Production of interleukin (IL)-10 and IL-12 by murine colonic dendritic cells in response to microbial stimuli

Abstract

Intestinal dendritic cells (DC) are likely to regulate immunity to gut microflora, but little is known about their responses to bacterial antigens. Therefore, DC from normal murine colon were characterized and their cytokine responses to components of Gram-negative and/or Gram-positive bacteria assessed. Cells were obtained by digestion of colonic tissue and contained DC that were identified by flow cytometry as CD11c⁺ major histocompatibility complex (MHC) class II⁺ cells. Purified DC were obtained by immunomagnetic separation plus cell sorting. DC had the morphology of immature myeloid cells, were endocytically active, expressed low levels of co-stimulatory molecules and stimulated a weak allogeneic mixed leucocyte reaction. Analysis of flow cytometry data by a sensitive subtraction method allowed measurement of production of interleukin (IL)-12 and IL-10 by small numbers of gut DC by intracellular staining. Fewer than 5% of unstimulated DC produced either IL-10 or IL-12. IL-10 production was significantly up-regulated following stimulation with Bifidobacteria longum, but not after exposure to lipopolysaccharide (LPS) or Streptococcus faecium. In contrast, colonic DC produced IL-12 in response to both LPS and B.longum. Thus, colonic DC can produce both IL-12 and IL-10 following bacterial stimulation. Cell wall components from different bacteria stimulate distinct responses and may direct immune responses differentially in the gut.

Introduction

A tightly regulated response allows the immune system to co-exist with the large amount of antigenic material present in the gastrointestinal tract in the form of commensal microrganisms and food antigens, while retaining the ability to respond to pathogens. Dendritic cells (DC) are sentinels of the immune system and are the only antigen-presenting cells known to be potent at initiating a primary immune response [1]. Functionally distinct DC subsets, producing cytokines such as interleukin (IL)-12, IL-10 and IL-4, influence the differentiation of T cells they activate [2–4]. Therefore DC may determine whether antigen encounter results in non-responsiveness or the generation of T helper (Th)1, Th2 or Th3/Treg responses.

DC in the gut mucosa potentially regulate tolerogenic, as well as stimulatory, responses but at present they have been characterized poorly. Small numbers of DC circulate continually through gut tissue with a transit time of a few days and recruitment is increased rapidly in response to inflammatory stimuli [5]. The immune system is not ignorant of luminal antigen [6] and intestinal DC can acquire orally administered antigens, such as ovalbumin, and upon isolation these DC can activate ovalbumin-specific T cells [7]. Flt3-ligand-treated mice exhibit enhanced oral tolerance; lower doses of ovalbumin are required to induce tolerance in Flt3-ligand-treated mice than in controls [8].

Several studies suggest that DC in intestinal lymphoid tissue are functionally different from DC at other sites. For instance, CD11b⁺ Peyer's patch DC secrete IL-10 and generate Th2 responses preferentially, whereas spleen DC produce IL-12 and favour Th1 generation [9,10]. DC from the intestine, but not from other sites, can produce transforming growth factor (TGF)-β which may have a regulatory function and contribute to IgA isotype switching in the intestine [11]. In addition, evidence from our laboratory demonstrates that DC from the mesenteric lymph nodes, but not those from peripheral nodes, induce α4β7 integrin expression on responding T cells [12]. Such cells stimulated by mucosal DC show preferential homing to the gut [13,14]. Gut DC may therefore influence both the differentiation and homing of lymphocytes that they activate. DC at other mucosal sites share some properties of gut DC. For instance, lung DC also stimulate Th2 responses when in the resting state and generate Th1 responses only after stimulation [15]. Thus a bias away from proinflammatory Th1 responses may be a general feature of mucosal DC.

Although several studies have shown that intestinal DC take up luminal antigens, the site or sites of this sampling have not been established definitively [16,17]. However, lamina propria DC can penetrate epithelial monolayers and express tight junction proteins, allowing them to send dendrites through epithelium and directly sample luminal bacteria, while maintaining epithelial integrity [16].

In order to avoid damaging immune responses, lamina propria DC may be biased towards production of regulatory or anti-inflammatory cytokines upon contact with bacteria. Alternatively, they may be inherently non-responsive to bacterial stimuli, or have become refractory to further activation by the nature of the persisting stimulus. Following systemic exposure to Toxoplasma gondii tachyzoite, DC produce IL-12 but become refractory to subsequent stimulation with the microbial stimulus in vivo[18]. In addition, prokaryotic determinants could actively inhibit mucosal responses to potentially proinflammatory stimuli [19]. Non-pathogenic bacteria attenuate the activation of NF-κB [20], a transcription factor which is phosphorylated in response to proinflammatory stimuli.

In this study, we have sought to identify and characterize colonic DC from normal animals without the use of growth factors such as Flt3-ligand. Cells were released from colonic tissue by rapid enzymatic digestion and we examined phenotypic and functional properties of DC. To determine whether colonic DC are capable of responding to bacterial antigens we assessed cytokine production after stimulation using cell wall components of either a Gram-negative bacterial strain (Escherichia coli) or Gram-positive strains (Bifidobacteria longum and Streptococcus faecium).

Materials and methods

Mice

Colons and spleens were obtained from unmanipulated specific pathogen-free female BALB/c mice aged 6–8 weeks (Harlan, Bicester, UK). Lymph node cells were obtained from specific pathogen-free mice of a similar age.

Processing of intestinal tissue

Colons were flushed out with phosphate-buffered saline (PBS) and cut into sections. Mucus and faeces were removed from the tissue using 1 mm dithiothreitol in Hanks's balanced salt solution (HBSS: Life Technologies, Paisley, UK). Epithelium was removed using 1 mm ethylene diamine tetra-acetic acid (EDTA) in calcium and magnesium-free HBSS. Collagenase digestion was performed in medium comprising 1 mg collagenase D (Roche Diagnostics Ltd, Lewes, UK) in 8·8 ml Hepes-buffered RPMI-1640 (Sigma-Aldrich, Poole, UK), supplemented with 2% fetal calf serum (FCS) and 2 µm DNase I (Roche Diagnostics). Cells released from the tissue were passed through a 100 µm cell strainer.

Cell surface labelling

Fc-block (CD16/CD32; BD Biosciences, Oxford, UK) was added to colonic cells prior to labelling with fluorescent-conjugated antibodies. Antibodies to CD11c, CD54 (ICAM-1), CD11b (Mac-1), CD8α (Ly-2), CD80 (B7-1), CD86 (B7-2), CD40, I-Ad and I-A/I-E (clone M5/114·15·2) were all purchased from BD Biosciences; DEC-205 (clone NLDC 145) was purchased from Serotec (Oxford, UK); and antimouse dendritic cells (clone 33D1) from Leinco Technologies (Stroud, UK). Four-colour flow-cytometry was carried out on Epics XL-MCL (Beckman Carlter, High Wycombe, UK) or FACSCalibur (BD Biosciences, Oxford, UK) machines. Data were acquired uncompensated and subsequently compensated offline using WinList software (Verity Software House, Topsham, ME, USA). Absolute cell counts were made by reference to a known volume of added flow count fluorospheres (Beckman Carlter, High Wycombe, UK). The level of staining was expressed as the geometric mean of the mean fluorescence intensity (MFI) of cells labelled with the test antibody minus the geometric mean MFI of cells stained with the isotype-matched control antibody.

Endocytosis assay

Endocytic activity was assessed by measuring uptake of the fluid phase marker fluorescein isothiocyanate-dextran (FITC-dextran; Sigma Aldrich Co. Ltd, Irvine, UK). Colonic cells were incubated for 30 min at 37°C in complete medium (RPMI-1640 Dutch modification (Sigma) supplemented with 10% FCS, 2 mm l-glutamine, 5 × 10⁻⁵m 2-mercaptoethanol, 100 µg/ml streptomycin and 100 U/ml penicillin) with or without the addition of 1 mg/ml FITC-dextran. Control incubations were performed at 4°C. Uptake of FITC-dextran was quantified by subtracting MFI from cells cultured without FITC-dextran from the MFI of cells cultured with the tracer.

Purification of DC

Colonic DC were purified by a two-stage process. First, the cells were enriched by immunomagnetic separation, and then cell-sorted to high purity by flow cytometry. Colonic cells were washed into mini-Macs buffer, and FCS, Fc-block and anti-CD11c microbeads added. CD11c⁺ cells were selected positively by passing through a MS column (Miltenyi Biotec, Bisley, UK) as described in the manufacturer's protocol. CD11c⁺ cells were purified further by cell-sorting on a FACSCalibur flow cytometer (Becton Dickinson) using a gate set on a plot of CD11c versus side scatter (vide infra). Cell sorting was performed in exclusion mode, and purity of the populations obtained confirmed by reanalysis of the sorted populations. CD11c⁺ cells comprised 20–70% of the cells after immunomagnetic separation and were >85% pure after cell sorting. Morphology of sorted populations was determined in Wright–Giemsa-stained cytospin preparations.

Preparation of mature spleen DC

Spleens were pressed through a metal strainer before overnight culture in complete medium: RPMI-1640 Dutch modification (Sigma Aldrich) supplemented with 10% FCS, 2 mm l-glutamine, 5 × 10⁻⁵m 2-mercaptoethanol, 100 µg/ml streptomycin and 100 U/ml penicillin. Non-adherent cells were then centrifuged over a 14·5% w/v metrizamide (Sigma) gradient, and DC-enriched low-density cells recovered from the interface. DC were then purified further by a combination of immunomagnetic separation and cell sorting, based on expression of CD11c as discussed above. Final purity was >85%.

Proliferation assays

Lymphocyte proliferation assays were performed in complete medium using 20 µl hanging drop cultures in inverted Terasaki plates. Graded numbers (100–3000 per well) of partially enriched or purified DC were used to stimulate allogeneic lymph node responder cells (12 500–100 000) from CBA mice in triplicate mixed leucocyte cultures. In control wells, responder cells were cultured in medium alone. After 3 days of culture at 37°C in humidified 5% CO₂ in air, the cells were pulsed with 1 µg/ml [³H]-thymidine with a specific activity of 2 Ci/mmol (Amersham Biosciences, Chalfont St Giles, UK) for 2 h. At the end of culture, cells were harvested under suction onto filter papers. The filters were washed with saline, 5% trichloroacetic acid and finally methanol before drying overnight. Thymidine uptake into proliferating cells was measured by imaging the filter paper on a PhosphorImager (Molecular Dynamics, Amersham Biosciences, Chalfont St Giles, UK).

Bacterial stimulation

Colonic cells were cultured in complete medium at a concentration of 1 × 10⁶/ml in 24-well plates. Monensin was added to half the wells at a final concentration of 3 µm. Lipopolysaccharide (LPS) (from Escherichia coli, Sigma) was added to colonic cells at a concentration of 1 µg/ml. B. longum or S. faecium cell walls, gifts from Karen Lammers (Dip. Medicina Interna e Gastroenterologia, Bologna, Italy), were added at a concentration equivalent to 1 × 10⁷ colony-forming units (CFU) per ml. In unstimulated cultures, mononuclear cells were cultured in medium alone.

Intracellular cytokine analysis

Production of cytokines was determined by intracellular staining and flow cytometry. This highly sensitive approach made it possible to analyse cytokine production by small numbers of DC present. The experimental approach and analysis of data has been described recently in detail by Panoskaltsis et al. [21]. In brief, cells harvested from unstimulated and bacterially stimulated cultures were surface labelled with monoclonal antibodies to indentify DC as described above and then permeabilized using Leucoperm A & B (Serotec) according to the manufacturer's instructions. Permeabilized cells were stained with phycoerythrin (PE)-labelled monoclonal antibodies to IL-12 or IL-10 (PharMingen). For analysis of flow cytometry data DC were gated as CD11c⁺ major histocompatibility complex (MHC) class II⁺ cells within the viable colonic cell population. Cytokine-positive DC were detected by comparing paired samples stained with the same anticytokine antibody, in which one of the pair contained cells that had been cultured in monensin and one had been cultured in medium alone. By using the same antibody to stain both samples in a comparison the potential problems associated with differences between antibodies with regard to non-specific binding to permealized cells (for instance, due to variations in free fluorochrome content) are avoided. The percentage of cytokine-positive cells was determined by subtracting a normalized histogram of staining of the non-monensin culture from a histogram of staining of the monensin-supplemented culture using WinList 5·0 software (Verity). The results are presented graphically as a histogram of the staining in the presence of monensin in which a shaded area represents the fraction of positive cells remaining after subtraction of the non-monensin labelling. In order to determine whether the two histograms in a pair were significantly different from one another, Kolomogorov–Smirnov (KS) statistics were used to calculate a D_crit value [22] according to the formula:

where D_max is the maximum value between test (monensin supplemented) and control samples after the two histograms have been converted into cumulative normalized histograms, n1 is the number of events in the test sample and n2 the number of events in the control sample. The P-value for D_crit was determined from a Kolmogorov–Smirnov statistical table. If the D_crit value was not statistically significant (i.e. P ≥ 0·05), cytokine staining was regarded as 0%.

To validate the technique, the specificity of any staining observed was confirmed in competition experiments using unlabelled antibodies of the same clone as the PE-labelled antibodies used for cytokine detection (vida infra). Pooled data from separate experiments were compared using either paired t-tests for normally distributed data or Wilcoxon signed-rank tests for non-normally distributed data and values for P of less than 0·05 were regarded as significant.

Results

Identification of Class II⁺ CD11c⁺ DC in the colon

To identify low-frequency DC in cells from colonic tissue we initially eliminated epithelial and other non-mononuclear cells by constructing a gate based on the forward-scatter and side-scatter properties of the cells (Fig. 1a). Four-colour flow cytometry of the colonic cells allowed the identification of a population of class II⁺ CD11c⁺ DC (Fig. 1b). The DC stained brightly for the allophycocyanin (APC)-conjugated anti-CD11c antibody. Levels of autofluorescence were low in the fourth fluorescence channel and therefore using the APC conjugate the DC could be distinguished easily. The matched isotype control for anti-CD11c (Fig. 1c) and anti-MHC class II (Fig. 1d) displayed low levels of non-specific binding, with no cells within the specific DC gate.

Fig. 1.

Identification of DC as CD11c⁺/class II⁺ cells. Flow-cytometric gating of colonic cells (a) revealed a population of CD11c⁺ and MHC class II⁺ putative DC (b). The matched isotype controls (c,d) displayed low levels of non-specific binding.

We calculated the numbers of DC by reference to known numbers of added fluorospheres. DC (2 × 10⁴ ± 2 × 10³) were recovered from each colon representing approximately 3% of total mononuclear cells.

Phenotype of freshly isolated murine colonic DC

Approximately 80% expressed the myeloid DC marker CD11b (Fig. 2a) and comprised CD11b⁺ and CD11b^–/lo subsets. Few colonic DC expressed the ‘lymphoid’ marker CD8α (Fig. 2b). Levels of expression of DEC205 were low (Fig. 2c), but almost all colonic DC were 33D1⁺ (Fig. 2d).

Fig. 2.

DC subsets in the murine colon. DC were gated as a CD11c⁺ MHC class II⁺ population in cells isolated from the colon (a–d) and Peyer's patch (e,f). Staining of gated DC with specific antibodies (filled histograms) and isotype-matched controls (open histograms) were examined using a third channel of the flow cytometer. Cells were stained as follows: (a,e) anti-CD11b; (b,f) anti CD8α; (c) anti-DEC-205; (d) 33D1.

Many of the DC prepared in parallel from the Peyer's patch were CD11b^– (Fig. 2e) and almost half were CD8α⁺ (Fig. 2f). A proportion of colonic lymphocytes was CD8α⁺. Together, these findings suggest that the failure to find substantial numbers of CD8α⁺ DC in the colon was not due to a failure of the methods used to extract them, or of a failure of antibody labelling.

Almost all colonic DC expressed low but detectable levels of CD86, CD40, CD80 and CD54 when freshly isolated and levels of expression were modestly up-regulated after overnight culture of colonic cells in complete medium (Fig. 3a; representative of three experiments). Both freshly isolated and cultured DC were endocytically active, showing temperature-dependent uptake of FITC-dextran (Fig. 3b). Lymphocytes analysed in parallel did not take up FITC-dextran.

Fig. 3.

Expression of co-stimulatory molecules and endocytic activity of freshly isolated and cultured murine colonic DC. (a) Co-stimulatory marker expression. DC were gated as a CD11c⁺ MHC class II⁺ population in freshly isolated colonic cells and colonic cells that had been cultured overnight in complete medium. Staining of gated DC with specific antibodies (filled histograms) and isotype-matched controls (open histograms) were examined using a third channel of the flow cytometer. Cells were labelled with anti-CD86, anti-CD40, anti-CD80 and anti-CD54. Values represent the geometric mean of the mean fluorescence intensity (MFI) of cells labelled with the test antibody minus the geometric mean MFI of cells stained with the isotype-matched control antibody. (b) Endocytic activity. Freshly isolated and cultured colonic cells were cultured with FITC-dextran or medium alone for 30 min at 37°C and at 4°C. Uptake of the tracer by CD11c⁺ MHC class II⁺ DC and lymphocytes (gated on light scatter) was calculated by subtracting mean fluorescence intensity (MFI) of cells cultured in medium alone from the MFI of cells cultured with FITC-dextran.

Stimulation of a mixed leucocyte reaction (MLR) by colonic DC

To investigate the stimulatory capacity of the colonic DC, enrichment of the population was initially carried out using immunomagnetic separation to positively select CD11c⁺ cells. This method served to enrich the DC population to approximately 70% (data not shown). The enriched population was gamma-irradiated and added to allogeneic lymph-node cells. This approach indicated that this freshly isolated CD11c-enriched population caused a dose-dependent proliferative response of allogeneic T cells and that this response was less potent than that stimulated by mature spleen DC (data not shown). To confirm that this stimulatory activity was attributable to the CD11c⁺ population and not to contaminating cells, we further purified the CD11c⁺ population by cell sorting (Fig. 4a). A high degree of purity was attained (>85%) but recovery was low (approximately 10% of the starting population). Purified CD11c⁺ cells again stimulated an allogeneic mixed leucocyte reaction. (Fig. 4b). The sorted cells had morphology consistent with that of immature myeloid cells (Fig. 5). In parallel with the up-regulation in co-stimulatory molecule expression, observed following overnight culture (Fig. 3a), CD11c⁺ cells separated after overnight culture displayed enhanced stimulatory capacity. Occasional cells in the cultured population had the morphology of mature DC (Fig. 5). Despite the up-regulation of stimulatory capacity with culture, these cultured intestinal DC were still less potent stimulators than mature spleen DC assayed in parallel (Fig. 4b).

Fig. 4.

Stimulatory capacity of purified colonic DC. (a) CD11c⁺ DC were purified by a combination of immunomagnetic separation and cell-sorting from freshly isolated (d0), and cultured (d1) colonic cells and from spleen low-density cells as a source of mature DC. Plots show staining with anti-CD11c (y-axis) against side-scatter (x-axis) for cells before and after purification. (b) The ability of each population to stimulate allogeneic T cells in a mixed leucocyte reaction was determined by [³H]-thymidine incorporation.

Fig. 5.

Morphology of sorted CD11c⁺ DC. Cytospins were made of CD11c⁺ DC from freshly isolated (d0) and cultured (d1) colonic cells, purified through cell sorting techniques. Slides were stained with Giemsa and viewed at ×1000 magnification.

Production of IL-10 and IL-12 by colonic DC

To assess ‘spontaneous’ cytokine production by colonic DC we cultured freshly isolated colonic cells in complete medium for 4 h in the presence or absence of monensin. To investigate the effect of bacterial antigens on cytokine production, LPS from E. coli or cell wall components of B. longum were added to the 4-h culture. Following permeabilization, intracellular cytokine was detected by flow cytometry and the proportion of cytokine positive DC determined. Figures 6a and 7a show representative results of staining for IL-10 (one of eight experiments) and IL-12 (one of 12 experiments), respectively. The shaded areas of the histograms represent the proportion of the DC becoming cytokine positive during the treatment with monensin.

Fig. 6.

IL-10 production by murine colonic DC. (a) Histograms show staining with anti-IL-10 antibody in the presence of monensin. DC were gated as CD11c⁺ MHC class II⁺ cells in colonic cells that had been cultured in medium alone or with LPS or B. longum. Numerical values represent the percentage of DC secreting IL-10, as indicated by the shaded area of the histogram, after subtraction of an identical labelling of cells cultured in the absence of monensin. Histograms are representative of a single experiment. (b) Data combined from eight to 12 experiments. Lines join paired samples of control (medium alone) and stimulated cultures within the same experiment. Wilcoxon's signed-rank test was used to derive P-values. (c) In a separate experiment, labelling of B. longum stimulated DC with anti-IL-10 was inhibited completely in the presence of an unlabelled blocking antibody of the same clone.

Fig. 7.

IL-12 production by murine colonic DC. (a) Histograms show staining with anti-IL-12 antibody in the presence of monensin. DC were gated as CD11c⁺ MHC class II⁺ cells in colonic cells that had been cultured in medium alone or with LPS or B. longum. Numerical values represent the percentage of DC secreting IL-12, as indicated by the shaded area of the histogram, after subtraction of an identical labelling of cells cultured in the absence of monensin. Histograms are representative of a single experiment. (b) Data combined from eight to 12 experiments. Lines join paired samples of control (medium alone) and stimulated cultures within the same experiment. Wilcoxon's signed-rank test was used to derive P-values. (c) In a separate experiment, labelling of B. longum stimulated DC with anti-IL-12 was inhibited completely in the presence of an unlabelled blocking antibody of the same clone.

In unstimulated cultures, few DC produced IL-10 (1·5 ± 2·2%; n = 8) or IL-12 (4·6 ± 5·6%; n = 12). As shown in Fig. 6a,b, B. longum stimulation generated a significant proportion of IL-10-producing DC when compared with unstimulated cultures (a mean of 35·75%versus 1·5%, P < 0·005; n = 8). In contrast, although significant levels of IL-10 were detected following LPS stimulation in some individual experiments, overall the proportion of DC producing IL-10 was not increased significantly under these conditions. Furthermore, in direct comparisons, B. longum stimulation generated significantly more IL-10-producing DC than LPS stimulation (P < 0·05; n = 8). Staining of IL-10 in B. longum-stimulated cultures was completely inhibited in the presence of an unlabelled anti-IL-10 antibody of the same clone, but the presence of an irrelevant antibody did not affect the staining significantly, confirming the specificity of the observed labelling (Fig. 6c).

Cell walls from a second Gram-positive commensal organism, S. faecium, did not stimulate IL-10 production by colonic DC, suggesting that the ability to induce IL-10 production by DC may be a characteristic of certain bacterial strains rather than a generalized property of Gram-positive bacteria (data not shown).

In contrast to the findings with IL-10, both LPS and B. longum generated a significant proportion of IL-12-producing DC compared with the proportion in unstimulated cultures (15·4%versus 4·6% for LPS, P < 0·05; 15·2%versus 4·6% for B. longum, P < 0·05). B. longum, but not LPS, generated significantly more IL-10-producing DC than IL-12-producing DC (P < 0·05). Representative data are shown in Fig. 7a and pooled data from replicate experiments in Fig. 7b. Staining for IL-12 was inhibited completely in the presence of an unlabelled antibody of the same clone, but the presence of an irrelevant antibody did not significantly affect the staining, confirming the specificity of the observed labelling (Fig. 7c).

In a series of experiments both LPS and B. longum were added simultaneously to the cultures in order to determine whether B. longum could inhibit LPS-induced IL-12 production. Results of these experiments were somewhat variable, perhaps reflecting the variability between individual experiments when a single stimulus was used. None the less, LPS-induced production of IL-12 was reduced when B. longum was added into the cultures (Fig. 8a). IL-10 production stimulated by B. longum was not affected by addition of LPS (Fig. 8b). IL-4-secreting DC were not detected under any culture conditions in either of two experiments.

Fig. 8.

IL-12 and IL-10 production by murine colonic DC following culture with both B. longum and LPS. Graphs show mean percentage of positive cytokine producing DC in unstimulated, LPS- and B. longum-stimulated cultures, and DC simultaneously exposed to both bacterial stimuli. Error bars represent standard error of the mean. LPS-induced production of IL-12 was reduced, but not to a significant degree, when B. longum was added to the cultures. No difference was seen in the percentage of IL-10-positive DC when LPS was added to B. longum-stimulated cultures.

Discussion

We have shown that functional CD11c⁺ DC can be identified and studied in cells from the normal mouse colon. DC comprised only a small proportion of the isolated cells. The low number of colonic DC probably reflects their frequency in situ rather than an inefficiency in the extraction process, as recent immunohistological analysis also indicates that CD11c⁺ cells are infrequent in the mouse colon [23,24]. Despite the low frequency of DC in colonic tissue we have been able to characterize these cells without expansion of the DC population with growth factors such as Flt3-ligand. Treatment of mice with Flt3-ligand expands numbers of DC in the intestine and other tissues [8,25,26]. However, it is not clear that this expanded population of immature DC is truly representative of the population that is normally resident in the tissues. Furthermore, recent in vitro evidence suggests that Flt3-ligand may act selectively on precursors of certain DC subsets [27]. Therefore, despite the challenges of low cell numbers we have analysed the DC present in unmanipulated mice. Due to the rarity of these cells and the potential for alteration in DC during prolonged separation procedures we have analysed colonic DC in a mixed population of cells using multicolour flow cytometry. The success of this approach was made possible by reducing problems of autofluorescence in the mixed cell sample by flow cytometry gating on the basis of light scatter combined with the use of DC-specific antibodies conjugated to the fluorochrome allophycocyanin (APC). The APC channel permitted a bright positive signal with minimal autofluorescence.

DC are able to sample antigen in the intestinal lumen. DC in Peyer's patches can acquire fed antigen [6] and possess functional characteristics such as production of IL-10 and TGF-β, which may underlie some of the immunological features of the intestine. However, the recent demonstration that DC in the lamina propria may be able to interact directly with the contents of the lumen [16] suggests that antigen sampling by lamina propria DC needs to be considered. The murine colon has been reported to contain immunostimulatory cells with the properties of DC [28], but these cells have not previously been characterized extensively.

Here we report that DC in the colon are CD8α^– and express a variable level of CD11b, a phenotype characteristic of ‘myeloid’ DC. The low expression of DEC205 is also suggestive of a myeloid population [29,30]. The CD8α⁺ DC population can be more difficult to extract from tissues [31]. However, the lack of CD8α⁺ DC in cells from the colon was not due to a failure to extract CD8α⁺ DC as this population was clearly evident in cells extracted in parallel from Peyer's patches. Thus, these finding are consistent with a model in which myeloid DC traffic through peripheral tissues into draining lymph nodes, whereas CD8α⁺ DC are normally resident in the T cell areas of lymphoid tissues.

Colonic DC were heterogeneous because both CD11b^hi and CD11b^–/lo subsets were evident. CD8α^–CD11b^– DC have been reported as a prominent feature of lymphoid tissue associated with the intestine, but not other sites [10]; the functional significance of this population is unknown. Because we obtained DC from full-thickness colon, there may be DC derived from colonic lymphoid follicles located in the submucosa as well as the scattered lamina propria population [32]. Future work will aim to determine whether this DC heterogeneity reflects functional specialization and/or anatomic derivation.

When freshly isolated, the colonic DC expressed detectable levels of co-stimulatory molecules were endocytically active and were weakly stimulatory in a mixed leucocyte reaction. Both co-stimulatory molecule expression and stimulatory activity were modestly up-regulated in DC isolated from cells cultured for 24 h, in the absence of exogenous stimulation. However, stimulatory activity remained lower than that of mature spleen DC and endocytic activity was maintained, suggesting that the DC cultured population was not fully mature. These findings are consistent with data on human colonic DC, which demonstrated that although culture in a mixed population of mononuclear cells overnight was sufficient to up-regulate co-stimulatory molecule expression on DC, it was a rather weak stimulus for DC maturation [33]. The purity of the sorted DC population used as stimulators in mixed leucocyte reaction was approximately 85%, judged by re-analysis on the flow cytometer and the majority of non-DC events were MHC class II^–, probably comprising tissue debris. Thus a contribution of non-DC to the stimulatory activity is unlikely. Low-level co-stimulatory molecule expression and stimulatory activity on freshly isolated cells may reflect activation during the isolation procedure despite our efforts to isolate the DC as quickly as possible. Alternatively, a low-level stimulatory capacity of DC in the steady state may be sufficient to activate T cells but favour the development of a regulatory population. It is notable that antigen-specific proliferation precedes the establishment of a tolerant state in models of oral tolerance [34].

We analysed the functional consequences of the interaction between colonic DC and bacteria in terms of cytokine production. The rarity of DC in intestinal tissue and the low yields of DC obtained following purification meant that we were unable to obtain sufficient pure DC for analysis by conventional assays. Therefore, we developed and validated an intracellular staining approach that combined multicolour flow cytometry with data analysis that detected low levels of antibody binding. Using this approach, rare DC could be identified rigorously and their cytokine production measured.

Colonic DC produced cytokine in response to bacterial stimulation. It is open to debate whether this reflects the response in vivo. None the less, the findings establish that DC from the colon are neither inherently non-responsive to bacterial stimulation nor have they been rendered refractory to restimulation due to persistent exposure to bacterial antigens in vivo. In the current study, colonic DC were able to make both IL-12 and IL-10 when stimulated with products of the commensal flora. Under appropriate stimulation DC can produce IL-4 [3], but no intracellular IL-4 was detected in the colonic DC population.

IL-10 can inhibit IL-12 production, down-regulate antigen presentation and is required for the action of some regulatory T-cells [35]. It is required for immune homeostasis in the intestine and IL-10 knock-out mice develop colitis [36,37]. IL-10 production by DC in the respiratory tract is crucial in mediating tolerance to respiratory antigens [11]. Elevated production of IL-12 and enhanced interferon (IFN)-γ production by Th1 cells is a feature of human Crohn's disease [38] and a number of animal models of colitis. Thus the balance of IL-10 and IL-12 production by DC is likely to be an important influence on intestinal immunity.

Under the conditions of the study, LPS and B. longum had differential effects of cytokine production by DC. LPS up-regulated both IL-12 and IL-10 production above the baseline. The low-level secretion of both IL-12 and IL-10 (<5% cytokine-positive DC) seen in unstimulated cultures may reflect activation during the isolation procedure or stimulation by residual intestinal bacteria. B. longum up-regulated both IL-10 and IL-12, but significantly more DC made IL-10 than IL-12. It is not possible from the current data to determine whether these differences between stimuli are absolute; stimulation at different doses or for different periods of time may reveal an altered pattern of cytokine production. None the less, the principle that colonic DC can respond differentially to microbial stimuli is established. A systematic study encompassing both commensals and pathogens will be required before general conclusions about differences between organisms can be drawn. The observation that another Gram-positive commensal organism, S. faecium, did not up-regulate DC production of IL-10 suggests that the effects on DC may be bacterial strain-specific and that a common cell wall component of Gram-negative bacteria, such as peptidoglycan, is unlikely to be responsible for the effects on DC. Analysis of Toll-like receptor expression and associated signalling pathways may reveal the molecular basis of the differential response of DC to bacterial stimuli. Future studies with neutralizing antibodies will help to determine whether the conflicting effects of LPS and bifidobacterial stimulation result from a direct cross-regulation between IL-10 and IL-12 or whether additional signals are involved. IL-10 down-regulates IL-12 production in DC [39].

In conclusion, colonic DC can produce immunoregulatory cytokines upon contact with commensal bacteria. The potential for different stimuli to elicit opposing responses suggests that a balance between different stimuli acting on DC may shape an immune response.