Active delivery of recombinant autoantigens or allergens at the intestinal mucosa by genetically modified Lactococcus lactis (LL) provides a novel therapeutic approach for the induction of tolerance. Celiac disease is associated with either HLA-DQ2- or HLA-DQ8-restricted responses to specific antigenic epitopes of gliadin, and may be treated by induction of Ag-specific tolerance. We investigated whether oral administration of LL-delivered DQ8-specific gliadin epitope induces Ag-specific tolerance. LL was engineered to secrete a deamidated DQ8 gliadin epitope (LL-eDQ8d) and the induction of Ag-specific tolerance was studied in NOD AB° DQ8 transgenic mice. Tolerance was assessed by delayed-type hypersensitivity reaction, cytokine measurements, eDQ8d-specific proliferation, and regulatory T cell analysis. Oral administration of LL-eDQ8d induced suppression of local and systemic DQ8-restricted T cell responses in NOD AB° DQ8 transgenic mice. Treatment resulted in an Ag-specific decrease of the proliferative capacity of inguinal lymph node (ILN) cells and lamina propria cells. Production of IL-10 and TGF-β and a significant induction of Foxp3+ regulatory T cells were associated with the eDQ8d-specific suppression induced by LL-eDQ8d. These data provide support for the development of effective therapeutic approaches for gluten-sensitive disorders using orally administered Ag-secreting LL. Such treatments may be effective even in the setting of established hypersensitivity.
The chronic small intestinal inflammation that defines celiac disease is characterized by flattening of the villous architecture and massive infiltration of T cells that release proinflammatory cytokines, such as IFN-γ and IL-2. Celiac disease is caused by a loss of tolerance to ingested dietary gliadin and is mediated by HLA-DQ2- or HLA-DQ8-restricted T cell response (1). Effective treatment can only be reached by a socially restrictive diet that requires lifelong abstinence from foods that contain gliadin, which is present in wheat and proteins from related cereals like rye or barley. Although a strict, gluten-free diet can lead to healing of the intestine, the intolerance to gluten is permanent (2) and better therapeutic options are needed (3).
Oral tolerance is defined as the induction of Ag-specific suppression of immune responses to an Ag by its prior oral feeding and is an attractive therapeutic approach for treatment of allergic, autoimmune, and inflammatory diseases. In contrast to most other autoimmune diseases, the trigger, the genetic association, and the highly specific humoral response for celiac disease have been well characterized (4). Because disease activity is strongly correlated to the presence and dosage of Ag, the induction of Ag-specific oral tolerance is an attractive therapeutic approach. Oral tolerance is mediated by multiple mechanisms such as anergy, deletion, and/or active suppression of Ag-specific effector T cells by regulatory T cells (Tregs).3 The efficacy of oral tolerance in preventing the induction of autoimmune and allergic diseases has been clearly demonstrated in several animal models, but unfortunately previous clinical attempts to induce oral tolerance for therapeutic purposes have failed (5, 6). These failures are related to the source, the purity, and the amount of (auto)antigen needed and the mode of presentation of the Ag to the mucosal immune system (7). Previously, a delivery system (TopAct; ActoGeniX), based on living Lactococcus lactis (LL) strains (ActoBiotics; ActoGeniX), for the oral administration of biopharmaceuticals has been described and validated in preclinical experiments as well as a clinical trial (8, 9, 10, 11). More recently, we reported that active in situ synthesis and mucosal delivery of OVA by genetically engineered LL induces Ag-specific oral tolerance in OVA TCR transgenic mice (DO11.10) by the induction of CD4+CD25− Tregs that function through a TGF-β-dependent mechanism (12). In the present study, we further developed this approach to investigate the possible induction of Ag-specific tolerance in a well established, genotypic-celiac disease mouse model (13). We therefore genetically engineered LL to secrete a deamidated DQ8-specific gliadin epitope (DQ8d) that is immunodominant for DQ8-mediated T cell responses (LL-eDQ8d) and subsequently studied its oral supplementation in DQ8d-immunized NOD AB° DQ8 transgenic mice. NOD AB° DQ8 is a mouse model that uses the NOD background, which contributes to autoimmunity and pathogenesis in combination with a human DQ8 MHC class II transgene, which contributes to the sensitivity to gliadin (13).
In this paper, we report that the active mucosal delivery of DQ8d by genetically modified LL induces suppression of systemic DQ8-specific T cell responses in NOD AB° DQ8 transgenic mice and provides a method for the induction of DQ8d Ag-specific tolerance. Moreover, this approach provides a method to deliver the right Ags in an adequate manner to the intestinal mucosal immune system in the context of a noncolonizing, nonpathogenic bacterium (7) and has the potential for being an effective and nontoxic treatment of celiac disease.
Materials and Methods
Bacteria and media
The genetically modified LL MG1363 strain was used throughout this study (14). Bacteria were cultured in GM17E medium, consisting of M17 broth (Difco) supplemented with 0.5% glucose and 5 μg/ml erythromycin (Abbott Laboratories). Stock suspensions of LL strains were stored at −20°C in 50% glycerol in GM17E medium. Stock suspensions were diluted 200-fold in GM17E medium and incubated at 30°C overnight. Within 16 h of culture, a saturation density of 2 × 109 colony-forming units per ml was reached. Bacteria were harvested by centrifugation and 10-fold concentrated in BM9 inoculation buffer at 2 × 109 bacteria/100 μl. The dose used in all experiments was 100 μl of this suspension administered daily by intragastric catheter.
Construction of genetically modified LL-eDQ8d
A genetically engineered LL strain, designated LL-eDQ8d, was constructed to secrete an immunodominant DQ8d (Fig. 1⇓). The sequence encoding the DQ8d was retrieved from published data (15). In summary, two glutamine residues within the α-gliadin peptide were changed into glutamic acids to stimulate the deamidated immunodominant α-gliadin response for DQ8-carrying celiac disease patients, and this epitope is recognized by T cells of NOD AB° DQ8 mice (16) (Table 1⇓). The DQ8d cDNA fragment was synthetically constructed (Operon Technologies) and amplified by PCR using the following forward and reverse primers: 5′-caatacccatcaggtgaaggttc-3′ and 5′-cgactagttaagcttgtgggttttcttgtgat-3′. For detection purposes an e-tag was attached to the fragment (Table. 1⇓). To add the e-tag to the 5′ end of DQ8d gene, we used the PCR product that was produced in step 1 (DQ8d) as template in a PCR with oligonucleotides 5′-ggtgctccagttccatacccagatccacttgaaccacgtcaatacccatca-3′ and 5′-cgactagttaagcttgtgggttttcttgtgat-3′. The amplified fragment was fused to the Usp45 secretion signal of the erythromycin-resistant pT1NX vector, downstream of the lactococcal P1 promotor (17, 18). MG1363 strains transformed with plasmids carrying eDQ8d cDNA were designated LL-eDQ8d. The LL-pT1NX, which is MG1363 containing the empty vector (pT1NX) served as control. The coding sequence of the circular vector was confirmed by sequence analysis and constitutive eDQ8d secretion did not alter the growth rate of LL.
Functional analysis of secreted epitopes
For functional analysis of the secreted eDQ8d epitope, a proliferation assay with human T cell clones derived from the intestines of celiac disease patients was performed. Bacteria were grown overnight as described before, diluted 1/50, and grown for another 4 or 6 h. T cell clones specific for gluten were generated from a small intestinal biopsy taken from an adult Dutch celiac patient that had been on a gluten-free diet for several years, as described (19). The patient gave informed consent to the study, which was approved by the ethics committee of Leiden University Medical Center. The patient was typed serologically to be HLA-DR3/4, DQ2/8, thus carrying both celiac disease-associated DQ dimers. T cell clone II29, when bound to HLA-DQ8, was found to respond to an α-gliadin-derived peptide with a minimal 9 amino acid core QGSFQPSQQ. Deamidation of the P1 and/or P9 glutamine residue into glutamic acid by the activity of tissue transglutaminase was found to substantially enhance the T cell stimulatory capacity of this gluten peptide (20). Proliferation assays were performed in duplicate or triplicate in 150 μl culture Iscoves medium in 96-well flat-bottom plates (BD Falcon) using 104 T cells stimulated with 105 HLA-DQ-matched and 3000 rad irradiated PBMC in the absence or presence of supernatant at several concentrations. After 48 h, cultures were pulsed with 0.5 μCi of [3H]thymidine, harvested 18 h thereafter upon which [3H]thymidine incorporation was determined as a measure for proliferation.
Transgenic mice that express HLA-DQ8 in an endogenous MHC II-deficient background (AB° DQ8) were backcrossed to NOD mice for 10 generations and intercrossed to produce congenic NOD AB° DQ8 mice, as described previously (13).
NOD AB° DQ8 transgenic and NOD AB° mice were previously generated and bred at Department of Immunology, Mayo Clinic (Rochester, MN). Seven- to sixteen-week-old mice were used for the experiments. Mice were weaned and maintained on gluten-free chow and were kept in a conventional animal facility until 8–12 wk of age. All experiments performed were approved by the Institutional Animal Care and Use Committee of the Mayo Clinic College of Medicine.
Ags and Abs
Oral feeding and delayed-type hypersensitivity (DTH) reaction
NOD AB° DQ8 mice on a gluten-free chow were sensitized by s.c. injection of 100 μg eDQ8d peptide in 100 μl of a 1:1 CFA saline solution (Difco) in the tail base at day 1 (23). The peptide used for the sensitization had the same sequence as the secreted epitope. Mice were fed BM9 as a negative control or LL-pT1NX or LL-eDQ8d dissolved in 100 μl BM9 at days 1–10. Feedings were performed by intragastric administrations of BM9 or bacterial suspensions using an 18-gauge stainless gavage needle. Ten days after immunization, Ag-specific DTH responses were assessed. Twenty-four hours thereafter, DTH measurements were performed. For measurement of Ag-specific DTH responses, baseline ear-thickness was measured using an engineer’s micrometer (Mitutoyo). Mice were then injected with 10 μg of eDQ8d in 10 μl of saline in the auricle of the ear. The ear-thickness was measured again in a blinded fashion at 24 h after challenge. DTH responses were expressed as the difference in the baseline ear-thickness and the ear-thickness 24 h after eDQ8d injection. Subsequently, mice were killed, spleen and lymph nodes were harvested, and cells were assessed for DQ8d-specific proliferation and cytokine production. To test for e-tag interference, NOD AB° DQ8 mice were immunized with 100 μg of DQ8d peptides with or without e-tag in 100 μl of 1:1 CFA solution in the tail base at day 1. At day 7, mouse DTH measurements were performed as described above with 10 μg of DQ8d with or without e-tag, corresponding to the peptide used for the immunization.
Isolation of lamina propria cells
Lamina propria cells were isolated using the protocol as previously described (24). Briefly, mice were euthanized by CO2 inhalation and the small intestine removed and cut longitudinally and laterally into 2-cm long pieces using a scalpel blade. The tissue was then washed six times using calcium/magnesium free/HEPES solution (1× Hanks’ balanced salt solution, 15 mM HEPES, 2% FBS) followed by 1 hour incubation in calcium/magnesium free/FBS/EDTA solution (1× Hanks’ balanced salt solution, 10% FBS, 15 mM HEPES, 5 mM EDTA, 100 μg penicillin/streptomycin) to remove epithelial cells. A subsequent collagenase (100 U/ml) digestion was performed for 1 hour to release the lamina propria lymphocytes.
Cell cultures, proliferation, and cytokine production assays
Cell suspensions of spleen and lymph nodes were prepared at day 11 of the experiment by homogenizing the tissue with a tissue grinder (VWR International) in 1× PBS. Erythrocytes were removed from the spleen cell suspensions by incubation with ammonium chloride/potassium lysis buffer. Cells were incubated in 96-well microtiter plates at 5 × 105 cells/well in 0.2-ml volumes at 37°C in RPMI 1640 (1.5% HEPES, 1% Penstrep, and 10% FBS) with supplements containing either medium alone, 10 μg/ml Con A, 50 μg/ml irrelevant peptide (HIV), or 50 μg/ml eDQ8d epitope. In a separate experiment IL-10, TGF-β, IL10 and TGF-β, or LAP-neutralizing Abs were added to splenocytes of LL-eDQ8d treated mice. After 24 h, proliferation was assessed by addition of 1 μCi/well [3H]thymidine for the last 24 h of culture. DNA-bound radioactivity was harvested onto glass fiber filter mats (PerkinElmer) and thymidine-incorporation measured on a scintillation counter (PerkinElmer). Results were expressed as mean cpm of triplicate wells. For the neutralizing Ab proliferation assay, the results were expressed as the percentage of LL-eDQ8d proliferation compared with the BM9-treated group. For cytokine measurements, supernatants of the cell cultures used in the different proliferation assays, described above, were collected after 24 h of culture and frozen at −20°C until cytokine analysis was performed. Cytokine production was quantified using the mouse inflammation cytometric bead assay (BD Biosciences).
T cell proliferation assay using lamina propria cells
Lamina propria cells were isolated from mice that had been administered medium (BM9), LL-pT1NX, or LL-eDQ8d. These cells (5 × 106 cells/ml) were then incubated with medium (RPMI 1640; Sigma-Aldrich), eDQ8d epitope (50 μg/ml), or irrelevant peptide (PLP-14) (50 μg/ml), for 24 h before the addition of [3H]thymidine for an additional 24 h. Cells were then harvested in a similar fashion as the splenocytes.
Flow cytometric analysis
Spleens and gut-associated lymph node tissue (GALT) of BM9-, LL-pT1NX-, or LL-eDQ8d-treated mice were isolated, prepared as described above, and stained for CD4, CD25, and Foxp3. Intracellular staining was performed for Foxp3 according to the manufacturer’s instructions (eBiosciences) and subsequently measured using flow cytometry on a BD Biosciences FACSCalibur (Mayo Clinic College of Medicine Flow Cytometry/Optical Morphology Core Facility). Cells were gated on CD4+CD25+ and CD4+CD25− subpopulations and within these populations Foxp3 histograms were used to determine mean fluorescence intensity (MFI).
Results from cytokine measurements are expressed as mean ± SEM. eDQ8d-specific proliferation, ear-thickness, and cytokine measurements were tested for significance using one-way ANOVA followed by Student’s t test comparison to determine the differences between individual groups. For Figs. 3⇓ and 4⇓, statistical significance is designated as follows; ∗, p < 0.05; ∗∗, p < 0.01; and ∗∗∗, p < 0.001.
Synthesis and secretion of functional DQ8d
In vitro synthesis and functionality of the secreted DQ8d was confirmed by a proliferation assay with human DQ8 T cell clones derived from the intestine of celiac disease patients. T cell clones derived from the intestine of celiac patients were stimulated with supernatant of a LL-eDQ8d culture at different concentrations (Fig. 2⇓A). Control (LL-pT1NX) supernatant did not induce proliferation (data not shown). The secreted immunodominant DQ8d peptide contains an amino-terminal e-tag for detection purposes. To exclude a possible interference of the e-tag with the functional properties of the peptide, NOD AB° DQ8 transgenic mice were immunized with DQ8d epitopes with or without e-tag at day 0. At day 7, DTH response measurements were performed with the injection of 10 μg of DQ8d epitope. The presence of an e-tag did not change the DTH response in DQ8d-immunized mice (Fig. 2⇓B). Immediately after the DTH measurement, bulk splenocytes and bulk ILN cells were isolated and stimulated ex vivo with DQ8d epitopes with or without e-tag. Again, the presence of the e-tag did not change the immune-stimulating properties of the DQ8d epitope (data not shown). These data demonstrate that LL-derived eDQ8d is fully bioactive and that the addition of the e-tag does not interfere with its functionality.
Suppression of the DTH and proliferative response by LL-eDQ8d
We investigated the effect of oral administration of LL-eDQ8d on the eDQ8d-induced DTH (25). NOD AB° DQ8 transgenic mice were immunized by eDQ8d and fed BM9 (inoculation buffer, as a negative control), LL-pT1NX (empty vector control), or LL-eDQ8d intragastrically for 10 consecutive days. On day 10, mouse ears were injected with 10 μg of eDQ8d and 24 h later ear-thickness measurements were performed. Control mice (fed BM9) were clearly immunized to eDQ8d. Daily intragastric administration of LL-eDQ8d significantly reduced the DTH response (13.1 × 10−2 mm vs 5.1 × 10−2 mm, p = 0.003; Fig. 3⇓A). Ear swelling was also slightly reduced in LL-pT1NX-treated mice compared with controls (9.3 × 10−2 mm vs 13.1 × 10−2 mm, p = 0.034) but to a much lesser degree than in LL-eDQ8d-treated mice. NOD AB° mice (without DQ8 transgene) showed only a minor increase in ear thickness (3.2 × 10−2 mm). These data indicate that orally administered LL-eDQ8d suppresses systemic inflammatory T cell responses in immunized NOD AB° DQ8 transgenic mice and that the secreted Ag is necessary for induction of a significant tolerogenic effect.
Peripheral immune responses were further analyzed by investigating eDQ8d-specific proliferation of spleen and draining ILN cells. Splenocytes of mice treated with BM9, LL-pT1NX, or LL-eDQ8d were isolated on day 11 after immunization, and the eDQ8d-specific proliferative response was assessed by ex vivo stimulation with eDQ8d peptides or irrelevant peptide. Splenocytes of immunized mice showed a high eDQ8d-specific proliferative response that was significantly suppressed by daily intragastric administration of LL-eDQ8d compared with the control BM9 fed mice (cpm: 13.1 × 103 vs 5.6 × 103, p = 0.048; Fig. 3⇑B). However, the ILN are the primary Ag recognition site in this immunization protocol; therefore, we also examined the proliferative capacity of these lymphocytes. Proliferation of ILN cells was significantly decreased in the LL-eDQ8d treated group as compared with the BM9 treated group (1.4 × 103 vs 2.8 × 103 cpm, p = 0.002; Fig. 3⇑C). Splenocytes and ILN cells did not show any proliferative response with the addition of an irrelevant control peptide in all three treatment groups.
To determine whether Ag-specific suppression exists in the lamina propria cells, these cells were isolated from mice administered LL-eDQ8d as well as mice administered BM9 alone or LL-pT1NX. These cells were then treated in vivo with medium, eDQ8d, or an irrelevant control peptide. A high background of proliferation was observed in lamina propria cells in all three groups of treated mice. However, the proliferation was suppressed with the addition of the eDQ8d peptide only in mice treated with LL-eDQ8d (Fig. 3⇑D). Proliferation in the lamina propria was not diminished with the addition of irrelevant control peptide. This demonstrates that in the lamina propria, the suppression induced by the feeding of the LL-eDQ8d is specific to eDQ8d peptide.
To investigate the mechanisms behind the reduction of Ag-induced T cell proliferation, we determined cytokine profiles of ex vivo stimulated splenocytes or ILN cells. Ex vivo eDQ8d-stimulated spleen cells showed a significant up-regulation of IL-10 (52.6 vs 5.4 pg/ml, p = 0.002) and a down-regulation of IL-12 production (8.0 vs 27.6 pg/ml) only in the LL-eDQ8d-treated group compared with the negative control (BM9; Fig. 3⇑, E and F). Moreover, LL-eDQ8d treatment significantly reduced the eDQ8d-induced IFN-γ production (380 vs 1328 pg/ml, p = 0.009) in the ILNs compared with the negative control (BM9)-treated mice (Fig. 3⇑G). There was no difference between the cytokine levels across treatment groups when we stimulated the splenocytes and ILN cells with irrelevant peptide. Addition of irrelevant peptide also did not change the levels of IL-10 or IL-12p70 from the medium in three treatment groups (BM9, LL-pT1NX, and LL-eDQ8d). Together, the proliferative and cytokine data indicate that LL-eDQ8d treatment is able to suppress T cell responses systemically in NOD AB° DQ8 transgenic mice in an Ag-specific manner.
Critical role for both TGF-β and IL-10 in LL-eDQ8d-mediated suppression
The functional importance of TGF-β, IL-10, and LAP (membrane-associated TGF-β) for the eDQ8d-specific splenic proliferative response of splenocytes from LL-eDQ8d treated mice compared with the BM9 control group was assessed using neutralizing Abs. The individual neutralization of IL-10, TGF-β, or LAP did not significantly interfere with the decreased splenic proliferative response of LL-eDQ8d-treated mice, but adding a combination of TGF-β and IL-10 neutralizing mAbs completely abolished the decreased eDQ8d-specific proliferative capacity of splenocytes (4.9 × 103 vs 1.6 × 103 cpm, p < 0.01; Fig. 4⇓). These data strongly suggest that the T cell activation suppression mediated by LL-eDQ8d treatment is dependent upon interplay between IL-10 and TGF-β.
Increase in Foxp3 expression by CD4+CD25+ and CD4+CD25− T cells
To analyze the role of Tregs in the induction of LLeDQ8d-induced tolerance, we investigated the expression of Foxp3 within the CD4+ T cell population by FACS analysis. A significant increase in the number of Foxp3+ CD4+CD25+ cells as well as Foxp3+ CD4+CD25− cells was seen in the spleen of LL-eDQ8d-treated mice as compared with the control group (BM9; MFI 171 vs 61 and 35 vs 6, respectively, Fig. 5⇓, A and B). An increased number of Foxp3+CD4+CD25− cells was also detected in the GALT of the LL-eDQ8d-treated mice as compared with the BM9-treated mice (MFI in M1 73 vs 30; Fig. 5⇓C), but not in the GALT CD4+CD25+ population. LL-pT1NX feeding also increased the number of Foxp3+CD4+CD25− cells in the spleen, but this was to a lesser extent than the LL-eDQ8d-treated mice (MFI 15 vs 35, respectively). These data suggest a strong association between the increase in Foxp3+ T cells and the suppression of the immune response to gliadin by the oral administration of the LL-eDQ8d bacteria.
Our data demonstrate that genetically modified LL can be used for mucosal delivery of functional immunodominant Ags, and that this approach suppresses inflammatory Ag-specific T cell responses in gliadin-sensitized NOD AB° DQ8 transgenic mice. Furthermore, our results suggest that the induced suppression is Ag-specific and mediated by Foxp3+ CD4+CD25+ and CD25− Tregs that possibly function through an IL-10- and TGF-β-dependent mechanism.
Successful clinical application of Ag-specific mucosal tolerance for the treatment of human diseases has been difficult to achieve and critically depends on several factors, including the purity, source, dose, and the mode of Ag presentation to the mucosal immune system (7). Several protocols for induction of oral tolerance, including oral administration of the Ag with IL-10, have been shown to induce Ag-specific Tr1 cells that suppress undesired immune responses toward self-Ags, allergens, and food Ags. Although a previous clinical attempt to induce tolerance in celiac disease by the administration of rhIL-10 in refractory celiac disease patients was ineffective (26), strategies to boost the number and/or function of Ag-specific Tr-1 cells may offer new therapeutic opportunities. This notion is supported by the finding that gliadin-specific mucosal Tregs from celiac disease patients are able to suppress proliferation of pathogenic Th0 cells (27, 28).
Both dendritic cells (DC) and Tregs are critically involved in tolerance induction (29, 30). We recently demonstrated that exposure to LL alters DC phenotype and function, which in the presence of simultaneous exposure to a DC-presented Ag might result in the generation of an Ag-specific Treg subset (12). We hypothesize that induction of Ag-specific Tregs in our experiments was mediated by altered presentation of the immunodominant peptide by DCs, and our observation that LL-eDQ8d treatment interfered with IL-12 production of splenocytes suggests that at least part of the tolerogenic effect is DC-mediated. However, it should be noted that the spleen is not the primary Ag-recognition site in our model and, alternatively, activation of Tregs may have resulted in a reduced activation of APCs. In fact, our results with the lamina propria cells demonstrate that the GALT is a site of induction of regulatory cells in response to mucosal administration of LL treatments.
We further demonstrated that LL-eDQ8d treatment reduced peripheral DTH responses as well as eDQ8d-specific proliferation of lamina propria cells, bulk splenocytes, and ILN cells. LL-pT1NX treatment also somewhat reduced the DTH and splenic proliferative capacity but was less pronounced than in the LL-eDQ8d treated mice. The LL-eDQ8d treatment-mediated IL-10 secretion and the reduction of IL-12 and IFN-γ production that was found after ex vivo stimulation of splenocytes and ILN cells respectively was not observed in the LL-pT1NX-treated mice. Moreover, splenocytes of sensitized LL-eDQ8d-treated mice were not sensitive to stimulation with an irrelevant Ag. These data confirm our previous findings that the tolerogenic effect, at least in part, is LL-mediated, but that the codelivery of low-dose Ag, in this case DQ8d, greatly enhances the induction of Ag-specific oral tolerance.
In recent years, it has become apparent that Tregs play a critical role in the induction and maintenance of oral tolerance (31, 32). Still, many questions need to be answered concerning the phenotype and complexity of Tregs, as well as the precise role and different overlaps in oral tolerance. Several phenotypically and functionally distinct Treg subsets have been implicated in suppression of intestinal inflammation and induction of oral tolerance, including adaptive Tregs, comprising Th3 and Tr1 cells, and naturally occurring Tregs (nTregs), which maintain tolerance to self-Ag under normal physiological conditions. Although it is probable that nTregs play a central role in regulating gut immune homeostasis, their precise function remains to be characterized (5, 33, 34). Furthermore, recently a separate category of Tregs has been described that acquires Foxp3 upon TGF-β stimulation. These so-called inducible Tregs have regulatory functions both in vitro and in vivo (35, 36). This recently discovered subset mimics the Tregs induced in this model, as both subsets are induced in the periphery, express Foxp3, and are critically dependent on TGF-β/IL-10.
To map the Tregs that mediated oral tolerance in our experiments, we studied the functional importance of TGF-β, IL-10, and LAP (membrane-associated TGF-β) on the eDQ8d-specific splenic proliferative response using neutralizing Abs. Interestingly, only the combined neutralization of IL-10 and TGF-β interfered with the proliferative capacity of the splenocytes, suggesting an interactive role for both anti-inflammatory cytokines. The exact mechanism by which these cytokines function is not completely understood but both cytokines frequently interact during regulatory Treg responses (37, 38, 39).
We found a significant Foxp3 up-regulation in both the mucosal and the splenic CD4+ T cell population. It is known that Ag-specific TGF-β-producing Th3 cells drive the differentiation of Ag-specific Foxp3+ regulatory cells in the periphery (40). Furthermore TGF-β-dependent conversion of peripheral CD4+CD25− T cells into CD25+, CD45RB−/low suppressor cells has been reported (35). Unfortunately, specific markers for distinguishing between nTregs and inducible Tregs are lacking. Our data therefore suggest that either mucosal CD4+CD25−Foxp3+ cells induced by LL-eDQ8d treatment play a regulatory role or that they can be eventually converted into CD4+CD25+ Tregs (41). It has been shown that oral tolerance induced by cholera toxin B subunit-conjugated Ag is associated with increased TGF-β production and the generation of Foxp3+CD25+CD4+ and both Foxp3+ and Foxp3−CD25−CD4+ Tregs (42). The relationship between thymus-derived CD4+CD25+Foxp3+ nTregs and other subpopulations induced in the periphery, such as peripherally generated CD4+ Foxp3+ cells, as well as Tr1 and Th3 cells, needs clarification, but these data suggest a significant overlap and interactive function in the induction of mucosal tolerance.
Direct in vivo induction of tolerogenic DC or Treg is a major target for immunotherapy for allergic, autoimmune, and several inflammatory diseases and can be achieved by exposing the mucosal immune system to low doses of Ag (7, 43, 44).
In this paper, we report that oral supplementation of a genetically modified LL-secreting DQ8d peptides greatly reduces systemic immune responses induced by Ag in DQ8d-immunized NOD AB° DQ8 transgenic mice. The suppression is mediated by the induction of Foxp3+ Tregs that are dependent on both TGF-β and IL-10. These observations support the development of gut delivered bacteria-enhanced oral tolerance for the treatment of both mucosal and systemic autoimmune, inflammatory, or allergic diseases by specific Ag-secreting LL. With such studies, this type of therapy may be considered a viable option for treating gluten-sensitive disorders such as celiac disease.
We thank Dr. Robin Patel and the Microbiology Research Facility at the Mayo Clinic Rochester for their expertise, help, and advice. We thank M. S. ten Brink and J. B. Daalhuisen for their technical support.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 The trademarks TopAct and ActoBiotics are used with the permission of ActoGeniX. This study was supported by National Institutes of Health Grant DK071003.
↵2 Address correspondence and reprint requests to Dr. Joseph A. Murray, Department of Medicine, Division of Gastroenterology and Hepatology, Mayo Clinic, 200 First Street, SW, Rochester, MN 55905. E-mail address:
↵3 Abbreviations used in this paper: Treg, regulatory T cell; DC, dendritic cell; DTH, delayed type hypersensitivity; DQ8d, deamidated DQ8-specific gliadin epitope; eDQ8d, deamidated DQ8-specific gliadin epitope that is immunodominant for DQ8-mediated T cell responses; GALT, gut-associated lymphoid tissue; ILN, inguinal lymph node; LAP, latency-associated peptide; LL, Lactococcus lactis; LL-pT1NX, LL MG1363 containing empty vector; LL-eDQ8d, LL-secreting eDQ8d; MFI, mean fluorescence intensity; nTregs, naturally occurring Tregs.
- Received September 2, 2008.
- Accepted May 25, 2009.
- Copyright © 2009 by The American Association of Immunologists, Inc.