Route of Antigen Presentation Can Determine the Selection of Foxp3-Dependent or Foxp3-Independent Dominant Immune Tolerance

Since the pioneering work of Medawar and colleagues, several protocols have been proposed to achieve transplantation tolerance in adults (1–3). One way to achieve robust tolerance is based on the induction of hematopoietic chimerism following transplantation of donor hematopoietic stem cells, leading to a recessive state of tolerance independent of Foxp3⁺ regulatory T (Treg) cells (4–6). Other approaches, namely the use of mAbs targeting T cell coreceptors (CD3 or CD4) or costimulation (CD154), could lead to dominant transplantation tolerance in several animal models (7–12). Such dominant tolerance was found to be associated with peripheral induction of Foxp3⁺ Treg cells (13, 14). In fact, it has been shown that Foxp3-deficient mice cannot be tolerized with protocols leading to dominant tolerance, such as nondepleting anti-CD4 (15).

Several mAbs effective in achieving transplantation tolerance were also shown able to induce tolerance in autoimmunity, often associated with the emergence of Foxp3⁺ Treg cells (16–20), as well as tolerance to allergens or model Ags (21, 22).

We became interested in the mechanisms leading to tolerance induction to Ags, and how adjuvants may influence the process, following our observations that it is possible to induce tolerance to a protein with anti-CD4 in a mouse model of allergic airway disease (21), but difficult to induce tolerance to factor VIII (FVIII) with the same mAb (or with anti-CD154) in a mouse model of hemophilia (23, 24). In fact, when the same Ag (FVIII) is expressed in tissues following gene therapy, a situation more similar to organ transplantation, it can be readily tolerized (25).

We therefore investigated the mechanisms leading to tolerance to a defined Ag (chicken OVA) when delivered as part of a skin graft (from OVA-actin transgenic mice) or delivered as OVA adsorbed to aluminum hydroxide (alum) i.p. We used OVA-specific MHC class II–restricted TCR-transgenic mice in two different genetic backgrounds (C57BL/6 and BALB/c) and the same tolerance-inducing method (nondepleting anti-CD4 mAb).

Given the known association of Foxp3⁺ Treg cells and dominant tolerance, it was surprising to find that, unlike tolerance to OVA-expressing skin grafts, the induction of tolerance to OVA-alum under CD4 blockade was Foxp3 independent. In this setting, tolerance induction triggered a reduction of the number of specific effector T cells, and simultaneously led to an IL-10–dependent dominant tolerance state. Remarkably, we showed that it was possible to take advantage of the Foxp3-independent tolerance protocol to boost the efficacy of tolerance induction, namely by achieving long-term survival of skin transplants in Foxp3-deficient hosts.

C57BL/6, BALB/c, C57BL/6-Tg(ACTB-OVA)916Jen/J (The Jackson Laboratory), OT-II.Rag1^−/− (Taconic), DO11.10.Rag1^−/− (The Jackson Laboratory), C57BL/6.IL-10^−/−, C57BL/6.Rag2^−/−, OT-II, DO11.10, and Wsh/Wsh mice were bred and maintained under specific pathogen-free facilities at the Instituto Gulbenkian de Ciência. Experimental animals were sex matched and between 6 and 8 wk of age. The T/B monoclonal mice (T-Bmc, 17/9 DO11.10.RAG1^−/−) bear monoclonal populations of T and B lymphocytes specific for chicken OVA 323–339 and hemagglutinin (HA) of influenza virus, respectively (26). These mice were kept under specific pathogen-free conditions at the Skirball Institute Central Animal Facility, New York University Medical Center.

Transplantation experiments were conducted by grafting full thickness skin on the lateral flank of recipient mice. Grafts were observed daily and considered rejected when no viable donor skin was present. Procedures were conducted in accordance with guidelines from the Animal Use and Institutional Ethical Committees and approved by Direcao Geral de Veterinaria or New York University School of Medicine Institutional Animal Care and Use Committee (New York).

Animals were sensitized, at the times described in the text, by i.p. injection of 10 μg of OVA (grade V; Sigma, St. Louis, MO) or β-lactoglobulin (β-Lg) (Sigma), previously run through a DetoxyGel column (Pierce, Rockford, IL) following the manufacturer’s instructions, and suspended in 2 mg of endotoxin-free alum (Alu-gel-S; Serva, Heidelberg, Germany). Mice were intranasally (i.n.) challenged with 50 μg of OVA in pyrogen-free saline and sacrificed 24 h after the last challenge.

Nondepleting anti-CD4 (YTS177) (27), anti–Il-10R (1.B1.2), anti–TGF-β (1D11), and the isotype control rat anti-dog CD4 (YKIX302) mAbs were produced in our laboratory using Integra CL1000 flasks (IBS, Chur, Switzerland), and purified from culture supernatants by 50% ammonium sulfate precipitation, dialyzed against PBS, and the purity checked by native and SDS gel electrophoresis. The hybridomas were generously provided by Prof. H. Waldmann (Oxford, U.K.).

The airways were washed through the trachea by slowly infusing and withdrawing 1 ml of cold PBS 2% BSA (Sigma) three times. Eosinophilia was quantified by flow cytometry using GR-1–FITC (eBioscience, San Diego, CA), CCR3-allophycocyanin (BD Pharmingen, San Diego, CA), and MHC class II–1Percp.Cy5.5 (eBioscience), with the eosinophils identified based on the side scatter/forward scatter profile and as the GR1^intMHC class II⁻CCR3⁻ cells (28).

Lungs were perfused with 4% formalin solution (Sigma), collected, and sectioned. Staining was performed using H&E, and mucus-containing cells were revealed using a periodic acid-Schiff stain. Photographs were taken using a Leica DM2500 microscope and a Leica DFC420 camera.

Serum titers of OVA-specific IgG1, IgG2a, and IgE were measured by ELISA using the following kits: IgG1 and IgG2a (SouthernBiotech, Birmingham, AL) with anti-OVA IgG1 standard from Serotec, Oxford, U.K.; IgE (BD Pharmingen) with anti-OVA IgE standard from Abcam (Cambridge, U.K.). Cytokine titers were determined in fresh lung tissue homogenate supernatants. Lung tissue was collected, homogenized in 100 mg/ml in HBSS (Life Technologies, Carlsbad, CA), centrifuged at 800 × g for 10 min, and the supernatant collected. The ELISA assays were performed according to the manufacturer’s instructions using the following kits: IL-13 (PeproTech, London, U.K.), IL-4, and IL-5 (BD Pharmingen).

Single-cell suspensions from spleens and lymph nodes (LN) from OT-II.Rag^−/− or DO11.10.Rag^−/− mice were resuspended at 5 × 10⁷ cells per ml and stained with 5 μM of CFSE or cell trace violet (CTV) (Invitrogen, Carlsbad, CA). The cells were washed, resuspended in saline, and injected i.v. in the tail vein of C57BL/6 or BALB/c mice at a total of 1 × 10⁷ cells per animal.

Single-cell suspensions from spleens and LNs from OT-II.Rag^−/− or DO11.10.Rag^−/− mice were labeled with CFSE or CTV. Cells were cocultured at 2 × 10⁴ cells per well with bone marrow–derived dendritic cells (DCs), generated following 7 d of culture with GM-CSF (PeproTech), as stimulators (2 × 10⁴ cells per well) in the presence of OVA_323–339 peptide (H2N-ISQAVHAAHAEINEAGR-OH; New England peptide).

Cells were stained for flow cytometric analysis with CD25-PE-Cy7 (PC61.5; eBioscience), CD3-PE-Cy7 (145-2C11; eBioscience), CD8–allophycocyanin–Alexa Fluor 750 (53-6.7;), CD4-PerCp (RM4-5; BD Pharmingen), CTLA-4-bio (UC10-4B9; eBioscience), the DO11.10 TCR-specific KJ1-26 mAb conjugated with PE or allophycocyanin (BD Pharmingen), and Vβ5.1/5.2 TCR-FITC (MR9-4; eBioscience) plus Vα2 TCR-eFluor 450 (B20.1; eBioscience) were used to determine the TCR transgenic populations in the OT-II mice. Cells were then washed and fixed using Foxp3 Staining Set (eBioscience). The cells were stained for intracellular cytokines and Foxp3 with Foxp3-allophycocyanin (FJK-16s; eBioscience), IFN-γ–FITC (XMG1.2; eBioscience), IL-13–PE (eBio13A; eBioscience), IL-4–FITC (BVD6-24G2; eBioscience), IL-5–allophycocyanin (TRF K5; BD Pharmingen) and IL-2 (JES6-5H4; BD Pharmingen). Quantification of cytokines ex vivo was performed without in vitro restimulation. Apoptotic cells were identified with annexin V–biotin (BD Pharmingen) and streptavidin-allophycocyanin-Cy7 (eBioscience) labeled in Annexin V Binding Buffer (BD Pharmingen) according to manufacturer protocol. Propidium iodide solution was added right before the cells were analyzed. Analyses were performed using an LSR Fortessa (BD Biosciences). The analysis gate was set on the forward and side scatters to eliminate cell debris and dead cells.

Statistical significance was determined using the two-tailed nonparametric Mann–Whitney U test and the log rank method. Statistical analysis was performed using Prism 4.0 (GraphPad, San Diego, CA) and p values <0.05 were deemed significant (*p < 0.05, **p < 0.01, ***p < 0.001).

Previous studies have shown that induction of transplantation tolerance with nondepleting anti-CD4 mAbs requires Foxp3⁺ Treg cells (15, 29). We used a different strain combination, important for our subsequent studies described in this study, with a similar outcome. We assessed whether in OVA-specific TCR-transgenic mice tolerance could be induced to OVA-expressing skin grafts. We transplanted C57BL/6 and OT-II.Rag^−/− mice with skin from OVA-actin transgenic mice (Fig. 1A, 1B). The animals were treated with 1 mg anti-CD4 (or isotype control) on days 0 and 2 after transplantation. We found that, both in C57BL/6 and OT-II.Rag^−/− mice, CD4 blockade induced long-term allograft survival, with OT-II.Rag^−/− mice (initially devoid of Treg cells) showing the emergence of peripherally induced Foxp3⁺ Treg cells in draining LN and spleen (Fig. 1C). To test the Foxp3 dependency of induced transplantation tolerance, we repeated the experiment with Foxp3-deficient OT-II.Rag^−/−.Foxp3^sc mice. We found that animals lacking functional Foxp3 could not be tolerized and readily rejected the skin allografts (Fig. 1B). We conclude that induction of transplantation tolerance with CD4 blockade depends on de novo Foxp3⁺ Treg generation.

We then investigated the induction of long-term tolerance to the same Ag (OVA), but provided systemically as OVA-alum i.p., using allergic airway disease as readout. We tested two different OVA-specific TCR-transgenic mice (OT-II and DO11.10) in distinct genetic backgrounds (C57BL/6 and BALB/c respectively) with similar results (Fig. 2, Supplemental Fig. 1). We were particularly interested in the DO11.10 mice as these share a genetic background with T-Bmc monoclonal mice used in Fig. 3. We found that CD4 blockade could prevent allergic airway disease, as Ab-treated mice were protected from peribronchiolar inflammatory infiltrates and hyperplasia of goblet cells, as well as airway eosinophilia (Fig. 2A–C). Furthermore, OVA-specific IgG1 and IgE were significantly reduced in anti–CD4-treated mice, when compared with control animals (Fig. 2D, 2E). We also found that CD4 blockade prevented an increase in the concentration of Th2-type cytokines (IL-4, IL-5, and IL-13; Fig. 2F). Overall, these results show that preconditioning with systemic administration of OVA under CD4 blockade prevents allergic airway disease, even in TCR transgenic mice where most T cells are specific to the Ag driving the pathology.

Given the high frequency of OVA-specific T cells in DO11.10Rag⁺ mice and the ability to identify those cells given their TCR usage, it was possible to monitor changes in the frequency of OVA-specific T cells with different functional characteristics. Mice treated with anti-CD4 mAb showed a reduced frequency of Th2-type OVA-specific T cells in the mediastinal LN (MLN) (Fig. 3A). In addition, those animals did not show more Th1 IFN-γ⁺ cells (Fig. 3A). We also found that the frequency of OVA-specific Foxp3⁺ Treg cells did not change with the tolerogenic treatment (Fig. 3B). Furthermore, DO11.10.Rag^−/− mice, devoid of thymic-derived Foxp3⁺ Treg cells (30) and subjected to the same tolerance induction protocol, did not develop detectable Foxp3⁺ T cells in the spleen or LNs (Fig. 3C). Note that the isotype of the anti-CD4 mAb is nondepleting as we (and others) have previously demonstrated. For instance, we showed that this mAb does not lead to alteration of the number of CD4 T cells in vivo that are not engaged in an immune response at the time of Ab treatment (20).

Previous studies in transplantation have implicated the de novo generation of Foxp3⁺ Treg cells, in a TGF-β–dependent process, in tolerance induced with anti-CD4 or costimulation blockade (13, 14). Importantly, mast cells were suggested to be critical for the establishment of such transplantation tolerance (31).

We used mast cell–deficient Wsh/Wsh mice, with a mutant allele at the W (c-kit) locus, to investigate whether mast cells also had a critical role in tolerance induction to proteins in alum (32, 33). We found that mast cell–deficient mice treated with anti-CD4 were protected from allergic airway inflammation to the same extent as littermates with normal numbers of mast cells (Fig. 3D). Furthermore we also neutralized TGF-β (using 1D11 mAb) during tolerance induction. This neutralization did not hamper protection from allergic airway disease (Fig. 3E). We confirmed that the dose of 1D11 mAb used effectively neutralized TGF-β as it was sufficient to prevent induction of transplantation tolerance (data not shown).

To definitely exclude a contribution of Foxp3⁺ Treg cells in tolerance induction with anti-CD4, we used T-Bmc mice deficient in Foxp3 (T-Bmc Foxp3^sf) (28). These mice bear OVA-specific TCR transgenic T cells and HA-specific BCR knock-in B cells in a RAG1 knockout background. They do not develop autoimmunity in the absence of Foxp3 because all their lymphocytes are monospecific and non–self reactive. These mice were sensitized with OVA-HA-alum i.p, and later exposed to OVA-HA i.n. Foxp3-deficient animals treated with anti-CD4 displayed similar protection from allergic airway disease as Foxp3 sufficient littermates (Fig. 3F). Foxp3–deficient animals were also prevented from generating IgE and did not display an increase in Th2 cells in MLN (Fig. 3G, 3H). Collectively these results point to a Foxp3⁺ Treg-independent tolerance induction to proteins in alum when anti-CD4 is administered.

To investigate the impact of anti-CD4 treatment in T cell proliferation and survival, CFSE-labeled OVA-specific T cells from DO11.10.Rag^−/− mice were adoptively transferred into BALB/c mice. The host animals were sensitized with OVA-alum i.p. treated or not, with anti-CD4 Abs. Lung, MLN, mesenteric LN, and spleen T cells were isolated at day 4 following treatment and analyzed by flow cytometry.

Consistent with a report on T cell response to OVA-alum administered i.p (34), MLN displayed the highest number of T cells responding to the Ag. Analysis of the CFSE dilution profile in OVA-specific T cells isolated from MLN suggests that anti-CD4 treatment prevents T cell proliferation, as the undivided cells are a greater proportion of the TCR-transgenic population unlike in mice injected with OVA alone (Fig. 4A). However, the number of undivided cells from anti-CD4–treated animals was remarkably similar to that obtained in mice treated with OVA alone, where extensive proliferation does occur, and significantly lower than in animals not treated with OVA. The difference between OVA and OVA + anti-CD4 treated groups is not in the undivided T cells but rather in the accumulation of T cells that underwent cell divisions. Given these data, either the anti-CD4 treatment is targeting the deletion of undivided cells or inducing cell death of T cells entering cell cycle.

The analysis of the TCR transgenic cells positive for the apoptosis marker annexin V showed that the frequency of apoptotic cells was similar between undivided cells from all experimental groups (Fig. 4C). Cells undergoing divisions from anti-CD4–treated but not control mice had a significant increase in the frequency of apoptotic cells. We conclude that, along with a possible effect on T cell proliferation, anti-CD4 treatment preferentially induces apoptosis of activated cells. Similar results were observed in the C57BL/6 genetic background (Supplemental Fig. 2).

However, some Ag-specific cells could still be isolated from tolerant mice, even when tolerance is induced in TCR-transgenic mice, suggesting that a dominant regulatory mechanism is in place. Because tolerance seems to be independent of Foxp3⁺ Treg cells, we sought to investigate if the Ag-specific T cells recovered from tolerized mice were Tr1, which is another well-characterized Treg population reported to express CD44 and high levels of LAG-3 and CD49b (35). Although some LAG-3⁺ cells could be found within OVA-specific cells from tolerant mice, a LAG-3⁺CD49b⁺ population was not present, suggesting that anti-CD4 treatment was not inducing Tr1 cells (Fig. 4D). It was also recently shown that tolerance induction can be mediated by T cells expressing low levels of IL-2 and high levels of CTLA-4 (36). We found that the recovered OVA-specific T cells from anti-CD4–treated mice displayed higher frequency of CTLA-4⁺IL-2^low T cells (Fig. 4E). Moreover, OVA-immunized mice had higher levels of IL-2 on those few cells that were CTLA-4⁺. We did not find changes in the level of CD25 expression on T cells from tolerized mice (Supplemental Fig. 3).

We therefore investigated whether the OVA-specific T cells that remain in tolerant mice displayed anergic behavior. By labeling TCR transgenic cells with CFSE and incubating them with OVA-loaded bone marrow–derived DCs, we could compare the proliferative ability of the OVA-specific T cells isolated from naive mice, mice sensitized with OVA-alum, or tolerized animals treated with anti-CD4 at the time of OVA-alum sensitization. We found that cells from tolerized mice proliferated less than T cells from control animals with a significant arrest in a nonproliferative state (Fig. 5A, 5B). When mice were treated with a neutralizing anti–IL-10R mAb, the anergic phenotype was partially abrogated.

To investigate whether those cells from tolerant mice, besides being anergic, are suppressive, we labeled OVA-specific T cells from naive mice with CTV (responder cells), and incubated them with TCR-transgenic cells from mice sensitized with OVA-alum or tolerized with OVA-alum + anti-CD4 in the presence of OVA-loaded DCs. We found that T cells from tolerant mice suppressed the proliferation of responder cells (Fig. 5C, 5D). However, the suppressive effect was abrogated when cells were collected from mice exposed to anti–IL-10R mAbs, or when anti–IL-10R mAbs were added to the culture (Fig. 5C–E, Supplemental Fig. 4).

To establish that the tolerance state could be transferred in vivo, we induced tolerance in OT-II mice. OT-II donors were exposed to OVA-alum i.p. and anti-CD4 (tolerant) or with an isotype control (immunized). A third group of mice were treated with OVA-alum with anti-CD4 and anti–IL-10R mAb (Fig. 5F). T cells from those groups of mice were adoptively transferred into recipient C57BL/6 mice, subsequently exposed to OVA, with some recipient mice also being treated with neutralizing anti–IL-10R mAbs (Fig. 5F). Production of OVA-specific IgG1 was quantified in the serum of recipient mice (Fig. 5G, 5H). We found that T cells from immunized mice exacerbated the OVA-specific response of the host, as anticipated as donor cells contributed to a secondary response. However, cells from tolerant mice did not contribute toward a secondary response and were able to suppress the immune response driven by the endogenous C57BL/6 cells (compared with mice that did not receive donor cells). The suppression was lost by IL-10 neutralization on the host mice, but also when anti–IL-10R mAbs were given to recipient mice (Fig. 5G, 5H, Supplemental Fig. 4). Taken together, these results show that 1) cells from tolerized mice can transfer suppression; and 2) IL-10 is necessary for immune suppression.

Finally, we tested whether CD4 blockade induces immune tolerance to OVA-alum i.p. in IL-10–deficient mice (Fig. 5I). Our results indicate that IL-10^−/− mice treated with anti-CD4 were not protected from allergic airway disease, as manifested by airway eosinophilia, a result anticipated from the data obtained with neutralizing mAbs.

Given our initial observation that transplantation tolerance required the induction of Foxp3⁺ Treg cells, we investigated whether a tolerance mechanism relying on IL-10 could support graft acceptance. C57BL/6 mice were tolerized to OVA-alum i.p. or the unrelated Ag β-Lg–alum by concomitant treatment with anti-CD4 mAb (Fig. 6A). At day 30 following the last administration of the Ag, all mice were transplanted with OVA-expressing skin grafts. We found that mice tolerized to OVA-alum i.p. accepted the OVA-expressing skin grafts (Fig. 6B). On the contrary, mice treated with anti-CD4 in the presence of an unrelated Ag (β-Lg) readily rejected the OVA-expressing skin grafts.

These data show that the tolerance state was Ag specific (as mice tolerized to β-Lg did not support survival of grafts bearing a different Ag), and show that tolerant mice remained competent to mount immune responses to unrelated Ags.

Finally, we investigated whether transplantation tolerance could be achieved in the absence of Foxp3⁺ Treg cells. OT-II.Rag^−/−.Foxp3^sc mice, as well as OT-II.Rag^−/− controls, were exposed to OVA-alum i.p. under the cover of anti-CD4, and later transplanted with OVA-expressing skin grafts at day 44 (as described in Fig. 6A). We found that OT-II.Rag^−/−.Foxp3^sc mice had a significant prolongation of graft survival, with several animals showing long-term graft survival (Fig. 6C). We also found that although TCR-transgenic mice treated with OVA-alum did not show evidence of peripheral induction of Foxp3⁺ Treg cells (see also Fig. 3C), subsequent skin transplantation led to the emergence of a Treg population, even without any further treatment (Fig. 6D).

Taken together, our data show that systemic administration of alloantigens prior to solid organ transplantation can provide an additional tolerogenic boost that may facilitate the induction of transplantation tolerance.

We found that the same tolerance-inducing agent (nondepleting anti-CD4 mAb) can trigger different mechanisms leading to dominant tolerance, depending on the context in which the Ag is presented to the immune system. By using TCR-transgenic mice it was possible to demonstrate that anti-CD4–mediated tolerance to an Ag in a skin graft requires the peripheral induction of Foxp3⁺ Treg cells. In contrast, anti-CD4–mediated tolerance to the same Ag administered by i.p. route with an adjuvant (alum) depends on a Foxp3-independent IL-10–dependent mechanism and results in apoptosis of effector cells and induction of a population of CTLA-4⁺IL-2^low T cells that can transfer the tolerance state.

Our data demonstrates the critical role of the context in which an Ag is perceived by the immune system for therapeutic induction of tolerance. In fact, it has been paradoxical that the same reagents (such as nondepleting anti-CD4 or anti-CD154) can readily induce transplantation tolerance to multiple alloantigens including major Ags (9, 37), but fail to achieve long-term tolerance to soluble proteins in the same mouse strains (23, 24). However, the same Ag (FVIII), which is not tolerizing when administered i.v. in the absence of adjuvants (23, 24), appears to induce tolerance more easily when expressed in tissues following transduction with viral vectors (25). Our previous findings showing that robust dominant tolerance can be induced to allergens (house dust mite) in alum (21) raised the hypothesis that it is necessary some adjuvant effect to achieve tolerance with mAbs that target T cell coreceptors or costimulation (14). Such an effect can be provided by tissue inflammation (in transplantation or viral transduction) or with the use of an adjuvant like alum.

When we investigated the mechanisms leading to immune tolerance induced with OVA-alum administration under CD4 blockade, it came as a surprise that the tolerance state was Foxp3 independent. In fact, most studies using anti-CD4 (or mAbs targeting other T cell molecules) have shown that tolerance was a consequence of peripheral Foxp3⁺ Treg induction and could not be induced in the absence of functional Foxp3 (15). Importantly, the tolerance state we achieved in this study is long term, as mice tolerized with OVA-alum + anti-CD4 accept indefinitely OVA-expressing skin grafts transplanted later. In addition, the tolerance state is Ag specific and not due to the persistence of the tolerance-inducing mAbs, because mice tolerized with β-LG–alum + anti-CD4 reject OVA-expressing skin grafts. These data also demonstrate that tolerant animals remain immunocompetent to mount immune responses toward unrelated Ags.

By performing adoptive cell transfers of OVA-specific T cells into wild-type (WT) mice, we could follow the fate of Ag-specific cells as tolerance is induced. We found that CD4 blockade at the time of exposure to the Ag has an effect not only on the inhibition of proliferation but also favoring apoptosis of the cells as they enter the cell cycle. We have recently investigated the impact of CD4 blockade in vitro at the time of activation using mathematical models. We found that although the impact on the rate of cell proliferation and apoptosis was small, such small effects led to a major difference in the outcome (38, 39). Previous studies with costimulation blockade had also shown that activation under those tolerogenic conditions lead to apoptosis of effector cells (20, 40–43), but in that case apoptosis occurred together with the induction of Treg cells (12, 44). We now found that apoptosis of the Ag-specific T cells is not complete, as some cells survive deletion. When we assessed those surviving Ag-specific cells, we found that they have a CTLA-4⁺IL-2^low anergic and immunosuppressive phenotype that is independent of TGFβ and Foxp3. This is consistent with studies suggesting that three factors can endow a conventional T cell population with a suppressive function: TCR triggering, IL-2 repression, and CTLA-4 expression (36).

We found that IL-10, unlike Foxp3, is key for the induction of immune tolerance triggered by Ag in alum in anti-CD4–treated mice. The implication of IL-10 in transplantation tolerance has been controversial. Several studies were able to demonstrate long-term transplantation tolerance in the absence of IL-10 (13, 45, 46). The majority of studies that have shown a strong IL-10 requirement for transplantation tolerance have used donor blood transfusion as part of the tolerance-inducing regimen (47, 48). It is possible that the systemic delivery of Ags mimics the boosting of an IL-10–dependent mechanism (in addition to the recessive mechanisms associated to apoptosis and inhibition of proliferation) that we described for OVA-alum.

Given that dominant transplantation tolerance can be extended to additional Ags present in new transplanted tissue by a mechanism that was termed linked suppression (49), it may be possible to use immune-dominant alloantigens to boost tolerance to tissues through a Foxp3-independent mechanism. But more important is the realization that effective tolerance to proteins, such as in allergic diseases or protein replacement therapies (namely in hemophilia or lysosomal storage diseases), may involve robust tolerance mechanisms that are independent of Foxp3.

We are grateful to Herman Waldmann for nondepleting anti-CD4 hybridoma, and Shimon Sakaguchi for helpful suggestions.

The authors have no financial conflicts of interest.