Oral Tolerance Induction with Antigen Conjugated to Cholera Toxin B Subunit Generates Both Foxp3⁺CD25⁺ and Foxp3⁻CD25⁻ CD4⁺ Regulatory T Cells¹

Oral administration of Ag can induce peripheral, so-called oral tolerance, characterized by a decreased ability to respond immunologically to systemic immunization with the same Ag (1, 2). Although when using excessive Ag doses oral tolerance may result from induction of anergy and even clonal deletion of T cells (3), under more physiological conditions oral tolerance is largely mediated through the mucosal induction of regulatory CD4⁺ T cells producing immunomodulatory cytokines such as IL-10 and TGF-β that can suppress the systemic response to the same Ag. Two regulatory T cells (T_reg)³ that contribute to oral tolerance, Tr1 cells, which secrete IL-10 (4), and Th3 cells, which secrete TGF-β (1), have been described. CD25⁺CD4⁺ T_reg (5) have been found in Peyer’s patches (PP) and mesenteric (LN) lymph node (MLN) within a short time after oral administration of high-dose OVA (6, 7, 8). The CD25⁺ T_reg population has a crucial role in maintaining immune tolerance to self-Ag, and mice depleted in T_reg cells or humans lacking T_reg develop severe autoimmune disease (9, 10, 11). CD25⁺ T_reg can produce either IL-10 or TGF-β or both (8, 11) and also express CTLA-4 (12). Although T_reg is anergic in vitro, it can be expanded in an Ag-specific manner in vivo after immunization (13). Importantly, these cells may also confer suppressor activity on CD4⁺ T cells by inducing the expression of the Foxp3 gene, which appears to be critical for the generation and suppressive function of CD25⁺ T_reg cells (14, 15, 16).

Despite many attractive features of oral tolerance induction as a means of potential Ag-specific immunotherapy in autoimmune and allergic disorders, it appears clear that to achieve this goal there is a need for development of Ag formulations and administration regimens with improved tolerance-inducing potency (1, 2). In earlier studies, we and others have shown (17) that even microgram amounts of relevant Ag conjugated to the nontoxic protein of cholera toxin (CT) B subunit (CTB) can induce oral tolerance with unique efficacy and in an Ag-specific manner, in experimental models, suppress development of autoimmune diseases (18, 19, 20, 21). Oral tolerance induction by CTB-conjugated Ag is associated with enhanced IL-10 and TGF-β1 mRNA expression and cytokine production by T_reg (19, 22). However, the extent to which the superior oral tolerance induced by Ag conjugated to CTB is associated with mucosal induction of Ag-specific T_reg or Foxp3 expression is largely unknown.

Therefore, in this study, in which we used OVA as a model Ag, we investigated the development and suppressive activity of Ag-specific CD25⁺ and CD25⁻ T_reg in response to orally administered OVA/CTB conjugate or OVA alone. We found that OVA/CTB treatment was much more efficient than OVA alone in inducing Ag-specific CD25⁺; however, we also found that CD25⁻ regulatory CD4⁺ T cells can strongly suppress effector T cell responses to Ag in vitro and in vivo. Both the CD25⁺ and the CD25⁻ suppressor T cell populations induced by OVA/CTB feeding expressed the Foxp3 gene, although to a different extent, and oral OVA/CTB treatment efficiently generated Foxp3⁺CD25⁺ T_reg cells from Foxp3⁻CD25⁻CD4⁺ T cells independent of natural CD25⁺ T_reg cells.

For the experiments we used four strains of 6- to 8-wk-old female mice: 1) BALB/c (B & K Universal); 2) DO11.10 OVA TCR transgenic (Tg) mice on BALB/c background (The Jackson Laboratory), a clone with nearly 50% of the CD4⁺ T cells expressing a TCR specific for the peptide_323–339 fragment of OVA; 3) OVA TCR Tg OT-II mice with the same TCR specificity as DO11.10, but on the C57BL/6 background (a gift from Dr. M. J. Wick, Göteborg University, Göteborg, Sweden); and 4) Rag1^−/− mice (The Jackson Laboratory). The mice were kept in ventilated cages under specific pathogen-free conditions at the Department of Experimental Biomedicine, Göteborg University, Sweden. The Göteborg University Ethical Committee for Animal Experimentation approved these studies.

OVA protein (grade VII) was purchased from Sigma-Aldrich, and OVA peptide_323–339 (ISQAVHAAHAEINEAGR) was obtained from TAG Copenhagen. CT was purchased from List Biological Laboratories. Highly purified recombinant CTB was provided by SBL Vaccines.

OVA protein was chemically coupled to CTB using N-succininmidyl (3-(2-pyridyl)-dithio)propionate (Pierce) as a bifunctional coupling reagent as described (17). Coupled OVA/CTB was purified and quantified by fast protein liquid chromatography (FPLC) gel filtration (Superdex 200 16/60 column; Pharmacia Biotech) using the Biologic Workstation FPLC system (Bio-Rad)). After use, the lots of purified conjugate were analyzed by GM1-ELISA using biotinylated anti-CTB mAbs and were shown to have retained GM1-binding activity (22). They were also shown to have similar, strong capacity to induce OVA-specific T cell proliferation when tested on DO11.10 splenocytes. In these assays, they were not significantly inhibited by preincubation and coculture with polymyxin (10 μg/ml; Sigma-Aldrich) but were completely inhibited by preincubation and coculture with GM1-ganglioside (10 nmol/ml).

BALB/c mice, DO11.10 OVA TCR Tg mice, BALB/c mice adoptively transferred with 1.5 × 10⁷ DO11.10 CD4⁺ T cells, Rag1^−/− mice adoptively transferred with CD25⁻, or CD25⁺CD4⁺ T cells from OT-II TCR Tg mice were given intragastrically (i.g.) either 20 mg of OVA, 200 μg of OVA/CTB conjugate, 200 μg of OVA/CTB mixed with 4 μg of CT, or only PBS up to three times at 2-day intervals. In some experiments, mice were also immunized once s.c. in a footpad with 50 or 100 μg of OVA emulsified 1/1 in CFA containing 100 μg of Mycobacterium tuberculosis H37Ra (Difco). At indicated days after i.g. treatment or s.c. immunization, the mice were sacrificed, and PP, draining LN (DLN; axillary, inguinal, MLN, or popliteal), and spleen were harvested for further in vitro studies.

Spleens and LN were pressed through nylon nets, and single-cell suspensions were prepared. Purified CD4⁺ T cells were isolated by negative selection using MACS microbeads labeled with various mAbs (Miltenyi Biotec) according to the manufacturer’s suggested protocol. To separate CD25⁺CD4⁺ T cells and CD25⁻CD4⁺ T cells, the isolated CD4⁺ T cells were further incubated with PE-conjugated anti-CD25 (10 μl/10⁷ cells) at 4°C for 10 min, whereafter anti-PE coated microbeads (Miltenyi Biotec) were added and incubated another 15 min at 4°C. Magnetic separation was performed with a positive selection column according to the manufacturer’s suggested protocol. CD25⁻CD4⁺ Tg T cells purified from LN or spleen from DO11.10 or OT-II mice were used as OVA Tg T effector cells (T_eff). In some experiments, CD25⁺CD4⁺ T cells and CD25⁻CD4⁺ T cells were further purified by flow cytometry on a FACStar cell sorter (BD Biosciences).

CD4⁺ T cells purified from DO11.10 mice were labeled at 3 × 10⁷ cells/ml with 5 μM CFSE (Molecular Probes) in PBS for 5 min followed by incubation with 5% FCS-PBS (5 mM EDTA) for 10 min at 37°C. After two washes, 5 × 10⁶ of the labeled cells in 200 μl of PBS were injected i.v. into the tail vein of recipient mice.

Spleen cells or the draining popliteal or inguinal LN cells were collected at indicated days after immunization. Cells were isolated and RBC were removed by lysis and then cultured. For proliferation of splenic and LN cells, 2.5 × 10⁵ or 5 × 10⁵ cells/well (in 200 μl) were cultured for 3 days in 96-well plates with or without 1–10 μg/ml OVA. [³H]Thymidine was added (1 μCi/well) for the last 16 h of culture. To test the regulatory function of isolated CD4⁺ T cells or separated CD25⁺CD4⁺ or CD25⁻CD4⁺ T cells from mice with different treatments, these cells were cocultured in different numbers with 1 × 10⁵ DO11.10 T_eff cells together with 5 × 10³ of bone marrow-derived dendritic cells (DC) prepared as described in Ref.23 and with 1 μg/ml OVA_323–339 peptide. These tests were performed using 96-well round-bottom plates (Nunc) for 3 days with [³H]thymidine added during the last 16 h. IL-2 production by cultured cells were assayed in 2-day culture supernatants using the cytokine cytometric bead array method (BD Biosciences). TGF-β in serum was measured by ELISA according to the manufacturer’s instructions (Duo Set kit; R&D Systems).

Freshly isolated splenic or LN cells were incubated with FITC-,or PE-, or allophycocyanin-labeled mAbs to mouse CD4, CD8α, CD25, CD62L, CD69, or KJ1–26 (BD Biosciences). For analysis of intracellular CTLA-4 and Foxp3, stained CD25⁺CD4⁺ or CD25⁻CD4⁺ T cells were fixed and permeabilized with Cytofix/Cytoperm solution (BD Pharmingen) according to the manufacturer’s suggested protocol and then incubated with PE- or allophycocyanin-conjugated anti-CTLA-4 (BD Biosciences) or anti-Foxp3 FLK-16 (Nordic BioSite) (0.5–1 μg/10⁶ cells) at 4°C for 30 min in the dark. Cells were then washed and analyzed by flow cytometry (FACSCalibur; BD Biosciences), which was also used to assess cell division in CFSE-labeled cells.

Total CD4⁺ T cells or CD25⁺CD4⁺ or CD25⁻CD4⁺ T cells from i.g. OVA/CTB-treated or PBS-treated BALB/c mice or OVA TCR Tg mice were injected i.v. into BALB/c mice (4 × 10⁵ cells/mouse). One day after transfer, the mice were immunized s.c. in the right footpad with 100 μg of OVA in 50 μl of CFA containing 100 μg of M. tuberculosis. Seven days after this immunization, mice received a s.c. challenge injection of 20 μg of OVA in PBS in the left footpad. Left footpad thickness was measured before and 24 h after OVA challenge in a blinded fashion using a caliper meter (Mitutoyo).

In separate experiments, purified CD25⁺CD4⁺ and CD25⁻CD4⁺ T cells from OT-II mice were injected i.v. into Rag1^−/− mice who were then treated i.g. with different regimens before receiving a s.c. immunization with OVA in CFA. One or 3 wk later, spleen CD25⁺ and CD25⁻CD4⁺ T cells were purified by FACS cell sorting and used for Foxp3 gene expression studies.

In some experiments, we first adoptively transferred CD25⁻CD4⁺ OVA Tg DO11.10 T cells to normal BALB/c mice. Then, starting 4 days later, we gave groups of mice three i.v. injections every second day with either anti-mouse TGF-β mAb (1D11.16.8) or sham treatment (normal mouse serum IgG). One day after each injection, mice in the different treatment groups were given either OVA/CTB conjugate (200 μg/dose) or PBS i.g. The anti-mouse TGF-β mAb was purified from a hybridoma supernatant by affinity chromatography and shown to give a single peak in FPLC and to contain <0.1 EU/mg of endotoxin.

In addition to the intracellular staining of Foxp3 protein (scurfin) as described above, Foxp3 gene expression was also studied by RT-PCR and real-time PCR methods.

RNA was purified by the total RNA extraction kit for mammalian RNA (Sigma-Aldrich), and DNase was treated by the DNase I Amp grade kit (Invitrogen Life Technologies) to remove residual genomic DNA. cDNA was synthesized by using an oligo(dT) primer and the Sensiscript RT-PCR Kit (Qiagen) as described by the manufacturer. Primers for mouse hypoxanthine phosphoribosyltransferase (Hprt), designed by primer3 software (〈http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi〉) and the Foxp3 primers described by Jason et al. (24) were from MGW-Biotech. Semiquantitative RT-PCR was performed using standard PCR conditions (1.5 mM MgCl₂ and annealing at 55°C) and forward and reverse primers (Foxp3, 5′-ggcccttctccaggacaga-3′ and 5′-gctgatcatggctgggttgt-3′; Hprt, 5′-atcagtcaacgggggacata-3′ and 5′-agaggtccttttcaccagca-3′). The reactions were amplified for 37 cycles and analyzed on 2% agarose gels stained with ethidium bromide.

Real-time quantitative PCR was performed on an ABI Prism 7500 thermal cycler (PerkinElmer) using SYBR Green master mix (Applied Biosystems) and the Foxp3 and Hprt primers, both at a final concentration of 200 nM. The specificity of the PCR was confirmed by the appearance of a single peak in the analysis of the dissociation curve, showing the predictable melting temperature of the primer pairs. No amplification was seen in the nontemplate control. A standard curve was generated from serial dilutions of purified primer product of Hprt and Foxp3. Normalized values for Foxp3 mRNA expression were calculated as the relative quantity of Foxp3 divided by the relative quantity of Hprt extrapolated from the standard curve calculated by the ABI Prism 7500 software (Applied Biosystems).

Results are expressed as mean ± SD. When not specified otherwise, we used the Student t test for determining statistical differences between experimental and control groups. Values of p < 0.05 or p < 0.01 are referred to as significant or highly significant, respectively.

In preliminary work, we found that i.g. administration to BALB/c (or C57BL/6) mice of a low-dose OVA/CTB conjugate, given either as a single 200-μg dose or as three 200-μg doses every second day, efficiently induced regulatory CD4⁺ T cells in MLN and spleen that could completely suppress OVA-specific proliferation of OVA TCR Tg DO11.10 or C57BL/6 OT-II T cells in coculture. These regimens of OVA/CTB also efficiently prevented the immunogenic effect of a s.c. OVA/CFA immunization 1 day after the i.g. treatment as manifested in a practically complete suppression of spleen or DLN T cells to OVA or OVA_323–339 peptide 7 or 21 days after immunization (Fig. 1 and data not shown). In contrast, i.g. administration of the same dose of OVA, whether alone or mixed with 100 μg of CTB, did not induce cells with any detectable regulatory activity in coculture with Tg T cells (data not shown), whereas cells with such suppressive activity could be found after three i.g. doses of a 100-fold higher amount of OVA (20 mg/feeding).

We wished to compare the relative efficiency of OVA/CTB conjugate and the high-dose unconjugated OVA to induce CD4⁺ T cells capable of suppressing OVA-specific stimulation of CD25⁻CD4⁺ Tg T_eff cells in vitro and to determine whether the suppressive cells were CD25⁺ or CD25⁻. We therefore gave either 200 μg of OVA/CTB or 20 mg of OVA or PBS i.g. on three occasions every second day to BALB/c mice. Three days after the last i.g. treatment, CD25⁺ and CD25⁻CD4⁺ T cells were isolated from pooled MLN and spleen mononuclear cells (MNC) using MACS microbeads, and their in vitro suppressive activity was tested in coculture experiments. In these experiments, we used isolated naive CD25⁻CD4⁺ DO11.10 cells as T_eff cells, DC as APC, and OVA_323–339 peptide for Ag-specific stimulation. As expected, CD25⁻CD4⁺ T cells isolated from the PBS-treated control mice had no suppressive effect on the proliferative response of DO11.10 T_eff cells to OVA_323–339 peptide. In contrast, CD25⁺CD4⁺ T cells or a 1:1 mixture of CD25⁺ and CD25⁻ CD4⁺ T cells from these mice could significantly suppress the T_eff response to OVA peptide (Fig. 2,A). A stronger suppressive effect was achieved using CD25⁺CD4⁺ T cells isolated from mice treated i.g. with high-dose OVA (Fig. 2,B) and was even further pronounced using CD25⁺CD4⁺ T cells isolated from the low-dose OVA/CTB-treated mice (Fig. 2, C and E). CD25⁻ cells from OVA- or OVA/CTB-treated mice in contrast had only a marginal suppressive effect under these conditions (Fig. 2, B and C).

Consistent with our previous work that coadministration of CT can abrogate oral tolerance induction (19, 20, 21), and in sharp contrast to the effects after feeding OVA/CTB conjugate without CT, neither CD25⁻ nor CD25⁺ CD4⁺ T cells isolated from BALB/c mice after feeding with a mixture of OVA/CTB and CT could effectively suppress the proliferative response of DO11.10 T_eff cells (Fig. 2 D).

Titration experiments using unfractionated pooled MLN and splenic CD4⁺ T cells from the different mouse treatment groups confirmed the findings with the isolated CD25⁺ and CD25⁻ cells. The CD4⁺ cells from the OVA/CTB-fed mice were suppressive even when added to the T_eff cells in a ratio between 1:8 and 1:16; in contrast, the CD4⁺ cells from OVA-treated mice only worked in a 1:1 ratio, and the PBS-treated CD4⁺ cells did not work at all even when tested in a 4:1 ratio with T_eff cells (data not shown).

Corresponding experiments in DO11.10 Tg mice confirmed the findings in normal BALB/c mice. Thus, three i.g. doses of OVA/CTB conjugate induced CD4⁺CD25⁺ T cells in MLN, which strongly suppressed T_eff cells in coculture (Fig. 4). Titration experiments also confirmed that the induced CD25⁺CD4⁺ suppressor T cells from OVA/CTB-fed Tg mice (Fig. 4,A) were far more potent in their suppressive capacity than CD25⁺CD4⁺ T cells (natural T_reg) from PBS-treated Tg mice (Fig. 4,C). Although the latter cells lost their suppressive effect when diluted to a T_reg:T_eff cell ratio of 1:3, T_reg from the OVA/CTB-fed mice significantly suppressed T_eff cells, even when used at a ratio of 1:81. Coadministration of CT with the OVA/CTB conjugate resulted in a loss of detectable suppressive activity of the CD4⁺CD25⁺ MLN T cells (Fig. 3 B).

In additional experiments, these results were also confirmed in BALB/c mice that had adoptively received DO11.10 Tg T cells. These mice were treated and tested as described in Fig. 1. The results (shown in Fig. 4,A) demonstrate that in much the same way as described for the normal and Tg mice, MLN cells from OVA/CTB-treated animals were strongly, almost completely, suppressed in their proliferative response to OVA_323–339 peptide. These cells could also strongly suppress the response of DO11.10 T_eff cells in coculture, and titration experiments showed that the isolated CD4⁺CD25⁺ T_reg (Fig. 4,B) or CD4⁺ MLN (Fig. 4 C) cells from the OVA/CTB-fed mice were much more potent in their suppressive capacity than the corresponding cells from PBS-treated mice.

Furthermore, in these mice, coadministration of CT together with OVA/CTB practically completely abrogated the oral tolerance induction (Fig. 4 A, last panel). In a separate experiment, we examined whether CT could break already established tolerance induced by prior i.g. treatment with OVA/CTB. To this end, we gave groups of BALB/c mice that had adoptively received DO11.10 Tg T cells, three i.g. doses of either 200 μg of OVA/CTB or PBS every second day, and 1 day after the last treatment we gave to half of the these mice 50 μg of OVA in CFA or 50 μg of OVA together with 2 μg of CT. One week later, we examined the spleen T cell proliferative response to stimulation with 1 μg/ml OVA_323–339 peptide. The results showed that while in the OVA/CFA-challenged group the feeding with OVA/CTB suppressed proliferation by 88 ± 11%, it only gave 45 ± 6% suppression in the OVA/CT-challenged mice (data not shown).

Our next goal was to determine whether a tolerizing i.g. administration regimen of OVA/CTB conjugate would enhance suppressive activity and expand the number of CD25⁺CD4⁺ T cells in different organs. We, therefore, again gave three i.g. doses of OVA/CTB, OVA alone, PBS, or OVA/CTB plus CT to groups of DO11.10 OVA TCR Tg mice as described in Fig. 2. Three days after the last treatment, PP, MLN, and spleen cells were collected, and we examined the frequency of total and OVA_323–339 peptide-specific (KJ1–26⁺) CD25⁺CD4⁺ T cells by flow cytometry. As shown in Table I, there was a consistent and significant, albeit quantitatively modest, increase in both total and KJ1–26⁺ CD25⁺CD4⁺ T cells in PP in mice treated with OVA/CTB compared with PBS-treated control mice. Mice fed a high dose of OVA had a much smaller increase in CD25⁺CD4⁺ T cells (Table I). CD25⁺CD4⁺ cells in MLN and spleen were also significantly increased in response to oral OVA/CTB treatment, although to a lesser extent than in PP. Coadministration of CT with the OVA/CTB conjugate abrogated the increase in both total and KJ1–26⁺ CD25⁺CD4⁺ cells in PP, MLN, and spleen (Table I).

CTLA-4 is considered to be a negative signaling molecule expressed on activated T cells and has been associated with suppressive capacity of natural T_reg cells (12). We found that CTLA-4 expression on CD25⁺CD4⁺ T cells was significantly increased in PP and MLN of i.g. OVA/CTB-treated mice compared with PBS-treated control mice, irrespective of whether they were DO11.10 Tg mice (Table I) or BALB/c mice that had adoptively received DO11.10 T cells (data not shown). Interestingly, in the latter group, expression of CTLA-4 was also increased on CD25⁻CD4⁺ T cells isolated from MLN cells after i.g OVA/CTB-treatment compared with PBS treatment resulting in a doubling of CTLA-4⁺ such cells (data not shown). In contrast, CTLA-4 expression on either CD25⁺ or CD25⁻ CD4⁺ T cells did not increase appreciably in either PP or MLN or spleen after treatment with a high dose of OVA or coadministration of CT together with OVA/CTB (Table I).

Expression of Foxp3 is a key marker for identification of functional T_reg cells in the CD4⁺CD25⁺ T cell population (14, 15, 16). By using the RT-PCR assay, we found that the Foxp3 gene was strongly expressed in purified CD25⁺CD4⁺ T cells from pooled MLNs and spleen from mice treated i.g. with OVA/CTB, OVA, PBS, or OVA/CTB plus CT (Fig. 5,A). More surprisingly, Foxp3 was also expressed in highly purified CD25⁻CD4⁺ T cells from the OVA/CTB-fed mice, although to a much lesser extent than in the CD25⁺ cells (Fig. 5,A). In contrast, no Foxp3 expression was seen in the CD25⁻ cells from mice fed with either the conjugate mixed with CT, with high-dose OVA, or with PBS alone (Fig. 5,A). Quantitative analyses by real-time PCR confirmed these findings (Fig. 5,B) and showed that Foxp3 expression in CD25⁺CD4⁺ T cells from OVA/CTB-treated mice was more than twice that of cells from PBS-treated mice and also higher than in cells from OVA-only treated mice. A very small increase in Foxp3 expression in CD25⁻CD4⁺ T cells after the OVA/CTB feeding was also confirmed (Fig. 5,B). The addition of CT to the OVA/CTB treatment completely abrogated the increase in Foxp3 activity and, in fact, led to even lesser activity than in the PBS-fed controls (Fig. 5,B). The increase in expression of Foxp3 was further confirmed by intracellular staining of Foxp3 protein in KJ1–26⁺ CD4⁺ cells using FACS analyses (Fig. 5 C).

We tested whether i.g. administration of OVA/CTB could induce cell division in adoptively transferred CFSE-labeled CD4⁺ DO11.10 T cells in wild-type BALB/c mice. The results, illustrated in Fig. 6, show that 3 days after a single 200-μg OVA/CTB feeding CD25⁺CD4⁺ T cells in MLN had undergone intense cell division, in sharp contrast to MLN cells from PBS-treated mice (Fig. 6,A). Consistent with this, it was further found that the OVA/CTB treatment also induced the expansion of Foxp3⁺ cells (Fig. 6 B).

We isolated total CD4⁺ T cells and separated the CD25⁺ and CD25⁻ CD4⁺ T cells from pooled MLN and spleen MNC of OVA/CTB-treated BALB/c mice 3 days after the three-dose i.g. treatment regimen described above and transferred 4 × 10⁵ of either type of cell i.v. into naive BALB/c mice. One day later, the latter mice were s.c. sensitized in their right rear footpad with OVA/CFA; 7 days later, half were challenged by a s.c. injection of OVA in the left rear footpad, and the DTH response in vivo was measured as the extent of footpad swelling. We collected pooled MLN and spleen T cells from the other half of the mice and examined them for proliferation in response to OVA stimulation in vitro. As shown in Fig. 7,A, there was significant suppression of DTH in mice that received either purified CD25⁻CD4⁺ T cells (>99% pure) or purified CD25⁺CD4⁺ T cells (>85% pure) from i.g. OVA/CTB-treated mice compared with the DTH response in control animals that received no cells or corresponding cells from PBS-treated mice. Furthermore, proliferative responses (Fig. 7 B) by MLN and spleen T cells in response to OVA stimulation were also significantly reduced.

In analogous experiments in which we used adoptive transfer of purified CD25⁺ or CD25⁻CD4⁺ T cells from similarly treated DO11.10 Tg mice, we found that DTH response in vivo (Fig. 7,C) and splenic T cell proliferation in vitro (Fig. 7,D) were also strongly suppressed in BALB/c recipients that had received either isolated CD25⁺ or CD25⁻CD4⁺ T cells from Tg mice fed with OVA/CTB; again, the cells from mice transferred with cells from OVA/CTB-treated donors had stronger suppressive function than the cells from mice that had received cells from PBS-treated mice. Furthermore, consistent with our previous in vitro findings, both CD25⁺ and CD25⁻CD4⁺ T cells isolated from Tg mice treated i.g. with a mixture of OVA/CTB and CT completely lacked suppressive capacity in vivo after adoptive transfer (Fig. 7, C and D).

Next, we tested to see whether suppressive T cells can be generated from natural CD25⁻CD4⁺ T cells after i.g. treatment with OVA/CTB. To this end, we isolated both CD25⁻ (>99% purity by FACS testing as shown in Fig. 8,A) and CD25⁺CD4⁺ T cells from naive OVA TCR Tg OT-II mice and adoptively transferred the CD25⁻ cells or a mixture of CD25⁻ and CD25⁺ cells (5:1 ratio) into Rag1^−/− mice. We then gave three doses of OVA/CTB conjugate (200 μg/dose) or PBS i.g. every second day, and then 1 day after the last administration, an immunization s.c. with OVA/CFA. Seven or 21 days after immunization, mice were sacrificed, and DLN, MLN, and spleen cells were collected, and MNCs were prepared and used for T cell proliferation tests and/or for FACS analyses of the frequency of total and Ag-specific CD25⁺CD4⁺ T cells. As shown in Table II, 7 days after the s.c. immunization, splenic T cell proliferation was completely suppressed in Rag1^−/− mice that had received mixed CD25⁻/CD25⁺CD4⁺ T cells and been fed with OVA/CTB compared with corresponding PBS-treated mice. Strikingly, however, the T cell proliferation was also markedly suppressed in Rag1^−/− mice that had received highly purified CD25⁻CD4⁺ OVA Tg T cells and then been treated i.g. with OVA/CTB before immunization (Table II). DLN T cell proliferation showed a similar pattern (Table II). Furthermore, the proliferation of OT-II CD25⁻CD4⁺ T_eff cells stimulated with OVA_323–339 peptide was also markedly suppressed when these cells were cocultured with splenic T cells obtained from Rag1^−/− mice that had received purified CD25⁻CD4⁺ OVA Tg T cells and then i.g. treatment with OVA/CTB whether followed by immunization with OVA/CFA or no immunization (data not shown).

Flow cytometric analyses revealed that mice that had received mixed CD25⁻/CD25⁺CD4⁺ T cells and mice given highly purified CD25⁻CD4⁺ OVA Tg T cells had significantly more CD25⁺CD4⁺ T cells in MLN (data not shown), spleen, and DLN after i.g. treatment with OVA/CTB compared with treatment with PBS (Table II).

Corresponding T cell proliferation and CD25 FACS staining analyses of MNC collected 21 days after immunization showed results similar to those described for the 7-day MNC (data not shown).

To test whether the CD25⁺ and/or the CD25⁻ T cells generated from the adoptively transferred CD25⁻CD4⁺ T cells by oral treatment with OVA/CTB expressed the Foxp3 gene, the spleen cells were stained with anti-CD4-FITC and anti-CD25-PE and further sorted into CD25⁺ and CD25⁻CD4⁺ T cells by a FACS sorter before examination for Foxp3 expression by RT-PCR. In the OVA Tg naive mice that provided donor cells, Foxp3, as expected, was only expressed in the CD25⁺ population and not in the CD25⁻ T cells. Of interest, the purified CD25⁺ population isolated from mice that had been recipients of Foxp3⁻CD25⁻ T cells and treated with OVA/CTB strongly expressed Foxp3 (Fig. 8,B) and also strongly suppressed the proliferation of CD25⁻CD4⁺ T_eff (Fig. 8,C). In contrast, the CD25⁺ population isolated from mice given CD25⁻ cells and treated with PBS, did not express Foxp3 gene and were also much less suppressive (Fig. 8,C); however, when the latter type of cells were examined 3 wk after immunization, they too expressed detectable Foxp3 (data not shown). The CD25⁻ T cells, in contrast, did not express any Foxp3 whether isolated from the OVA/CTB- or PBS-treated groups even though the CD25⁻ cells from the OVA/CTB-treated group, although being slightly less effective than the CD25⁺ cells, efficiently suppressed OT-II T_eff Ag-specific proliferation when tested in coculture (Fig. 8 C).

Finally, we sought to examine whether TGF-β levels in serum are increased and whether TGF-β is important in the generation of T_reg in response to i.g. administration of OVA/CTB conjugate. Therefore, we adoptively transferred CD25⁻CD4⁺ OVA Tg DO11.10 T cells to normal BALB/c mice. Then, starting 4 days later, we gave groups of mice three i.v. injections every second day of either anti-mouse TGF-β mAb or as sham treatment normal mouse serum IgG. One day after each injection, different treatment groups also received either OVA/CTB conjugate (200 μg/dose) or PBS i.g. Three days after the last i.g. treatment, serum and MLN cells were collected for determination of TGF-β1 levels in serum by ELISA, frequency of CD25⁺CD4⁺ T cells among total and KJ1–26⁺ lymphocytes in MLN by flow cytometry, and the MLN T cell proliferative response to OVA_323–339 peptide. As shown in Table III, the treatment with anti-TGF-β mAb decreased the levels of TGF-β1 in serum by ∼80% compared with the levels in mice sham-treated with normal mouse IgG. In the PBS-fed group, the treatment with anti-TGF-β was also associated with a small but significant (p < 0.05) increase in CD25⁺CD4⁺ Ag-specific (KJ1–26⁺) T cells in MLN (group 2 vs group 4). Of note, in the control IgG-treated mice, the oral administration of OVA/CTB conjugate not only led to a 90% increase in the frequency of CD25⁺ T cells among KJ1–26⁺ T cells in MLN, but also to a marked increase in serum TGF-β1 levels and evidence of functional suppression reflected in an almost complete suppression of MLN cell proliferation to OVA_323–339 peptide (groups 3 vs 4). In sharp contrast, in the mice depleted of TGF-β by specific Ab, the i.g. administration of OVA/CTB conjugate did not result in any increased frequency of CD25⁺ T cells in MLN, even though the OVA/CTB conjugate feeding of these mice was associated with a significant, although in absolute quantities small, increase in serum TGF-β1 (groups 1 vs 2) (Table III). The anti-TGF-β treatment also largely abrogated the ability of i.g. OVA/CTB treatment to induce functional suppression of the MLN cell proliferative response (p > 0.20 for difference to PBS-treated mice, group 1 vs group 2), even though there was a tendency to a small reduction in the response. We conclude that depletion of TGF-β by specific Ab treatment appears to completely abrogate the induced development of OVA-specific CD25⁺CD4⁺ T cells in response to oral administration of OVA/CTB conjugate associated with a loss in suppression of Ag-specific T cell reactivity.

Our results show that i.g. administration of OVA/CTB conjugate effectively induces OVA-specific peripheral immunological tolerance, whether tested in normal BALB/c mice, OVA TCR Tg (DO11.10) mice, or in either BALB/c mice or Rag1^−/− mice transferred with OVA TCR Tg T cells. Consistent with our previous findings in other systems (17, 18, 19, 20, 21, 22), oral tolerance induction with Ag/CTB conjugate was superior to that induced with free Ag both in markedly increasing the efficacy of tolerization and in dramatically decreasing the effective amounts of Ag. The oral tolerance by i.g. Ag/CTB is associated with strong induction of CD25⁺ T_reg as well as to a lesser extent CD25⁻CD4⁺ T_reg that independently can transfer tolerance adoptively to naive recipients. Coadministration of CT, in contrast, effectively abrogates both the induction of oral tolerance and of T_reg by i.g. OVA/CTB. The CD25⁺ T_reg cells induced by i.g. administration of OVA/CTB in the absence of CT strongly express the T_reg cell-controlling gene Foxp3 as well as CTLA-4, and depletion of TGF-β by specific Ab treatment completely inhibits the generation of CD25⁺CD4⁺ T_reg cells in response to i.g. administration of OVA/CTB.

In both OVA Tg and normal BALB/c mice, OVA/CTB-induced CD25⁺ cells efficiently suppressed OVA-specific CD25⁻ T_eff cell proliferative responses in vitro as well as DTH response in vivo. Consistent with recent reports showing that food Ag could induce mucosal CD25⁺ T_reg cells (6, 7, 8, 25), these CD25⁺ cells in PPs or MLNs were significantly increased after i.g. treatment with OVA/CTB, and studies with adoptively transferred CFSE-labeled CD4⁺ T cells confirmed the expansion of CD25⁺CD4⁺ T_reg. However, although the increase in T_reg cell number was relatively modest, at most 2- or 3-fold, the suppressive activity of these cells on Ag-specific T_eff proliferation in coculture experiments in contrast increased >20-fold as determined in cell titration studies. Our results indicate that tolerance induced by OVA/CTB is not only associated with an increase in the number of CD25⁺ T_reg cells, but also and mainly with increased suppressive activity and Foxp3 activation on a per cell basis.

TGF-β plays an important role in the induction and suppressive action of natural T_reg (11, 14, 26), and also in the generation of Ag-induced peripheral T_reg cells (7, 13, 27), but it is not known whether TGF-β is required for oral tolerance generation of mucosal T_reg cells. In our study, oral tolerance induction by i.g. administration of OVA/CTB led to a markedly elevated level of TGF-β1 in serum associated with an increase in MLN T_reg cells. Conversely, in vivo depletion of TGF-β by specific mAb treatment completely suppressed the generation of CD25⁺ T_reg cells in response to the i.g. OVA/CTB treatment and effectively removed most, if not all, of the ability of i.g. OVA/CTB treatment to induce functional suppression of T cell responses. It is notable that mucosal Ag/CTB treatment not only increases TGF-β production, as shown in this study and in previous reports (19, 22), but also almost completely suppresses proinflammatory IL-6 production (21). In light of the recent finding by Bettelli et al. (27) that TGF-β in the absence of IL-6 strongly promotes T_reg generation but when combined with IL-6 instead, induces IL-17-producing, so-called Th17 effector T cells, the combined increase in TGF-β and decrease in IL-6 is put forward as an important explanation for the remarkably efficient oral tolerance induction and T_reg generation by Ag/CTB treatment.

Of importance, and consistent with a recent report by Hauet-Broere et al. (25), we found that in addition to the strong induction of CD25⁺CD4⁺ T_reg, i.g. treatment with OVA/CTB conjugate also induced CD25⁻CD4⁺ T cells that displayed suppressive function both in vitro and in vivo after adoptive transfer of these highly purified cells in recipients immunized s.c. with OVA/CFA. It is notable that the latter cell population also expressed the Foxp3 gene, although to a much lesser extent than the CD25⁺ cells, and that these cells, at least in part, may therefore be under differentiation toward increased expression of Foxp3 as well as expression of the IL-2α receptor (CD25). In support of this kind of differentiation pathway, Zelenay et al. (28) recently reported that Foxp3⁺CD25⁻CD4 T cells constitute a reservoir of committed regulatory cells that regain CD25 expression upon homeostatic expansion. Of note, CTLA-4, which plays an essential role in the function of CD25⁺ T cells in controlling inflammation (12), was found to be increased not only on OVA/CTB-induced CD25⁺ but to a lesser extent also on OVA/CTB-induced CD25⁻ T cells that had been sorted by FACS to rule out any contamination by CD25⁺ T cells.

Naturally occurring CD25⁺CD4⁺ T_reg play crucial roles in the maintenance of immunological self-tolerance and negative control of various immune responses (9, 10). A recent study (29) also demonstrated a strong impairment in oral tolerance in naive mice depleted in vivo of natural CD25⁺CD4⁺ cells by Ab treatment. This raised a question as to whether tolerance induction by i.g. OVA/CTB would be critically dependent on natural CD25⁺CD4⁺ cells. To address this question, we adoptively transferred highly purified CD25⁻ OT-II T cells and, for comparison, a mixture of CD25⁺ and CD25⁻ OT-II T cells to Rag1^−/− mice and then tested the induction of oral tolerance after i.g. treatment of the recipient mice with OVA/CTB conjugate. We found that this treatment strongly suppressed both splenic and DLN T cell proliferation to subsequent systemic immunization with OVA/CFA in both groups. When the CD4⁺ T cell populations of the treated mice were further examined, it was evident that in Rag1^−/− mice that had received only the highly purified CD25⁻ population, the i.g. treatment with OVA/CTB had induced a significant generation of CD25⁺ T cells in MLN, spleen, and DLN similar to that seen in OVA/CTB-treated recipients of CD25⁺/CD25⁻ cells and significantly above the levels in PBS-treated controls. Furthermore, highly purified (cell sorted) OVA/CTB-induced CD25⁺ T cells isolated at the end of the experiment from mice that initially received CD25⁻ cells strongly expressed Foxp3 in contrast to the CD25⁺ population from corresponding PBS-treated mice. However, when the cells were isolated and examined at 3 wk rather than 7 days after treatment, the CD25⁺ cells from the PBS-treated animals also had detectable expressed Foxp3. Our results indicate that mucosal treatment with OVA/CTB independent of natural CD25⁺ T_reg cells can promote Ag-induced generation of at least two populations of T_reg from CD25⁻ precursors: one comprising strongly suppressive Foxp3⁺CD25⁺ T_reg, and one comprising CD25⁻CD4⁺ T cells that have acquired regulatory-suppressive function without activation of the Foxp3 gene system.

The efficient generation and functional activation of CD25⁺ T_reg by i.g. treatment with OVA/CTB conjugate is probably largely explained by the effects of the conjugate on gut mucosal APC. Previous studies have shown that conjugation of Ag to CTB, a molecule that binds with high affinity to the GM1 ganglioside receptors present on most cells (30), greatly facilitates Ag uptake and MHC class II-restricted Ag presentation by CD11c⁺ DC as well as other types of APC, such as B cells and macrophages (23). One possibility is that mucosally administered Ag/CTB conjugate preferentially binds to and is taken up by tolerogenic subsets of mucosal DC or other APC. Consistent with this, we found that the increase in frequency of CD25⁺CD4⁺ T cells in PP and MLN correlated closely with an increase of CD11c⁺CD8α⁺B220⁺ DC between 2 h and 2 days after i.g. treatment with OVA/CTB (our unpublished data). This agrees fully with the recent findings by Anjuere et al. (31) of a selective increase in this DC subset in MLN following CTB feeding, and the ability of this DC population from CTB-fed mice to support the differentiation of CD4⁺ Ag-specific T_reg producing TGF-β and IL-10.

CT is a powerful mucosal adjuvant for most coadministered protein Ags including OVA and stimulates strongly enhanced Th1, Th2, and CTL responses (2, 32). Oral administration of CT can in parallel with its adjuvant action prevent induction of oral tolerance to a coadministered Ag (17, 18, 19, 33), although CT has also been reported (34) to promote the induction of T_reg (Tr1 cells) in vitro. In this study, we found that coadministration of even very small amounts of CT together with the OVA/CTB conjugate effectively prevented the induction of CD25⁺CD4⁺ suppressor T cells. Furthermore, CT also inhibited the normal suppressive function and Foxp3 gene expression of the mucosal CD25⁺ T_reg cells in the OVA/CTB-treated mice in comparison with PBS-fed mice. However, in line with the previously reported inability of CT to abrogate already established oral tolerance (35), our findings showed only a relatively small effect of CT given after, rather than together, with a tolerizing OVA/CTB feeding regimen.

The mechanism(s) by which CT exerts these effects remains to be defined. One obvious mechanism could be that CT inhibits the generation or function of a normally predominant tolerogenic mucosal APC population. In support of this, Anjuere et al. (36) found that oral administration of CT to mice results in a marked increase in PP and MLN of CD11c⁺CD8^int DC with potent immunological APC capacity together with inhibition of the normal development of tolerogenic CD11c⁺CD8α⁺B220⁺ (plasmacytoid) DC in response to CTB administration. We have also seen that i.g. coadministration of CT with OVA/CTB strongly decreases the frequency of tolerogenic CD11c⁺CD8⁺B220⁺ DC compared with both OVA/CTB-fed mice and unfed controls (our unpublished data). Of note, CT, different from CTB, is a potent inducer for production of both IL-1β and IL-6 by both DC and other APC (32); these cytokines have recently been reported to activate Foxp3⁻ effector/memory T cells and attenuate Foxp3⁺CD4⁺CD25⁺ T_reg cell function (27, 37). In addition, CT down-regulates expression of TGF-β (19) as well as IL-2 and IL-2R/CD25 (38). Such effects may also be important, because TGF-β and IL-2 are essential for maintenance of CD25⁺ T_reg cells in vivo (11, 14, 26, 27, 39).

In conclusion, our findings show that the exceptionally efficient induction of peripheral tolerance by orally administered Ag coupled to CTB is associated with increases in both the frequency and suppressive activity of Ag-specific CD25⁺CD4⁺ T_reg cells that express Foxp3 as well as CTLA-4 proteins, and also leads to the additional generation of both Foxp3⁺ and Foxp3⁻CD25⁻CD4⁺ T_reg. The predominant induction and activation of CD25⁺ T_reg cells by oral Ag/CTB conjugate treatment is associated with increased induction of TGF-β1. Coadministration of CT with normally tolerizing Ag/CTB treatment regimens completely blocks the induction of oral tolerance and T_regs, probably largely through the differential effects of CT compared with CTB on APC that can then translate opposing effects on T_reg and T_eff cells.

While this manuscript was undergoing the revision process, another study from this laboratory also described the induction of tolerance associated with but not critically dependent on CD25⁺ cell expansion after oral administration of an influenza virus hemagglutinin-derived peptide/CTB gene fusion protein (40).

We thank Ulf Yrlid for help with the CFSE-labeling experiments and constructive suggestions, Margareta Blomquist, Veronica Johansson, and Bin-Ling Li for skilled general technical assistance, and Marianne Lindblad for expertly preparing and characterizing OVA/CTB conjugate lots.

J. Holmgren is an inventor on a patent which is titled, “Immunological tolerance-inducing agent” and whose proprietor is Duotol AB, Sweden. J. Holmgren also has equity in Duotol AB, which has interests in CTB-based oral tolerance immunotherapy.