Induction of antigen-specific regulatory T cells in the liver-draining celiac lymph node following oral antigen administration

Abstract

Regulatory T cells are induced by oral administration of an antigen, but the physiological requirements and localization of the inductive sites are largely unknown. Using an adoptive transfer system of cells transgenic for ovalbumin T-cell receptor (OVA TCR tg), we found that antigen-specific CD4⁺ T cells were activated in the liver-draining celiac lymph node (CLN) shortly after ovalbumin feeding, and that a significantly higher proportion of the T cells in the CLN developed into the putative regulatory phenotype [co-expressing CD25 with the glucocortico-induced tumour necrosis factor (TNF) receptor family related gene (GITR), cytotoxic T-lymphocyte antigen (CTLA)-4 and CD103] than in Peyer's patches, the mesenteric and peripheral lymph nodes and the spleen. In addition, a particularly high level of expression of CD103 on the OVA-specific T cells in the CLN may favour homing to the epithelium of the intestine. While equally suppressive, OVA tg T cells isolated from the CLN of OVA-fed DO11·10 mice were less dependent on transforming growth factor (TGF)-β for suppression than cells isolated from the peripheral and mesenteric lymph nodes, which indicates the involvement of an additional suppressive mechanism. The expression of FoxP3 was not up-regulated in any of the lymph node compartments studied. Our phenotypic and functional findings suggest that the induction of regulatory T cells in the CLN may be relevant in the control of the immune response to dietary antigens.

Introduction

Stringent regulation of the extensive and potent immune system of the gut is vital, as breakdowns in tolerance to luminal antigens may lead to inflammatory bowel disease (IBD) or food intolerance.^1,2 Our knowledge of the mechanisms that are involved in the induction and maintenance of tolerance at mucosal sites is still incomplete. However, a central role for CD4⁺ regulatory T cells in this process appears to be likely.^3–5

Natural CD4⁺ CD25⁺ regulatory T cells (Treg), which are believed to be specific for self antigens, are dependent on thymic function for their development.⁶ However, other subtypes of Treg can be induced from naïve CD4⁺ T cells in a thymus-independent manner, and these Tregs are probably important in the induction of tolerance to non-self antigens. Regulatory T cell type 1 (Tr1) cells⁷ and T helper 3 (Th3) cells⁴ are examples of peripherally induced T cells with regulatory properties. They share many phenotypic characteristics with natural thymus-derived Tregs, i.e. the expression of CD25, cytotoxic T-lymphocyte antigen (CTLA)-4 and the glucocortico-induced tumour necrosis factor (TNF) receptor family related gene (GITR).⁸ However, unlike natural Tregs, their suppressive ability seems to depend primarily on the secretion of suppressive cytokines such as interleukin (IL)-10 and transforming growth factor (TGF)-β.⁹ While the surface markers that are associated with Tregs are also expressed by activated conventional CD4⁺ T cells, the expression of the transcriptional repressor FoxP3 appears to be exclusive to Tregs. To date, most studies have suggested that only natural thymus-derived Tregs express FoxP3. A connection between the homing molecule CD103 (αEβ7) and a population of Tregs has also been described.^10–12 The CD103 integrin mediates adhesion to epithelial cells through its binding to E-cadherin, which is expressed selectively on epithelial cells,^13,14 and it is therefore important for the localization of Tregs to the skin and intestine. The subpopulation of Tregs that expresses CD103 has recently been revealed to have a preferential capacity to prevent IBD.¹⁵

Regulatory T cells can be induced in the periphery by various protocols, including intravenous,¹⁶ oral^4,16,17 and intranasal¹⁸ antigen administration. In vitro, naïve CD4⁺ T cells differentiate into the regulatory phenotype by activation in the presence of IL-10 alone,⁷ with IL-10 in combination with TGF-β¹⁹ or by stimulation with dendritic cells (DCs) that express low levels of co-stimulatory molecules. DCs with a tolerogenic phenotype can be generated in vitro by several different protocols, including treatment with IL-10²⁰ or prostaglandin (PGE).^2,21

Peripheral induction of Tregs is dependent on the interactions between naïve T cells and DCs that carry the relevant antigenic peptide. It is important to determine the location and circumstances of this interaction, as maturation of DCs into the relevant tolerogenic phenotype may be determined regionally. In vivo, the liver provides a microenvironment that is exceptionally well suited for this type of conditioning. The liver is heavily trafficked by DCs²² which are in constant contact with antigenic material derived from the intestine via the portal blood. Thus, ovalbumin administered orally has been found to co-localize with major histocompatibility complex (MHC) class II⁺ CD11c⁺ cells in the liver.²³ Under homeostatic non-inflammatory conditions, the liver DCs mature in the tolerogenic environment provided by Kuppfer cells in response to physiological concentrations of endotoxin in the portal blood (reviewed in ref. 24). Liver-derived DC progenitors produce IL-10 and are poor T-cell stimulators.²⁵ Accordingly, it has been reported that portal drainage through the liver is a prerequisite for the establishment of oral tolerance²⁶ and that portal administration of transplantation antigens is a very effective way of inducing graft tolerance.^27,28

As DCs in the liver have been shown to migrate to the liver-draining celiac lymph node (CLN),²⁹ we set out to investigate whether this is an important site for the interactions between naïve CD4⁺ T cells and gut-derived antigens. Several earlier studies have dealt with the process of activation of antigen-specific T cells and the induction of tolerance after oral antigen administration, but they have focused mainly on the events in the gut-associated lymphoid tissues (GALT), i.e. Peyer's patches and the mesenteric lymph node (MLN).^17,30–33 In this paper, we investigated the liver-draining CLN as a potential key player in the response to dietary antigens.

Materials and methods

Mice

BALB/c mice were purchased from B & K (Stockholm, Sweden). Ovalbumin (OVA) T-cell receptor (TCR) transgenic (Tg) mice on the BALB/c background, clone DO11·10 were used, and were originally obtained from The Jackson Laboratory (Bar Harbor, ME). Male and female mice 10–14 weeks of age were used in the study. All mice were housed under specific pathogen-free conditions in the animal facility of Göteborg University. Experiments with animals were performed with the permission of the regional Ethics Committee of Göteborg University (DNR 328-2003).

Antibodies for fluorescence-activated cell sorter (FACS) analysis

The following antibodies used for staining were purchased from BD Pharmingen (San Diego, CA): PerCP-conjugated anti-CD4 (RM4-5) and anti-CD8 (53–6·7), phycoerythrin (PE)-conjugated or biotinylated anti-CD25 (PC61 and 7D4, respectively), PE-conjugated anti-CD69 (H1·2F3), anti-CD103 (M290), anti-CD122 (TM-b1), anti-CTLA-4 (UC10–4F10-11), anti-interferon (IFN)-γ (XMG1·2), and anti-IL-4 (11B11). The following corresponding isotype control antibodies from BD Pharmingen were used: PE-conjugated rat immunoglobulin (Ig) G1 (R3-34), hamster IgG1 (G235-2356), rat IgG2a (R35-95), rat IgG2b (A95-1), and biotinylated rat IgMκ (R4-22). Biotinylated purified polyclonal goat anti-mGITR (cat. no. AF524) and goat IgG isotype control (cat. no. AB-108-C) antibodies were purchased from R & D Systems (Abingdon, UK). Biotinylated anti-TGF-β (1D11) was obtained from ATCC (Manassas, UA), while the corresponding biotinylated isotype control antibody (Ab) was purchased from BD Pharmingen (mouse IgG1, MOPC-31C). Biotinylated Abs were revealed with antigen-presenting cell (APC)-labelled streptavidin (BD Pharmingen). FITC-conjugated anti-KJ1·26 monoclonal antibody (mAb), recognizing the transgenic TCR of the DO11·10 mouse, was a kind gift from K. Schön (Department of Clinical Immunology, Göteborg University, Göteborg, Sweden).

Flow cytometry in the adoptive transfer system

An adoptive transfer system was used for FACS analysis of the kinetics and phenotype of antigen-specific CD4⁺ T cells following oral antigen administration (Fig. 1a).

Figure 1.

Experimental design. (a) An adoptive transfer system was used for flow cytometry analysis of the phenotype of antigen-specific CD4⁺ T cells at indicated time points. (b) OVA-fed DO11·10 mice were used for analysis of FoxP3 by RT-PCR and for functional studies including suppression assays and analysis of cytokine production i supernatants, at the timepoint indicated.

Adoptive cell transfer

Spleens and lymph nodes from DO11·10 mice (10–14 weeks of age) were harvested, and a single-cell suspension was prepared. Erythrocytes of the spleens were lysed in ammonium chloride buffer and the remaining live cells were washed and resuspended in phosphate-buffered saline (PBS). An aliquot of the cell suspension was used to determine the percentage of CD4⁺ KJ1·26⁺ cells and the baseline activation level using flow cytometry. The remaining cells were injected intravenously (i.v.) into sex- and age-matched BALC/c recipients. The recipient mice were injected with 4 × 10⁶ OVA TCR tg T cells, as evaluated by percentage of total cells.

Induction of oral tolerance

Twenty hours after adoptive cell transfer and following an 8-hr starvation period, the BALB/c mice were fed intragastrically with 10 mg OVA (Sigma Chemical Co., St. Louis, MO) in 0·3 ml of PBS. From this time-point until the analysis, the OVA-fed group received a diet that contained 6% OVA (AnalyCen, Lidköping, Sweden), which gives an approximate daily intake of 180 mg of OVA. Control mice were tube-fed with saline, and maintained on a standard diet throughout the experiment.

Flow cytometric analysis

Three- or four-colour cytometry was performed on a FACSCalibur that was equipped with dual lasers (BD Biosciences, Stockholm, Sweden). FACS analysis were performed on cells from OVA-fed and control BALB/c recipients of OVA tg cells (Fig. 1a). A total of 0·4–1·0 × 10⁶ cells from Peyer's patches, spleens, or the celiac, mesenteric and peripheral lymph nodes were incubated with Fc-block for 5 min at room temperature. The cells were then stained with biotinylated, PE-, FITC- and PerCP-conjugated mAb for 20 min at 4° and washed twice. The biotinylated samples were then incubated with APC-conjugated streptavidin (BD Pharmingen) for 20 min at 4° and washed twice before immediate analysis by flow cytometry. For the analysis of intracellular CTLA-4, IL-10, IFN-γ and IL-4, the cells were surface-stained as described above, and then fixed with Cytofix/Cytoperm solution (BD Pharmingen) for 20 min at 4°, washed twice with PermWash solution (BD Pharmingen) and incubated with Fc-block for 10 min at 4° before intracellular staining with PE-labelled antibody for 20 min at 4°. For the detection of intracellular TGF-β, the cells were surface-stained, fixed, and washed as above, before incubation with biotinylated anti-TGF-β antibody for 20 min at 4°. The cells were then washed twice and stained with APC-conjugated streptavidin for 20 min at 4°. The samples were washed twice in PermWash solution before analysis.

Functional studies using DO11·10 mice

As the yield of cells in the adoptive transfer system did not allow functional studies, they were performed on OVA-fed OVA TCR tg DO11·10 mice (Fig. 1b). The DO11·10 mice were maintained on an OVA-containing diet for 10 days prior to the experiments.

Cell purification

Peyer's patches, spleens and lymph nodes (mesenteric, celiac and peripheral) were collected and pooled from three OVA-fed and three control DO11·10 mice in each experiment. The peripheral lymph nodes (PLNs) included the inguinal, axillary and poplital lymph nodes. Single-cell suspensions were prepared and the splenic erythrocytes were lysed in ammonium chloride buffer. For the isolation of OVA TCR tg T cells, cells were incubated with Fc-block for 5 min and then stained with FITC-conjugated anti-KJ1·26 Ab for 20 min at 4°. The samples were then incubated with anti-FITC magnetic beads (MACS; Miltenyi Biotec, GmbH, Bergisch Gladbach, Germany) and separated in a magnetic column according to the manufacturer's protocol. Purity was determined by flow cytometry; lymph node cells were found to be >85% KJ1·26⁺ T cells. The cells isolated from the spleens and Peyer's patches were less pure and were therefore excluded from further analyses. As reference material for the polymerase chain reaction (PCR) analysis, total CD4⁺ T-cell populations and CD4⁺ T cells that were enriched for or depleted of CD25⁺ cells were purified from the peripheral and MLNs of naïve BALB/c mice using mouse CD4 (L3T4) microbeads or the mouse CD4⁺ CD25⁺ regulatory T-cell isolation kit, both of which use the MACS system (Miltenyi Biotec).

Proliferation assay

KJ1·26⁺ T cells purified from the MLN and PLN of naïve DO11·10 mice were used as responder cells. For this purpose, 4 × 10⁴ cells/well were cultured for 5 days with 1 × 10⁵ KJ1·26-depleted and irradiated splenocytes from naïve DO11·10 mice as APCs. In co-cultures, KJ1·26⁺ cells that were isolated from different lymphoid compartments of OVA-fed or control DO11·10 mice were added in different ratios to the responder cells, as indicated in figure 5. The cells were cultured in U-bottomed 96-well plates in Iscove's medium supplemented with 10% heat-inactivated fetal calf serum (FCS), 2 mm l-glutamine, 50 µg/ml gentamycin, and 50 µm 2-mercaptoethanol, and stimulated with 500 µg/ml OVA (Sigma Chemical Co.). In some experiments, neutralizing anti-IL-10 (10 µg/ml, JES5-2A5; BD Pharmingen), anti-TGF-β (100 µg/ml, 1D11; ATCC) or control IgG [anti-toxic shock syndrome toxin (TSST), 100 µg/ml, 49A3·7; a kind gift from A. Tarkowski, Göteborg University, Göteborg, Sweden] was added to the co-cultures. At the end of the culture periods, the cells were pulsed overnight with 1 Ci[³H]thymidine (Amersham Pharmacia Biotech, Little Chalfont, UK).

Figure 5.

In vitro suppression assay with ovalbumin (OVA) T-cell receptor (TCR) transgenic (tg) cells that were isolated from the different lymph nodes of OVA-fed DO11·10 mice. (a) KJ1·26⁺ responder cells were isolated from the mesenteric lymph node (MLN) and peripheral lymph node (PLN) of naïve DO11·10 mice and cultured. Typical counts for OVA-stimulated responder cells (proliferation 100%) were 20 000 counts per minute (c.p.m.). The number of responder cells was also doubled as a crowding control (third bar from the left). Suppressor KJ1·26⁺ T cells were isolated from the MLN, PLN and celiac lymph node (CLN) of tolerized DO11·10 mice that were fed an OVA-containing diet. The suppressor cells were added to the responder cultures in ratios of 1 : 1, 1 : 2, 1 : 5, 1 : 10, 1 : 20, and 1 : 50. The data shown are the mean values of six separate experiments, and the error bars indicate standard deviation (SD). Differences between the tests and controls were analysed with the Mann–Whitney U-test. (b) To obtain a single value for comparing the suppressive efficiency of cells that were isolated from the different lymph nodes of OVA-fed mice, the dilution that allowed 50% of the baseline proliferation (Prol₅₀) was calculated for the different dilution series in each experiment. The data are presented as box plots with the median values as indicated (n = 6).

Analysis of cytokines in culture supernatants

The concentrations of cytokines in the culture supernatants at the end of the culture period were measured by Cytometric Bead Array (Inflammation Kit for IL-6, IL-10, monocyte chemotactic protein 1 (MCP-1), IFN-γ, TNF-α and IL-12 p70, and Th1/Th2 Cytokine Kit for IL-2, IL-4, IL-5, IFN-γ and TNF-α; BD Pharmingen) and analysed on the FACSCalibur. The concentrations of activated TGF-β1 were determined using the human TGF-β1 Quantikine immunoassay (R & D Systems), according to the manufacturer's instructions.

FoxP3 mRNA expression

The expression of FoxP3 mRNA was determined by real-time reverse transcriptase–polymerase chain reaction (RT-PCR) and relative quantification using hypoxanthine-guanin phosphoribosyltransferase (HPRT) as the reference gene. Before the experiment, control mice were maintained on a standard diet while the OVA-fed group received an OVA-containing diet (6%) for 10 days. Lymph node cells from non-fed DO11·10 mice were used as the control. The MLNs, PLNs and CLNs of three OVA-fed mice were pooled for each experiment. KJ1·26⁺ T cells were isolated from a lymph node pool of control DO11·10 mice and the different lymph nodes of OVA-fed DO11·10 mice. Total RNA was extracted from 4–5 × 10⁵ bead-sorted cells using the RNeasy Micro Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The RNA was treated with DNA-free (Ambion, Austin TX) to remove any contaminating genomic DNA. cDNA was prepared in a random hexamer-primed Superscript RT reaction (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. The FoxP3 mRNA levels were measured in duplicate using the LightCycler (Roche Diagnostics, Mannheim, Germany), reagents from the LightCycler FastStart DNA Master SYBR Green I kit (Roche Diagnostics) and the following primer pair: 5′-CAGGAAAGACAGCAACCTT-3′ and 5′-TGCTTGGCAGTGCTTGA-3′. HPRT was used as the endogenous reference gene for relative quantification and was detected using the primer pair 5′-GTTGGATACAGGCCAGACTTTGTTG-3′ and 5′-GATTCAACTTGCGCTCATCTTAGGC-3′. Primers (Tib Molbiol, Berlin, Germany) were designed so as not to amplify genomic DNA. The PCR cycling conditions were: 95° for 10 min, followed by 45 cycles of 95° for 15 seconds, 58° for 7 seconds, and 72° for 8 seconds. A melting curve analysis was performed for each run to ensure specificity of the primers. The data were collected using LightCycler Data Analysis Software (Roche Diagnostics), and the LightCycler Relative Quantification Software (Roche Diagnostics) was used for calibrator normalized relative quantification. FoxP3-expressing CD4⁺ CD25⁺ T cells were used as the positive control.

Statistical analysis

Statistic analyses were performed with the Kruskall–Wallis method to confirm significant differences, followed by the Mann–Whitney U-test using the statview software (SAS Institute, Cary, NC). P-values ≤0·05 were regarded as significant (*P < 0·05; **P < 0·01).

Results

Early, simultaneous engagement of antigen-specific cells in the lymphoid compartments after oral antigen administration

Using an adoptive transfer system, we monitored the activation process of antigen-specific CD4⁺ T cells following oral antigen administration. Cells from the spleens and lymph nodes of naïve DO11·10 OVA TCR tg mice were isolated and transferred intravenously into congenic sex- and age-matched BALB/c mice. The recipients received 4 × 10⁶ CD4⁺ OVA TCR tg T cells, which were identified by the KJ1·26 monoclonal antibody. Twenty hours later, the OVA-fed group were tube-fed with 10 mg of OVA, and were subsequently maintained on an OVA-containing diet until the end of the study. The remaining recipient mice (non-fed control group) were tube-fed with saline, and maintained on a standard OVA-free diet throughout the experiment.

We found clear and reproducible differences in the activation patterns of different lymphoid compartments (Fig. 2). In the Peyer's patches, transferred OVA TCR tg CD4⁺ T cells were activated within 2 hr of antigen administration, as determined by the up-regulation of CD69 (Fig. 2a). Already at 6 hr after antigen feeding, T-cell activation was induced in the CLN and MLN, and also at peripheral sites. At this time-point the Peyer's patch still expressed significantly more CD69 and CD25 than any other compartment (P ≤ 0·01; Figs 2a and b).

Figure 2.

Expression of activation markers on ovalbumin (OVA) T-cell receptor (TCR) transgenic (tg) CD4⁺ T cells from different lymphoid organs at different time-points following oral antigen administration. BALB/c mice received 4 × 10⁶ KJ1·26⁺ cells from naïve DO11·10 donors and were given an OVA-containing diet or a control diet. Transgenic CD4⁺ T cells from different lymphoid compartments of the OVA-fed mice were analysed for the expression of the activation markers (a) CD69, (b) CD25 and (c) CD122, at intervals following antigen introduction. For simplicity, measurements for the different compartments of the non-fed control mice are presented as a mean value, as no differences were detected amongst the organs. The data are presented as mean values (n = 4–6) and the error bars indicate the standard error of the mean (SEM). Statistic analyses were performed with the Mann–Whitney U-test. PP, Peyer's patch; MLN, mesenteric lymph node; CLN, celiac lymph node; PLN, peripheral lymph node.

The CLN and MLN followed similar kinetic patterns, i.e. CD69 and CD25 expression peaked after 24 hr and CD122 expression peaked after 48 hr. However, the ratio of CD25-positive cells was significantly higher in the CLN than in the MLN after 24 hr (P = 0·03), and also from day 3, 4, 5 and 7 of antigen exposure (P = 0·02, 0·004, 0·02, and 0·009, respectively) (Fig. 2b). Maximum expression of CD25 occurred after 24 hr, followed by a burst of proliferation in the CLN, as there was a significant rise in the percentage of transgenic cells between days 2 and 3 after the onset of OVA feeding.

The expression of the early activation marker CD69 revealed that the activation kinetics of the OVA TCR tg CD4⁺ T cells in the PLN was similar to that in the GALT and CLN. However, up-regulation of CD25 was significantly slower in the PLN than in the CLN and was lower after 24 hr of antigen exposure (P = 0·009). In the PLN, the level of CD25 peaked after 48 hr of antigen administration, and the level of CD122 peaked 1 day later. Despite this, CD25 expression in the PLN reached the same level as in the CLN after 3 days of antigen exposure and remained at this high level throughout the study.

The OVA TCR tg CD4⁺ T cells in the spleen were also engaged in the response to the fed antigen with a similar kinetic but a lower expression of the activation markers studied.

A high proportion of CD25^bright OVA TCR tg T cells co-expressing GITR, CTLA-4 and the integrin αEβ7 (CD103) in the CLN

While the levels of induction of CD25⁺ OVA TCR tg T cells were similar in the GALT and CLN at early time-points, a significantly higher proportion of CD25⁺ T cells was found in the CLN and PLN than in the GALT or spleen at the end of the study (Fig. 2b). This was a result of the appearance of a second wave of OVA TCR tg cells with high levels of expression of CD25 (CD25^bright). After 7 days of feeding, large proportions of CD25^bright cells were found in the CLN and PLN (29% and 25%, respectively), which were significantly higher than those found in the GALT and spleen (Figs 3c, 3d and 4a; P = 0·009). In order to distinguish cells with a regulatory phenotype from activated cells with intermediate CD25 expression, we chose to concentrate on the CD25^bright cells for further analysis.

Figure 3.

Expression of CD25 on ovalbumin (OVA) T-cell receptor (TCR) transgenic (tg) CD4⁺ T cells from the celiac lymph node (CLN) (a, c, e) and mesenteric lymph node (MLN) (b, d, f) of BALB/c recipients, after 3 days (a, b) and 7 days (c, d) of OVA administration and on day 7 for the non-fed control group (e, f). The R3 gate represents all cells stained positive for CD25, while the R4 gate only includes the CD25^bright cells. Data shown are from one representative experiment out of four (a, b, e, f) or six (c, d) independent experiments.

Figure 4.

Expression of high levels of CD25 (CD25^bright) (a) and co-expression of CD25^bright with the glucocortico-induced tumour necrosis factor (TNF) receptor family related gene (GITR) (b), CD103 (c), and cytotoxic T-lymphocyte antigen (CTLA)-4 (d) on KJ1·26⁺ CD4⁺ T cells isolated from different lymphoid compartments following adoptive transfer and oral antigen exposure. (a) The percentages of CD25^bright cells at different time-points following antigen introduction. The data are the mean values of four to six observations, and the error bars indicate the standard error of the mean (SEM). (b, d) ovalbumin (OVA) T-cell receptor (TCR) transgenic (tg) CD4⁺ T cells co-expressed CD25^bright with other markers of regulatory T cells after 7 days of antigen exposure: (b) CD25^bright GITR^bright, (c) CD25^bright CD103⁺ and (d) CD25^bright CTLA-4⁺ populations. The data are presented as box plots with the median values indicated (n = 5). PP, Peyer's patch; MLN, mesenteric lymph node; CLN, celiac lymph node; PLN, peripheral lymph node.

The percentage of cells that co-expressed CD25^bright with high GITR expression was determined after 7 days of antigen exposure (Fig. 4b). The OVA TCR tg cells in the CLN and PLN contained a significantly higher proportion of cells that stained positive for CD25^bright GITR^bright than the lymphoid compartments of the GALT and spleen (P = 0·009), mirroring the results obtained for the CD25^bright cells.

Interestingly, when the CD25^bright T cells were analysed for co-expression of CD103 and CTLA-4, the CLN was found to contain a significantly higher proportion of CD25^bright CD103⁺ transgenic cells than any other compartment analysed (P = 0·009; Fig. 4c), and also significantly more CD25^bright CTLA-4⁺ cells than the MLN, Peyer's patch and spleen (P = 0·007, P = 0·004 and P = 0·004, respectively; Fig. 4d). However, expression of CD103 was not found in the PLN. Apart from the CLN, only the MLN harboured a small fraction of transgenic cells that co-expressed CD25^bright together with CD103. The high level of CD103 expression in the CLN may indicate that a higher proportion of CLN tg T cells home to the mucosal site from this lymph node, while activated cells in the PLN seem to lack mucosal homing capacity.

Virtually all CD25^bright cells in the CLN were GITR^bright (97%) and the vast majority were CTLA-4⁺ (86%). More than half of the CD25^bright cells in the CLN were also CD103⁺ (59%). The CD25^bright transgenic cells in the CLN showed a memory phenotype, being CD45RB^low (data not shown). Thus, the phenotype of the second burst of CD25^bright cells in the CLN indicates an induction of Tregs.

Efficient suppression by tolerized transgenic cells but no difference in suppressive capacity amongst OVA TCR tg cells isolated from different lymph nodes

We were unable to use the adoptive transfer system to investigate the functional suppressive capacity of the tolerized transgenic cells, because the yield of transgenic cells after transfer was too low for suppression assays. Instead, DO11·10 OVA TCR tg mice were fed an OVA-containing diet for 10 days before the experiments. The kinetics of up-regulation of CD25 was slower in DO11·10 mice than in the adoptive transfer system, and the longer period of antigen exposure was therefore chosen. Expression of CD25 of KJ1·26⁺ T cells from OVA-fed DO11·10 mice was similar in all lymphoid compartments, with approximately 5% CD25^bright transgenic T cells (data not shown).

Responder KJ1·26⁺ OVA TCR tg cells were isolated from the the PLN and MLN of naïve DO11·10 mice and stimulated in vitro with OVA. Irradiated and KJ1·26-depleted spleen cells of naïve DO11·10 mice were used as APCs. In co-cultures, KJ1·26⁺ OVA TCR tg cells isolated from the PLN, MLN or CLN (purity > 85%) were added in a 1 : 1 ratio and in gradually lower ratios.

KJ1·26⁺ OVA TCR tg cells isolated from OVA-fed DO11·10 mice efficiently suppressed the responder cell culture in a dose-dependent manner (Fig. 5a). Cells from the CLN could be used at a lower ratio (1 : 10) than cells from the MLN (1 : 5) or PLN (1 : 2) and still significantly suppress the responder culture, in comparison with the responder control culture. However, in order to obtain one comparable result from each lymph node in each individual experiment, the dilution was plotted against the percentage of proliferation. The ratio that retained a reduction in proliferation of 50% was used as the Prol₅₀ value, and this parameter was compared between the transgenic cells from different lymph nodes in a total of six experiments (Fig. 5b). Using this approach, there was no significant difference in suppressive capacity between transgenic cells that were isolated from the different lymph nodes, as analysed by the Kruskall–Wallis test. Thus, although there was a tendency for the CLN-derived cells to have a greater overall suppressive capacity than cells from the other lymphoid sites, this difference did not reach statistical significance.

Cytometric Bead Array analysis of supernatants from different co-cultures did not reveal any differences in the production of IL-10 and IFN-γ (data not shown). For the analysis of TGF-β1 production, a sandwich enzyme-linked immunosorbent assay (ELISA) was performed on supernatants from the proliferation assays and the different co-cultures. However, it was not possible to detect levels of TGF-β1 that were above the background level, probably because of the serum content of the cell culture medium.

The suppressive activities of OVA TCR tg cells isolated from different lymph nodes are differentially blocked by antibodies to TGF-β1

In some of the suppression assays, neutralizing antibodies against TGF-β1 and IL-10 were added to the co-cultures (Fig. 6a). The neutralizing anti-IL-10 antibodies had no effect on suppression in any of the cultures. However, neutralizing anti-TGF-β1 antibodies almost completely reversed the suppression in co-cultures with tolerized transgenic cells from the PLN (83% proliferation), while the neutralizing effect of the anti-TGF-β1 antibody was slightly weaker in co-cultures with cells isolated from the MLN (69%) (Fig. 6a). Surprisingly, neutralizing anti-TGF-β1 antibodies had a weaker effect on the suppression in co-cultures with cells isolated from the CLN (45%), which indicates the influence of some other suppressive mechanism at this site.

Figure 6.

(a)Impact of neutralizing anti-transforming growth factor (TGF)-β1 and anti-interleukin (IL)-10 antibodies on the suppressive effects of ovalbumin (OVA) T-cell receptor (TCR) transgenic (tg) T cells isolated from lymph nodes of OVA-fed DO11·10 mice. KJ1·26⁺ responder cells were isolated from the mesenteric lymph node (MLN) and peripheral lymph node (PLN) of naïve DO11·10 mice and cultured. Suppressor KJ1·26⁺ T cells were isolated from the MLN, PLN, and celiac lymph node (CLN) of tolerized DO11·10 mice that were fed an OVA-containing diet and added to the responder cultures in a 1 : 1 ratio. Neutralizing anti-TGF-β1 antibody (Ab) (100 µg/ml), anti-IL-10 Ab (10 µg/ml), a combination of the two or a control Ab [anti-toxic shock syndrome toxin (TSST); 100 µg/ml] was added to the different co-cultures. The data shown are the mean values of two experiments and the error bars indicate the standard deviation (SD). (b) Intracellular expression of TGF-β1 on OVA tg CD4⁺ T cells from different lymph nodes after 7 days of oral antigen exposure, following the adoptive transfer of OVA tg cells into BALB/c mice. The data are presented as box plots, with the median values indicated (n = 5).

OVA tg CD4⁺ T cells that were adoptively transferred into BALB/c mice were analysed for intracellular cytokine expression after 7 days of oral administration. A higher percentage of TGF-β1⁺ cells were observed in the PLN than in the MLN or CLN (Fig. 6b), which supports the notion of a greater impact of TGF-β1-neutralizing antibodies on suppressor cells that originate from the PLN. However, this difference was not significant. The transgenic cells were also analysed for the expression of IL-4 and IFN-γ on day 7 of antigen exposure; very few of the total OVA TCR tg cells stained positive for IFN-γ (< 5%) or IL-4 (< 3%) (data not shown). There was no correlation between the expression of any of the cytokines and that of CD25.

No up-regulation of FoxP3 expression in the transgenic T cells isolated from the different lymph node compartments in OVA-fed DO11·10 mice

As there have been some indications of FoxP3 expression by Tregs induced in the periphery,^34,35 we investigated the expression of FoxP3 in our system. KJ1·26⁺ OVA TCR tg cells were isolated from DO11·10 mice that were maintained on a standard diet or that were fed an OVA-containing diet for 4 or 10 days. RT-PCR was performed on the cells from naïve mice or from different lymph nodes of OVA-fed mice, using the expression of HPRT as a reference (Fig. 7). Total CD4⁺, CD4⁺ CD25⁺ and CD4⁺ CD25^– T cells that were isolated from the lymph nodes of BALB/c mice were used as reference control cells (on the left in Fig. 7). However, we could not detect any up-regulation of FoxP3 in the suppressive transgenic cells from any of the lymph nodes after either 4 days (data not shown) or 10 days of feeding.

Figure 7.

Expression of Foxp3 by KJ1·26⁺ cells isolated from control DO11·10 mice or from different lymph nodes of ovalbumin (OVA)-fed DO11·10 mice. KJ1·26⁺ T cells were isolated from the lymph nodes and the expression of FoxP3 mRNA of 4–5 × 10⁵ bead-sorted cells was determined by reverse transcriptase–polymerase chain reaction (RT-PCR) and relative quantification, using hypoxanthine-guanin phosphoribosyltransferase (HPRT) as the reference gene. Bead-sorted CD4⁺, CD4⁺ CD25⁺ and CD4⁺ CD25^– cells were used as reference material. n = 3 for reference cells, n = 7 for non-fed control lymph node cells, and n = 4 for KJ1·26⁺ cells that were isolated from different lymph nodes of OVA-fed mice.

Discussion

In this paper, we report that the liver-draining CLN is an important site of T-cell priming to orally administered antigens, and is potentially involved in the development of tolerance to orally administered antigens. While the bulk of studies regarding CD4⁺ T-cell involvement in responses to oral antigens have concentrated on the local lymphoid tissues in the GALT, i.e. the Peyer's patches and MLNs,^17,31–33 a few, including a study by Smith et al., have also investigated the engagement of PLNs.³⁰ Interestingly, the same group has also recently shown that the magnitude of activation of antigen-specific T cells after oral administration of an antigen is very similar regardless of whether the antigen is administered as a tolerizing or as a priming regimen.³⁶ It is therefore necessary to try to carefully monitor the quality of the T-cell response in order to reveal the regulatory events induced by oral antigen administration. The present study indicates that the CLN is as engaged in the process of T-cell priming to fed protein antigens as the lymphoid compartments of the GALT, and that the CLN harbours a high proportion of antigen-specific T cells with a regulatory phenotype and gut homing potential after antigen feeding. Accordingly, oral OVA administration directly to OVA transgenic DO11·10 mice has been shown to result in the formation of OVA-specific Tregs locally in the liver.³⁷

The rapid and distinct activation of OVA-specific T cells in the CLN, as shown in the present paper by the up-regulation of CD69, was similar to that found in the GALT and PLN. The simultaneous engagement of T cells in the GALT and the periphery has previously been demonstrated, but in previous studies the liver-draining CLN was not included.³⁰ Interestingly, the expression of CD25 in the CLN exceeded that seen in the other lymphoid compartments. Further, a significantly higher proportion of the antigen-specific T cells developed a regulatory phenotype (CD25^bright, GITR^bright, CTLA-4⁺ and CD103⁺) in the CLN than in any other lymphoid compartment, including the GALT, which indicates that the CLN is a major site for induction of Tregs specific for dietary antigens. In the PLN, the response appears to be rapidly down-regulated, as suggested by the induction of T cells with a regulatory phenotype in the PLN. In contrast to other studies, the antigen used here was incorporated as a natural constituent into the diet, and thus administered ad libitum on a daily basis. This protocol of antigen administration was chosen to mirror a natural food intake, and can therefore serve as a relevant model for studying the natural regulatory mechanisms involved in oral tolerance. Unfortunately, the yield of transgenic cells from the adoptive transfer system does not allow functional studies of suppressive capacity in vitro.

The transgenic cells isolated from OVA-fed DO11·10 mice were found to be highly suppressive in the in vitro assay. However, despite the phenotypic distinctions revealed by the adoptive transfer system, we could not detect significant differences in suppressive efficiency between the transgenic cells from different lymph nodes of OVA-fed OVA TCR tg mice. It is possible to attribute this finding to the nature of the system studied. The DO11·10 mouse represents a less physiologically relevant model, with a high proportion (80%) of OVA TCR tg CD4⁺ T cells in the lymph nodes, as compared with a corresponding value of only about 1% in the adoptive transfer system used in this study. The high proportion of antigen-specific cells in the DO11·10 mouse is likely to alter the priming process. However, as we also in this system induce T cells with a regulatory phenotype, we still think it is important to gain information about their functional suppressive capacity.

Interestingly, we found specific up-regulation of the integrin αEβ7 (CD103) on CD25^bright cells in the CLN, and a comparably lower proportion of cells that expressed this marker in the MLN. This may indicate preferential induction of Tregs in the CLN that are destined for the intestine, as this integrin mediates adhesion to E-cadherin.^13,14,38 E-cadherin is highly expressed preferentially on epithelial cells. Thus, it is possible that transgenic cells primed in the CLN or MLN are more effective than peripherally primed cells in an in vivo readout system of suppressive function that is local to the gut, while the standard in vitro suppression assay does not reveal this possible functional difference.

Although evidence has emerged that only thymic-dependent natural CD4⁺ CD25⁺ Tregs express FoxP3, some studies have shown that a population of Tregs that express FoxP3 can also be induced in the periphery. Thus, a protocol has recently been described to convert naïve antigen-specific CD4⁺ T cells into FoxP3-expressing Tregs in vivo by prolonged subcutaneous infusion of the relevant peptide,³⁴ and FoxP3-expressing Tregs have also been induced by non-depleting antibodies in a model of transplantation tolerance.³⁵ However, acquired cytokine-dependent Tregs, such as Tr1 and Th3 cells, have not been shown to express FoxP3 in most studies.³⁹ In our model, transgenic cells from OVA-fed DO11·10 mice did not express higher levels of FoxP3 than transgenic cells from naïve mice. Thus, although the T cells were functionally suppressive, FoxP3 was not induced by antigen feeding. FoxP3 can be induced in vitro in naïve CD4⁺ T cells that are converted into Tregs by TCR stimulation in the presence of TGF-β.⁴⁰ It is possible that this conversion takes place locally in the intestinal mucosa where there is a high TGF-β drive.

Interestingly, the neutralizing anti-TGF-β1 antibody almost completely reversed the suppression provided by cells from the PLN, while this Ab had a slightly smaller effect on suppression by cells from the MLN. TGF-β-dependent suppression is consistent with the induction of Th3 cells after oral antigen administration⁴ and has been shown to be of major importance in the suppressive mechanism at PLNs in orally tolerized animals.⁴¹ However, blocking TGF-β1 only partially reversed the suppression by cells isolated from the CLN, and the neutralizing anti-IL-10 Ab had no effect in any of the lymphoid compartments studied, which suggests the participation of an additional suppressive mechanism. The role of IL-10 in oral tolerance is disputed and appears to be complex.^31,42 One possibility is that IL-10 production takes place at a later stage in the local environment of the effector site (e.g. the intestinal mucosa).

The fundamental processes involved in the establishment and maintenance of oral tolerance remain to be fully elucidated. However, there is accumulating evidence from various experimental models that tolerogenic DCs play a role in peripheral tolerance (reviewed in refs ⁴³ and ⁴⁴). Different mechanisms have been reported to underlie the capacity of DCs to arrest T-cell responses, including the induction of anergy and activation-induced cell death and induction of Tregs.⁴⁵ Although passive regulatory mechanisms are certainly involved in oral tolerance, it is now well established that an active process involving Tregs is essential (reviewed in ref. 46), which is underlined by the fact that tolerance can be transmitted to naïve recipients by adoptive transfer of CD4⁺ T cells from tolerant donors.^4,17

The liver receives not only self material but also exogenous antigens from the gut.²³ We have recently described a tolerance-inducing system for orally administered antigens, whereby antigens are sampled from the contents of the small intestine by the intestinal epithelial cells (IEC), assembled with MHC II, and loaded onto vesicular exosome structures which are termed tolerosomes.⁴⁷ They are released by the basolateral membrane⁴⁸ and enter the portal circulation, from where they can be isolated. The liver contains DCs that effectively clear the blood of particulate matter of similar size to tolerosomes (about 40 nm).⁴⁹

Liver-derived DCs have been shown to produce the anti-inflammatory cytokine IL-10^{25, 50} and induce T cells, which in turn produce IL-10.^23,50 This might be explained by the unique microenvironment in the liver. The Kuppfer cells, which reside in the liver sinusoids, respond to physiological concentrations (ranging from 10 pg/ml to 1 ng/ml) of lipopolysaccharide (LPS) in the portal-venous blood by producing IL-10, TNF-α, TGF-β and prostanoids,²⁴ which have all been shown to promote the development of tolerance-inducing APCs.^51–54 In addition, natural killer T (NKT) cells, which are abundant in the liver, have been shown to produce large amounts of IL-10 and IL-4, presumably in response to certain bacterial products that are presented via the CD1 molecules on DCs.⁵⁵ Thus, the unique microenvironment in the liver might cause ‘differential’ activation of resident DCs, thereby favouring the induction of tolerance to self antigens as well as microbial and dietary antigens. Thus, the engagement of transgenic T cells and their superior priming for switching to the regulatory phenotype, which we noted in the CLN in response to an orally administered antigen, may well be explained by the unique conditioning of DCs in the liver. Furthermore, the expression of the gut-trophic integrin CD103 was found almost exclusively on Treg-like cells in the CLN, which indicates specific homing of these Treg to the intestinal mucosa. Finally, transgenic T cells isolated from the CLN of OVA-fed DO11·10 mice were less dependent on TGF-β1 for their suppression than cells isolated from other lymph nodes. Thus, both the phenotypic and functional findings reported here indicate an alternative mechanism of Treg priming in the CLN as compared with cells primed in other lymphoid compartments following the feeding of antigen.

In conclusion, we have shown that the CLN is an important compartment involved in the immune response to orally administered antigens, and should be taken into account when exploring the mechanisms of oral tolerance.

Abbreviations

APC

antigen-presenting cell

CLN

celiac lymph node

GALT

gut-associated lymphoid tissues

GITR

glucocortico-induced TNF receptor family related gene

IBD

inflammatory bowel disease

IFN

interferon

Ig

immunoglobulin

MLN

mesenteric lymph node

OVA

ovalbumin

PE

phycoerythrin

PLN

peripheral lymph node

PP

Peyer's patches

Tg

transgenic

TNF

tumour necrosis factor

Tr1

regulatory T cell type 1

Treg

regulatory T cell