Gp130 Signaling Promotes Development of Acute Experimental Colitis by Facilitating Early Neutrophil/Macrophage Recruitment and Activation¹

The pathogenesis of inflammatory bowel disease (IBD)³ like Crohn’s disease and ulcerative colitis, which are characterized by chronic recurrent inflammation of the gastrointestinal tract, is complex and the detailed molecular mechanisms are still only poorly understood. Crohn’s disease is thought to be mediated by an inappropriate response to resident microbes, leading to excessive release of proinflammatory cytokines like IL-1β, TNF-α, IL-23, and IL-6 (1).

IL-6 has been repeatedly implicated in IBD, and both experimental and clinical evidence for its involvement in the immunopathology of the disease have been presented in several reports (2, 3, 4). Enhanced IL-6 production and signaling, namely STAT3 activation, has been found in patients suffering from Crohn’s disease (5, 6). The critical involvement of IL-6 in chronic stages of T cell-dependent colitis models has been well-characterized (4, 5, 7). In contrast, the effects of IL-6 cytokines on acute colitis or initiation of early stages of inflammation are less well understood. Dextran sulfate sodium (DSS)-induced experimental colitis causes strong phosphorylation of STAT3 (P-STAT3) in the colonic mucosa of mice. IL-6-deficient animals show a milder course of DSS colitis and reduced levels of P-STAT3 (8). However, the IL-6 cytokine family comprises nine different cytokines with pleiotropic and overlapping effects due to the high degree of redundancy mediated by their common signal transducer glycoprotein of 130kDa (gp130) (9). Six of the IL-6 cytokines require a specific α-chain in addition to gp130, also called β-chain. gp130 is abundantly expressed in nearly all tissues and its activation by association with an α-chain/ligand complex activates two distinct signaling pathways, the JAK/STAT1/3 or Ras/MAPK cascade, respectively (9, 10). At present, the role of gp130 in the development of colonic inflammation has not been defined in detail. Additionally, the establishment of acute DSS-induced colitis does not require T cells, as it can be established in SCID mice (11), indicating a crucial role of nonlymphoid cells.

Leukocyte infiltration of the intestinal mucosa represents a common characteristic in both human and murine colitis, since macrophages and neutrophils are potent sources of inflammatory mediators that cause mucosal injury (12, 13). Blockage of leukocyte recruitment by eliminating them from the peripheral blood by apharesis reduces mucosal inflammation and injury and has proven an effective strategy for treating severe ulcerative colitis (14). Furthermore, there is clear evidence that recruitment of leukocytes to sites of inflammation is impaired in the absence of IL-6 due to reduced production of chemokines and expression of adhesion molecules (15).

In the light of these data, we thus addressed the effects of gp130-dependent signaling on acute colonic inflammation with a special focus on leukocyte infiltration. Since mice lacking gp130 are not viable, we chose a model of inducible gp130 deletion using the Cre-recombinase/loxP system (MxCre/gp130^−/−) (16). To further dissect the cellular mechanisms of IL-6 family cytokines in intestinal inflammation and their impact on leukocyte functions, we generated neutrophil/macrophage-specific gp130-deleted mice (LysCre/gp130^−/−) and bone marrow chimeric mice. We demonstrate that gp130 signaling was required for the development of acute severe DSS-induced colitis by facilitating leukocyte recruitment and activation. This was dependent on direct signals on effector cells as well as indirect effects mediated by parenchymal cells via adhesion molecule and chemokine expression.

C57BL6/J mice carrying loxP sites flanking exon 16 coding for the gp130 transmembrane domain were crossed with transgenic (tg) mice expressing Cre-recombinase as described previously (16, 17). Cre-recombinase expression was controlled by type I IFN-sensible Mx1 promotor (MxCre) (16). This was activated by i.p. injection of 80 μg of poly(I:C) (Sigma-Aldrich) 15 and 10 days before start of experiment. IFN-induced activation of Mx1 promotor led to expression of Cre and subsequent deletion of gp130. Animals that were negative for MxCre expression but carried loxP sites in both gp130 alleles (gp130^f/f) served as controls and were treated equally.

Macrophage/neutrophil-specific gp130-deficient animals were generated by crossing gp130^f/f animals with lysozyme M-promotor Cre (LysCre) tg mice. LysCre-negative (gp130^f/f) animals served as controls.

Bone marrow chimeric mice were generated by transplanting freshly isolated bone marrow from GFP tg (β-actin/GFP) C57BL6/J donors. Wild-type (wt) (gp130^f/f) or MxCre/gp130^−/− recipients were both injected twice with 80 μg of poly(I:C) (Sigma-Aldrich) on days 15 and 10 before the experiment and received 1 × 10⁶ donor cells after whole body irradiation (2 × 6 Gy). The animals received antibiotic-treated drinking water for 14 days. After 4 wk, the bone marrow was fully engrafted and the animals received drinking water containing 3% DSS for 7 days to induce acute colitis. All experiments were approved by the local Animal Care Committee. All experiments were conducted in agreement with the “Guide for the Care and Use of Laboratory Animals” (National Institutes of Health Publication 86-23, revised 1985).

Experimental colitis was induced by feeding the mice drinking water supplemented with 3% DSS for 7 days, which caused severe erosive colitis as described previously (18). Body weight and disease activity index (DAI) were assessed on a daily basis. DAI was calculated as described earlier (19), combining weight loss, stool consistency, and stool blood content/rectal bleeding. Mice were sacrificed at day 7 and colons were removed for further analysis.

Wild-type, MxCre/gp130, and LysCre/gp130 were injected with 1 ml of 3% thioglycolate broth (Sigma-Aldrich) i.p. or PBS as control. Animals were sacrificed after 18 h and peritoneal cells were collected by peritoneal lavage. Total cell content was counted manually and expressed as cells per cavity. For further phenotypic analysis, cells were stained for CD45, CD11b, and F4/80 and analyzed by flow cytometry.

For histopathological analyses, colons were fixed in 4% paraformaldehyde, embedded in paraffin, cut into 5-μm sections, and subsequently stained with H&E. Histological colitis score was determined as previously described (18). In brief, histological sections were scored as follows; epithelium: Normal morphology (0), loss of goblet cells (1), loss of goblet cells in large areas (2), loss of crypts (3), and loss of crypts in large areas (4); infiltration: no infiltrate (0), infiltrate around crypts (1), infiltrate reaching the lamina muscularis mucosae (2), extensive infiltration reaching the lamina muscularis mucosae and thickening of the mucosa (3), and infiltration of the submucosal layer (4). Total histological score represents the sum of both scores and ranges from 0 to 8. For each sample, 10 fields were randomly selected and mean grade was calculated.

For immunofluorescence analysis, cryosections of 5 μm were stained with rat anti-mouse CD11b mAb (BD Biosciences), rat anti-mouse CD68 mAb (BD Pharmingen), rabbit anti-mouse Ki67 pAb (eBioscience), and rabbit anti-mouse P-STAT3 pAb (Cell Signaling Technology).

Colonic immune cells were isolated by mechanic-enzymatic digestion as described previously (20). In brief, colons were cut longitudinally and washed vigorously in HBSS. Epithelial cells were removed by incubation in HBSS supplemented with EDTA (0.5 mM) for 20 min at 37°C. Remaining tissue was then minced into 1-mm pieces and digested in RPMI containing collagenase D (400 U/L) (Roche) and DNase I (0.01 mg/ml) (Boehringer Mannheim) in a shaking waterbath at 37° for 30 min. The suspension was subsequently filtered through 70-μm nylon filters and washed twice in HBSS. The single-cell suspension was then stained with fluorochrome-conjugated Abs (CD45 and Gr1 both BD Biosciences; F4/80 Serotec; CD11b, CD3, CD4, CD8, CD11c, and Ly6G all eBioscience) and subjected to flow cytometry using a BD Canto II (BD Biosciences). Data were analyzed using FlowJo software (Tree Star). Calculations of colonic cell populations were done as follows: numbers given in dot plots express percentage of CD45⁺ cells. Numbers in graphs give percentage of total cells (percentage of specific population × percentage of CD45⁺).

White blood cells (WBC) were counted automatically. After RBC lysis (PharmLyse; BD Biosciences), cells were stained for CD11b, CD115, Ly6G, CD3, CD4, CD8, CD19 (all eBioscience), and Gr1 (BD Biosciences) and subjected to flow cytometry.

Bone marrow cells were harvested from explanted femurs, filtered, washed, and stained with Abs given above and analyzed by flow cytometry.

Mesenteric lymph nodes were explanted, disrupted mechanically, filtered through cell strainer (70 μm), and analyzed by flow cytometry using Abs given above.

The lower abdomen of anesthetized mice was carefully opened and a flanged tubing was inserted in the proximal colon immediately below the cecum. A second tube was inserted through the rectum and secured by ligature to allow drainage through the rectum. The colon segment was gently flushed and perfused at a rate of 30 ml/h with 150 mmol/L NaCl for 5 min, followed by perfusion with 1% Evans Blue in NaCl for 10 min. Then the lumen was washed with 6 mM acetylcysteine for 5 min followed by NaCl for 10 min. After sacrifice of the animals, colons were removed, rinsed with saline and placed in N,N-dimethyl-formamide overnight for extraction of Evans blue. Dye concentration was measured at 620 nm and expressed as absorbance/cm colon.

Colon samples were collected and snap frozen in liquid nitrogen. RNA was isolated using RNeasy columns (Qiagen) following the manufacturer’s instructions. First-strand synthesis was performed with oligo(dT) primers and reverse transcription with M-MLV Reverse Transcriptase (Invitrogen). Quantitative real-time PCR was performed using SYBR Green Reagent (Invitrogen) in Applied Biosystems Prism 7300 real-time PCR system (Applied Bioscience). Reactions were done in triplicate and GAPDH values were used to normalize gene expression. Expression levels were expressed in times vs control basal expression (“fold induction”), which was set to “1.”

Frozen colon samples were homogenized in ice-cold lysis buffer containing 10 mM HEPES, 2 mM EDTA (pH 8), 5 mM DTT, 1 mM Pefabloc, and 1 tablet of mixture of proteinase inhibitors (Roche). TNF and IL-6 concentrations were measured in whole tissue extracts by ELISA following the manufacturer’s instructions (R&D Systems) and expressed as pg per mg of total protein.

Protein extracts were separated on a 10% SDS-polyacrylamide-gel and transferred to nitrocellulose membranes (Whatman). Membranes were blocked with 5% milk in PBS and probed with the following primary Abs: P-STAT3 (Cell Signaling Technology), ICAM-1 (Santa Cruz Biotechnology), and α-tubulin (Sigma-Aldrich) as loading control.

Systemic deletion of gp130 was induced by administration of 2 × 80 μg poly(I:C) in MxCre/gp130^−/−. Control animals (gp130^f/f) either received equal amounts of poly(I:C) or PBS. Acute DSS colitis was induced by treating gp130^f/f and MxCre/gp130^−/− animals with 3% DSS for 7 days. Treatment of gp130^f/f mice with poly(I:C) 15 and 10 days before start of DSS administration did not affect the development of colitis (Fig. 1). Control gp130^f/f mice displayed symptoms of severe colitis, whereas MxCre/gp130^−/− mice showed a delayed onset and markedly milder course of disease (Fig. 1, A and B). This was evidenced by significantly lower body weight loss and a lower DAI of MxCre/gp130^−/− animals while gp130^f/f littermates suffered from dramatic wasting syndrome (Fig. 1, A and B). Macroscopic and histological analysis after 7 days of DSS treatment revealed a severe destructive colitis in gp130^f/f animals. In contrast, colonic inflammation was significantly attenuated in gp130-deleted animals as reflected by less colonic shortening and milder histological changes (Fig. 1,C, D, and F). Furthermore, gp130-deficient animals showed an improved survival compared with wt mice, when DSS treatment was continued for 10 days (90 vs 20%, data not shown). These results suggested a central role of gp130-dependent signals in the establishment of acute mucosal inflammation. Since increased epithelial permeability is central in the establishment of DSS-induced colitis, we tested whether deletion of gp130 a priori altered intestinal permeability in vivo. As shown in Fig. 1 E basal colonic permeability was not changed in gp130^f/f and MxCre/gp130^−/− animals.

To characterize cell populations involved in mediating acute colitis, we next analyzed the mucosal inflammatory infiltrate. For this purpose, single-cell suspensions from colonic tissues of gp130^f/f and MxCre/gp130^−/− mice were isolated and analyzed by flow cytometry. Additionally, we performed immunostainings of colon cryo-sections. Before colitis induction, no major differences were observed between the experimental groups (Fig. 2,A). However, MxCre/gp130^−/− showed significantly higher numbers of CD11c⁺ cells before DSS treatment. Pretreatment with poly(I:C) for induction of Cre expression and subsequent gp130-deletion in MxCre/gp130^−/− did not alter mucosal immune cell content in gp130^f/f compared with PBS-treated controls (Fig. 2,A, C, and D). However, 7 days after induction of DSS colitis the proportion of intestinal CD45⁺ immune cells was significantly lower in cell suspensions derived from MxCre/gp130^−/− compared with gp130^f/f animals (15.1 ± 3.1 vs 8.2 ± 2.6%). In acute DSS colitis, macrophages (F4/80⁺) and neutrophils (CD11b⁺/Ly6G⁺) represented the predominant mucosal cell populations (Fig. 2). In contrast to gp130^f/f mice, colonic tissue from MxCre/gp130^−/− contained significantly lower numbers of macrophages and neutrophils (Fig. 2, A and C). We furthermore observed a reduced population of CD11c⁺ and CD3⁺CD8⁺ cells in colonic mucosa of gp130-deleted mice, whereas the proportion of CD3⁺CD4⁺ cells was equally low in both groups. These results were further confirmed by immunofluorescence staining, which revealed a pronounced infiltration of myeloid cells (CD11b⁺) in gp130^f/f mice with a high proportion of CD68⁺ macrophages, while only moderate mucosal infiltration was observed in the MxCre/gp130^−/− group (Fig. 2,D). Basal leukocyte content was equal in all experimental groups (Fig. 2,A, C, and D). Local cell proliferation was assessed by Ki67 staining of colon sections. High proliferative activity was observed in the epithelial compartment, whereas CD11b/Ki67 costaining was virtually absent in all mice, largely excluding local proliferation of neutrophils and macrophages. To further determine whether reduced infiltration of inflammatory cells correlated with reduced production of leukocytes, bone marrow and peripheral blood of gp130^f/f and MxCre/gp130^−/− was analyzed. Untreated MxCre/gp130^−/− showed a marked neutrophilia as described previously (16), that remained largely unchanged after DSS treatment, whereas no differences were observed for “inflammatory” CD115⁺Gr1^high monocytes (Fig. 1,E). No major differences were observed in the bone marrow or in mesenteric lymph nodes (Fig. 1, B and F). These experiments demonstrated that reduced disease activity in MxCre/gp130^−/− animals directly correlated with impaired recruitment of monocytes/macrophages and neutrophils to the intestinal mucosa.

A number of proinflammatory cytokines are known to mediate intestinal inflammation. We therefore compared cytokine expression in colon samples from gp130^f/f and MxCre/gp130^−/− before and after DSS treatment by real-time PCR and ELISA. IL-6, TNF, and IL-17 expression was strongly induced in the course of DSS colitis in wt animals. However, in colonic mucosa of MxCre/gp130^−/− mice, IL-6 and IL-17 expression levels remained largely unaltered compared with baseline levels before DSS treatment (Fig. 3, A and B), indicating that mucosal IL-6 production was gp130 dependent. Subsequently IL-17 production was equally abrogated, which is in line with recent data that demonstrate a requirement of IL-6 (and TGF-β) for the induction of Th17 cells in mice (21). In contrast, TNF induction appeared to be independent of gp130 signals since it was strongly induced after 4 (not shown) and 7 days of DSS in both groups (Fig. 3,A). However, after 7 days, it was significantly reduced in MxCre/gp130^−/− on the protein level (Fig. 3,B), in concordance with the lower mucosal infiltration of macrophages (Fig. 2).

We next studied the mRNA expression of key chemokines that are known to control migration of neutrophils and monocytes. The induction of mucosal CXCL2 (=MIP2β) and in parts CXCL1 (=KC) expression was markedly diminished in MxCre/gp130^−/− animals compared with gp130^f/f controls (Fig. 3,C). Basal levels of CXCL1 were significantly reduced in the mucosa of gp130-deficient mice (Fig. 3,C). Transcripts of CCL2 (MCP-1), a potent monocyte chemoattractant, were significantly elevated in gp130^f/f mucosa samples (Fig. 3 C) compared with MxCre/gp130^−/−, whereas no significant differences in expression were found for CCL20 (=MIP3-α) and CX3CL1 (=Fractalkine).

Intestinal mucosa of patients suffering from IBD has been shown to exhibit significantly higher amounts of P-STAT3 (5). Gp130 signaling induces prominent STAT-3 activation. As shown in Fig. 4, STAT-3 phosphorylation was strongly induced in colons of gp130^f/f animals after onset of colitis (Fig. 4, A and C). Immunofluorescence analysis of colonic sections revealed prominent P-STAT3 staining of infiltrating cells as well as parenchymal cells such as smooth muscle cells and epithelial cells (Fig. 4,C). As expected, STAT3 activation was significantly blunted in MxCre/gp130^−/− and virtually absent in untreated animals of both genotypes. Residual STAT3 activation in MxCre/gp130^−/− animals was mostly found in epithelial cells of colon crypts (Fig. 4,C). Besides local chemokine production, expression of adhesion molecules on endothelial and parenchymal cells represents a crucial step for the recruitment of leukocytes to sites of inflammation. It has previously been shown that endothelial expression of ICAM-1 can be strongly induced by IL-6 in a gp130/STAT3-dependent manner (15). DSS treatment augmented ICAM-1 expression on mRNA and protein level, whereas this response was diminished in gp130-deleted mice (Fig. 4 B).

Since protection of gp130-deleted mice was associated with a markedly decreased mucosal infiltration of neutrophils and macrophages, we sought to analyze the direct effects of gp130 related cytokines on these cells in vivo. We thus generated monocyte/macrophage/neutrophil-specific gp130-deficient mice by crossing C57BL6 gp130^f/f mice with Lys(lysozyme M)-Cre mice. LysCre/gp130^−/− mice displayed a complete gp130 knockout in macrophages and neutrophils. This was demonstrated by specific RT-PCR of peritoneal macrophages and neutrophils elicited by thioglycollate (not shown).

Macrophage/neutrophil specific gp130 deficiency rendered the animals less susceptible to DSS induced acute colitis compared with gp130^f/f littermates as evidenced by a clearly delayed and reduced weight loss and lower DAI as well as reduced colonic shortening (Fig. 5, A–C) and histological pathologies (not shown). However, protection was not as effective as in systemically gp130-deleted mice (Fig. 1). Moreover, the recruitment of leukocytes to the colonic mucosa was only marginally decreased compared with control animals (Fig. 5, C and D). In contrast, mucosal cytokine production was clearly attenuated, as exemplified by reduced IL-6 and TNF mRNA levels (Fig. 5,F), indicating a defective activation of the infiltrating leukocytes. However, circulating leukocyte numbers were lower in LysCre/gp130^−/− most likely reflecting the reduced systemic inflammatory response, whereas the composition of bone marrow cells was unchanged (Fig. 5, G and H).

The data obtained in LysCre/gp130^−/− animals suggested that leukocyte recruitment was crucially dependent on gp130-mediated effects in parenchymal cells, whereas activation of leukocytes in the mucosa seemed to directly involve gp130 on myeloid cells. We therefore generated bone marrow chimeras to further test this hypothesis. MxCre/gp130^−/− or gp130^f/f animals were whole body irradiated and received bone marrow grafts from GFP tg gp130^f/f (GFP/gp130^f/f) donors. Engraftment was monitored by FACS analysis of peripheral blood leukocytes. Four weeks after transplantation, bone marrow was fully engrafted and experimental colitis was induced. In the early phase of colitis, no differences in the disease course between the two groups was evident. However, in the later phase (days 4–7), MxCre/gp130^−/− recipients were protected from severe wasting (Fig. 6, A and B). Bone marrow transplantation itself did not alter the susceptibility to DSS colitis (Fig. 6, A and B). Flow cytometric and histological analysis of the mucosal cell content 7 days after colitis induction revealed a significantly diminished mucosal infiltration of macrophages and neutrophils with a lower proportion of CD11b⁺/GFP⁺ (=donor) cells in gp130-deleted recipient mice compared with gp130^f/f mice (Fig. 6, C and D). Circulating leukocyte numbers were not significantly altered between the groups (Fig. 6 E). These data indicated a crucial involvement of gp130-dependent signals in parenchymal cells for the recruitment of CD11b⁺ leukocytes to the colonic mucosa, thereby facilitating the development and maintenance of colitis.

Our observations in the DSS model of the experimental intestinal inflammation model prompted us to test whether these results could be transferred to other inflammatory conditions. For this purpose, we used thioglycollate-induced peritonitis, which serves as a well-established model to study leukocyte migration in vivo. Gp130^f/f, LysCre/gp130^−/−, and MxCre/gp130^−/− mice were injected with 1 ml 3% thioglycollate and peritoneal cells were harvested 18 h after injection. Cell counts and FACS analysis revealed a profound impairment of neutrophil and macrophage recruitment in MxCre/gp130^−/− compared with gp130^f/f and LysCre/gp130^−/− mice (Fig. 7). LysCre/gp130^−/− showed reduced neutrophil numbers, whereas macrophage recruitment was not altered compared with wt, which is in accordance to the findings in the DSS model (Fig. 5 C). This experiment demonstrated the relevance of gp130-dependent mechanisms for effective migration of neutrophils and macrophages to sites of inflammation.

Earlier reports have demonstrated an important contribution of IL-6 and IL-6 trans-signaling for disease progression of T cell-dependent chronic colitis (4, 5). In contrast, the role of the common signal transducer gp130 in colitis models has been less well defined. Gp130 is ubiquitously expressed and serves as essential (co-) receptor for a family of cytokines including IL-6, Oncostatin-M, leukemia inhibitory factor, neuropoetin, cardiotrophin-1, cardiotrophin-like-cytokine, ciliary neurotrophic factor, IL-11 and IL-27 (22). Specific inhibition of IL-6 trans-signaling, which involves binding of IL-6/sIL-6R complexes to target cell expressed gp130, has been shown to render mucosal T cells resistant to apoptosis, thereby promoting inflammation in experimental colitis (5). Another recent study demonstrated the inhibitory effects of IL-6 signaling on FoxP3 expression in naive CD4⁺CD25⁻ T cells, inhibiting the generation of Tregs and thereby promoting colonic inflammation in a T cell transfer model of murine colitis (23). The impact of gp130-dependent signals on acute, T cell-independent colitis is unclear. Several animal models of IBD, like DSS-induced colitis, and most probably human Crohn’s disease as well, are triggered by bacterial activation of innate immune cells (1). We therefore decided to investigate the contribution of gp130-derived signals for leukocyte recruitment and activation in an acute, bacteria-dependent, T cell-independent model of murine colitis. We show here that systemic gp130 deletion protected the animals from severe colitis and wasting. Gp130 deficiency markedly decreased the influx of macrophages and neutrophils and reduced production of proinflammatory cytokines. Notably, the basal number of circulating leukocytes is significantly higher in MxCre/gp130^−/− animals (Fig. 2 E), (16).

Interestingly, IL-6 mRNA induction was completely blocked upon gp130 deletion. This finding was unexpected and is at present unexplained. It was recently reported that gp130-mediated STAT3 phosphorylation augments STAT3 gene transcription, which consequently accumulates in its unphosphorylated form (U-STAT3). U-STAT3 was shown to subsequently induce a second wave of P-STAT3 independent genes, among those NFκB-dependent genes like IL-6 (24). Our data indicate that this mechanism could also be relevant in vivo. Production of IL-17, known to be central in the pathogenesis of autoimmunity and important in host defense, is dependent on IL-6 (25). Concordantly, we found that the strong induction of IL-17 in inflamed mucosa was abolished in MxCre/gp130^−/− mice. TNF levels were only marginally reduced, which is in line with a previous study that suggested a minor or even protective role for TNF in the acute phase of DSS colitis (26). Along with lower IL-6 levels, we observed a markedly decreased transcription of leukocyte attracting chemokines and adhesion proteins. Decreased mucosal expression of CXCL1, CXCL2, CCL2, and ICAM-1 was accompanied by a reduced infiltration of leukocytes in the colon of MxCre/gp130^−/− mice.

To further dissect the cellular mechanisms that govern the gp130-dependent recruitment of leukocytes to the colonic mucosa, we generated LysCre/gp130^−/− animals and established gp130 chimeras using GFP-tg gp130^f/f mice as bone marrow donors. Selective gp130 deletion in neutrophils and macrophages only marginally altered the mucosal accumulation of these cells after 7 days of DSS treatment. However, initiation of the disease was markedly delayed and reduced. LysCre/gp130^−/− animals did not show any signs of colitis in the first 4 days of DSS treatment. This finding might indicate defective activation of resident and invading leukocytes in the initiation phase of colitis, whereas the influx of leukocyte was only weakly affected and, thus, over time was able to partially restore disease activity in a gp130-independent fashion. These observations suggested that besides myeloid cells also gp130-dependent signals in parenchymal cells were involved in controlling disease activity in the DSS model. The transcription of many chemokines and adhesion molecules is known to be P-STAT3 dependent in these cells (15, 27). Using bone marrow chimeric mice, we could demonstrate that the influx of gp130^f/f leukocytes was significantly reduced in gp130-deficient recipient mice, which indicated a requirement for gp130 in non-bone marrow derived cells for mucosal inflammatory cell recruitment. This hypothesis was further supported by the time course and activity of the disease in the chimeric animals. In the first days during the initiation phase, weight loss and DAI were comparable between both groups. However, at later time points disease activity was markedly milder in gp130 bone marrow chimeric mice. During this phase of the disease, we observed diminished inflammatory infiltration in the gp130-deleted recipients. Defective migration of leukocytes to sites of inflammation in gp130-deficient mice was furthermore demonstrated in a sterile peritonitis model, which underscored the previous findings. In this study, we found that systemic gp130 deletion strongly impaired peritoneal leukocyte accumulation, whereas this effect was much weaker in LysCre/gp130^−/− animals. Together, our data indicate a potent role of gp130-derived signals in different cell compartments in leukocyte homeostasis during acute inflammation.

Considering the role of STAT3 in intestinal mucosa, it has so far been attributed anti-inflammatory properties. Takeda et al. (28) described that macrophage/neutrophil-specific STAT3 deletion (LysCre/STAT3^−/−) resulted in spontaneous severe enterocolitis. However, the role of STAT3 appears to be more complex since both IL-6 family cytokines via gp130 and IL-10 activate STAT3 (29). Thus, deletion of STAT3 results in unresponsiveness to IL-10. Systemic IL-10 deficiency also leads to spontaneous enterocolitis development, which could explain the phenotype in LysCre/STAT3^−/− animals.

In contrast, we provide evidence that gp130-dependent STAT3 activation triggers mucosal inflammation, which is in line with a report that showed that gp130-dependent STAT3 signaling promotes gastric inflammation and carcinogenesis (30). We observed prominent STAT3 activation in infiltrating leukocytes, parenchymal cells, and in colonic epithelial cells following DSS treatment. P-STAT3 was first seen in the epithelial compartment and could still be found in MxCre/gp130^−/− mice. A previous study proposed a protective role of gp130-dependent STAT3 activation in mucosal injury, suggesting a protection from colitis after total deletion of gp130 (31), which we indeed were able to show here. However, the present data might need further confirmation in other models of experimental colitis.

In summary, our data contribute to the understanding of the function of gp130/IL-6 family cytokines in the context of mucosal inflammation and regulation of innate immune cells. We propose a dual mechanism of action mediated by gp130 signals in the early stages of inflammation, involving chemokine and adhesion molecule induction in parenchymal cells, which promotes leukocyte recruitment as well as directly activating effects on invading myeloid cells. Anti-IL-6 or anti-gp130 strategies might therefore prove very efficient for both chronic and acute inflammatory conditions.

We thank Gernot Sellge and Thomas Gebhardt for helpful discussion and advice, and Malika Al-Masaoudi for technical assistance.

The authors have no financial conflict of interest.