The Effector T Cell Response to Ryegrass Pollen Is Counterregulated by Simultaneous Induction of Regulatory T Cells

Thymus-derived natural regulatory T cells (Tregs) maintain the balance between defense against pathogens and evolving immunopathology or autoimmune disease. They control self-Ag–specific autoreactive T cells that have escaped depletion in the thymus by suppression of effector T cell (Teff) proliferation and cytokine production (1). Murine and human natural Tregs are CD4⁺ and characterized by constitutive expression of high levels of the IL-2Rα–chain (CD25) and the transcription factor Foxp3 and low expression of the IL-7R (CD127) (2, 3).

Allergens are innocuous environmental non–self-Ags, but allergic individuals develop pathological immune responses to these Ags resulting in rhinitis, conjunctivitis, asthma, and airway remodeling. Susceptibility to allergy is associated with genetic predisposition as well as with environmental factors at the time of Ag exposure, but the underlying mechanisms are not clear (4). In contrast to self-Ags, foreign Ags, such as allergens, are not expressed in the thymus. Therefore, specific depletion of Teffs or generation of natural Ag-specific Tregs in the thymus cannot be used to establish tolerance to foreign Ags. Although tolerance is not desired to foreign Ags derived from pathogens, lack of tolerance is a problem in the case of harmless environmental Ags, such as allergens. Furthermore, immune responses to pathogens also need to be regulated to limit associated tissue damage. In a murine model, allergen-specific Tregs were demonstrated to reverse allergic airway inflammation and prevent development of airway remodeling (5). However, it is unknown whether Tregs specific for foreign Ags are derived from the thymus, perhaps due to cross-reactivity of foreign Ags with self-Ags, or by conversion from non-Tregs in the periphery.

Human studies have mostly identified Tregs based on high expression of CD25 and Foxp3 and, in some cases, low expression of CD127 (2, 3). However, although the expression pattern of these markers differs between Tregs and resting Teffs, it changes toward the Treg phenotype upon activation of Teffs. Therefore, these markers cannot be used to identify Treg poststimulation in vitro. Furthermore, activated T cells with upregulated CD25 and Foxp3 are likely to circulate in peripheral blood in response to inflammation elicited by pathogens or after exposure to pollen allergens during the pollen season. Recently, expression of the ectonucleotidase CD39 on Tregs was described, which, together with CD73, promotes their regulatory function by hydrolysis of ATP to the immunosuppressive molecule adenosine (6, 7).

Only limited studies have examined the frequency or function of Tregs in allergic individuals. Karlsson et al. (8) showed that depletion of CD4⁺CD25⁺ cells from PBMCs of children with cow’s milk allergy increased T cell proliferation to β-lactoglobulin. Ling et al. (9) found that the suppressor function of CD4⁺CD25⁺ cells isolated from the blood of allergic subjects was reduced compared with healthy controls, implying a role for Tregs in regulating responses to allergens. However, these CD25⁺ cells could have been contaminated with in vivo-activated Teffs that would dilute Tregs. During the pollen season, pollen exposure can lead to increased activation and expansion of allergen-specific Teffs in vivo prior to venesection of the subjects. Therefore, contamination of the purified Tregs with activated Teffs rather than impaired function of the Treg could explain the observed reduction in suppressor function in vitro. A more recent study clearly demonstrated that allergen-specific Foxp3⁺ Tregs are present and equally functional in PBMCs from both allergic and nonallergic subjects (10). In this study, however, suppressive function was tested after expansion of Tregs isolated from human PBMC for 15 d with allergen-loaded monocyte-derived dendritic cells and exogenous IL-2. This procedure may have preferentially expanded functional allergen-specific Treg.

Recently, it was shown that Foxp3 expression could be induced in CD4⁺CD25⁻Foxp3⁻ T cells by stimulation with anti-CD3 and anti-CD28 in the presence of TGF-β and IL-2, but it was not clear whether these converted cells exerted suppressor function (11, 12). Moreover, the extent of the TCR stimulation and concentrations of added cytokines were high compared with in vivo situations. These experiments were also performed using purified CD4⁺ T cells, removing potential contributions of other cell types to the microenvironment, including bystander effects of secreted cytokines.

Because of the similar phenotype of activated T cells and natural Tregs, we hypothesized that a subpopulation of Ag-activated T cells might represent adaptive Tregs. The generation of such allergen-specific Tregs in the periphery by conversion of non-Treg could be another mechanism for regulating immune responses to harmless non–self–Ags, such as allergens. To investigate the potential for generating Tregs in the periphery by expansion or conversion in a setting more closely resembling the in vivo situation, we stimulated whole PBMC from ryegrass pollen (RGP)-allergic subjects and nonallergic controls with RGP extract (RGPE) in the absence of exogenous cytokines. We observed the induction of a subset of dividing CD4⁺ T cells that expressed high levels of Foxp3 and had regulatory function.

Twenty-two RGP-allergic atopic donors (mean age 31.2 ± 7.8 y, 6 out of 22 female) and 12 nonatopic controls (mean age 42.1 ± 15.9 y, 6 out of 12 female) were recruited at the Alfred Hospital Allergy Clinic (Melbourne, Australia). A diagnosis of allergy to RGP was based on a clinical history of seasonal rhinitis (18 out of 18 patients) and/or asthma (7 out of 18 patients) during the grass pollen season, which in Australia runs from August to February and evidence of specific sensitization by positive skin prick test to RGPE and/or RGPE-specific IgE CAP-FEIA score (Phadia, Uppsala, Sweden) greater than 0.70 kU/l (class 2). Blood samples were taken outside the pollen season except for six of the atopic subjects used in experiments presented in Fig. 4A and 4B. No donor was suffering from symptoms of allergic rhinitis or asthma at the time of blood sampling. All patients gave written informed consent with approval by the Alfred Hospital and Monash University Ethics Committees (Melbourne, Australia).

RGP was purchased from Greer Laboratories (Lenoir, NC) as dry, nondefatted pollen. For RGP extraction, 4 g pollen was suspended in 40 ml PBS for 30 min at 37°C under constant rotation, followed by centrifugation at 5000 × g at room temperature. Supernatants were sterile filtered through a 0.2-μm pore-size filter (Sartorius, Goettingen, Germany).

PBMCs were isolated from heparinized blood using Ficoll-Paque (Pharmacia, Uppsala, Sweden) density gradient centrifugation. Cells from allergic and nonallergic donors used to compare induction of Foxp3^hi cells were cryopreserved and thawed as described by Maecker et al. (13). The cells were frozen in 10% DMSO in FCS at 10⁷ cells/ml and stored in liquid nitrogen for subsequent parallel testing. The recovery on thawing was 70–80%. Freshly isolated PBMCs were used for the experiments shown in Figs. 3, 4C, 4D, and 5.

CFSE labeling of PBMCs or sorted cells was performed at a concentration of 1 μM CFSE and 10⁷ cells/ml in PBS. Cells were stimulated with RGPE at a protein concentration of 20 μg/ml in complete medium (RPMI 1640 supplemented with 2 mM glutamine, 100 IU/ml penicillin-streptomycin (Life Technologies, Invitrogen, Auckland, New Zealand), and 5% screened, heat-inactivated human AB serum (Sigma-Aldrich, St. Louis, MO) in 24-well tissue culture plates (Greiner Labortechnik, Frickenhausen, Germany) with 2.5 × 10⁶ PBMCs per well in 2 ml volume at 37°C in 5% CO₂ in air in a Heraeus incubator (Heraeus, South Plainfield, NJ) for 7 d or longer as indicated. PKH-26 labeling of PBMC or sorted cells was performed according to the manufacturer’s instructions (Sigma-Aldrich).

For labeling of surface markers, cells were incubated with specific fluorochrome-conjugated Abs to CD4, CD25, CD127, CD73 (BD Biosciences, San Jose, CA), and CD39 (AbD Serotec, Kidlington, U.K.) for 20 min at 4°C. For Foxp3 staining, surface marker-stained cells were fixed, permeabilized, and stained for intranuclear Foxp3 using the Foxp3 staining kit (eBioscience, San Diego, CA) according to the manufacturer’s instructions.

For intracellular cytokine staining, 7 d cultured PBMCs were restimulated with PMA and ionomycin (Sigma-Aldrich) or with immobilized anti-CD3 mAb (OKT-3, in-house, purified from cell culture supernatant, coated at 4 μg/ml) and 2 μg/ml soluble anti-CD28 (BD Biosciences) for 5 h (IL-4, IFN-γ, IL-2) or 16 h (IL-17) in the presence of 10 μg/ml brefeldin A (Sigma-Aldrich) for the last 4 h. Staining with Abs specific for IL-4, IFN-γ (BD Biosciences), IL-2 (Invitrogen), or IL-17 (eBioscience) was performed simultaneously with staining for Foxp3. Staining for IL-10–secreting cells was performed after 16 h restimulation with anti-CD3/anti-CD28 using a cytokine secretion assay kit (Miltenyi Biotec, Bergisch-Gladbach, Germany) according to the manufacturer’s instructions.

Labeled cells were analyzed by flow cytometry on an FACSCalibur, FACSAria, or LSRII (BD Biosciences), and flow cytometry-based cell sorting was performed on an FACSAria (BD Biosciences).

PKH-26–labeled PBMCs of three RGP-allergic donors were stimulated with RGPE. After 7 d, CD4⁺ T cells were enriched by MACS followed by flow cytometry-based sorting of dividing CD4⁺CD25⁺CD39⁺ or CD39⁻ and nondividing CD4⁺CD25⁻CD39⁻ cells. Foxp3^hi cells were enriched in the CD39⁺ fraction and reduced in the CD39⁻ fraction.

Sorted T cell subsets were cocultured with CFSE-labeled PBMCs of the same donors at ratios as indicated maintaining a constant total cell number of 5 × 10⁴ cells per well in duplicate in a 96-well plate. Poststimulation with plate-bound anti-CD3 for 4 d cells were harvested and stained for CD3 and CD4 for analysis of proliferation of CD4⁺ T cells based on CFSE dilution. Cell culture supernatants were analyzed for secreted IFN-γ by ELISA (OptEIA kit, BD Biosciences).

CD4⁺ T cells were isolated from PBMCs by negative selection magnetic bead sorting using the human CD4⁺ T cell isolation kit II (Miltenyi Biotec) and then stained with fluorescently labeled Abs to CD4, CD25, and CD127 for further sorting by flow cytometry into CD4⁺CD25⁺CD127^lo pre-existing Tregs and CD4⁺CD25⁻CD127^hi non-Tregs. Sorted pre-existing Tregs labeled with CFSE and non-Tregs labeled with PKH-26 were cocultured for 7 d while being stimulated with RGPE or plate-bound anti-CD3. After 7 d, pre-existing Tregs and non-Tregs could still be distinguished based on CFSE or PKH-26 labeling. Proliferation as well as surface marker and Foxp3 expression were analyzed by flow cytometry (schematic overview in Fig. 5A).

Total RNA was isolated from cultured PBMCs using the RNeasy kit (Qiagen, Hilden, Germany) according to the manufacturer’s recommendations. cDNA was synthesized using multiscribe reverse transcriptase and random hexamer primers in the presence of RNase inhibitor (Applied Biosystems, Foster City, CA) at 42°C for 15 min followed by 5 min at 99°C. IL-10, TGF-β, and 18S rRNA-specific primers and TaqMan probes were designed using Primer Express software (Applied Biosystems) and synthesized by GeneWorks (Hindmarsh, South Australia, Australia; primers) and Applied Biosystems (probes) (Table I). Quantitative real-time PCR was performed in duplicate on an ABI Prism 7900 instrument (Applied Biosystems) as follows: 50°C for 2 min, 95°C for 10 min followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Results for target genes were analyzed using standard curves for each gene and normalized to 18S rRNA expression.

Supernatants were collected at indicated time points and stored at −20°C until assayed for IFN-γ using the the IFN-γ ELISA OptEIA kit (BD Biosciences). TGF-β was analyzed using the Duo Set ELISA kit (R&D Systems, Minneapolis, MN). IL-10 was determined by Bioplex assay (Bio-Rad, Hercules, CA).

Statistical analysis was carried out using GraphPad Prism software version 5 (GraphPad, San Diego, CA). p values were determined by either Mann-Whitney U test (nonparametric) for independent values or two-tailed paired Student t test for normalized paired values. Correlation analysis was performed using Spearman’s rank correlation for nonparametric data. A p value <0.05 was considered significant.

Freshly isolated PBMCs were stained for Treg markers CD25, CD127, and Foxp3 (Fig. 1A), and Tregs were identified as CD4⁺Foxp3⁺CD25⁺CD127^lo. To observe the induction of Tregs by allergen stimulation, CFSE-labeled PBMCs were cultured with RGPE for 7 d in complete medium without addition of exogenous cytokines. RGPE induced proliferation of CD4⁺ T cells in all RGP-allergic subjects tested (n = 18) as well as in all nonatopic subjects (n = 10). Dividing and nondividing cells were distinguished by CFSE dilution and analyzed for expression of the Treg markers CD25, CD127^lo, and Foxp3 (Fig. 1B).

In the nondividing population a subset of CD25⁺CD127^loFoxp3⁺CD4⁺ T cells similar to Tregs in unstimulated PBMCs was found. These are likely to be pre-existing Tregs that did not proliferate in response to RGPE stimulation, indicating anergy or specificity for different Ags. In the dividing CD4⁺ T cell population, two subsets expressing either intermediate or high levels of Foxp3 were observed (Fig. 1B). These results were confirmed using a different anti-Foxp3 Ab (clone 259D, Biolegend, San Diego, CA; data not shown). The Foxp3^int and Foxp3^hi cells did not differentially express CD25 or CD127. Instead, all dividing CD4⁺ T cells acquired a Treg-like phenotype as they upregulated CD25 and mostly downregulated CD127 to similar levels as found on the Treg ex vivo (Fig. 1A, 1B). High expression of Foxp3 in a subset of the dividing CD4⁺ T cells indicates the potential for a regulatory or suppressive function of these cells, which may be important in controlling the immune response to allergens. However, others have shown that the induction of Foxp3 expression by activating human CD4⁺ T cells is not necessarily associated with regulatory/suppressive function (12). Therefore, we next investigated whether Teff or Treg properties were associated with the two Foxp3 subsets of dividing CD4⁺ T cells.

Upon stimulation, Teffs are expected to produce their characteristic signature cytokines, such as IFN-γ and IL-2 by Th1, IL-4 by Th2, and IL-17 by Th17 cells, whereas Tregs are anergic and therefore do not produce these cytokines. To examine the cytokine profile of the Foxp3^hi and Foxp3^int cells, we stimulated PBMCs with RGPE for 7 d, restimulated with PMA/ionomycin for 5 or 16 h, and analyzed cytokine production by intracellular cytokine staining. RGPE-responsive Foxp3^hi dividing CD4⁺ T cells produced none of the effector cytokines IL-4, IFN-γ, IL-2, and IL-17, consistent with the anergic phenotype of Tregs (Fig. 2). Foxp3^int cells produced effector cytokines IL-4, IFN-γ, and IL-2, and to a lesser extent IL-17, confirming that they were Teffs. Similar results were obtained after TCR-specific restimulation with anti-CD3/anti-CD28 (data not shown).

IL-10 production was detected in a small percentage of Foxp3^int cells but not in Foxp3^hi cells by cytokine secretion assay postrestimulation with anti-CD3 and anti-CD28 for 16 h (Fig. 2A). IL-10 mRNA expression analyzed by real-time PCR in total PBMCs varied during the 7-d time course, but was not significantly increased by RGPE stimulation at any time point (Fig. 2B, Table I). Small amounts of secreted IL-10 were detected in RGPE-stimulated cultures of PBMCs after 5 and 7 d compared with unstimulated controls (Fig. 2B).

TGF-β mRNA expression and cytokine secretion were analyzed in total cultured PBMCs. Neither TGF-β mRNA expression levels nor TGF-β secretion into the cell culture supernatants was increased in response to RGPE stimulation compared with unstimulated controls (Fig. 2C).

The hallmark of Tregs is their ability to suppress proliferation of CD4⁺ Teffs. Therefore, we tested the capacity of RGPE-induced Foxp3^hi cells to suppress CD4⁺ T cell proliferation and production of IFN-γ. Total dividing CD4⁺ T cells are a heterogeneous mix of not only Foxp3^hi cells but also activated Foxp3^int Teffs. Cytokines released by Teff are likely to enhance activation and proliferation of responder T cells (Tresps) and thereby counteract the regulation by Foxp3^hi cells. Therefore, accurate testing of their suppressor function requires purification of the Foxp3^hi cells. Surface markers associated with a Treg phenotype, CD25 and CD127^lo, were expressed similarly on all dividing CD4⁺ T cells and not strongly associated with Foxp3^int or Foxp3^hi expression. In our hands, CD127 was downregulated on activated T cells. Thus, surface-marker expression of activated T cells was similar to that of Tregs (Fig. 1). As a consequence, none of these markers could be used to isolate the Foxp3^hi cells. Screening of further markers revealed an association of CD39 with cells expressing high levels of Foxp3 (data not shown). The ectonucleotidase activity of CD39 has recently been described as facilitating the suppressive function of Tregs (6, 7). Therefore, we sorted CD4⁺CFSE^loCD25⁺CD39⁺ cells to enrich live dividing CD4⁺Foxp3^hi cells induced by RGPE stimulation to test their suppressor function in five nonatopic donors (Fig. 3).

Postenrichment of CD4⁺ T cells by negative selection, we sorted CD4⁺ T cells by flow cytometry to enrich Foxp3^hi cells in the CD25⁺CD39⁺ fraction and Foxp3^int cells in the CD25⁻CD39⁻ fraction (Fig. 3A, 3B). This resulted in varying frequencies of Foxp3^hi cells in both fractions for each donor (Table II) depending on the starting frequency of induced Foxp3^hi in the dividing CD4⁺ T cells and the strength of association between high Foxp3 and CD39 expression.

Compared to the nondividing CD25⁻CD39⁻ fractions, enriched Foxp3^hi dividing CD4⁺CD25⁺CD39⁺ cells significantly suppressed proliferation of responder CD4⁺ T cells in anti-CD3–stimulated CFSE-labeled PBMCs (Fig. 3D). The only nonresponding subject had the lowest frequency of Foxp3^hi cells in the enriched Foxp3^hi dividing CD4⁺CD25⁺CD39⁺ cells (Table II). Suppression was calculated as the percent reduction of the frequency of CFSE^lo T cells in the total CFSE⁺CD4⁺ T cells compared with coculture with nondividing CD4⁺CD25⁻CD39⁻ T cells. Moreover, secretion of IFN-γ into the supernatant of the coculture was suppressed in the presence of the CD39⁺ fraction compared with the nondividing CD4⁺CD25⁻CD39⁻ T cells in three of four subjects tested. In the one subject in whom no suppression of CD4⁺ T cell proliferation was observed (Fig. 3D, ▴), IFN- γ secretion was also suppressed, but only at the 1:1 ratio of Tregs plus Tresps (data not shown).

We directly examined the suppressive function of dividing CD4⁺Foxp3^hi cells on proliferation and cytokine production by coinduced dividing CD4⁺Foxp3^int Teffs in an allergen-specific setting by testing correlation with Foxp3^hi cell frequency. This does not require Treg purification or enrichment (14). The frequency of dividing CD4⁺ T cells and the proportion of Foxp3^hi cells in the dividing CD4⁺ T cells were inversely related (Fig. 4A). We also studied the relationship between induction of Foxp3^hi cells and IL-4–producing Th2 effector cells. After RGPE stimulation for 7 d, CFSE-labeled PBMCs were restimulated with anti-CD3 and anti-CD28 for 6 h, harvested, and stained for CD4, Foxp3, and IL-4. The proportion of Foxp3^hi cells within the dividing population of CD4⁺ T cells was inversely related to the percentage of IL-4⁺ dividing CD4⁺Foxp3^int cells for all donors (Fig. 4B). No differences were observed for data from allergic donor samples collected during or outside the pollen season. These findings further confirm that Foxp3^hi cells can suppress proliferation and cytokine production of Teffs, if present in sufficient numbers.

Induction of Foxp3 in activated T cells has been shown previously to be only transient. We tested the stability of the Foxp3^hi subset over a period of 28 d. Although a decrease in mean fluorescence intensity of the Foxp3^hi cells was observed over time, a distinct population of Foxp3^hi cells was present over the whole observation period (Fig. 4C). A kinetic analysis of the induction of Foxp3^int and Foxp3^hi cells in relation to proliferation in three allergic donors showed very different profiles (Fig. 4D), but nevertheless as on day 7 (Fig. 4A), the frequency of Foxp3^hi cells was inversely related to proliferation over the whole period.

We investigated whether the Foxp3^hi Tregs were derived from division of pre-existing Tregs or generated by conversion of CD4⁺CD25⁻Foxp3⁻ non-Tregs. Pre-existing Tregs were isolated from PBMCs by negative selection of CD4⁺ T cells and flow cytometric sorting based on CD25⁺ and CD127^lo surface marker expression. They were labeled with CFSE and cocultured with the remaining Treg-depleted PBMCs labeled with PKH-26 with RGPE or anti-CD3 stimulation for 7 d (Fig. 5A).

Dividing CD4⁺Foxp3^hi cells were induced from PKH-26–labeled non-Tregs by allergen stimulation as well as nonspecific TCR stimulation with anti-CD3. The frequency of conversion to Foxp3^hi cells was independent of the presence or absence of pre-existing Tregs (Fig. 5B, 5C). However, alongside conversion of non-Tregs, division of CFSE-labeled Tregs was observed. Expanded Tregs contained a markedly higher proportion of Foxp3^hi cells than dividing PKH-26–labeled non-Tregs. Therefore, an increased frequency of Foxp3^hi cells in the total dividing CD4⁺ cells was observed in the presence of Tregs compared with PBMCs depleted of pre-existing Tregs. After allergen-specific stimulation with RGPE, a trend toward higher contribution of dividing non-Tregs than dividing Tregs to the Foxp3^hi cells was observed in three out of three donors tested (Fig. 5B). In contrast, proliferation was higher poststimulation with anti-CD3 than RGPE, resulting in a trend for higher overall contribution of dividing pre-existing Tregs to the Foxp3^hi cells in two out of two donors tested (Fig. 5C).

There was no significant difference in the frequency of pre-existing Tregs in allergic compared with nonatopic donor PBMCs (Fig. 6A). To investigate whether the ability to generate Foxp3^hi Tregs in response to allergen stimulation is altered in allergic subjects, we compared a larger set of PBMCs from allergic donors to nonatopic controls, all collected outside the pollen season. The frequency of RGPE-induced Foxp3^hi cells within the dividing CD4⁺ T cell population was significantly decreased in allergic subjects compared with nonatopic subjects (Fig. 6B). These results indicate that induction of dividing CD4⁺Foxp3^hi cells may have an important role in counterregulating the immune response to allergen in healthy subjects and that this mechanism is impaired in allergic subjects.

We studied the potential for induction of Tregs in human PBMCs during an allergen-specific immune response. We observed a subset of CD4⁺ T cells that divided in response to stimulation with RGPE and expressed high levels of Foxp3. In contrast, Teffs that expanded in parallel expressed low or intermediate levels of Foxp3. The Foxp3^hi cells exhibited phenotypic and functional properties of Tregs, and their induction in response to allergen stimulation was reduced in allergic compared with healthy individuals. Induction of Foxp3 expression in response to T cell activation has been described, but it is not clear if this denotes acquired regulatory function. Unlike in our study, two subsets of Foxp3^int and Foxp3^hi cells were not identified in previous studies (12, 15, 16). The two Foxp3 subpopulations may not have been apparent when analyzing undistinguished dividing and nondividing human CD4⁺ T cells after polyclonal activation because the Foxp3^hi cell subset in bulk CD4⁺ T cells is obscured by nondividing Foxp3⁺ T cells and cannot be detected as a distinct population. We distinguished Foxp3^hi and Foxp3^int subsets by separately analyzing the dividing cells identified by dilution of CFSE or PKH-26 dye.

We tested the dividing Foxp3^hi and Foxp3^int cells for their ability to produce the Teff cytokines IL-4, IFN-γ, IL-2, and IL-17 and found that these were secreted by Foxp3^int but not Foxp3^hi cells. The Foxp3^hi cells did not produce IL-10 or TGF-β in response to RGPE stimulation and therefore do not resemble Tr1 or Th3 cells. Low levels of IL-10 were secreted poststimulation of PBMCs with RGPE for 5 and 7 d. Given the low frequency of IL-10⁺ CD4⁺ T cells, the IL-10 is likely to be derived from another cell subset. Assessment of suppressor function of the Foxp3^hi cells is hampered by the lack of an exclusive Treg marker that distinguishes Treg from activated cells; activated T cells upregulate the Treg markers CD25 and Foxp3 and downregulate CD127, a marker of resting Tregs. We overcame this problem by enriching dividing CD4⁺CD25⁺Foxp3^hi T cells for the recently described Treg marker CD39 (an ectoenzyme that, together with CD73, hydrolyzes ATP to adenosine and contributes to the mechanism of suppression by Tregs) (6, 7). Dividing Foxp3^hi cells are associated with high CD39 expression allowing enrichment of Foxp3^hi cells by sorting CD25⁺CD39⁺ cells. However, there is still a strong contamination with Foxp3^int cytokine-producing Teffs that are likely to enhance proliferation and thereby interfere with the suppression by Foxp3^hi cells. Nevertheless, in four out of five subjects tested, the enriched Foxp3^hi T cells suppressed proliferation of CD4⁺ T cells. The Foxp3^hi cell content postenrichment was lowest in the subject in whom only suppression of IFN-γ production but not proliferation was observed.

We investigated whether Foxp3^hi dividing CD4⁺ T cells were generated by expansion of Tregs that pre-existed in peripheral blood or by conversion of CD4⁺CD25⁻Foxp3⁻ T cells. It has been demonstrated that natural Tregs can be expanded in vivo (17) and in vitro poststimulation with anti-CD3 and anti-CD28 in the presence of high concentrations of exogenous IL-2 (18). Conversion of CD4⁺CD25⁻Foxp3⁻ to CD25⁺Foxp3⁺ T cells upon activation in the presence of IL-2 and TGF-β has also been demonstrated (19–22). We did not add exogenous IL-2 or TGF-β to our cultures, but low concentrations of TGF-β were present in the human serum used for cell culture (data not shown), and low amounts of endogenous IL-2 and TGF-β were produced by the RGPE-stimulated PBMCs (Fig. 2). However, blocking TGF-β with anti–TGF-β Ab did not significantly alter the frequency of induced Foxp3^hi cells (data not shown).

To study the contribution of expansion and conversion to the induction of dividing Foxp3^hi cells, we tracked Tregs and non-Tregs separately in culture by labeling them with CFSE and PKH-26, respectively. After RGPE stimulation, more Foxp3^hi cells were derived from conversion of non-Tregs than from expansion of pre-existing Tregs; the opposite trend was found for stimulation with anti-CD3. However, our study design did not allow determination of whether the expanding pre-existing Tregs were thymus-derived natural Tregs or non-Tregs converted in the periphery as a result of allergen exposure. Foxp3⁺ cells in the nondividing CD4⁺ population after RGPE stimulation might represent pre-existing natural Tregs that are not specific for RGPE and therefore do not expand.

Previous studies on the regulatory function and stability of activation-induced Foxp3 expression in human CD4⁺CD25⁻ cells are conflicting, demonstrating either an association (15, 16, 23) or no association (12, 24, 25) between induced Foxp3 expression and suppressor function. This may be due to the different experimental methods, such as polyclonal stimulation of purified T cells, to convert the majority of the CD4⁺CD25⁺Foxp3⁻ T cells. Also, stimulated cells were used in the suppression assays without further purification to remove activated T cells that could counteract the suppressor effect of the converted Foxp3⁺ T cells. In the majority of reports, the induction of Foxp3 has been studied in response to anti-CD3 and anti-CD28 stimulation. Such strong nonspecific stimulation results in activation involving the majority of the T cells, creating an artificially enhanced response that is then harder to suppress. These conditions could also result in proliferation of Tregs, which has been shown to be associated with temporary abrogation of suppressive function. Nevertheless, using allergen-specific activation of total PBMCs, we found a weak inverse correlation of the frequency of Foxp3^hi cells with proliferation of CD4⁺ T cells and with the frequency of IL-4–producing dividing CD4⁺ T cells.

The lack of suppression by induced Foxp3⁺ T cells in other studies was attributed to the observation that Foxp3 induction and potential suppressive function can be only transient effects of human CD4⁺ T cell activation (15, 16). Consistent, stable, and high Foxp3 expression induced by lentiviral gene transfer into human CD4⁺ T cells was shown to confer suppressor Treg function (26, 27). These findings agree with our demonstration that only a high level of Foxp3 expression is associated with suppressive function. The expression of Foxp3 in Foxp3^hi cells was stable over at least 28 d. Dividing and thus activated Teffs express Foxp3 at a higher level compared with resting T cells but still lower than activated Tregs.

Both allergic and nonallergic individuals have allergen-specific Teffs that can be expanded in vitro by stimulation with allergen as shown previously (28, 29). We have now shown that Tregs are also induced as part of the immune response to allergen, but at a decreased frequency in allergic compared with nonallergic individuals. Thus, allergic individuals may be distinguished from nonallergic by an impaired allergen-induced regulatory mechanism that controls allergen-specific Teffs. The association of high expression of Foxp3 with CD39 expression was only significant in nonatopic subjects, which further emphasizes that allergen-expanded T cells from allergic and nonatopic subjects differ in their expression of molecules associated with regulation of T cell responses (data not shown). Mantel et al. (30) recently showed that a Th2 environment and, more specifically, the presence of IL-4 and expression of GATA3 inhibits Foxp3 upregulation in human CD4⁺ T cells and thereby development of Tregs, which could explain the reduced frequency of dividing Foxp3^hi cells in allergic individuals.

In summary, we demonstrate that functional Tregs are induced in response to stimulation by allergen, representing a self-regulatory mechanism that normally limits T cell activation and allergen-induced inflammation. This mechanism is a target for allergen-specific immunotherapy for allergic subjects (e.g., by coadministration of agents that specifically enhance generation of Foxp3^hi cells).

We thank Nicola Leung for advice on the IL-17 staining protocol and James Dromey for advice on coculture of PKH-26– and CFSE-labeled cells and critical discussion of the manuscript.

Disclosures The authors have no financial conflicts of interest.