Chronic colitis induces expression of β-defensins in murine intestinal epithelial cells

Abstract

Anti-microbial peptides are important effectors in innate immunity. In the gut they defend against pathogens, shape the commensal microbiota and probably control intestinal homeostasis. Ulcerative colitis (UC), but not Crohn's disease, shows increased expression of inducible β-defensins (hBD-2, hBD-3 and hBD-4) in colonic epithelial cells. Does inducible defensin production precede the chronic intestinal inflammation characteristic of UC, or is it a consequence of the T cell-driven chronic inflammation? The aim was to analyse defensin mRNA and protein expression in colonic epithelial cells in two colitis mouse models resembling UC, the interleukin (IL)-2^−/− mouse and the dextran sulphate sodium (DSS)-induced colitis mouse. Defensin mRNA was assayed by in situ hybridization and quantitative real-time reverse transcription–polymerase chain reaction (RT–PCR). Defensin peptide was assayed by immunohistochemistry. Mouse β-defensin 3 (mBD-3, orthologue to hBD-2) was up-regulated strongly in colonic epithelium of 15-week-old IL-2^−/− mice and DSS-induced colitis mice with chronic bowel inflammation, but not in apparently healthy IL-2^−/− 5-week-old mice, IL-2^+/− 15-week-old mice or in acute stage DSS mice. Up-regulation was seen both at the mRNA- and at the protein level (only mBD-3 investigated). IL-17, but not several other cytokines, including interferon (IFN)-γ, induced mBD-3 mRNA expression in mouse colon carcinoma cells. The mRNA expression level of the constitutively expressed α-defensin, cryptdin-4, was up-regulated marginally in acute stage DSS-colitis mice and in IL-2^−/− mice before signs of colitis. Inducible β-defensin expression in colonic epithelium is the consequence of the chronic bowel inflammation caused by activated T cells releasing cytokines including IL-17.

Introduction

The colitis in interleukin-2 gene-inactivated mice (IL-2^−/− mice) that develops in all animals that are not struck by early lethal anaemia is highly reminiscent of human ulcerative colitis (UC) with diarrhoea, rectal bleeding and general wasting [1]. It is a T cell-dependent disease, and the development of colitis is reliant on antigens from the microbial flora. Thus, mice raised under germ-free conditions do not develop colitis or show a much milder form of the disease [1–3]. A second colitis model with striking immunological and histopathological similarities to UC is the chronic dextran sulphate sodium (DSS) model in C57BL/6 mice, developed by Melgar and associates [4]. The chronically inflamed colon contains dense infiltrates of mononuclear cells including plasma cells, irregular epithelial structure and few goblet cells. Moreover, a prominent cytokine production, including IL-1β, IL-12, IL-17 and interferon (IFN)-γ, is seen in the inflamed colon tissue. However, little is known about the epithelial innate immune reactions in the two colitis models.

Anti-microbial peptides (AMPs) are important effectors in innate immunity [5–8]. Three AMP classes have been identified: α- and β-defensins and cathelicidin. The intestinal mucosa of humans and mice produces several α- and β-defensins and one cathelicidin, LL-37/cathelicidin-related anti-microbial peptide (CRAMP). While α-defensins are produced by specialized epithelial cells, the Paneth cells present at the bottom of the crypts of Lieberkühn in the small intestine, β-defensins and cathelicidin are produced by the ordinary columnar epithelial cells both in small and large intestine. Humans express two epithelial α-defensins, HD-5 and HD-6. Furthermore, humans, but not mice, express α-defensins in neutrophils, termed human neutrophil proteins (HNP) 1–4. In the mouse there are six abundantly expressed α-defensins, termed cryptdins (Crp 1–6). The α-defensins are stored as inactive propeptides in secretory granules of Paneth cells or neutrophils and need proteolytic cleavage in order to gain anti-microbial activity. Cryptdins are activated by metalloproteinase 7 within the granules, while HD-5 and HD-6 are activated by trypsin after their release. α-Defensins are expressed constitutively in high amounts. They play a role in the defence against pathogenic bacteria, influence the balance in the microbial ecosystem and probably control intestinal homeostasis [9]. Both humans and mice express a relatively large number of β-defensins, of which six human (hBD1-6) and five mouse (mBD1-5) defensins have been isolated and characterized. hBD-1 and its orthologue mBD-1 are expressed constitutively, while other β-defensins including hBD-2 and its orthologue mBD-3 are induced by bacteria and cytokines. mBD-3, like certain human β-defensins, possesses the ability to attract immune cells chemotactically [10].

UC, in contrast to Crohn's colitis, is associated with induction of hBD-2, hBD-3 and hBD-4 in the colonic epithelium [11–14]. In UC there is also ectopic development of α-defensin-producing Paneth cells in colon [11,14,15]. Is the increased defensin production part of the aetiology or a consequence of an established inflammation caused by activated T cells?

To gain insight into the role of defensins in UC, we have studied the expression of the inducible β-defensins mBD-3 and mBD-4 and the constitutively expressed α-defensin Crp-4 in relation to the disease process in two mouse models, the IL-2^−/− colitis mouse and the DSS-induced colitis mouse.

Materials and methods

Mice

Mice were kept in a temperature-controlled, air-conditioned room with an artificial 12-h light/dark cycle, fed a standard pellet diet and drinking water ad libitum.

IL-2^+/− mice on a C57BL/6 background were obtained from Jackson Laboratories (Bar Harbor, ME, USA). A breeding colony was established. The litters were genotyped for IL-2 gene disruption by polymerase chain reaction (PCR) analysis of DNA, as described previously, and classified as homozygous knock-out (IL-2^−/−), heterozygous (IL-2^+/−) and wild-type (IL-2^+/+) [16].

Groups with C57BL/6 mice (Bomholtgard Breeding and Research Centre, Ry, Denmark), matched for sex and age (8–15 weeks), were either administered 3·5% weight/volume DSS (mol wt 44 000; TDB Consultancy, Sweden) in the drinking water ad libitum for 6 days and then killed (acute colitis) or administered 3·5% weight/volume DSS in their drinking water ad libitum for 5 days and then normal drinking water ad libitum for 3 weeks, at which time the mice were killed (chronic colitis). The latter procedure is a well-documented regimen to establish a colitis in C57BL/6 mice that greatly resembles UC [4]. The control group was left untreated for 4 weeks.

Isolation of epithelial cells

Epithelial cells were isolated from colon and small intestine tissue specimens by a procedure adopted from isolation of human intestinal epithelial cells [17]. Briefly, the specimen was opened longitudinally, washed, cut into small pieces and treated with 1·5 mM dithiothreitol at 20°C for 20 min with gentle shaking. After centrifugation, the pellet was suspended in HEPES-buffered Earl's balanced salt solution (EBSS; Invitrogen, Carlsbad, CA, USA) with 10% fetal calf serum (FCS) and vortexed for 4 min. The supernatant containing free luminal epithelial cells and intraepithelial lymphocytes was collected. The sediments of tissue pieces were treated twice with a mixture of collagenase type II and IV (290 U/ml; Worthington, Freehold, NJ, USA) at 37°C for 40 min. Liberated cells were passed through a stainless steel grid (60-mesh) resulting in a cell fraction containing epithelial cells, lamina propria leukocytes and stromal cells. The two cell fractions were subjected to density gradient centrifugation with 70, 40 and 20% Percoll (Pharmacia, Uppsala, Sweden) in EBSS. Epithelial cells, enriched by the interface between 40 and 20% Percoll, were collected from both cell fractions and pooled. The cells were then treated with sheep anti-rat immunoglobulin (Ig)G-coated paramagnetic beads (Invitrogen Dynal, Oslo, Norway), charged with a rat anti-mouse CD45 monoclonal antibody (clone IBL-3/16, IgG1; AbD Serotec, Kidlington, Oxford, UK) and a magnet to remove all types of leucocytes. Unbound cells were collected, washed twice with HEPES-buffered EBSS with 5% FCS, washed once with RNase-free 0·15 M phosphate-buffered saline (PBS; pH 7·2) and kept frozen at −80°C in 25 mM sodium citrate (pH 7·0) containing 4 M guanidine thiocyanate, 0·5% sarcocyl and 0·1 M 2-mercaptoethanol until RNA extraction.

In vitro induction of defensin by cytokines

The colon cancer cell line CMT93 (DZMC, Braunschweig, Germany) was grown in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% FCS, penicillin (100 IU/ml), streptomycin (100 µg/ml) and l-glutamine (2 mmol/l) at 37°C in 5% CO₂. Cells were trypsinized, washed in DMEM, seeded into 12-well tissue culture plates at 0·25 × 10⁶ cells/ml and grown to confluence. Cells were incubated for 24 h with the following recombinant murine cytokines: tumour necrosis factor (TNF)-α, IL-17, IFN-γ, IL-1β and IL-6 (all from PromoKine, Heidelberg, Germany) or with Escherichia coli lipopolysaccharide (LPS)_. Cytokines were tested in three or more experiments with serial dilutions collectively covering a concentration range of 20–720 ng/ml and LPS was tested in serial dilutions from 9 to 1 µg/ml. Cells were harvested by trypsinization, washed, counted and frozen for RNA extraction.

RNA extraction and real-time quantitative reverse transcription–polymerase chain reaction (qRT–PCR)

Total RNA was extracted from primary intestinal epithelial cells and CMT93 cells by the acid guanidinium thiocyanate–phenol–chloroform method [18]. The ribonuclease inhibitor rRNasin (Promega, Madison, WI, USA) was added to each sample (1 U/ml) and samples were stored at −80°C until use.

A real-time qRT–PCR assay for mBD-3 mRNA was constructed using the Taqman-EZ™ technology (Perkin-Elmer Applied Biosystems, Sundbyberg, Sweden). Specific primers were placed in different exons and an internal probe labelled with 6-carboxyfluorescein as 5′-reporter dye was placed over the exon boundary within the amplicon to prevent amplification of genomic DNA. The sequences were: 5′-primer, 5′-CTTTGCATTTCTCCTGGTGC-3′; 3′-primer, 5′-GCCTCCTTTCCTCAAACAACT-3′ and probe 5′-CTGTCTCCACCTGCAGCTTTTAGCAAAA-3′. An ABI Prism 7700 Sequence Detection System (Perkin-Elmer Applied Biosystems) was used to monitor release of reporter dye. Samples were analysed in triplicate and concentration of mRNA copies was determined from an external RNA copy standard curve. The assay showed a linear relation between concentration of standard RNA and number of PCR cycles over a range of more than 5 logs. For Crp-4 mRNA determinations a commercially available assay was used (Applied Biosystems). RNA copy standards for both assays were prepared from the amplicons as described previously [11]. All samples were analysed for their content of the housekeeping gene 18S rRNA using the primers and probe supplied by the manufacturer (Applied Biosystems) and the results were expressed as mBD-3 or Crp-4 mRNA copies per unit of 18S rRNA, corresponding approximately to copies/epithelial cell [11]. Verification of PCR products in the two assays was performed by cloning and sequencing as described [11].

In situ hybridization

Anti-sense and sense RNA probes were prepared from cloned mBD-3 and mBD-4 RT–PCR products using the digoxigenin (DIG) RNA labelling kit (Roche Diagnostics, Mannheim, Germany). The sizes of the probes were 165 base pairs (bp) and 196 bp for mBD-3 and mBD-4, respectively. The primer sequences were: 5′-primer; 5′-CCTTCTCTTTGCATTTCTCCTGG-3′ and 3′primer; 5′-CATTTGAGGAAAGGAACTCCACAA-3′ (mBD-3 probe) and 5′-primer; 5′-TCTTCACATTTCTCCTGGTGCTGCTG-3′; 3′-primer 5′-TTGCTGGTTCTTCATCTTTTTATCT-3′ (mBD-4 probe). In situ hybridization was performed on 10-µm thick cryosections fixed in 4% paraformaldehyde using the DIG-labelled RNA probes and alkaline phosphatase-labelled Fab fragments of sheep anti-DIG antibody (Roche Diagnostics, Basel, Switzerland), as described previously [19]. Hybridization was visualized by incubation with nitroblue tetrazolium, 5-bromo-4-chloro-3-indolyl phosphate and levamisole in 100 mM Tris-HCl (pH 9·5), 100 mM NaCl, 50 mM MgCl₂. The corresponding sense DIG-labelled RNA probes were used as negative controls.

Immunohistochemistry

Fresh tissue samples were snap-frozen in liquid nitrogen and stored at −80°C. Ten-µm-thick sections were air-dried, fixed in 4% paraformaldehyde in phosphate buffer (pH 7·4) for 15 min and then rinsed in cold 0·02 M PBS (pH 7·2). Free aldehyde groups were blocked with 0·1 M glycine in PBS for 10 min at room temperature. Sections were then subjected to antigen retrieval by boiling three times in 10 mM sodium citrate solution supplemented with 0·05% Tween 20 for 5 min using a microwave oven. Sections were cooled at room temperature and incubated with PBS, 1% bovine serum albumin (BSA) and 0·05% saponin for 1 h. Affinity purified rabbit anti-mBD-3 IgG (Alpha Diagnostic, San Antonio, TX, USA) diluted in PBS/saponin was added and sections were incubated at 4°C overnight. Sections were then stained with peroxidase-labelled anti-rabbit Ig using the Immpress reagent kit (Vector Laboratories, Burlingame, CA, USA). The brown-coloured end product was developed using 0·05% 3,3′-diaminobenzidine tetrahydrochoride and 0·03% H₂O₂ in 0·05 M Tris HCl buffer (pH 7·6) and counterstained with methyl green. Sections incubated with the IgG fraction of normal rabbit serum (Dakocytometion, Burlingame, CA, USA) served as negative controls.

Statistical analysis

Results were checked for Gaussian distribution by the Kologorov–Smirnov normality test. Statistical significance of differences between experimental groups was performed using one-way analysis of variance (anova), with Bonferroni multiple-comparison post-test for data sets with Gaussian distribution and Kruskal–Wallis non-parametric anova with Dunn's multiple-comparison post-test if one or more data sets did not show Gaussian distribution (Prism 5, GraphPad Software, San Diego, CA, USA). Two-sided analyses were used throughout and P-values <0·05 were considered statistically significant.

Ethical considerations

The local ethics committee on animal experiments, Northern Sweden, approved the study.

Results

Fifteen weeks or older IL-2^−/− mice (IL-2^−/− 15-week-old) displayed pronounced clinical colitis symptoms, while young IL-2^−/− mice (IL-2^−/− 5-week-old) did not. IL-2^+/− and IL-2^+/+ mice were also symptom-free at the age when colitis had established in the IL-2^−/− mice. DSS-treated mice showed weight loss, rectal bleeding, loose stools and diarrhoea in the acute phase and loose stools in the chronic phase, as described previously [4]. Examination of haematoxylin and eosin-stained colon tissue sections of both the IL-2^−/− and chronic DSS colitis mice confirmed the induction of chronic inflammation (data not shown).

Chronic colitis leads to up-regulation of β-defensins in both mouse models

We first investigated whether expression of β-defensin mRNA could be detected in the epithelium of severely sick IL-2^−/− mice. Figure 1a and b shows an example of in situ hybridization with anti-sense and sense probes for mBD-3 mRNA on colons of IL-2^−/− 15-week-old mice. Colon epithelium cells showed positive staining with the anti-sense probe but not with the sense probe. Scattered lamina propria cells were also stained with the mBD-3 anti-sense probe (arrows in Fig. 1a). The mBD-3 anti-sense probe was also used to stain colons from IL-2^−/− 5-week-old mice and 15-week-old IL-2^+/− and IL-2^+/+ 15-week-old mice. No, or only very faint, epithelium staining was seen (data not shown). Additionally, mBD-4 mRNA anti-sense and sense probes were utilized to stain colonic epithelium of IL-2^−/− 15-week-old mice. A weak but specific staining was obtained with anti-sense probe (data not shown).

Fig. 1.

Mouse β-defensin 3 (mBD-3) mRNA (a, b) and peptide (c–f) is expressed in the chronically inflamed colonic mucosa of interleukin (IL)-2^−/− 15-week-old mice (a–d) and chronic dextran sulphate sodium (DSS)-induced colitis mice (e, f) as revealed by in situ hybridization and immunohistochemistry, respectively. (a) In situ hybridization with mBD-3 anti-sense probe; arrowheads indicate epithelial cell staining. Arrows indicate stained cells in the lamina propria. (b) In situ hybridization with negative control mBD-3 sense probe; (c) anti-mBD-3 peptide immunoperoxidase staining of IL-2^−/− 15-week-old mice colon; (d) anti-mBD-3 immunoperoxidase staining of IL-2^+/+ mouse colon; (e) anti-mBD-3 immunoperoxidase staining of colon from a mouse with chronic DSS-induced colitis; (f) negative immunohistochemistry control, colon from mouse with chronic DSS-induced colitis incubated with immunoglobulin G fraction of normal rabbit serum. Arrows in (c) and (e) indicate epithelial staining of colonic tissues. Original magnification: (a,b,c) × 252; (d,e,f) × 126.

Because mBD-3 mRNA was expressed more strongly than mBD-4 mRNA we developed a specific real-time qRT–PCR assay for this β-defensin mRNA and quantified mBD-3 mRNA in isolated colon epithelial cells from IL-2^−/− 15-week-old, IL-2^−/− 5-week-old, IL-2^+/− 15-week-old, IL-2^+/+ 15-week-old and 5-week-old old IL-2^+/+ mice and from DSS mice in acute and chronic disease states and sham-treated wild-type mice (Fig. 2a and b). mBD-3 mRNA was detected only occasionally in 15-week-old control mice, and then at very low levels, and not at all in 5-week-old control mice. In contrast, in severely ill IL-2^−/− 15-week-old mice and in mice with chronic DSS colitis, mBD-3 mRNA was expressed at elevated levels. Compared to controls, these differences were highly significant. IL-2^−/− 5-week-old mice, i.e. mice without clinical symptoms, did not show significantly increased mBD-3 mRNA levels compared to controls. The same was true for DSS mice in an acute disease state. mBD-3 mRNA was not expressed in epithelial cells from apparently healthy small intestine of the severely sick IL-2^−/− 15-week-old colitis mice nor in small intestinal epithelial cells from control mice (Fig. 2c).

Fig. 2.

Mouse β-defensin 3 (mBD-3) (a,b,c) and cryptdin-4 (Crp-4) (d,e,f) mRNA levels in isolated epithelial cells from colonic (a,b,d,e) and small intestinal (c,f) mucosa of interleukin (IL)-2^−/− 15-week-old mice, IL-2^−/− 5-week-old mice and IL-2^+/− 15-week-old mice (a,c,d,f) and acute and chronic DSS-induced colitis mice (b,e) and 5- and 15-week-old control mice (indicated by IL-2^+/+ 5-week-old mice and IL-2^+/+ 15-week-old mice or control, respectively; a–f), as determined by real-time quantitative reverse transcription–polymerase chain reaction. Each dot represents one mouse. Horizontal bars depict median values. Statistically significant differences are indicated.

To investigate whether mBD-3 was expressed at the protein level, immunoadsorbent purified polyclonal rabbit anti-mBD-3 antibody was used to stain colon tissue sections from the colitis mice. Well-defined epithelial cell staining was seen in colons from IL-2^−/− 15-week-old mice and chronic DSS colitis mice (Fig. 1c and e). In the wild-type mice staining was barely detectable (Fig. 1d). The antibody control showed no staining (Fig. 1f).

IL-17 stimulates mBD-3 expression in a murine colon carcinoma cell line

We investigated whether cytokines would induce mBD-3 mRNA expression in a murine colon carcinoma cell line, CMT93. The proinflammatory cytokines, TNF-α, IL-17, IFN-γ, IL-1β and IL-6 were studied. Stimulation of cells with E. coli LPS served as a positive control. Figure 3 shows the results of three independent experiments. Contrary to normal mouse colon epithelium the cell line expressed high levels of mBD-3 mRNA in the absence of any added stimulant. IL-17 induced a strong, significant (P < 0·01) up-regulation of mBD-3 mRNA, while no such effect was seen with TNF-α, IFN-γ, IL-1β or IL-6 (Fig. 3, and data not shown).

Fig. 3.

Interleukin (IL)-17 but not tumour necrosis factor (TNF)-α or interferon (IFN)-γ induces significant increase in mouse β-defensin 3 (mBD-3) mRNA levels in the mouse colon carcinoma cell line, CMT93. CMT93 cells were incubated with serial dilutions of the three cytokines, Escherichia coli-derived lipopolysaccharide (3 µg/ml) or medium alone (–) and mBD-3 mRNA levels were determined by real-time quantitative reverse transcription–polymerase chain reaction. Bars indicate mean values of three experiments and whiskers, 1 standard deviation.

Expression of the α-defensin, Crp-4 mRNA, in colon and small intestine epithelium is affected only marginally by inflammation

In mice, Crp-4 mRNA is expressed at high levels in the normal healthy small intestine (Fig. 2f). However, in apparent contrast to human α-defensins, Crp-4 is also expressed at significant, albeit lower, levels in normal mouse colon (Fig. 2d and e). Sequencing of the PCR product from five normal mouse colon epithelial cell samples showed that it indeed was Crp-4 mRNA. On average, the Crp-4 expression levels were higher in 15-week-old compared to 5-week-old control mice, although this difference did not reach statistical significance (Fig. 2d). We also noted that the expression levels in colon of the two groups of 15-week-old control mice used here differed from each other, although the mice had the same genetic background, i.e. C57BL/6 (Fig. 2d and e). However, there was a difference in husbandry between the groups.

Analysis of colonic epithelial cells from diseased IL-2^−/− 15-week-old mice and the clinically healthy, IL-2^−/− 5-week-old mice and from acute and chronic DSS colitis mice revealed that Crp-4 mRNA was elevated significantly in IL-2^−/− 5-week-old compared to IL-2^+/+ 15-week-old mice and also slightly elevated in IL-2^−/− 15-week-old mice and mice with acute DSS colitis (Fig. 2d and e). No increase was seen in mice with DSS-induced chronic colitis; instead, there was a tendency for a decrease from the acute to the chronic stage (Fig. 2e). Similarly, the Crp-4 levels in IL-2^−/− 15-week-old mice tended to be lower than in IL-2^−/− 5-week-old mice (Fig. 2d). Collectively, these results suggest that establishment of chronic inflammation is preceded by a transient production of α-defensin in colon.

Discussion

The main finding in this study is that inducible β-defensin mBD-3 is up-regulated strongly in colon epithelial cells of mice with chronic, but not acute, colitis. It is perhaps surprising that no up-regulation is seen in the acute stage of DSS-induced colitis, considering that exposure to microorganisms in colon must be substantial. Because the analyses were performed on isolated epithelial cells, it can be excluded that loss of epithelium at the acute stage is responsible for the low mBD-3 mRNA values. Moreover, Crp-4 mRNA levels were actually increased at the acute stage. We favour the explanation that mBD-3 is up-regulated as a consequence of activation of certain T cell populations as colitis progresses into a chronic state. In the chronic DSS-induced colitis model of Melgar and co-workers [4], it was shown that IL-17 protein levels increased gradually as the disease progressed, reaching its highest level when the experiment was terminated, i.e. at 33 days. Similarly, IL-1β, active IL-12 and IFN-γ increased with time. As our analysis of defensin mRNA expression in mouse colon carcinoma cells demonstrated that treatment with IL-17, but not IL-1β, TNF-α or IFN-γ, induced mBD-3 it is likely that IL-17 is responsible for synthesis of the β-defensin by epithelial cells in vivo. T helper type 17 (Th17) cells and/or Th1/Th17 cells in the inflamed mucosa probably produce this cytokine. Such IL-17-producing T cell types are increased in the inflamed mucosa of UC colon [20]. Moreover, it has been shown that IL-17, with or without IL-22, up-regulate hBD-2 strongly in human keratinocytes [21].

Expression of mBD-3 in epithelial cells may be the consequence of signal transduction via the nuclear factor (NF)-κB transcription system controlling numerous innate and adaptive immune response molecules [22]. The 5′ flanking region of the mBD-3 gene and the hBD-2 gene contain a binding site for NF-κB [23]. The NF-κB transcription system can be activated directly through Toll-like receptors (TLRs) on epithelial cells or indirectly through cytokines/chemokines produced by immune cells. Activation of epithelial cells through the interaction of microorganisms with TLR on these cells is expected to be a fast process, therefore one would expect that induction of defensin expression should have occurred during the 5 days period of the acute DSS model. However, this was not observed. In contrast, we observed that during the acute phase (day 2) of cholera and enterotoxigenic E. coli (ETEC) infection in humans, hBD-2 is up-regulated strongly (Shirin et al., submitted for publication). It is possible that commensals in the DSS-treated mouse intestine do not have the ability to act directly on epithelial cells, in contrast to the pathogens. Whether the NF-κB system is involved in the indirect induction of mBD-3 is unclear. The finding that hBD-2 is up-regulated in colon from patients with UC parallels the result with mBD-3 obtained here, and indicates that in UC hBD-2 induction is also an indirect effect of IL-17-producing T cells, possibly in an attempt to cope with microbial insult in the diseased colon [11,13,20].

One additional similarity between UC and the IL-2^−/− model was that hBD-3 and hBD-4 and mBD-4, all three β-defensins that lack NF-kB sites [10,24,25], are up-regulated in the epithelial cells of the inflamed colon mucosa in UC patients [12] and mice with chronic colitis. These results underscore the notion that other pathways than via NF-κB are also involved in the regulation of defensin production in colitis.

We noted that mBD-3-positive cells were detected in the lamina propria of IL-2^−/− mice with chronic inflammation. Whether these are β-defensin-producing plasma cells, as we observed in the colonic mucosa of UC patients [26], is currently under investigation.

The results for the mRNA levels of the α-defensin, Crp-4, in the mouse models differ from those obtained in UC analysing the α-defensins HD-5 and HD-6. In UC both α-defensin mRNAs were up-regulated in diseased colon tissue due to that metaplastic Paneth cells were induced [11,15]. This phenomenon probably did not occur in the two colitis models, as the Crp-4 mRNA levels were not increased significantly in mice with chronic colitis if compared with the same-age control mice. Interestingly, there was a slight increase at the acute phase in DSS-colitis and in 5-week-old IL-2^−/− mice, indicating a transient ectopic production of α-defensin possibly contributing to the establishment of chronic disease. HD-5 was shown to be chemotactic for macrophages, memory and naive T cells and mast cells [27]. It is tempting to speculate that also Crp-4 is chemotactic for these cell types and that increased Crp-4 production could be a contributing factor to the recruitment of macrophages, T cells and granulocytes to the colon mucosa that is seen in acute DSS-induced colitis [16]. Chemoattraction of macrophages by HD-5 was inhibited by hBD-2 [27]. If the orthologues have the same functions, mBD-3 would be expected to inhibit chemotaxis by Crp-4. Furthermore, both Crp-4 and HD-5 were shown to inhibit IL-1β release from macrophages [28]. Thus, it is possible that precipitation of UC includes attacks by noxious agents at the epithelial lining, e.g. DSS used here, and/or an epithelium with insufficient epithelial transport, as in the IL-2^−/− mice [29], breakage of the colonic homeostasis by causing α-defensin production by the epithelium with consequent recruitment of immune cells whose functions are moderated locally by α-defensins. These immune cells might, however, cause induction of β-defensin production and thereby inhibit the effects of the α-defensins. The proportional magnitude of the response might be decisive for whether one sees an accelerating acute inflammation or develops a chronic inflammation with undulating disease activity. Whether long-standing chronic inflammation will result in metaplastic Paneth cells in the colon also in the mouse models is yet to be determined.

The main conclusion from this study is that β-defensin induction in colonic epithelial cells is the consequence of chronic inflammation in both mouse models, which also may apply to UC in humans.

Disclosure

The authors declare that there are no conflicts of interest.