Low-Dose IL-2 Induces Regulatory T Cell–Mediated Control of Experimental Food Allergy

Thymic-derived CD4⁺CD25⁺FOXP3⁺ regulatory T cells (Tregs) account for ∼5–10% of circulating CD4⁺ T cells, suppress autoreactive lymphocytes, and control innate and adaptive immune responses in mice and humans (1, 2). Treg impairment is associated with loss of tolerance, autoimmunity, and also allergy (3). Indeed, allergy is one of the clinical manifestations of patients with the immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome, a Treg deficit caused by mutations in FOXP3 (4). Dysfunctional Tregs have been also identified in allergic individuals (5–7). Both decreased Treg numbers and function have been reported, particularly in children at high risk for allergic diseases based on maternal atopy compared with children at low risk (8). Studies investigating the pathophysiology of food allergy point to a functional role of Tregs in the development of normal tolerance to food allergens and the spontaneous resolution of milk allergy (9, 10).

In support for a role of Tregs in allergy, the induction of Tregs has been also demonstrated in successful allergen-specific immunotherapy (AIT), where they suppressed almost all aspects of Th2-mediated hypersensitivity (11, 12). In addition, transfer of Tregs inhibited and reversed airway hyperresponsiveness, gut inflammation, and related Th2 responses in murine models (13–15). Collectively, these data highlight that allergic diseases result from an inappropriate balance between allergen-activated effector Th2 cells and Tregs, and that treatments to enhance Treg responses might be useful in prevention and treatment of allergic diseases.

IL-2 is critical for Treg development, expansion, activity, and survival (16). IL-2 was originally discovered as a growth factor for activated T cells in vitro and has been extensively used to boost immune responses in patients with tumors (17) or chronic infection such as HIV (18). More recently, it was shown that ld-IL-2 increase specifically Treg numbers and function, and may be of therapeutic benefit in the treatment of autoimmune and inflammatory diseases. Pioneer clinical trials revealed the potential benefit of ld-IL-2 administration in alleviating chronic graft versus host disease (19, 20) and chronic hepatitis C–mediated vasculitis (21). This was then confirmed in alopecia areata (22) and other autoimmune diseases including systemic lupus erythematosus (23), confirming the potential of this approach. IL-2 is not only required for the survival of Foxp3⁺ Tregs in the periphery, but also appears to be critical for maintenance of their suppressive functions (24). Thus, we sought to evaluate ld-IL-2 treatment in allergy control. Our results show that ld-IL-2 induces specific Treg expansion and activation, including in gut and lymphoid-associated tissues, which elicit protection against clinical manifestations of food allergy in mice. We demonstrated both preventive and therapeutic long-term effects of ld-IL-2. The mechanisms underlying this efficacy depend on the expansion and activation of Tregs that modify the Th1/Th2 balance and inhibit mast cell activation.

Seven-week-old BALB/C (AnNR/J) female mice were purchased from Elevage Janvier (Le Genest-Saint-Isle, France). Mice were maintained in our animal facility under specific pathogen-free conditions in agreement with current European Union legislation on animal care, housing, and scientific experimentation. All procedures were approved by the local ethics animal committee. Mice were treated daily by i.p. or oral (p.o.) administration with the human rIL-2 (Proleukin; Novartis) during 5 consecutive days at the dose of 50,000 IU/injection in 200 or 100 μl, respectively.

Based on published procedures (15) with specific modifications, BALB/c mice were sensitized twice at 1-wk interval by i.p. injection of 10 μg OVA (Ova A5503; Sigma-Aldrich) mixed with 500 μg aluminum hydroxide (Alum) gel (AlH₃O₃; Sigma) or PBS mixed with Alum as control (nonsensitized mice [NS]). Allergy was then induced 10 d later by OVA p.o. administrations (20 mg/mouse), five times within a 10-d period. Severity of allergic response was assessed 30–45 min after the last induction by measuring changes in body temperature using a rectal thermometer (Bioseb, Vitrolles, France) and by evaluating clinical score (component of the severity of diarrhea graded from 0 to 3 points and appearance of hirsute pelage marked out 0–2 points).

Crude peanut extract was prepared from defatted raw flour as previously described (25). Mice were sensitized i.p. three times at 1-wk interval with 500 μg peanut extract in 500 μg Alum. Control groups received PBS plus 500 μg Alum. Mice were then orally challenged with peanut extract (15 mg) every 2 d for a total of seven challenges. Severity of allergic response was assessed 30–45 min after the last challenge by measuring body temperature drops and clinical symptoms including: 1) repetitive mouth/ear scratching, 2) piloerection, 3) puffiness around eyes or mouth/eye edema, 4) decreased activity/lethargy, 5) diarrhea, and 6) death. Each clinical sign was noted with a score of 1; the maximum score per mouse is 6.

Mice received two i.p. injections of 250 μg PC61 Ab (anti-CD25; Bio X Cell) at 3-d interval before OVA sensitization or challenge. As control, mice were treated with PBS or isotype (Antibodies Online). Blood samples were collected at day 10 post-PC61 treatment to control Treg depletion by flow cytometry.

Spleens were isolated and dissociated in PBS 5% FCS. RBCs were lysed with 1 ml ammonium-chloride-potassium buffer and then washed with 9 ml PBS 5% FCS. Mesenteric lymph nodes (MLNs) and Peyer’s patches (PP) were isolated and then digested with 0.46 mg/ml Liberase TM and 0.1 mg/ml DNase (both from Roche Applied Science) in complete RPMI 1640 20% FCS medium for 10 min and filtered. Lamina propria lymphocytes from small intestine (SI) were isolated as previously described with minor modifications (26). In brief, SI was harvested and PP were removed. The tissue was then cut in small pieces and incubated with 10 μg/ml Liberase TM, 0.1 mg/ml DNase, and 0.5 mg/ml hyaluronidase (all from Roche Applied Science) for 25 min at 37°C in a shaker (150 rpm). The tissues were smashed, passed through a 70-μm cell strainer, and washed in complete RPMI 1640 3% FCS. Cells were centrifuged at 1300 rpm for 7 min at 4°C, washed again, passed through a 40-μm cell strainer, and finally suspended in complete RPMI 1640 20% FCS medium.

Isolated cells were stained with the following Abs at predetermined optimal dilutions for 20 min at 4°C: CD45-PETR, B220-FITC, IgM-allophycocyanin, IgD-PeCy7, SiglecF-PE, streptavidin-e780 (eBioscience), and IgE-biotin (Southern Biotech) to analyze B cells/eosinophils; Live Dead-e780, CD45-PETR, CD49b-FITC, FCεR1-PeCy7, c-Kit–allophycocyanin, B220-e450, SiglecF-PE, CD3-HorizonV500, NkP46-AlexFluor700 (eBioscience) to analyze mast cells and NK cells; and CD45-PETR, CD3-PE, CD8-Alexa700, CD4-HorizonV500, CD25-PeCy7, GITR-PE (eBioscience), and ICOS-allophycocyanin (Biolegend) to analyze T cells. Intracellular detection of Foxp3-e450 (eBioscience), CTLA4-biotin (eBioscience), and Ki67-FITC (eBioscience) were performed on fixed and permeabilized cells using a CytoFix/CytoPerm kit buffer (eBioscience). Cells were acquired on an LSR II (Becton Dickinson) and analyzed with FlowJo software (Tree Star). Dead cells were excluded by forward/side scatter gating. Tregs were defined as Foxp3⁺CD25⁺CD4⁺ cells, activated Tregs as ICOS⁺Foxp3⁺CD25⁺CD4⁺ cells, and activated effector T cells (Teff) as CD25⁺Foxp3⁻CD4⁺ cells. Memory and naive B cells were defined as IgM⁻IgD⁻ and IgM⁺IgD⁺ cells among B220⁺ cells, respectively. Eosinophils and mast cells were characterized as SSC^hiCD45⁺SiglecF⁺ and FceR1⁺c-Kit⁺CD49b⁻, respectively.

Heparinized blood sample was collected 2 d after the last OVA p.o. administrations and erythrocytes were lysed. Cells were stimulated with RPMI 1640 (negative control), 500 ng/ml anti-IgE (positive control; Southern Biotech), or 1 μg/ml OVA for 2 h at 37°C. Cells were then stained with CD45-PETR, CD49b-FITC, FcεR1-PeCy7, and IgE-biotin to identify the basophil population and CD63-allophycocyanin (eBioscience) to measure degranulation.

OVA-specific production of IFN-γ, IL-4, and IL-5 were determined by murine ELISPOT assay (Mabtech, Nacka Strand, Sweden). In brief, 5 × 10⁵ cells/well (or 2 × 10⁵ for IFN-γ) were restimulated for 48 h in duplicates at 37°C in 5% CO₂ with 100 μg OVA protein (0.5 mg/ml final). Medium alone and Con A (Sigma) at 2 μg/ml final were used as negative and positive controls, respectively. Spots were counted with an AID ELISPOT reader (ELR03; AID Autoimmun Diagnostika, Strassberg, Germany). Results, expressed as spot-forming units (SFU) per 10⁶ cells, represent the mean of duplicates.

To measure murine mast cell protease type 1 (mMCPT-1) levels, we harvested blood samples 30 min after the last OVA challenge and isolated sera by centrifugation. Samples were frozen and kept at −20°C until use. Level of mMCPT-1 in serum was measured according to the manufacturer’s recommendations using specific mouse ELISA kits (eBioscience).

IFN-γ, IL-4, and IL-5 secretion levels were measured by murine ELISA assays (eBioscience) in supernatants of cells (5 × 10⁵ cells/well) cultured during 72 h in 96-well plates (Nunc, Denmark) in complete RPMI 1640 with 100 μg OVA (0.5 mg/ml final).

OVA-specific IgE titers were quantified in serum of sensitized mice by ELISA. OVA (500 ng/well) was coated in flat-bottom, 96-well plates (Nunc, Denmark) at 4°C overnight. After washing and blocking nonspecific binding sites with PBS + 1% BSA for 1 h at room temperature, 100 μl diluted serum samples were added to each well and incubated for 2 h at room temperature. OVA-specific IgE were detected with biotinylated anti-IgE (Southern Biotech) and revealed with peroxidase-conjugated streptavidin (Sigma-Aldrich) and TMB at room temperature for 5 min. The reaction was stopped by adding 100 μl HCl (1M), and the OD was read at 450 nm using an automatic ELISA plate reader (DTX 880, Multimode detector; Beckman Coulter).

Segments of duodenum were flushed with PBS 5% FCS and immersed in 1 ml RNA-later (Qiagen, Hilden, Germany) and stored at −80°C. For RNA extraction, total tissues were transferred in 1 ml TRIzol Reagent (Life Technologies, France), and disruption and homogenization were done with TissueLyser II (Qiagen). Total RNA was isolated by using RNeasy Mini Kits (Qiagen). cDNA was generated using SuperScript III (Life Technologies) according to the manufacturer’s instructions. Quantitative PCR was performed using the 7500 Fast Real-Time PCR System (Life Technologies) with Fast Master mix (Life Technologies). The probe IDs are: m1,TNF-α: Mm00443258_m1, IFN-γ: Mm01168134_m1, IL-4: Mm00445259_m1, IL-5: Mm00439646_m1, IL-6: Mm00446190_m1, IL-10: Mm00439614_m1, IL-12p40: Mm01288992_m1, IL-13: Mm00434204_m1, 18S: Mm03928990_g1, and GAPDH: Mm99999915_g1. PCRs were performed in triplicate, and the mRNA levels were normalized to that of GAPDH and 18S mRNA levels. Gene expressions in control and ld-IL-2–treated groups were expressed as relative transcript abundance, relative to the GAPDH reference gene.

One centimeter length of duodenum was removed and immersed in 4% formaldehyde buffer. Tissues were further dehydrated in alcohol solutions using an automatic tissue processor (STP 120; Microm Microtech, Francheville, France) and included in paraffin. Transverse sections were cut at 4 mm by HM340E microtome (Thermo Scientific). Sections were stained with toluidine blue, and mast cells were visualized with a Leica DMRBE microscope. A total of four sections per group were randomly analyzed (original magnification ×400), and results were expressed as number of mast cells per millimeter.

Statistical significances were evaluated between the treated groups and control subjects using GraphPad Prism (GraphPad Software, San Diego, CA) using the Mann–Whitney U test (comparison of rank, unpaired test, nonparametric test, two-tailed p value) with *p < 0.05 taken as statistical significance (**p < 0.01, ***p < 0.001). Statistical significances for long-term ld-IL-2 therapeutic effect were evaluated using one-way ANOVA test with *p < 0.05 taken as statistical significance (***p < 0.001).

We evaluated the effect of ld-IL-2 on immune cell populations at various locations, focusing mainly on the GALTs. BALB/c mice were injected daily for 5 d with 50,000 IU IL-2 per injection or PBS as a negative control. CD4⁺, CD8⁺ T cells, B cells, Tregs, and Teff were monitored in spleen, MLN, PP, and SI mucosa. We observed that ld-IL-2 had no significant impact on CD4⁺, CD8⁺ T cell, and B cell populations in spleen, PP, and SI in terms of percentages and absolute numbers (Fig. 1A). Only a significant expansion of total cells in MLN was observed in ld-IL-2–treated mice, which did not affect the relative values of B, CD4⁺, and CD8⁺ T cells (Fig. 1A). As expected, ld-IL-2 induced expansion of total CD4⁺FOXP3⁺ cells (Fig. 1B, upper panels), Tregs (CD4⁺CD25⁺FOXP3⁺; data not shown), and CD8⁺ Tregs (CD8⁺FOXP3⁺; Supplemental Fig. 1) in the different organs. Interestingly, the highly suppressive ICOS⁺ Tregs (27) were robustly expanded in percentage (Fig. 1B, 1C) and absolute number (data not shown) in local and peripheral lymphoid tissues. Ld-IL-2 induced both Treg proliferation and activation as shown by a significant increase of the proportion of Ki67⁺, ICOS^hi, and CTLA-4⁺ Tregs in spleen and MLN (Fig. 1E). In contrast, ld-IL-2 had no significant impact on CD4⁺CD25⁺FOXP3⁻ Teff, except a moderate but significant increase in spleen (Fig. 1C). Nevertheless, the Treg/Teff ratio was significantly increased in all organs of treated mice as compared with control subjects (Fig. 1D). Importantly, other cell populations such as mast cells, eosinophils, type 2 innate lymphoid cells (ILC2), naive B cells, or IgE⁺ B cells are not significantly expended under ld-IL-2 except for moderate eosinophilia in spleen and NK cells (Supplemental Fig. 1). A small increase of IgE⁺ B cells in spleen and mast cells in SI were also observed in IL-2–treated mice but at lower levels than allergic mice. Taken together, these results show that ld-IL-2 induces expansion and activation of Tregs in naive mice, notably in gut-associated tissues.

A murine food allergy model with high degree of anaphylaxis was established by repeated p.o. administration of OVA protein in OVA-sensitized mice (Fig. 2). Severe allergic reactions were observed in sensitized mice and scored based on diarrhea and fur characteristics (Fig. 2A). A significant decrease in body temperature was also induced in OVA-sensitized mice and was positively correlated with the severity of allergic symptoms (Fig. 2A, 2B). Elevated mMCPT-1 levels were also detected in serum of sensitized mice after OVA challenges and can be related with an increase of mast cells in the gut (Fig. 2C). The robustness of this food allergy model was also evidenced by the severe allergic reactions observed in sensitized mice when they were challenged 30 or 60 d after the last sensitization (data not shown). Moreover, a strong systemic and local Th2 bias was observed in OVA-sensitized mice as shown by high levels of IL-4– and IL-5–secreting cells in spleen, MLN, and PP as compared with the NS (Fig. 2D and data not shown). We also detected OVA-specific IgE Abs in serum and feces in sensitized mice that can be related to a significant increase of IgE⁺ memory B cells in allergic mice compared with NS (Fig. 2E). Taken together, these results show that our protocol is effective to establish food allergy with a high degree of anaphylaxis and strong Th2-biased immune responses.

We first investigated the preventive efficacy of ld-IL-2 in the OVA-specific allergy model. Mice were daily i.p. injected with IL-2 or PBS as control for 5 consecutive days and then sensitized and challenged with OVA according to our protocol. We observed that i.p. ld-IL-2 prevents food allergy because 90% of IL-2–treated mice were fully protected (score ≤2), whereas control mice had severe allergic reactions with a median score of 5 (Fig. 3A). The protective effect of ld-IL-2 was confirmed by the prevention of body temperature decline in IL-2–treated mice in contrast with PBS-treated controls (Fig. 3A) and a significant decrease of serum levels of mMCPT-1 (Fig. 3B) that were both positively correlated (p < 0.001, r = 0.746). Notably, the protective effect was robust because IL-2–treated mice were still protected against severe hypothermia (temperature declines >2°C) after repeated OVA challenges (Fig. 3C).

Protective effect of ld-IL-2 was also demonstrated in a second experimental model of food allergy using peanut protein extract as allergen and using specific scoring system based on anaphylactic symptoms to measure the severity of allergic reactions (Fig. 3D and 2Materials and Methods). We observed that IL-2–treated mice were protected against peanut allergy because very few mice presented anaphylactic symptoms, whereas pruritus, piloerection, swollen eyes, lethargy, and diarrhea were observed in the PBS group after peanut extract challenge. Absence of temperature decline in IL-2–treated mice confirmed the preventive effect of IL-2 (Fig. 3D).

To evaluate the role of Tregs in ld-IL-2 treatment efficacy, we performed similar experiments in which Tregs were depleted. Mice were first treated by i.p. ld-IL-2 or PBS during 5 d and then injected or not with PC61 mAb (anti-CD25 IgG1) to eliminate Tregs. Mice were then sensitized and challenged with OVA to induce food allergy. PC61-mediated specific Treg depletion was documented by the major loss of CD4⁺FOXP3⁺CD25⁺ cells in blood of treated mice at day 10 and confirmed by the significant decrease of total Foxp3⁺ cells in comparison with PBS or isotype controls (Supplemental Fig. 2). Notably, no variations in blood eosinophil and NK cell counts or ILC2 cells in PP were observed (Supplemental Fig. 2).

After OVA challenges, we observed that sensitized control mice in which Tregs have been depleted after PC61 treatment had more severe allergic symptoms than controls, highlighting that Tregs downmodulate allergic immune responses in PBS-treated control mice (Fig. 3E). More importantly, we demonstrated that Treg depletion abrogates the efficacy of ld-IL-2 as shown by severe allergic symptoms, an important decrease in body temperature (Fig. 3E) and significant increase of mMCPT1 levels (Fig. 3G) in Treg-depleted IL-2–treated mice. Interestingly, the Treg counts observed before mice sensitization (pre-S) can be correlated with the severity of the allergic reactions (Fig. 3F). However, no significant variation of OVA-specific Abs was observed in serum of treated mice at this early time point, including when OVA-specific IgE, IgG2a, IgG1, and IgA isotypes were measured (Fig. 3H and data not shown).

To formally demonstrate that the protective effect induced by ld-IL-2 is Treg mediated, we realized experiments in which Tregs were depleted in protected mice, 1 mo after ld-IL-2 treatment. After additional OVA challenges, we observed that late PC61 treatment abolished protection because treated mice presented severe allergic reactions with clinical scores comparable with PBS control groups (Fig. 3I).

To better characterize the mechanisms associated to the clinical protection induced by ld-IL-2, we analyzed the immune profile of treated and untreated mice 2 d after the last OVA challenge. While ld-IL-2–treated mice were protected from allergic responses, the percentages of total CD4⁺FOXP3⁺ T cells or activated CD4⁺CD25⁺FOXP3⁺ICOS⁺ Tregs (Fig. 4A) were not increased in the different organs as compared with control mice. We also did not observe variation of IL-4– and IL-5–secreting cell numbers measured by ELISPOT in spleen, PP (data not shown), and MLN (Fig. 4B, upper panels). However, when IL-4 and IL-5 productions were measured by ELISA in supernatants after Ag-specific restimulation, we reported a decrease of Th2 cytokine secretions in MLN of IL-2–treated mice (Fig. 4B, lower panels). In contrast, we observed an increase of OVA-specific IFN-γ–secreting cell numbers and IFN-γ production in MLN and PP of IL-2–treated mice compared with control mice (Fig. 4B and data not shown). As a result, ld-IL-2 treatment led to a modulation of Th2/Th1 balance in draining lymphoid organs highlighted with the decline of the IL-4/IFN-γ ratio in PP (data not shown) and MLN (Fig. 4C). We also observed specific local modifications of Th2 cytokine expression levels assessed in SI by quantitative RT-PCR (qRT-PCR). IL-4, IL-5, IL-6, and IL-13 gene expression were decreased in the duodenum of the IL-2–treated group (Fig. 4D), whereas IFN-γ, IL-12, TNF-α, and IL-10 gene expression levels were increased (Fig. 4D).

We also investigated the effect of ld-IL-2 treatment on the different effector cells of allergic disease. We observed that ld-IL-2 prevented the eosinophil and basophil recruitment as shown by the lower cell counts in blood of treated mice compared with the PBS group (Fig. 5A, 5B). Interestingly, we showed in ex vivo assay that ld-IL-2 reduced basophil degranulation as shown by the low expression of CD63 in IL-2–treated mice compared with control mice. More importantly, we also observed in SI of treated mice lower frequency of mast cells that can be correlated with limitation of temperature decline (Fig. 5C). The control of mast cell infiltrate induced by IL-2 is Treg mediated because the percentages of mast cells were high, reaching similar levels than allergic animals in PC61-treated mice (Fig. 5C). These results show that ld-IL-2 prevents food allergy by a Th1 shift of the Th1/Th2 balance and reduced activation of effector cells.

To further evaluate ld-IL-2 efficacy, we investigated the long-term effect of IL-2 treatment for food allergy protection and evaluated its preventive and therapeutic effects. In addition to the 5-d ld-IL-2 course performed pre-S, a 5-d ld-IL-2 course between mice sensitization and challenge (pre-C) and a therapeutic 5-d ld-IL-2 course after p.o. challenge were also tested (Fig. 6).

When long-term efficacy of the pre-S treatment was evaluated, we confirmed that IL-2–treated mice are immediately protected from allergic symptoms (p < 0.05, Mann–Whitney compared with PBS group at day 23; Fig. 6A) and remains protected against four additional cycles of OVA challenges (between days 36 and 100; Fig. 6A). After another 5-d ld-IL-2 course initiated 3 mo after the first one, long-term treatment efficacy was observed because mice resisted to 10 additional OVA challenge cycles. Altogether, we observed that ld-IL-2 treatment could control allergic reactions over a 10-mo period, with no progressive loss of efficacy over time (Fig. 6A) that was confirmed by absence of body temperature variation (Supplemental Fig. 3).

A between mice sensitization and challenge treatment showed that protection was not immediately effective (score >2 at day 23; Fig. 6B) but was elicited as from the second cycle of OVA challenges (day 36). As previously and because of a partial loss of the protective effect at days 79 and 100, a new 5-d ld-IL-2 course was performed, and the long-term efficacy of the treatment was then tested. We observed that IL-2 could control allergic reactions over a 7-mo period with high efficacy because >80% of mice were fully protected (score ≤2; Fig. 6B) and no significant temperature decline was observed (Supplemental Fig. 3).

In the same experiment, we observed that allergic mice treated with IL-2 after p.o. challenge were also not immediately protected from allergic reactions induced by OVA challenges (day 36 and day 56; Fig. 6C). However, mice became resistant to next OVA challenges (day 79 and day 100). Additional cure of ld-IL-2 was performed like other groups 3 mo after the sensitization, and long-term efficacy was investigated over an extra period of 7 mo. We observed that IL-2 could control allergic reactions with similar efficacy and long-term maintenance than the pre-S treatment (Fig. 6A, 6C). Importantly, therapeutic efficacy of IL-2 was confirmed by absence of body temperature variation and low level of MPCT1 in IL-2–treated mice compared with control mice (Supplemental Fig. 3). However, no significant variations of OVA-specific IgE titer were observed at early (day 73) or late (day 320) time points in IL-2–treated mice compared with control mice (Supplemental Fig. 3).

We wondered whether alternative and more convenient routes of administration could be used to perform ld-IL-2 treatment. We demonstrated that ld-IL-2 p.o. treatment induced protection against allergy as shown by the lower susceptibility of IL-2–treated mice to allergy induction compared with control mice (Fig. 7A). A robust protection was observed because mice were fully protected after three cycles of challenges over a 2-mo period (data not shown). We also demonstrated that the protective effect is Treg-mediated because protection was abolished when Tregs have been depleted after IL-2 p.o. treatment (Fig. 7A). Notably, IL-2 p.o. induced protection is associated with a decrease of mast cell frequency in the duodenum, whereas large infiltrates were shown in Treg-depleted mice after OVA challenges (Fig. 7B).

Altogether, these results show that both preventive and curative administration of IL-2 provide long-term clinical protection against food allergy, including when IL-2 is administrated by more convenient route.

Previous investigations have established that Tregs play a central role both in the development of natural p.o. tolerance (28, 29) and in the mechanism of AIT efficacy (12, 30, 31). However, specific Treg-based immunotherapies have not been evaluated in the context of food allergy. For this purpose, we established an experimental model based on previous work (15), comprising sensitization with OVA/alum i.p. injections followed by administration of high amounts of OVA by repeated intragastric delivery. This induces gastrointestinal hypersensitivity characterized notably by temperature decline, acute diarrhea, local recruitment, and degranulation of effector cells, associated with a Th2 immune response profile as shown by high levels of IL-4– and IL-5–secreting cells in peripheral and gut-associated tissues. This Th2 profile was confirmed by high levels of OVA-specific IgE and IgG1 Abs after allergen challenges (Fig. 2 and data not shown). As discussed recently by Liu et al. (32), the genetic background, the nature of allergen, and the route of allergen exposure greatly influence the phenotype of mouse models. We have tested different inbred mouse strains and selected the BALB/c AnN mice in which we succeed to induce severe allergic reactions. Indeed, in contrast with a previous report (33), we observed that a significant proportion of sensitized BALB/c AnN mice were diarrheic and were hypothermic after a limited number of OVA challenges (Figs. 2, 3) whatever the animal provider (Janvier Labs or Charles River; data not shown).

Using this mouse model, we showed that ld-IL-2 treatment efficiently prevents and controls food allergy reactions that can be explained by the capacity of Tregs to rebalance the Th2-dominant local immune responses. Indeed, we observed in protected mice a significant decrease of Th2-associated gene expression and cytokine production in gut and lymphoid-associated tissues, whereas the Ag-specific Th1 immune responses in MLN and PP were amplified (Fig. 4 and data not shown). Because IL-4 has been shown to exacerbate anaphylaxis (34), the reduction of IL-4 levels induced by ld-IL-2 during the effector phase of immediate hypersensitivity may rapidly alleviate allergic symptoms. Interestingly, the Th1/Th2 immune deviation cannot be related with the potential immunogenicity of human IL-2 used in these experiments because control of Th2 allergic responses was also observed in mice treated with murine IL-2 (Supplemental Fig. 4 and data not shown). In contrast, we showed that mice treated with heat-denatured human IL-2 were not protected against Th2-mediated allergic reaction. A switch to Th1 immune responses was already reported in AIT (35), including in food allergy treatment (36). Even though Th1 redirection induced by AIT with allergen extracts or peptides continues to be reported in more recent publications (37, 38), other groups, however, did not observe this effect (39). More recently, the induction of allergen-specific Tregs producing IL-10 has been considered as a main event causing peripheral T cell tolerance (35). Importantly, the mechanisms of immunodeviation and immunoregulation should not be considered as alternative and mutually exclusive events. Indeed, it was reported that Th2 to Th1 recommitment of T cells correlates with the induction of CD4⁺CD25^high and IL-10–producing Tregs in wasp venom immunotherapy (40). Accumulation of IL-10⁺ Tregs and Th1 cells in allergen-challenged skin sites was also shown after tolerogenic vaccination with Fel d 1 peptides (41). In our results, we observed an increase of both IL-10 and IFN-γ responses in PP and duodenum (data not shown). Regulatory IFN-γ/IL-10 double-positive Foxp3⁻ CD4⁺ T cells that have been described in mice undergoing chronic Ag stimulation (42) might be induced after ld-IL-2 treatment, thus explaining the unexpected absence of Foxp3⁺ Tregs expansion after p.o. challenges in the protected mice (Fig. 4A). In this line, Wilson et al. (43) report that the preventive protective effect of IL-2:anti–IL-2–based therapy in the mouse asthma model was correlated with an increase of IL-10⁺CD4⁺ cells. There are other lines of evidence suggesting that IL-10 and Tregs play interrelated roles in the regulation of effector phase of the allergic responses. IL-10 produced by increased numbers of induced Tregs during AIT downregulates eosinophil function and activity, and suppresses IL-5 production by Th2 cells (44, 45). As a result, local IL-10 production might have a major contribution in ld-IL-2 treatment efficacy, as suggested by the established correlation between IL-10 secretion in PP and the control of degranulation revealed by low mMCPT-1 concentration in serum of ld-IL-2–treated mice (data not shown).

Nevertheless, we can state that ld-IL-2 acts primarily through Treg expansion and activation. The fundamental role of Tregs in protective phenotype was demonstrated by the abolition of protection after anti-CD25 PC61 mAb treatment, which induced Treg depletion but preserved ILC2 and eosinophil populations that are known to promote allergic reactions. Interestingly, we observed that ld-IL-2 induced initial Treg expansion in systemic (spleen) and local compartments (MLN, PP) including SI where Tregs express a lower level of CD25, whereas their proliferation occurred mainly at a systemic level as shown by the increased percentages of KI67⁺, ICOS^hi, and CTLA4^hi Tregs in the spleen (Fig. 1). In contrast, a few weeks after the end of ld-IL-2 treatment, we did not observe any sustained CD4⁺Foxp3⁺ or CD4⁺CD25⁺Foxp3⁺ICOS⁺ Treg expansion in the different organs of protected mice after OVA challenges (Fig. 4A). These observations suggest that: 1) alternative local immune-suppressive mechanisms might be induced by IL-2–activated Tregs, including suppressive Foxp3⁻ T cells; or 2) Ag-specific Tregs might be selected and persisted at low frequencies. Interestingly, a relationship between the ICOS^hi Tregs and the suppressive IL-10⁺Foxp3⁻ Tr1 cells has been already described (46). However, the loss of protection observed when PC61 mAb was administered between two OVA challenges (Fig. 3I) highlights the role of Tregs in the long-term protected phenotype. Further investigations are necessary to address this specific issue. We especially plan to investigate whether association of ld-IL-2 with allergen could improve therapeutic efficacy by generating allergen-specific Tregs. Alternatively, specific IL-2 treatment regimens to maintain a high level of Tregs over a long-term period could be proposed. Importantly, we previously demonstrated that sustained stimulation and expansion of Tregs does not induce general immunosuppression because we showed that long-term IL-2 treatments control immune disorders in mice without impairing immune responses to infection, vaccination, and cancer (47). Alternative and more convenient routes of ld-IL-2 administration can be proposed. In this article, we have shown for the first time, to our knowledge, that p.o. administration of IL-2 can induce Treg activation and immune protection against food allergy that can be related to a local modification of Th1/Th2 balance (i.e., decrease of IL-4 and increase of IFN-γ responses in PP of IL-2–treated mice; data not shown) and a control of mast cell infiltration in gut (Fig. 7). Accordingly with previous work (48), our results emphasize the suppressive function of Tregs on mast cell compartment because Treg depletion induces an intestinal mastocytosis that can be correlated with the severity of allergic reactions (Figs. 5C, 7B). Whatever the route of IL-2 administration, we have shown that ld-IL-2 induced a Treg-mediated control of allergic diseases that may be explained by Th2 and mast cell suppression (34, 48).

Altogether, the present data show for the first time, to our knowledge, the therapeutic potential of ld-IL-2 for the treatment of food allergy and should prompt further investigations of this potential to re-establish long-term immune tolerance to allergens.

We thank the staff at Centre d’Expérimentation et Formation (Groupe Hospitalier Pitié-Salpêtrière) for taking care of the mice and Raphael Jeger-Madiot, Pierre-Axel Vinot, and Simon Brunel for technical assistance.

The authors have no financial conflicts of interest.