Proinflammatory cytokine synthesis by mucosal fibroblasts from mouse colitis is enhanced by interferon-γ-mediated up-regulation of CD40 signalling

Authors

  • T. De L. Karlson,

    Corresponding author
    1. Department of Microbiology and Immunology, Institute of Biomedicine, Gothenburg University, Gothenburg, Sweden, and
    Search for more papers by this author
  • C. V. Whiting,

    1. Department of Clinical Veterinary Science, University of Bristol, Langford, Bristol, UK
    Search for more papers by this author
  • P. W. Bland

    1. Department of Microbiology and Immunology, Institute of Biomedicine, Gothenburg University, Gothenburg, Sweden, and
    2. Department of Clinical Veterinary Science, University of Bristol, Langford, Bristol, UK
    Search for more papers by this author

Errata

This article is corrected by:

  1. Errata: Corrigenda Volume 148, Issue 1, 188, Article first published online: 6 March 2007

Tanya Karlson, Department of Microbiology and Immunology, Institute of Biomedicine, Gothenburg University, Box 435, 405 30 Gothenburg, Sweden.
E-mail: Tanya.karlson@immuno.gu.se

Summary

Gut mesenchymal fibroblasts form complex phenotypical and functional populations. They participate actively in homeostatic maintenance of the extracellular matrix, epithelial barrier function, repair mechanisms and leucocyte migration. In inflammation, they become activated, change matrix expression and synthesize proinflammatory mediators. Subpopulations of mucosal fibroblasts express CD40 and the aim of this study was to define its role in their proinflammatory function. Stable primary fibroblast lines derived from normal mouse colon and inflamed colon from CD4+ CD45RBhigh-transplanted SCID mice were used as models to explore the role of mucosal fibroblast CD40 in the inflammatory process. Phenotype correlated with in situ fibroblast phenotype in the tissues of origin. Lines from both sources co-expressed CD40 and Thy1·2 independently of α-smooth muscle actin. A subpopulation of CD40+ fibroblasts from normal colon expressed CD40 at high levels and expression was enhanced by interferon (IFN)-γ treatment, whereas all CD40+ fibroblasts from colitis expressed at low levels and expression was unaffected by IFN-γ treatment. Despite lower-level expression of CD40 by cells from colitis, they secreted constitutively interleukin (IL)-6 and C-C chemokine (CCL)2. Ligation of CD40 enhanced secretion of these mediators and induced secretion of CCL3. CD40 in cells from colitis was more responsive to ligation than CD40 on cells from normal tissue and this sensitivity was amplified selectively by the action of IFN-γ. We conclude that the inflammatory milieu in colitis induces long-lasting changes in phenotype and proinflammatory function in colonic fibroblasts. In particular, proinflammatory signalling from fibroblast CD40 is amplified synergistically by the Th1 effector T cell cytokine, IFN-γ.

Introduction

The pathogenesis of the chronic inflammatory bowel disease, Crohn's disease, is characterized by high levels of mucosal interferon (IFN)-γ synthesized by local effector T cells, which then induces macrophage activation and synthesis of tumour necrosis factor (TNF)-α. A feature of the chronicity of the disease, however, is the involvement of connective tissue cells, which enhances disease persistence and leads to serious consequences through fibrosis. Less is known about how mesenchymal cells become activated, or the mechanisms whereby they contribute to chronicity of the inflammatory process.

Fibroblasts are heterogeneous, differing from organ to organ and even showing different tissue characteristics, such as cytokine production, extracellular matrix components and proliferation [1–3]. It is becoming clear that their previous characterization as the matrix-building component of solid tissues was an over-simplification and that they are involved in many normal homeostatic and pathological processes [4]. They regulate innate immunity through chemokine synthesis [5,6]; control the proliferation and differentiation of adjacent epithelium [7,8]; maintain chronic inflammatory diseases, such as rheumatoid arthritis [9], Graves' disease-associated opthalmopathy [10] and inflammatory bowel disease [11]; initiate neoplasia [12]; and orchestrate the wound-healing response [13].

Although fibroblasts are heterogeneous within and between tissues, and these multiple roles obviously result from the interaction of fibroblast subsets with agonists triggering a range of transcription pathways, fibroblast CD40 is emerging as a receptor capable of inducing both activation- and suppression-linked functions [5,14]. CD40 is a phosphorylated glycoprotein belonging to the tumour necrosis factor receptor family [15], is a crucial factor driving germinal centre formation and the terminal differentiation of B cells [16], and is the major receptor activating dendritic cells (DC) and macrophages [17]. Its ligand, CD154 [CD40 ligand (CD40L)] is expressed by activated T cells, mast cells, human vascular endothelial cells, smooth muscle cells, human eosinophils and macrophages [18–20]. Interactions between CD40 and CD40L result in alterations in processing and presentation of antigens by antigen-presenting cells; cytokine and chemokine synthesis; proliferation; and up-regulation of cell surface proteins [21]. Importance of the CD40/CD40L interaction in activation events in inflammation has been shown by CD40L blockade, which effectively prevents the onset of inflammation and local synthesis of interleukin (IL)-2, IFN-γ, IL-12 and TNF-α[22]. Fibroblasts from many tissues have been shown to respond to CD40 ligation by transcribing multiple mediators through the nuclear factor kappa B (NFκB) and mitogen-activated protein kinase (MAPK) pathways [11,23].

In the gut mesenchyme, CD40 has been described on fibroblasts in inflamed colon in inflammatory bowel disease patients [5]. As human patients present with well-established disease, it is difficult to use clinical material to define the role of fibroblast CD40 in pathogenesis versus maintenance of disease. We have therefore analysed colonic mucosal fibroblasts in a mouse model of Crohn's disease, in established disease as a first step. In a previous study, in which we used this model system to define the role of transforming growth factor (TGF)-β and its major receptors in regulation of inflammation and wound-healing in the gut [24], we defined two major mesenchyme phenotypes: α-smooth muscle actin (SMA)+vimentin+RII+type I collagen myofibroblasts, which increase in prominence in colitis; and α-SMAvimentin+RII+type I collagen+ lamina propria fibroblasts.

In this study, we have derived primary fibroblast lines from normal and inflamed mouse colon, characterized them in terms of CD40 expression and their representation of in situ fibroblasts in the tissue of origin, and have examined their comparative proinflammatory potential on CD40 ligation. We demonstrate an activated, proinflammatory phenotype in fibroblasts from inflamed colon, despite their lower levels of CD40 expression, and describe potentiation of CD40 signalling by IFN-γ in inflamed cells. We propose that the CD40+ fibroblast population in chronically inflamed colonic mucosa undergoes a permanent change in phenotype which enables it to contribute directly to the chronicity of colitis.

Materials and methods

Cell lines

Fibroblast cell lines were derived by outgrowth in culture from normal Balb/c colon (normal) and colon tissue from a CD4+ CD45RBhigh-transplanted C.B-17 (congenic with Balb/c) SCID mouse (inflamed), as described previously [24]. The cells were grown in α-minimum essential medium (MEM) supplemented with heat-inactivated 10% fetal calf serum (FCS), penicillin/streptomycin (100 U/ml; 100 μg/ml), gentamicin (40 μg/ml) and 200 mM l-glutamine (all Gibco, Invitrogen, Stockholm, Sweden) in uncoated Falcon tissue culture flasks at 37°C under 5% CO2 95% air until confluent, between 5 and 7 days. Confluent cells were treated with trypsin (0·025%) and ethylenediamine tetraacetic acid (EDTA) (0·54 mM) to allow dissociation and reseeded at 1 in 20. Lines were used in the study from passages 5–25.

Flow cytometry

Normal and inflamed fibroblasts were seeded in 25 mm2 culture flasks and allowed to grow until confluent between 5 and 7 days. They were stimulated with 0, 100 or 200 U/ml of mouse recombinant IFN-γ (R&D systems, Novakemi, Stockholm, Sweden) for 24 h. After incubation, cells were treated with trypsin/EDTA, resuspended in medium and washed by centrifugation (treatment determined in preliminary experiments to have no effect on CD40 expression). Aliquots of 105 cells/100 µl were stained with fluorescein isothiocyanate (FITC)-conjugated hamster anti-mouse CD40 monoclonal antibody (MoAb) (100 μg/ml) (clone HM40-3) (BD Biosciences, Stockholm, Sweden), or with the same concentration of appropriate isotype control for 60 min at 4°C. Cells were washed with ice-cold phosphate-buffered saline (PBS) × 3 and 10 000 cells were analysed for CD40 expression using a fluorescence-activated cell sorter (FACScan) flow cytometer (Becton Dickinson, Stockholm, Sweden).

Immunohistochemistry

Cryostat sections (5–6 µm) of colon tissue from normal Balb/c mice, non-transplanted C.B-17 SCID mice and C.B-17 SCID mice 6 weeks after transfer of 4 × 105 CD4+ CD45RBhigh Balb/c spleen cells were air-dried and fixed at 4°C in 100% ice-cold acetone for 10 min. The slides were air-dried for 5 min followed by 5 min re-hydration in PBS. Slides were incubated for 30 min with 10% normal donkey serum and 10% normal goat serum in PBS for 30 min to block non-specific binding, washed three times and blocked with avidin/biotin (Vector Laboratories, Inc., Peterborough, UK). Tissues were double-stained with rat anti-mouse CD40 (20 µg/ml) (clone 3/23, Serotec, Oxford, UK), isotype control rat IgG2a and rabbit anti-mouse collagen I (1 : 100) (Novotec, Lyon, France) or rabbit IgG as control, all diluted in PBS with 2% bovine serum albumin (BSA) and incubated overnight at 4°C, followed by washing. Tissues were then incubated with biotinylated donkey anti-rat (1 : 200) (Stratech Scientific, Cambridge, UK) for 1 h at room temperature, washed and incubated with goat anti-rabbit-FITC (1 : 200) (Stratech Scientific) and streptavidin-Texas Red (Vector Laboratories Inc.) for 1 h at room temperature. The slides were washed and mounted with Vectashield (Vector Laboratories, Inc.).

Cytokine production after ligation with CD40

Cells from normal or inflamed colon were seeded at 2000 cells/well in a 96-well plate in α-MEM supplemented with 10% FCS. After confluence at 5–7 days, cells were incubated with or without 200 U/ml rmIFN-γ in 0·1% FCS medium for 24 h. Cells were washed three times with PBS and murine soluble CD40L (Peprotech EC Ltd, London, UK) was added at 0; 0·1; 1·0 and 10·0 µg/ml for 24 h in 0·1% FCS supplemented medium. After 24 h, supernatants were removed, centrifuged to remove debris and stored at −70°C.

Cytokine production was detected in cell culture supernatants using a 96-well plate assay mouse cytokine/chemokine LINCOplex kit (Linco Research, Inc., Missouri, USA) for IL-6, C-C chemokine (CCL)2, CCL3, CCL5, IL-12 and TNF-α. Standards and controls were diluted in the same medium as culture supernatant (α-MEM, 0·1% FCS). Standards, controls, blanks and samples were all run in duplicate. In brief, a 96-well filter plate was blocked with 200 µl of assay buffer on a shaker for 10 min at room temperature. Assay buffer was removed by vacuum and appropriate standards, controls, blanks, samples and mixed beads were added to a final volume of 200 µl per well. The plate was incubated on a plate shaker overnight at 4°C. After incubation, fluid was removed by vacuum filtration and the plate was washed twice. After removal of washing buffer by vacuum filtration, a detection antibody cocktail was added to each well and the plate was incubated for 60 min at room temperature. Streptavidin–phycoerythrin 25 μl was added directly to each well and incubated on a plate shaker for 30 min at room temperature. After washing and vacuum filtration, 100 µl of sheath fluid was added to each well and shaken on a plate shaker for 5 min. Samples were read and analysed using a Bio-Plex Manager system (Bio-Rad Laboratories, Sundbyberg, Sweden).

Statistics

Results were analysed by analysis of variance (anova) within groups and by the non-parametric independent sample Mann–Whitney test between normal and inflamed groups, using spss 12·0.1 software. A P-value < 0·05 was considered statistically significant.

Results

In situ CD40 expression

To define the expression of CD40 by mucosal fibroblasts in situ, tissue from four normal Balb/c, three non-transplanted SCID mice and eight transplanted SCID mice was examined on four separate occasions, the co-expression of type I collagen (a product of most mucosal fibroblasts [25,26] and therefore a putative fibroblast marker) and CD40 was determined. In non-transplanted SCID mouse colon tissue (Fig. 1a) and in normal Balb/c colon tissue (Fig. 1b), many cells expressing type I collagen co-expressed CD40 at a sufficiently high level to be detected by immunohistochemistry. There were also CD40 single-positive cells, presumably CD40+ mucosal DC and macrophages. In the inflamed colon, however (Fig. 1c), there were many CD40+ single-positive cells, indicating the known influx of activated DC in this model [27] and in human Crohn's disease [28–30], but none of the type I collagen+ cells expressed CD40 at an intensity sufficiently high to be detected by immunohistochemistry.

Figure 1.

In situ CD40 expression by fibroblasts in normal and inflamed mouse colon. Frozen sections of non-transplanted SCID mouse colon (a), Balb/c mouse colon (b) and inflamed, T cell-transplanted SCID mouse colon (c) were stained for CD40 (rat anti-mouse CD40; donkey anti-rat biotin; streptavidin–Texas red) and type I collagen [rabbit anti-mouse type I collagen; goat anti-rabbit fluorescein isothiocyanate (FITC)]. Frequent double-stained (yellow, arrowheads) CD40+ putative fibroblasts are visible in the lamina propria of non-inflamed colons (a,b), but all type I collagen+ cells in inflamed colon (c) are CD40.

Within the limitations of immunohistochemistry, this indicates that fibroblast CD40 in the mouse colon is expressed at high intensity on the SMA type I collagen+ subpopulation described previously [24], but that these CD40high fibroblasts are absent from colitic tissue.

Colonic fibroblast lines are stable and representative of colonic fibroblasts in situ

Phenotypic analysis of the cell lines, derived by outgrowth from normal and inflamed colon, showed a remarkably stable phenotype, with type I and type IV collagens, vimentin, laminin, TGF-βRI and TGF-βRII, and CD40 all retaining stable expression up to passage 25. The only significant shift in phenotype was in expression of α-SMA, which was expressed by only 30% of early passage cells, but after passage eight was expressed by the majority of cells. Figure 2 shows representative images of double staining with α-SMA, and CD40 or Thy1·2 (CD90). The results show clearly that although the majority of cells of either origin expressed α-SMA, the CD40+ cells were SMA, correlating with the in situ observation above.

Figure 2.

Phenotype of mucosal fibroblast cell lines. Representative lines from normal colon growing in chamber slides were stained for α-smooth muscle actin (SMA) (red) and either Thy1 [(a) CD 90 or (b) CD40].

Inflamed mouse colonic fibroblasts express lower levels of CD40 than normal colonic fibroblasts

The complexity of fibroblast lines from normal and inflamed colon was analysed in detail by flow cytometry on two lines derived from normal colon and two from inflamed colon. Analysis of the data showed consistent reproducibility even at passage numbers from 15 to 25. The results shown here consist of three different passages and were carried out in triplicate. Comparison of CD40 expression profiles showed comparability within normal lines and within inflamed lines, but differences between lines from normal and inflamed sources (Fig. 3). In particular, whereas IFN-γ stimulated expression of high level CD40 expression in a subpopulation of normal tissue-derived lines (Fig. 3a,c), it had no significant effect on colitis-derived lines (Fig. 3b,d). CD40 expression was examined in greater detail in two representative cell lines, one derived from normal Balb/c colon, CFN3, and one derived from moderately inflamed colon (defined using a six-parameter severity score [24] on tissue from a CD4+ CD45RBhigh-transferred SCID mouse 6 weeks after transfer), CFI5. Forward- and side-scatter analysis of CFN3 showed two distinct subpopulations, based on size and granularity, that were gated as shown in Fig. 4a. Although the CFI5 line was more homogeneous, it was still possible to distinguish two populations which were gated using the same parameters as for CFN3 (Fig. 4b). Ungated analysis of CD40 expression on CFN3 and CFI5 showed that both expressed CD40 constitutively (Fig. 5). Within this overall expression there was further complexity. Thus, CFN3 expressed CD40 constitutively at three levels of intensity, giving three definable, but overlapping populations based on peak fluorescence intensity, CD40low, CD40med and CD40high (Fig. 5a, also see Fig. 3a,c). CFI5 cells from colitis, however, expressed mainly CD40 constitutively at CD40low, with no distinct CD40med or CD40high populations (Fig. 5b). Although this was the characteristic profile of CD40 expression in colitis-derived lines, subpopulations were very occasionally observed (Fig. 3d). Independent analysis of the two gated populations in Fig. 4 showed that the CD40high subpopulation of CFN3 was present only in gate 1 (Fig. 6a), with gate 2 cells expressing mainly at CD40low level (Fig. 6b). The CD40low/med mixture of CFI5 was found in both gates (Figs 6c,d).

Figure 3.

CD40 expression by subpopulations of fibroblast lines from normal and inflamed mouse colon. Normal cells lines CFN3 (a), CFN4 (c) and inflamed lines CFI5 (b), CFI6 (d) were stained with hamster anti-mouse CD40 fluorescein isothiocyanate (FITC) [filled histogram, untreated; open histograms, treated with 200 U/ml interferon (IFN)-γ for 24 h], isotype control (dotted line), unstained control (dashed line).

Figure 4.

Forward- and side-scatter characteristics of fibroblast lines from normal and inflamed mouse colon. Normal [ (a) CFN3] and inflamed [ (b) CFI5] cell lines were removed at confluence from plates and analysed by flow cytometry for forward- and side-scatter characteristics. Gates 1 and 2 were set for major definable populations of normal cells and also applied to inflamed cells for analysis.

Figure 5.

Subpopulation analysis of CD40 expression on fibroblast lines from normal and inflamed mouse colon. Confluent cells were removed from plates and analysed by flow cytometry for CD40 expression using hamster anti-mouse CD40 fluorescein isothiocyanate (FITC) (filled histogram), isotype control (open histogram) or unstained (dotted line histogram). Major subpopulations are indicated by arrows. Results are representative of three separate experiments.

Figure 6.

Detailed CD40 expression by gated subpopulations of fibroblast lines from normal and inflamed mouse colon. Normal cells (a,b) and inflamed cells (c,d) were stained with hamster anti-mouse CD40 fluorescein isothiocyanate (FITC) (filled histogram), isotype control (open histogram) or unstained (dotted line histogram) and gated populations defined in Fig. 1[gate 1 (a,c), gate 2 (b,d)] were examined. Representative of three separate experiments.

Mucosal fibroblast CD40 is modulated differentially by IFN-γ[31]. We therefore measured its effect on expression of CD40 by subpopulations within the cell lines. Analysis of the total cell population showed that the CD40high subpopulation of CFN3 incurred increases in both peak fluorescence intensity (Fig. 3a, Table 1a) and median CD40 expression (Table 1b) after treatment with IFN-γ. Treatment induced a reduction in CD40low frequency from 26% to 11% at 200 U/ml IFN-γ and, inversely, from 24% to 42% in the CD40high subpopulation. In inflamed tissue-derived CFI5 cells, there were very slight increases in intensity of CD40 expression at all levels (Fig. 3b, Table 1d,e), but no clear shift of frequency of cells from the CD40low/med population to CD40high (Fig. 3b, Table 1f).

Table 1.  CD40 expression by subpopulations of fibroblasts from normal and inflamed mouse colon. Peak fluorescence intensity (a,d), median fluorescence intensity (b,e) and frequency of expression (c,f) of CD40 by fibroblast lines from normal (a,b,c) and inflamed (d,e,f) colon. Confluent cells were stimulated for 24 h with 0, 100 or 200 U/ml interferon (IFN)-γ. Results are representative of three experiments in triplicate.
 Peak fluorescence intensityMedian fluorescence intensityFrequency of expression
PeakCD40lowCD40medCD40highMedianCD40lowCD40medCD40high%CD40lowCD40medCD40high
Normal(a)(b)(c)
0 IFN-γ17392570 IFN-γ15493400 IFN-γ264124
100 IFN-γ1734421100 IFN-γ1851422100 IFN-γ154034
200 IFN-γ2242557200 IFN-γ1854509200 IFN-γ113742
Inflamed(d)(e)(f)
0 IFN-γ8241660 IFN-γ9442660 IFN-γ343210
100 IFN-γ1243212100 IFN-γ1044310100 IFN-γ323518
200 IFN-γ1526232200 IFN-γ1145291200 IFN-γ294119

Analysis of gated subpopulations showed that CFN3 in gate 1 were shifted from CD40med to CD40high expression by IFN-γ, particularly at the highest concentration of 200 U/ml of IFN-γ (Fig. 7a). On the other hand, CD40low and CD40med populations in gate 2 were not affected significantly (Fig. 7b). CFI5 fibroblasts showed no up-regulation of CD40 in gate 1 (Fig. 7c), but a slight up-regulation in gate 2 after 100 U/ml of IFN-γ treatment (Fig. 7d). The relative insensitivity to IFN-γ of cells from inflamed tissue was confirmed in other inflamed cell lines (Fig. 3d).

Figure 7.

Effect of interferon (IFN)-γ treatment on CD40 expression by gated subpopulations of fibroblast lines from normal and inflamed mouse colon. Cells were grown to confluence and stimulated for 24 h with 0, 100 and 200 U/ml IFN-γ, released into suspension culture and stained for CD40 expression with hamster anti-mouse CD40 fluorescein isothiocyanate (FITC) (filled histogram, untreated; open histograms, treated with 100 or 200 U/ml IFN-γ for 24 h), isotype control (dashed line), or unstained (dotted line). Gated populations, as defined in Fig. 1, were examined for normal cells [gate 1 (a), gate 2 (b)] and inflamed cells [gate 1 (c), gate 2 (d)]. Results are representative of three separate experiments.

Thus, more cells from normal colon tissue expressed CD40 at high levels than from inflamed colon, and they were more susceptible to IFN-γ induction than inflamed cells. Whereas all the IFN-γ-induced effects on CD40 expression in cells from normal tissue were confined to cells in gate 1, inflamed cells in both gates showed a slight up-regulation in CD40 expression. In summary, although cell lines from both tissue sources expressed CD40 constitutively, only a subpopulation of normal tissue-derived cells expressed at a high level. Also, normal tissue-derived cells were more responsive to IFN-γ up-regulation of CD40, in particular by shifting small cells from medium to high CD40 expression. Cells derived from inflamed tissue were less responsive to IFN-γ treatment in terms of CD40 expression.

Inflamed mucosal fibroblasts have enhanced proinflammatory responses to CD40L, despite lower levels of CD40 expression

The demonstration of CD40 on both normal and inflamed colon fibroblasts caused us to question the responses of these cells to ligation with soluble CD40L (sCD40L). In addition, the lower level of CD40 expression in inflamed tissue-derived fibroblasts suggested altered susceptibility to CD40 ligation in an inflammatory milieu. We tested this hypothesis by in vitro treatment of the cell lines with sCD40L, with or without pretreatment with IFN-γ.

Both cell lines were stimulated with 200 U/ml of IFN-γ for 24 h, or left untreated, before stimulation with sCD40L at concentrations of 0, 0·1, 1·0 or 10 µg/ml for a further 24 h. Supernatants were analysed for IL-6, CCL2, CCL3, CCL5, TNF-α and IL-12. The results (Fig. 8) show that, in general, CFN3 had lower constitutive cytokine secretion and lower responses to CD40 ligation − surprisingly, given their higher level of CD40 expression.

Figure 8.

Secretion of proinflammatory cytokines and chemokines by fibroblast lines from normal and inflamed mouse colon and effects of CD40 ligation, with and without pretreatment with interferon (IFN)-γ. Cells were grown to confluence and stimulated with (+, solid bars) or without (–, open bars) IFN-γ (200 U/ml) and with or without sCD40L (0; 0·1, 1·0 and 10 μg/ml) for 24 h. Secretion of interleukin (IL)-6, CCL2, 3 and 5 was measured simultaneously in cell supernatants by multiplex enzyme-linked immunosorbent assay. (a) Interleukin (IL)-6 synthesis by normal and inflamed fibroblasts; (b) CCL5 synthesis by normal and inflamed fibroblasts; (c) CCL2 synthesis by normal and inflamed fibroblasts; (d) CCL3 synthesis by normal and inflamed fibroblasts. Bars represent means ± s.d. of duplicate values from three separate experiments. *P < 0·05, **P < 0·01, ***P < 0·001; n.s., not significantly different.

Constitutive secretion of IL-6 by CFN3 cells (Fig. 8a) was very low and was only enhanced significantly at the highest CD40L concentration and after pretreatment with IFN-γ. However, CFI5 cells secreted higher levels of IL-6 constitutively and CD40 ligation had no significant effect without pretreatment with IFN-γ, which increased levels of IL-6 in a CD40L dose-dependent manner (Fig. 8a).

CCL5 expression showed similar kinetics to IL-6, but levels of expression were all very low. Thus, CFN3 cells did not show constitutive expression, and were responsive to CD40 ligation only at the highest concentration of sCD40L, with no significant added effect of IFN-γ treatment. CFI5 cells showed no constitutive secretion and responded only at the highest CD40L concentration. IFN-γ alone significantly increased CCL5 synthesis and synergized with CD40L to produce CCL5 in a CD40L dose-dependent manner (Fig. 8b).

The kinetics of CCL2 secretion were different. In this case, CFN3 cells secreted CCL2 constitutively, although at very low levels, and reacted only to the highest concentration of CD40L, independently of pretreatment with IFN-γ. CFI5 cells, on the other hand, responded differently. In this case, CCL2 was secreted constitutively at levels similar to those from CD40L-treated CFN3 cells. This constitutive secretion was not affected significantly by CD40 ligation, but was amplified significantly by IFN-γ pretreatment, although without apparent CD40 dose-dependency (Fig. 8c).

The CCL3 response differed from these, in that both cell lines behaved identically: there was no constitutive secretion; there was a response to the highest concentration of CD40 ligation; and there was a significant potentiation by IFN-γ treatment only in CFN3 cells (Fig. 8d).

Synthesis of TNF-α or IL-12 by either cell line was not detected (data not shown).

In summary, normal tissue-derived fibroblasts did not constitutively synthesize IL-6, CCL3 or CCL5 and constitutively produced only very low levels of CCL2. Secretion of CCL2 and CCL5 were insensitive to CD40 ligation; and there was no added effect of IFN-γ pretreatment in contrast to the IL-6 and CCL3 responses, which were CD40L-responsive and amplified by IFN-γ. In contrast, inflamed tissue-derived fibroblasts synthesized constitutively significantly higher levels of IL-6 and CCL2 than normal cells; showed significantly greater IL-6, CCL2 and CCL5 responses to CD40 ligation; and demonstrated synergistic amplification of these responses by IFN-γ pretreatment. Interestingly, CCL3 responses were almost identical in normal and inflamed fibroblasts.

Discussion

Fibroblasts were considered previously to be concerned only with the regulation of the extracellular matrix. This assumption has been challenged recently and several studies have shown that fibroblasts in a variety of tissues − lung, spleen, breast and colon − can produce proinflammatory cytokines. The exact role of the fibroblast in inflammation has not been clarified fully, but there is evidence that fibroblasts from inflamed tissue are phenotypically different from fibroblasts from normal tissue and also that their cytokine profile is distinct [32]. It has been suggested that fibroblasts may be involved in the regulation of the switch from acute to chronic persistent inflammation through the involvement of the transcription factor NFκB [33]. A principal signalling pathway for activation of NFκB is through the receptor CD40, which has been shown to be important in fibroblast activation. Thus, it has been shown [11] that interaction between CD40 and its ligand CD154, in fibroblasts taken from normal or Crohn's disease (CD) colon, results in an augmented inflammatory response and, in intestinal lesions of CD, cells expressing CD40 accumulated at the site of inflammation [29].

It is well recognized that CD40 ligation in a wide range of cell types results in proinflammatory signalling [34–36]. However, the mechanisms of such signalling in fibroblasts have not been established. As this is difficult to control in recently isolated cells, we have used fibroblast lines established from normal mouse colon and from inflamed colon from the CD4+ T cell-transplanted SCID mouse model of colitis.

The results presented here show, for the first time, the existence of a population of colonic fibroblasts from inflamed tissue that have acquired the capacity to produce higher concentrations of proinflammatory cytokines/chemokines compared to normal cells. These populations of inflamed cells were also to able increase their production of cytokines/chemokines through CD40/CD40L interactions and stimulation with IFN-γ, despite lower expression of CD40 compared to normal cells.

Flow cytometry analyses of our cell lines, combined with in situ staining, showed that normal colonic fibroblasts express CD40 constitutively. This is consistent with results from other laboratories, which have shown that normal human fibroblasts from tissues including gut also express CD40 constitutively [3,31,37,38]. Expression of the receptor by the normal tissue-derived cell lines was complex, as it shows apparent subpopulations with distinct densities of expression of CD40. This probably reflects the situation within normal mucosa, in which we have described previously complex subpopulations of fibroblasts in the colonic lamina propria [24]. Although we have not excluded the possibility that the isolation procedure alters CD40 expression on colonic fibroblasts we have demonstrated, within the constraints of acceptable methodology, that our model normal colon cell lines contain CD40high fibroblasts, which are also present in the parent tissue; and that our colitis-derived lines do not, and these are also absent from inflamed colon. These data correlate with the complex responsiveness of CD40 induction by IFN-γ reported by Gelbmann et al. [11] in some lines from normal human colon, but more uniform responsiveness to IFN-γ by other lines in their study and, as reported by Vogel et al. [5], presumably reflects the relative ease by which the complex mucosal fibroblast populations can be isolated from mouse colon versus human colon. Our previous work [24] shows that mouse colonic mucosal fibroblasts have complexity of phenotype in situ and the ex vivo expression of CD40 and its modulation by IFN-γ probably represent this complexity fairly. Inflamed fibroblasts also expressed CD40 constitutively, but at a lower level of receptor density. Given the proinflammatory milieu in which these cells reside in situ, particularly rich in IFN-γ[39] which is a known inducer of CD40 [10], this was surprising. Although studies of human intestinal mucosal fibroblasts [5,11] have demonstrated CD40 expression by fibroblast lines from normal tissue, there have been no other comparative studies of CD40 expression on lines from normal and inflamed mucosa. However, further analysis showed that induction of CD40 by IFN-γ on inflamed cells was much less efficient than on fibroblasts from normal tissue. We speculate that there are differences in signalling from the IFN-γR which result in differential CD40 transcription between fibroblasts from normal versus fibroblasts from inflamed tissue.

CD40 ligation in both sets of cell lines triggered proinflammatory signalling, as expected, and as shown in human studies [11]. However, there were crucial and surprising differences between cells from normal colon and cells from inflamed colon. Thus, cells from inflamed colon showed a constitutively activated phenotype, as evidenced by their constitutive secretion of higher levels of IL-6 and CCL2 than in normal cells. This is consistent with our previous observation of mucosal fibroblast activation in this model [24]. Furthermore, despite the relatively low level of expression of CD40 on inflamed cells, they synthesized higher levels of IL-6, CCL2 and CCL5 in response to CD40L than normal cells and these CD40 responses were further amplified by IFN-γ. This amplification of CD40 cytokine responses by mucosal fibroblasts has also been shown in human studies [5,11], but the synergistic action of IFN-γ on CD40 responses in inflamed tissue fibroblasts has not previously been demonstrated. CCL3 secretion was similar for cell lines from both sources, with no constitutive secretion, stimulation at high levels of CD40L and little evidence of IFN-γ potentiation. These results suggest that different proinflammatory signalling pathways − predominantly NFκB for IL-6, CCL2 and CCL5 and nuclear factor activated T cell (NFAT) for CCL3 − are affected in different ways by CD40L and IFN-γ stimulation in cells which have been preconditioned by a Th1 inflammatory milieu. This points future studies towards differential transactivation of proinflammatory transcription factors, as has been shown for C/EBPβ and NFκB [40–42], and indicates a possible mechanism whereby fibroblasts in the gut may contribute to the persistence of inflammatory disease. Overall, levels of proinflammatory cytokine/chemokine secretion by the fibroblast lines were in line with, or in excess of, levels reported for activated mouse dendritic cells − for example, IL-6 [43], CCL3 [44], supporting our view that mucosal fibroblasts may be crucial in directing chronic inflammatory pathology in the mucosa.

In conclusion, the evidence presented here shows the existence within chronic mucosal inflammation of subpopulations of fibroblasts with an activated phenotype, permitting them to produce cytokines/chemokines constitutively and to be more sensitive to CD40L and IFN-γ in the milieu, thereby increasing the risk of chronicity and relapse. Use of such cell lines derived from model colitis will permit further exploration of the regulation of the pathogenesis and chronicity of mucosal inflammation.

Acknowledgements

This work was supported by Vetenskapsrådet no. K2002–06X-14233, kungl och Hvitfeldtska Stiftelsen, Sahlgrenska Universitetssjukhuset stiftelse, Wilhelm och Martina Lundgrens Vetenskapsfonder, Stiftelsen professor Nanna Svartz fond and Wellcome Trust Biomedical Research Collaboration grant no. AL069896.

Ancillary