrial epithelial cells exposed to bacteria, LPS, or IL-1α (53). Polarized secretion of cytokines has been poorly investigated; therefore, the relevance of such process in the orchestration of the immune response can only be hypothesized. However, given that our results strongly suggest that IL-6 production by IECs is low in comparison with immune cells (e.g., macrophages), we speculate that its release could be aimed to have an autocrine, rather than a paracrine, effect with specific roles in the local innate immune response. Furthermore, because of the variety of cytokine receptors expressed in the different IECs along the intestinal crypt, STAT3 activation by cytokines could be aimed to target cells with different degrees of differentiation, and therefore, different outcomes should be expected. Investigating the mechanism of STAT3 activation in IECs along the crypt axis could, therefore, be important to understand the real function of STAT3 signaling in the development of IBD and colorectal cancer (11, 47).

In conclusion, we demonstrated a new role for IEC polarization in the compartmentalization of the IL-6/STAT3 signaling in the colonic mucosa during inflammation. This new mechanism could be important for regulating the immune response in the colonic mucosa during colitis development.

We thank Norma Trejo and Octavio López-Méndez for excellent technical assistance. The authors thank Dr. Angeles Hernandez-Cueto for histopathological evaluation of colonic samples.

The authors have no financial conflicts of interest.