Essential crosstalk between myeloid and lymphoid cells for development of chronic colitis in myeloid-specific signal transducer and activator of transcription 3-deficient mice

Abstract

Dysregulated cytokine responsiveness by myeloid cells can be a trigger for the development of chronic inflammation as well as inflammatory bowel disease. Thus, mice with a myeloid-specific defect in signal transducer and activator of transcription (Stat) 3 develop spontaneous colitis secondary to the inability of myeloid cells to respond to the immunosuppressive cytokine interleukin-10. We now examined whether the inflammation caused by Stat3-deficient macrophages is cell autonomous or dependent on their interaction with lymphocytes. For this purpose, myeloid-specific Stat3-deficient mice (LysMcre/Stat3^flox mice) were intercrossed with RAG-1 knockout mice to generate LysMcre/Stat3^flox RAG^–/– mice. In these mutants and LysMcre/Stat3^flox littermate control mice we determined the onset and severity of spontaneous chronic enterocolitis, and the reaction to dextran sodium sulphate (DSS)-induced epithelial damage, as well as to lipopolysaccharide (LPS) challenge. In contrast to LysMcre/Stat3^flox mice, LysMcre/Stat3^flox RAG^–/– animals are protected from chronic enterocolitis. Although they respond to oral dextran sulphate with acute colitis symptoms, the inflammation heals similarly to wild type mice whereas LysMcre/Stat3^flox mice exhibit continued colitis pathology. In addition, the hyperreactivity of LysMcre/Stat3^flox mice to LPS-challenge in vivo was less severe in the absence of lymphocytes. Despite clear differences in the strength of inflammatory responses, macrophages of both LysMcre/Stat3^flox mice and LysMcre/Stat3^flox RAG^–/– animals exhibited increased costimulatory capacity. In conclusion, our findings demonstrate that Stat3-deficient myeloid cells alone are not capable of inducing the severe pathology seen in LysMcre/Stat3^flox mice. Yet when these cells can interact with lymphocytes their increased costimulatory potential will trigger an overshooting inflammatory response.

Introduction

Macrophages are widely distributed throughout the mucosal surfaces of the body. They are part of the innate immune system and play an important role as first line of defence against invading micro-organisms. In addition, they are highly effective bone-marrow derived antigen-presenting cells, working in tight contact with the lymphoid cells of the adaptive immune system, partly orchestrating the immune response of these cells.

The signal transducer and activator of transcription (Stat)3 is a cytoplasmic transcription factor essential for the signal transduction pathway downstream of gp130 containing receptors for the interleukin (IL)-6 family of cytokines, the IL-10 receptor, and granulocyte colony-stimulating factor receptor^1–4. Because the IL-6 family comprises several cytokines which are required during early embryogenesis, complete deletion of Stat3 in vivo is lethal.⁵ In order to further elucidate the role of myeloid cells in mucosal immunity, we explored the function of Stat3 using the Cre/loxP recombination system. By placing Cre under control of the murine lysozyme M gene regulatory region (LysMcre mice), the Stat3 locus is effectively targeted in myeloid cells as lysozyme M is expressed in murine monocytes, macrophages and neutrophils.⁶ In these cells Stat3 is indispensable for IL-10 signalling and is also a key factor downstream of the IL-6 receptor.^7,8 We have previously shown that in LysMcre/Stat3^flox animals deletion of Stat3 in myeloid cells leads to development of spontaneous colitis;⁹ this is despite the fact that in these animals only Stat3 signalling in myeloid cells is targeted, unlike IL-10 knockout animals where IL-10 is completely absent. We now demonstrate that LysMcre/Stat3^flox animals are protected from colitis, when backcrossed to a recombination activating gene (RAG)1-deficient background, suggesting that the development of intestinal pathology in these mice is not a macrophage/granulocyte autonomous process but rather depends on myeloid cell/lymphocyte interactions. In line with these results, Stat3-deficient bone marrow macrophages present more costimulatory molecules and hold an increased potential to costimulate resting T cells.

Materials and methods

Mice

Conditional Stat3 knockout mice (LysMcre/Stat3^flox) on a mixed genetic background (129/Ola × C57BL/6) were used for this study.⁹ As LysMcre/Stat3^flox/flox reacted essentially as LysMcre/Stat3^flox/– mice, we used both types of mice interchangeably. To indicate this fact we refer to this mice as LysMcre/Stat3^flox animals. T- and B-cell deficient animals were obtained by crossing these mice into a recombination activating gene (RAG1)-deficient background (LysMcre/Stat3^flox RAG^–/–).¹⁰ In all experiments littermate wild type or heterozygous RAG1^+/– animals were used as controls. All animals were housed under specific pathogen-free conditions in the animal facility of the Institute of Medical Microbiology, Immunology and Hygiene, Munich.

DSS-induced colitis

Mice (aged 6–8 weeks) were fed with 5% dextran sodium sulphate (DSS) dissolved in the drinking water for 7 days, followed by either a 2 or a 14 day observation period with untreated drinking water. After this time mice were killed by cervical dislocation and inflammation of the colon was examined.

Histopathological evaluation

The entire colon was removed and a 2 cm piece of the ascending colon was fixed in 10% buffered formalin for histological analysis. Six to eight transverse sections were paraffin-embedded and 4 µm sections were stained with haematoxylin/eosin solution according to standard protocols. Scoring was performed in a blinded fashion using a combined score of inflammatory cell infiltration (0–3) and tissue damage (0–3) as described previously.¹¹

Organ culture

A colon segment was excised, opened and washed repeatedly with sterile phosphate-buffered saline and finally placed into a 24-well flat bottom culture plate containing 1 ml of RPMI-1640 supplemented with penicillin, streptomycin and 10% heat inactivated fetal calf serum (FCS). The tissue specimen were incubated for 36 hr at 37°. At the end of the culture period the wet weight of the specimen was determined and the culture supernatants were harvested for cytokine analysis. Cytokine levels were calculated as picogram per gram wet weight.

Cytokine measurements

Cytokine levels for IL-6, IL-10, IL-12p70, IFN-γ and tumour necrosis factor-α (TNF-α) were measured using the cytometric bead array for mouse inflammatory cytokines (BD Pharmingen, Heidelberg, Germany) according to the manufacturer's instructions.

Preparation and stimulation of peritoneal macrophages

For isolation of peritoneal macrophages mice were injected intraperitoneally (i.p.) with 2 ml of 4% thioglycollate (Sigma, Munich, Germany). Cells were harvested from the peritoneal cavity 5 days after injection. Cells were incubated for 2 hr and then washed three times to remove non-adherent cells. Macrophages were cultured with RPMI-1640, supplemented with penicillin, streptomycin and 10% FCS. Cells were stimulated for 24 hr with 1 µg/ml lipopolysaccharide (LPS; Escherichia coli serotype O127:B8 (Sigma, Munich, Germany)).

TNF-α and IL-12p70 concentration were determined by enzyme-linked immunosorbent assay (ELISA) according to the manufacturer's instructions (R & D Systems, Wiesbaden, Germany).

Co-stimulation assay

This assay has been described in detail before.¹² Briefly, 96-well flat-bottom plates were coated with anti CD3 at a concentration of 10 µg/ml antibody at 4° overnight in carbonate–bicarbonate buffer. After that the wells were washed once with sterile PBS and blocked with complete culture medium. Naïve splenic CD4⁺ T cells (5 × 10⁴) were added to each well. Macrophages (1 × 10⁴) were added and incubated at 37° in an incubator with 5% CO₂. Proliferation was assessed by cell-titre proliferation assay (Promega, Munich, Germany) according to the manufacturer's protocol at day 3.

Results

LysMcre/Stat3^flox RAG^–/– animals thrive normally while LysMcre/Stat3^flox mice show impaired weight gain

LysMcre/Stat3^flox mice develop clinical signs of spontaneous colitis at the age of 12–20 weeks.⁹ From this time on the animals develop episodes of loose stools and bleeding, which usually resolve spontaneously, indicating a phased disease process comparable to chronic inflammatory bowel disease in humans. To investigate evolution of colitis, animals were weighed regularly during a period of four months. Comparison of weight gain in an age- and sex-matched group of LysMcre/Stat3^flox, LysMcre/Stat3^flox RAG^–/– and wild type animals showed normal growth in LysMcre/Stat3^flox RAG^–/– mice while LysMcre/Stat3^flox mice failed to thrive appropriately. This process starts before the age of 3 months and becomes more prominent with age, finally resulting in comparable weight gain in LysMcre/Stat3^flox RAG^–/– and wild type mice and a 27% lower weight gain in the LysMcre/Stat3^flox group (Fig. 1).

Figure 1.

LysMcre/Stat3^flox (MCre Stat3) mice failed to thrive, while LysMcre/Stat3^flox RAG^–/– (MCre Stat3 RAG) animals gained weight like their wild type littermates. *P < 0·05 versus wild type mice (n = 8 animals for wild type, n = 8 animals for MCre Stat3 RAG, and n = 7 animals for MCre Stat3 per experiment; one representative out of two independent experiments is shown).

Macroscopic and histological evaluation shows no evidence of chronic colitis in LysMcre/Stat3^flox RAG^–/– mice

At the age of 24 weeks, the mice were killed for evaluation. In LysMcre/Stat3^flox RAG^–/– mice macroscopic analysis of the large intestine exhibited no gross abnormalities, while in LysMcre/Stat3^flox mice the entire colon showed marked thickening and shortening (data not shown). No macroscopic signs of small bowel involvement could be detected in any of these animals. Histological analysis of colonic sections from LysMcre/Stat3^flox mice showed the presence of extensive mononuclear infiltration in the mucosa and submucosa. In 30% of the animals this infiltration was found to be transmural (Fig. 2a, b). Immunohistochemical staining of sections revealed that the majority of the infiltrating cells were F4/80⁺ macrophages and CD3⁺ lymphocytes (Fig. 2c and data not shown). Epithelial erosions and ulcerations were seen in all animals. In some mice pleomorphism of enterocytes indicates long-standing chronic inflammation. In LysMcre/Stat3^flox RAG^–/– animals, no or only sparse focal infiltrates of macrophages and neutrophils were noted, which never extended beyond the mucosal layer. No ulceration or epithelial cells damage could be detected (Fig. 2d, e). Using a histological scoring system described in Materials and methods, we assessed the severity of colitis in LysMcre/Stat3^flox, LysMcre/Stat3^flox RAG^–/– and wild type control animals, respectively. LysMcre/Stat3^flox animals had a mean score of 3·9 out of a maximum score of 6 (intensity ranging from 3·0 to 5·5) reflecting the high inflammatory activity in these samples. LysMcre/Stat3^flox RAG^–/– animals had an average score of 1, with values ranging from 0 to 1·5. In the wild type control group only one animal showed focal inflammation with a score of 1·5 (Fig. 2g).

Figure 2.

Development of spontaneous colitis in MCre Stat3 but not MCre Stat3 RAG mice. At 24 weeks of age mice were killed and histological sections of colon specimen were analysed by HE and immunohistological staining. (a, b) MCre Stat3 animals showed extensive mononuclear infiltration in the mucosa and submucosa; (c) staining for F4/80revealed that many of the infiltrating cells were macrophages. (d, e) In MCre Stat3 RAG animals no infiltrate could be found and (f) by staining with the F4/80 antibody only few macrophages were detected. (g) Mean ± SEM of the histology score (see Materials and methods for description) for n = 6 animals per group. Magnification is 50× for (a) and (d) and 100× for (b, c, e and f). *P < 0·001 versus wild type mice.

Both LysMcre/Stat3^flox RAG^–/– and LysMcre/Stat3^flox mice are susceptible to acute DSS-induced colitis

To evaluate whether LysMcre/Stat3^flox RAG^–/– mice are generally resistant to the induction of colitis, 6–8-week-old animals were challenged with 5% DSS in the drinking water. While LysMcre/Stat3^flox RAG^–/– mice did not develop spontaneous colitis with age they were equally susceptible to DSS-induced colitis as LysMcre/Stat3^flox or wild type littermates. When challenged for 7 days with 5% DSS in the drinking water all mice started to lose weight after 2 days and showed similar histology scores (3·8–4·7 out of a maximum of 6) at the end of the experiment (Fig. 3). These findings demonstrate, that LysMcre/Stat3^flox RAG^–/– animals can develop a pronounced inflammation despite the fact that they are lacking T and B cells. Thus, we can exclude that a lack of effector cells is the reason for LysMcre/Stat3^flox RAG^–/– mice being protected from spontaneous colitis.

Figure 3.

Acute challenge with 5% DSS induced colitis of similar intensity in MCre Stat3, MCre Stat3 RAG and wild type animals (age 6–8 weeks). Animals were challenged with 5% DSS for 7 days, and five or more colonic sections were used for histological evaluation. Shown is the mean ± SEM of the histology score (see Materials and methods for description) One representative out of two independent experiments is depicted (n = 2, respectively, five animals per group in each experiment). Representative (b, c and d) histological sections are shown (HE stain, magnification 50×).

LysMcre/Stat3^flox RAG^–/– mice recover from DSS induced colitis while LysMcre/Stat3^flox mice develop chronic colitis

Spontaneous colitis becomes manifest by the age of 12–20 weeks in LysMcre/Stat3^flox mice. We tested if a chronic or prolonged colitis could be induced in these animals earlier in life using an exogenous short-term challenge. Therefore we treated wild type, LysMcre/Stat3^flox and LysMcre/Stat3^flox RAG^–/– mice for 7 days with DSS and evaluated the course of the colitis for the following 2 weeks. LysMcre/Stat3^flox RAG^–/– mice were able to recover from DSS-induced colitis like wild type animals while LysMcre/Stat3^flox mice developed signs of persistent colitis after a single 7-day course of oral DSS. LysMcre/Stat3^flox RAG^–/– mice regained weight considerably faster than LysMcre/Stat3^flox littermates after DSS challenge was ended. Two weeks after the end of DSS treatment LysMcre/Stat3^flox RAG^–/– and wild type mice had reached approximately their initial weight while LysMcre/Stat3^flox mice were still substantially lighter (Fig. 4a). In accordance with this clinical finding LysMcre/Stat3^flox mice also showed histological signs of ongoing colitis (Fig. 4b) 2 weeks after DSS treatment was stopped. To assess the local mucosal cytokine milieu, we obtained small fragments of mucosa 2 weeks after DSS treatment, cultured these in tissue culture medium and determined the cytokine levels in the supernatant 36 hr later. Using the organ culture assay we found high levels of interferon-γ (IFN-γ) (mean of 5600 pg/ml/g) in LysMcre/Stat3^flox mice and a significantly lower level in LysMcre/Stat3^flox RAG^–/– and wild type animals (mean of 1700 and 63 pg/ml/g, respectively) (Fig. 5a). No difference could be detected for TNF-α. The production of IL-6 (28 ng/ml/g) was increased in the wild type mice 2 weeks after DSS exposure (Fig. 5b). In the LysMcre/Stat3^flox RAG^–/– as well as in the LysMcre/Stat3^flox or the wild type animals the amounts of IL-10 detected in the organ cultures were near the detection limit of the used ELISA assay.

Figure 4.

MCre Stat3 mice developed chronic colitis after a single, 7 day exposure to DSS, while MCre Stat3 RAG and wild type mice recovered. (a) Weight curves of all DSS treated animals showed a similar weight loss while treated with DSS but only MCre Stat3 RAG and wild type mice regained weight and recovered 2 weeks after cessation of treatment. *P < 0·05 versus wild type mice (n = 5 animals for wild type and MCre Stat3, n = 7 animals for MCre Stat3 Rag, one representative out of two experiments is shown). (b) The presence of chronic colitis in MCreStat3 mice is also reflected by the histology scores from mice 2 weeks after DSS treatment. (n = 4 mice per group).

Figure 5.

Cytokine production in organ cultures of colon samples two weeks after DSS treatment. (a, b) After a 7 day course of 5% DSS, mice were fed with normal drinking water for 14 days before analysis. Pieces of colonic mucosa were incubated in medium for36 rnd the concentration of(a) IFN-γ TNF-α IL-10 and (b) IL-6 measured in the supernatant. Bars are mean ± SEM and represent the pooled data of two independent experiments with n = 5 animals for wild type, n = 4 animals for Mcre Stat3 Rag, and n = 3 animals for Mcre Stat3 animals per group.

Enhanced susceptibility of Stat3-deficient macrophages to LPS challenge in vivo is reduced in the absence of lymphocytes

We wanted to determine whether the dependence on lymphocyte interaction is not only crucial for the intestinal inflammation but also relevant for systemic inflammation. To that means, we stimulated mice with LPS in vivo (1 µg LPS per gram body weight i.p) and demonstrated that survival of LysMcre/Stat3^flox RAG^–/– animals was clearly improved (Fig. 6) compared to the highly susceptible LysMcre/Stat3^flox animals. To compare the serum kinetics of cytokines during LPS induced septic shock, blood samples were taken two days before, and 1·5 and 6 hr after i.p. LPS injection. Baseline cytokine levels were identical in all groups of animals. After LPS injection LysMcre/Stat3^flox RAG^–/– and LysMcre/Stat3^flox mice showed higher levels of TNF-α, IL-12 p70 and IFN-γ compared to wild type controls (Fig. 7a, b, c). The high number of natural killer (NK) cells generally present in RAG-deficient mice may be the reason for the high serum level of IFN-γ seen in LysMcre/Stat3^flox RAG^–/– animals after in vivo LPS challenge. In LysMcre/Stat3^flox animals we found high levels of IL-10 and IL-6 (Fig. 7d, e). Yet even this massive production of IL-10 could not protect the animals from the development of septic shock because of the inability of the Stat3-deficient myeloid cells to respond to IL-10.

Figure 6.

MCre Stat3 mice are far more susceptible to in vivo LPS challenge than MCreStat3 RAG mice. While all of the animals in the MCre Stat3 group died after they had been challenged with 1 µg/g body weight LPS, all of the wild type and 70% of the MCre Stat3 RAG mice survived. (n = 5 animals for wild type, n = 6 animals for MCre Stat3 RAG, and n = 4 animals for MCreStat3). Shown is one representative out of two experiments.

Figure 7.

Serum levels of cytokines after in vivo LPS challenge. Mice were bled before treatment (0 hr) or 1·5 and 6 hr after LPS challenge and the concentrations of IFN-γ (a), TNF-α (b), IL-12p70 (c), IL-10 (d) and IL-6 (e) were determined by the cytometric bead assay. In untreated animals no elevated cytokine levels were detected, indicating that there was no constitutive preactivation of inflammatory cells in any group. (n = 6 animals for wild type, and n = 4 animals for MCre Stat3 RAG and for MCre Stat3).

LysMcre/Stat3^flox macrophages provide more effective costimulatory support to naïve CD4 T cells than wild type cells

We analysed the expression of CD86 and MHC class II molecules on bone marrow macrophages of wild type, LysMcre/Stat3^flox and LysMcre/Stat3^flox RAG^–/– animals. Furthermore, we measured the expression of CD86 on macrophages isolated ex vivo from the mesenteric lymph node of wild type, LysMcre/Stat3^flox and LysMcre/Stat3^flox RAG^–/– mice. In bone-marrow derived macrophages as well as in resident lymph node macrophages we could demonstrate a persistently enhanced expression of CD86 in LysMcre/Stat3^flox as well as in LysMcre/Stat3^flox RAG^–/– animals (Fig. 8a, b). CD86 plays a major role in T-cell costimulation, yet its overexpression might not be sufficient to induce immunopathology.¹³ Therefore, to determine whether an enhanced costimulatory potential of Stat3-deficient macrophages accounts for the development of spontaneous colitis we used an in vitro coculture model. Bone-marrow derived macrophages from LysMcre/Stat3^flox and wild type mice were cocultivated with naïve CD4⁺ T cells from wild type animals. Suboptimal concentrations of plate-bound anti-CD3 antibodies were used to provide the first signal to the lymphocytes, so the proliferative response depended highly on the costimulatory signals provided by the macrophages. The analysis showed significantly increased proliferation in the T cells cocultured with Stat3-deficient macrophages compared to wild type macrophages (Fig. 9). This finding, as well as the increased expression of the proliferation marker Ki67 in the intestinal mucosa of LysMcre/Stat3^flox mice seen by immunohistology (data not shown) demonstrate the functional relevance of the enhanced expression of CD86 and major histocompatibility complex (MHC) class II molecules on Stat3-deficient macrophages.

Figure 8.

CD 86 and MHC II expression on bone marrow derived macrophages (a) and CD86 expression on macrophages isolated from mesenteric lymph nodes (b). Bone-marrow derived macrophages from in vitro cultures and freshly isolated macrophages were analysed for CD86 and IAb or CD86, respectively. Gates were set for macrophages/monocytes in FSC/SSC dotplots by backgating on CD11b positive cells. The experiments are representative for four individual cultures and two isolations.

Figure 9.

Stat3-deficient macrophages provide enhanced costimulation to resting T cells. Bone-marrow derived macrophages from wild type or LysMcre/Stat3^flox animals were cocultured with naïve splenic CD4⁺ T cells and a suboptimal concentration of plate-bound anti-CD3 and proliferation was measured. All experiments were done in triplicates. The experiment shown is representative of four individual experiments.

Discussion

Macrophages are part of the mucosal immune system, screening the lamina propria for conserved patterns of carbohydrate or lipid structures on invading micro-organisms. Tight regulatory mechanisms prevent the self-perpetuation of an inflammatory response despite the high antigen load at mucosal surfaces. In a previous study we demonstrated that cell-type specific disruption of Stat3 signalling in macrophages and neutrophils leads to the development of spontaneous colitis and renders the animals highly susceptible to LPS-mediated septic shock.⁹ The present study shows that in the absence of macrophage–lymphocyte interactions neither spontaneous colitis nor the enhanced hypersensitivity to LPS develop.

Stat3 is an important cytoplasmic transcription factor downstream of the IL-6 cytokine family and IL-10-signal in macrophages.^3,14 Therefore, a defect in Stat3-mediated signalling interrupts the regulatory feedback loop of macrophage-derived IL-10.¹⁵ However, as only the signal transduction downstream of the IL-10 receptor is affected, macrophages still produce IL-10 in response to LPS challenge. In our experiments the level of circulating IL-10 in acute LPS-induced shock was even higher than in wild type animals (Fig. 7d) yet it did not result in adequate down-regulation of the inflammatory response in LysMcre/Stat3^flox animals. IL-10 has the capacity to induce the development of CD4⁺ CD25⁺ regulatory T cells⁸ as Stat3 signalling was not affected in T cells. Regulatory T cells have been shown to suppress innate immune pathology¹⁶ and the development of chronic colitis¹⁷ through cytokine dependent mechanisms. Yet, in our model CD4⁺ CD25⁺ regulatory T cells were unable to prevent neither chronic colitis nor LPS-induced septic shock. We detected massive IFN-γ expression in organ cultures from LysMcre/Stat3^flox mice after short-term DSS challenge, while LysMcre/Stat3^flox RAG^–/– and wild type mice produced only small amounts upon DSS treatment. One can speculate that in wild type animals, IFN-γ production is suppressed because IL-10 inhibits IL-12 production by macrophages/monocytes. Similar findings have been reported by Kobayashi and coworkers, showing that IL-12p40-mediated induction of T helper 1 cells is crucial for the development of chronic enterocolitis in myeloid-cell specific, Stat3-deficient mice.¹⁸ As macrophages recognize the pathogen-associated molecular pattern of invading pathogens, they produce an array of chemokines and cytokines to attract and activate lymphocytes. LysMcre/Stat3^flox macrophages produce significantly higher amounts of TNF-α and NO than wild type macrophages yet not significantly more than macrophages isolated from LysMcre/Stat3^flox RAG^–/– animals (data not shown). This observation was also made after LPS-induced septic shock in vivo (Fig. 7a, b, c). Yet, in this case IFN-γ levels in LysMcre/Stat3^flox RAG^–/– animals were nine times higher than in LysMcre/Stat3^flox mice and about 16 times higher than in wild type mice. This could be the result of an increased activity of NK cells in RAG-deficient animals, and might also partly be caused by an increased macrophage-derived IFN-γ secretion using non-Stat3-dependent pathways.

While various animal models for inflammatory bowel disease demonstrate that the transfer of activated T cells,¹⁹ or the presence of T cells with a regulatory defect,^20,21 is sufficient to induce chronic colitis we could demonstrate that crossing LysMcre/Stat3^flox mice on a RAG1 background completely prevented colitis. Hence, the sole presence of Stat3-deficient macrophages seems not to be sufficient to trigger chronic colitis neither spontaneous, nor after DSS challenge (Figs 2b and 4a). In contrast, acute colitis could be induced in LysMcre/Stat3^flox RAG^–/– mice in the same way as in LysMcre/Stat3^flox mice, demonstrating that massive leucocyte infiltration can occur also in the absence of lymphocytes.

Our data showed that the costimulatory capability of Stat3-deficient macrophages was significantly higher than that of wild type cells (Fig. 9). This could be explained by increased Stat1 mediated signalling found in these animals²² and, consequently, a higher level of Stat1-induced cytokines. Furthermore, in wild type macrophages the IL-10-mediated activation of Stat3 induces a strong antiproliferative effect, reduces the expression of costimulatory molecules and inhibits the production of a variety of cytokines and chemokines, while the synthesis of cytokine antagonists^8,23 is enhanced. Accordingly, we detected increased CD86 and MHC II expression in Stat3-deficient macrophages from bone marrow cultures and freshly isolated tissue macrophages.

The role of IL-6 in the regulation of chronic colitis remains enigmatic since it may act as a pro- as well as anti-inflammatory cytokine.^24,25 We observed that IL-6 was up-regulated in the colon of wild type mice after recovery from DSS-induced colitis. It is possible that IL-6 has an anti-inflammatory function in the colon of wild type mice while Stat3-deficient cells respond to IL-6 with an IFN-γ-like pro-inflammatory response²⁶. The latter is also observed in mice deficient for the suppressor of cytokine signalling (SOCS)3 which is normally also induced by IL-6.^7,27

In summary, these results underline the importance of lymphocytes for the chronic inflammatory processes. In the LysMcre/Stat3^flox model, costimulatory properties of the affected myeloid cells are markedly enhanced, and additionally the feedback inhibition of myeloid cells through IL-10 is abrogated in the absence of Stat3. Despite this signalling defect, myeloid cells depend on lymphoid cells to perpetuate the immune response and ultimately induce chronic colitis. Thus, concomitant activation of both the innate and the adaptive immune systems appears to be responsible for the development of chronic colitis in myeloid-specific Stat3-deficient animals.