IL-33 Induces Antigen-Specific IL-5⁺ T Cells and Promotes Allergic-Induced Airway Inflammation Independent of IL-4¹

It is generally accepted that CD4⁺ T cells that express type 2 cytokines can develop from at least two pathways: the IL-4-dependent and -independent pathways (1, 2, 3, 4, 5, 6, 7, 8, 9). Although the IL-4-dependent classical Th2 cell development pathway has been extensively studied, the mechanism of IL-4-independent Th2 differentiation remains obscure. These cells, originally identified in IL-4 null mice, express IL-5 and/or IL-13 and play an important role in helminthic infection (2, 3, 4, 5, 6), vaccination (7), and asthma (8, 9).

We have previously reported that IL-1R-related molecule (ST2),³ a member of the IL-1 receptor superfamily, is preferentially expressed on Th2 but not Th1 cells (10). Subsequent reports extended this finding and showed that soluble ST2 (sST2), a decoy receptor of ST2 ligand, could suppress type 2 responses and airway inflammation (11, 12). Interestingly, although IL-4 is capable of inducing ST2, it is not required for ST2 expression on CD4⁺ cells (12). In line with the above, ST2^−/− mice developed impaired type 2 functions but had normal Th2 cells, suggesting potential redundancy and heterogeneity of T cell populations involved in type 2 immunity (13, 14, 15). Recently, IL-33 has been identified as the ligand of ST2 (16). IL-33, a member of the IL-1 family, signals via a heterodimeric receptor complex consisting of ST2 and IL-1R accessory protein (16, 17, 18) and triggers the activation of NF-κB and all three MAPKs: p38, ERK1/2, and JNK1/2 in mast cells (16, 19, 20). IL-33 is expressed by a variety of cell types (16). Administration of IL-33 into naive mice induced innate type 2 immune response associated with airway hypersensitivity (16, 21). Moreover, it was found that both human and murine mast cells when stimulated in vitro with IL-33 produced a wide spectrum of cytokines and chemokines (19, 20, 22). In addition, IL-33 enhanced IL-4-driven Th2 cell responses (16, 21, 23, 24) and acted as a selective chemoattractant for Th2 cell recruitment (25). However, the potential role of IL-33 in the differentiation of uncommitted Ag-specific CD4⁺ T cells and contribution of these cells to allergen-induced airway inflammation in an IL-4-free environment is unknown.

We report here that IL-33-polarized murine and human naive CD4⁺ T cells into CD4⁺ T cells that produce IL-5 and IL-13 independent of IL-4. In vivo, ST2^−/− mice produce less IL-5 and display less inflammation in the lungs compared with wild-type (WT) mice in a model of OVA-induced allergic airway inflammation. Conversely, administration of IL-33 during Ag priming markedly increased the percentage of CD4⁺IL-5⁺ and exacerbated OVA-specific airway inflammation in WT and IL-4^−/− mice. Finally, adoptive transfer of IL-33-differentiated IL-5⁺IL-4⁻T cells triggered airway inflammation in naive IL-4^−/− mice. Thus, we demonstrate here that IL-33 induces IL-5-producing T cells and promotes eosinophilic airway inflammation independent of IL-4.

BALB/c, C57BL/6 mice (Harlan Olac), and STAT6^−/− mice (The Jackson Laboratory) were used. ST2^−/− mice on the BALB/c background (26) were backcrossed with D0.11.10 for 10 generations to generate ST2^−/−/D0.11.10 mice. IL-4R^−/− mice on the BALB/c background and IL-4^−/− on the D0.11.10 background were provided by Dr. J. Alexander (University of Strathclyde, Glasgow, U.K.). Mice were kept at the Biological Services facilities of the University of Glasgow in accordance with the U.K. Home Office guidelines. IL-4^−/− mice were also bred and housed at the animal facility of the Instituto de Ciências Biológicas, Universidade Federal de Minas Gerais (Belo Horizonte, Brazil).

IL-33 proteins were produced and purified as previously described (25). Endotoxin was removed by purification with polymyxin B columns. The purity of IL-33 was >97%. Endotoxin levels were < 0.1 endotoxin units/μg protein according to the Limulus amebocyte lysate QCL-1000 pyrogen test (Cambrex).

Cord blood was obtained from informed consenting mothers. Human and murine CD4⁺ T cells were purified by negative selection (AutoMACS; Miltenyi Biotec). T cells (purity ≥98%) were cultured in RPMI 1640 supplemented with 10% FCS, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 0.05 M 2-ME. CD4⁺ T cells (2 × 10⁶ cells/ml) were activated with plate-bound anti-CD3 Abs (2–3 μg/ml; BD Biosciences), IL-4, IL-12 (all from PeproTech), and different doses of IL-33 or a combination of the cytokines. CD4⁺ T cells from D0.11.10 mice were cultured with mitomycin C-treated APC (spleens from ST2KO mice), OVA peptide (10 nM), and cytokines as above. After 36 or 72 h, RNA and supernatants were collected for PCR and ELISA, respectively. To test the stability of IL-33-induced IL-5-producing T cells, cells stimulated with anti-CD3 Abs with or without IL-33 for 6 days were washed and restimulated with anti-CD3 Abs alone. After 72 h, supernatants were collected and ELISA was performed. To evaluate intracellular cytokine expression by IL-33-stimulated naive T cells, purified naive CD4⁺CD62L⁺ T cells (1.2 × 10⁶) were cultured with bone marrow-derived myeloid dendritic cells (4 × 10⁵) (27), OVA peptide (10 nM), and IL-33 (10 ng/ml) in 24-well plates. After 3 days, fresh medium supplemented with IL-2 (10 U/ml) was added to the culture for another 3 days. Next, cells were washed and stimulated again under the same conditions. Cells after the first and second round of stimulation were washed and activated with PMA (500 ng/ml) and ionomycin (50 ng/ml; both from Sigma-Aldrich) in the absence of IL-33 for 4 h followed by intracellular cytokine staining. To test the role of NF-κB and MAPK, CD4⁺ cells from BALB/c mice were stimulated with anti-CD3 Abs (4 μg/ml) and IL-33 (10 ng/ml). After 5 days, cells were restimulated with immobilized anti-CD3 Abs for 24 h. Cells were then washed and rested for 2 h and incubated with IL-33 alone or in the presence of the following inhibitors or their controls: SN50M, SN50 (18 μM) SB203560 (10 μM) PD098059 (60 μM), JNK II (50 μM), or DMSO.

Murine and human cytokines IL-4, IL-5, IL-6, IL-10, IL-13, IFN-γ, and IL-17 were analyzed by ELISA using paired Abs (BD Biosciences). Serum levels of IgE and IgG1 were measured using an OptEIA ELISA kit (BD Biosciences).

Cultured or freshly isolated cells from draining lymph nodes (DLN) or bronchoalveolar lavage (BAL) were stimulated with PMA/ionomycin for 4 h; GolgiStop was added during the final 3 h. The cells were incubated with anti-mouse CD16/32 Abs (BD Biosciences) followed by PerCP-conjugated anti-CD4 (BD Biosciences) and FITC-conjugated anti-ST2L (MD Biosciences) or F4/80 Ab and CCR3 Ab or appropriate isotype controls. Cells were then fixed with Cytofix/Cytoperm buffer (BD Biosciences), permeabilized with Perm/Wash buffer (BD Biosciences), and incubated with FITC-conjugated anti-IFN-γ, PE-conjugated anti-IL-4, allophycocyanin-conjugated anti-IL-5 (all from BD Biosciences), or rat anti-IL-33 Ab or biotinylated goat anti-IL-13 Ab (R&D Systems) or isotype controls followed by incubation with secondary Abs or streptavidin if necessary.

CD4⁺ T cells were stimulated with immobilized anti-CD3 Abs (4 μg/ml) in the presence of IL-33 (10 ng/ml). After 6 days, cells were restimulated with immobilized anti-CD3 Abs for 24 h and rested for 2 h. The cells were then incubated with IL-33 at different time points followed by extraction of cytoplasmic and nuclear proteins using Nuclear Extract kits (Active Motif). Cytoplasmic proteins were used in Western blotting with anti-phospho-p65, phospho-ERK1/2, phospho-p38, phospho-JNK, and p38 Abs (Cell Signaling Technology). Nuclear proteins were used to evaluate NF-κB activity with a TransAM NF-κΒ p62 kit (Active Motif).

This was conducted as described previously (28, 29).

All mice were sensitized i.p. with 100 μg of OVA in 2% alum (aluminum hydroxide gel adjuvant; Brenntag) on day 0, then challenged intranasally (i.n.) on days 8–10 with 10 μg of OVA or PBS (30).

To detect the effect of exogenous IL-33, the dose of OVA was reduced. Mice were immunized i.p. with 10 μg of OVA in 2% alum on day 0. On days 0–2, these mice were injected i.p. with IL-33 (2 μg/mouse) or PBS. Control mice were given PBS or IL-33 alone. Mice were challenged i.n. on days 8–10 with 2 μg of OVA, except the PBS group which received i.n. PBS instead of OVA. All mice were sacrificed on day 11 and serum, BAL, lungs, and DLN were analyzed as described previously (31). The lung sections were stained with H&E or periodic acid-Schiff (PAS) and examined under light microscopy. Peribronchial and perivascular inflammation was scored using a semiquantitative scoring system assessing the degree of eosinophilic inflammation: 0 = no eosinophils; 1 = eosinophils make up <10% of total infiltrate or total infiltrate is <20 cells; 2 = eosinophils make up 10–50% of total infiltrate; and 3 = eosinophils make up >50% of total infiltrate. Immunohistochemical staining was performed on frozen lung sections using anti-mouse IL-33 Ab (Nessy-1; Axxora Life Science) with a biotinylated pan-specific secondary Ab followed by detection with the avidin-biotin complex/diaminobenzidine system (both Vector Laboratories). For the quantitative analysis of mucus area, images covering 81.754 μm² of lung sections were captured and mucus (red-stained areas) was measured with the software Image Pro-Plus (Media Cybernetics). To estimate eosinophil levels in the lungs, the eosinophil peroxidase (EPO) colorimetric assay was performed as described previously (32).

CD4⁺ cells were isolated from IL-4^−/− mice sensitized with OVA or OVA plus IL-33 and challenged with OVA as described above (protocol 2). These T cells (2 × 10⁶) or PBS were injected i.v. into naive IL-4^−/− recipients on day 0. On days 1–3, mice were challenged with OVA (10 μg/mouse). Mice were sacrificed 24 h after the last challenge.

ANOVA followed by Tukey’s test or Student’s t test was applied to in vitro studies. Analysis between individual in vivo groups was examined by ANOVA followed by the Student-Newman-Keuls test or Student’s t test. Experiments were performed at least twice. All data are expressed as mean ± SEM. A value of p < 0.05 was considered to be significant.

Although IL-33 induces IL-5 and IL-13 production by IL-4-differentiated Th2 cells (16, 21), its role in activation of uncommitted naive T cells is unknown. To address this issue, naive CD4⁺ T cells from WT or ST2^−/− mice were cultured with anti-CD3 Ab with or without IL-33 for 72 h. WT but not ST2^−/− cells produced markedly higher concentrations of IL-5 and IL-13 when stimulated with IL-33 compared with cells cultured with anti-CD3 Ab alone (Fig. 1,A). Unexpectedly, IL-33 failed to induce any IL-4 synthesis, the signature cytokine for classical Th2 cells. IL-33 was also unable to induce IFN-γ, IL-17, or IL-10 production (Fig. 1,A and data not shown). We further confirmed this observation by using CD4⁺CD62L⁺ naive T cells from WT or ST2^−/− OVATcR-transgenic mice. In the presence of APC and OVA peptide, IL-33 enhanced the synthesis of IL-5 and IL-13 (but not IL-4 or IFN-γ) by WT but not ST2^−/−OVATcR-transgenic T cells (Fig. 1,B). Furthermore, human cord blood CD4⁺ T cells, which are mainly comprised of naive cells, also produced enhanced levels of IL-5 and IL-13 but not IFN-γ or IL-4 when cultured with anti-CD3 and IL-33 compared with anti-CD3 alone (Fig. 1,C). Importantly, CD4⁺T cells stimulated with anti-CD3 and IL-33 in the first round of stimulation still produced an increased amount of IL-5 and IL-13 but not IL-4 and IFN-γ after restimulation with anti-CD3 alone (Fig. 1,D). Consistent with the cytokine ELISA data, intracellular staining shows that IL-33-polarized CD4⁺ T cell population selectively expressed IL-5 and IL-13 but not IL-4 and IFN-γ after the first and second round of stimulation (Fig. 1, E and F, and data not shown). IL-33 also enhanced ST2 protein and mRNA expression in CD4⁺ T cells (Fig. 1, G–I), suggesting a self-amplification circuit in the induction of IL-5-producing cells. Thus, IL-33, in the presence of Ag, preferentially polarizes a population of IL-5⁺IL-13⁺IL-4⁻ T cells.

IL-4, IL-4R, and STAT6 are essential for classical Th2 cell development (33, 34, 35). To further characterize the IL-33-polarized IL-5⁺T cells, we tested whether IL-33 is able to activate CD4⁺ T cells in the absence of IL-4, IL-4Rα, or STAT6. As shown in Fig. 2,A, IL-33 increased IL-5 and IL-13 production in both WT and IL-4^−/− cells, although the level of IL-13 remained slightly lower in IL-4^−/− compared with WT. Similar results were observed with CD4⁺ T cells from IL-4Rα^−/− mice (data not shown) and STAT6^−/− mice (Fig. 2,B). Invariant NKT (iNKT) cells can also express CD4 and have been shown to be involved in type 2 immunity (36). To exclude a role for iNKT cells in this context, we polarized CD4⁺ T cells from iNKT-deficient Jα18^−/− mice (37). CD4⁺ T cells from WT and Jα18^−/− mice produced comparable levels of IL-5 and IL-13 when activated with IL-33 and anti-CD3 (Fig. 2 C), indicating that NKT cells did not contribute to the IL-33-induced IL-5-producing T cells. IL-33 did not significantly affect the production of IFN-γ or IL-4 in IL-4^−/−, IL-4R^−/−, STAT6^−/−, or Jα18^−/− mice (data not shown). Taken together, these results demonstrate that IL-33 triggers IL-5 and IL-13 production from naive CD4⁺ T cells independent of IL-4, IL-4Rα, and STAT6 signaling.

We also investigated the involvement of GATA3 and T-bet in IL-33-induced T cell polarization. IL-4 and IL-12 induced the transcription of GATA3 and T-bet, respectively (38, 39). In contrast, IL-33 failed to induce GATA3 or T-bet even at high concentrations (Fig. 2, D and E). These results indicate that IL-33 polarizes the IL-5-producing T cell via a distinct signaling pathway from Th1 or classical Th2 cells.

As a member of the IL-1 family, IL-33 may signal via the MyD88-dependent and -independent pathways (21, 40, 41). We therefore investigated the role of MyD88 and Toll/IL-1R domain-containing adaptor-inducing IFN-β (TRIF) in IL-33-mediated T cell polarization. As shown in Fig. 3,A, IL-33 failed to induce IL-5 and IL-13 secretion by MyD88^−/− T cells. In contrast, IL-33 stimulated normal levels of IL-5 and IL-13 production by CD4⁺ cells from TRIF^−/− mice (Fig. 3,B). Thus, IL-33 polarization of IL-5- producing T cells was MyD88 dependent but TRIF independent. We further explored the involvement of the key signaling components in IL-33 signaling. It has been reported that IL-33 triggers all three MAPK pathways and the NF-κB pathway in mast cells (16, 20). Little, however, is known about the signaling pathways triggered by IL-33 in T cells and their contribution to cytokine production. We found that IL-33 induced the phosphorylation of NF-κBp65 and the MAPK p38, JNK1/2, and ERK1/2 (Fig. 3,C). The NF-κB activity induced by IL-33 was further confirmed by nuclear translocation of NF-κBp65 (Fig. 3,D). IL-33-induced IL-5 synthesis was suppressed by the inhibitors of p38 (SB203560), ERK1/2 (PD098059 and U0126), and JNK1/2 (JNK II) but not by the inhibitor of NF-κB (SN50) (Fig. 3,E and data not shown). In contrast, the production of IL-13 was suppressed by p38, NF-κB, and, to a lesser extent, by ERK1/2 and JNK1/2 inhibitors (Fig. 3 E and data not shown). Thus, IL-33 polarizes IL-5- producing T cells via the ST2, MyD88, MAPK, and NF-κB pathways. Interestingly, IL-33 appears to induce IL-5 and IL-13 production by different signaling mechanisms.

The role of ST2 in the asthma model is controversial (10, 11, 13, 14, 15, 42). We sought to further determine the role of ST2/IL-33 in type 2 responses by using WT and ST2^−/− BALB/c mice in the OVA-specific model of airway inflammation (30). WT and ST2^−/− mice were sensitized once with OVA (100 μg), then challenged i.n. with 10 μg of OVA for 3 subsequent days. ST2^−/− mice showed significantly less severe airway inflammation compared with WT mice. This was evident in a reduced number of total cells, eosinophils, and macrophages in the BAL of the ST2^−/− mice compared with WT mice (Fig. 4,A). The BAL from ST2^−/− mice also contained significantly less IL-5 than that of the WT mice, but IL-13 and IL-4 levels were not significantly different (Fig. 4,B). The serum IgE and IgG levels were similar in the ST2^−/− and WT mice (data not shown). ST2^−/− mice also exhibited reduced lung pathology after Ag challenge, showing significantly less inflammatory cell infiltration in the peribronchial and perivascular areas of the lungs than that of the WT mice (Fig. 4, C and D). The contribution of endogenous IL-33 in OVA-induced airway inflammation was further supported by our immunohistochemical analysis which clearly demonstrated the presence of endogenous IL-33 in the lungs of mice sensitized and challenged with OVA (Fig. 4,E). Since the morphological appearance of IL-33-positive cells suggested that these might be macrophages, we isolated alveolar macrophages and stained them with anti-IL-33-specific Ab. Indeed, as shown in Fig. 4 F, F4/80⁺CCR3⁻ alveolar macrophages abundantly expressed IL-33. Altogether these results therefore demonstrate that IL-33/ST2 signals are required for an optimal allergen induced-airway inflammation and IL-5 but not IL-4 production.

Our result from ST2^−/− mice, as well as previous data showing the beneficial effect of soluble ST2 in airway inflammation (11), suggested that IL-33 participates in Ag-specific airway inflammation. We therefore further investigated whether IL-33 could induce IL-5-producing T cells in vivo and exacerbate OVA-triggered airway inflammation. To be able to define the effect of exogenous IL-33, we reduced the dose of OVA in our model. As shown in Fig. 5,A, mice sensitized (10 μg) and challenged (2 μg) with OVA alone exhibited a low inflammatory response in the BAL. In contrast, mice sensitized with OVA plus IL-33 displayed markedly elevated total cell numbers, eosinophils, and macrophages in the BAL (Fig. 5,A). The IL-33-enhanced inflammation was not evident in ST2^−/− mice. Importantly, OVA-nonsensitized WT mice injected with IL-33 and challenged with OVA showed no increase in the total cell count and eosinophilia compared with the PBS group. We then analyzed the ST2 expression and cytokine profile on DLN T cells. IL-33 administration without Ag sensitization slightly increased the percentage of CD4⁺ ST2⁺ cells compared with the PBS group (1.6 and 0.9%, respectively; Fig. 5,B). Cells from WT mice sensitized and challenged with OVA showed a substantial increase in the percentage of ST2⁺CD4⁺ cells (9.4%). This was further increased by the IL-33 treatment at sensitization (16.7%). Intracellular staining of DLN CD4⁺ T cells revealed that IL-33 alone had no effect on the type 2 cytokine profile, but increases the percentage of IFN-γ⁺ cell population compared with the PBS group. In contrast, CD4⁺ cells from OVA-sensitized and challenged mice showed an increase in percentage of IL-4⁺ cells but not IL-5⁺ cells compared with either IL-33 alone or the PBS-treated groups. Importantly, the administration of IL-33 during the OVA sensitization markedly enhanced the percentage of IL-5⁺ T cells but decreased the percentage of IL-4⁺ T cells and did not change the percentage of IFN-γ⁺ cells compared with the OVA group (Fig. 5,C). The injection of IL-33 in the OVA- sensitizing phase also triggered IL-13 production in WT mice (66.6 ± 4.4 pg/ml), which was undetectable in other groups. IL-33 did not affect the cytokine production by CD4⁺ T cells in ST2^−/− mice (data not shown). Thus, IL-33 markedly and specifically increased the number of CD4⁺IL-5⁺ T cells in vivo during Ag priming via ST2. Lung histology analysis showed that IL-33 given during OVA sensitization significantly increased inflammatory cell infiltration in the peribronchial and perivascular areas of the lungs compared with that from mice injected with OVA alone (Fig. 5 D). Mice given PBS or IL-33 without Ag did not have any detectable histological changes (data not shown). Together, these results suggest that IL-33 is a pathogenic factor for Ag-specific airway inflammation accompanied by the polarization of CD4⁺IL-5⁺ T cells in vivo.

To further investigate the relationship between IL-33 and IL-4 in the polarization of IL-5-producing T cells and allergic response, we determined whether IL-33 can induce allergic airway inflammation in IL-4^−/− mice. IL-4^−/− and WT mice were sensitized with OVA or OVA plus IL-33 and challenged with OVA as before. IL-4^−/− mice developed less severe allergic inflammation as demonstrated by the number of total cells, eosinophils, and macrophages in the BAL compared with those in WT mice when sensitized and challenged with OVA alone. This cellular deficit was completely reversed by the presence of IL-33 at sensitization (Fig. 6,A). The airway inflammation in OVA plus IL-33-treated WT and IL-4^−/− mice was accompanied by elevated IL-5 and IL-13 synthesis in DLN cells when restimulated with OVA peptide in vitro (Fig. 6,B and data not shown). Intracellular cytokine staining confirmed that DLN cells from IL-4^−/− mice sensitized and challenged with OVA alone produced limited amounts of IL-5. Treatment with IL-33 during OVA sensitization significantly increased the percentage of IL-5⁺ cells (Fig. 6,C). Gating on the CD4⁺ cell population revealed the presence of an increased percentage of CD4⁺IL-5⁺ T cells in IL-4^−/− mice treated with OVA and IL-33 compared with mice treated with OVA alone (Fig. 6,D). Furthermore, IL-33 treatment increased cellular infiltration in the peribronchial and perivascular areas of the lungs of the IL-4^−/− mice to a level comparable to that in the WT mice (Fig. 6,E). The administration of IL-33 during sensitization also enhanced goblet cell hyperplasia and mucus production in the airways of WT and IL-4^−/− mice (Fig. 6, F and G). Taken together, these results demonstrate that in the presence of allergen, IL-33 can induce a population of IL-5-producing T cells and airway pathology in vivo in the absence of IL-4.

To confirm that IL-5-producing T cells can mediate the allergic airway inflammation in IL-4^−/− mice, CD4⁺ cells from IL-4^−/− mice sensitized and challenged with OVA, with or without IL-33 at sensitization, were purified and adoptively transferred into naive IL-4^−/− recipients. Control mice were injected with PBS. All recipients were challenged with OVA peptide i.n. the day after cell transfer for 3 subsequent days. Adoptive transfer of T cells from OVA or OVA plus IL-33-treated mice increased the total number of inflammatory cells in the BAL of the recipients (Fig. 7,A). However, the number of cells in the BAL was significantly higher in mice receiving CD4⁺ cells from OVA plus IL-33-treated mice compared with those with T cells from mice given OVA alone. Differential cell counts revealed that whereas transfer of T cells from OVA-treated donors induced a modest increase in eosinophil and macrophage influx into the BAL, a significantly higher number of eosinophils and, to a lesser extent, macrophages were found in recipients receiving T cells from OVA plus IL-33-treated donors. Histological examination showed that injection of T cells from OVA plus IL-33-treated mice induced significantly increased cellular infiltration in the peribronchial and perivascular areas of the lungs of the recipients compared with mice that received PBS or T cells from OVA-treated mice (Fig. 7,B). Assay for EPO confirmed the presence of eosinophils in the lung tissue (Fig. 7,C). In addition, mice receiving CD4⁺ T cells from OVA plus IL-33-treated mice developed an increased mucus production compared with other groups (Fig. 7 D). Thus, IL-33-polarized IL-5⁺IL-4⁻ CD4⁺ T cells are able to induce allergic airway inflammation in the complete absence of IL-4.

Data presented here demonstrate that the newly discovered cytokine IL-33 is a potent driver of a population of CD4⁺ T cells producing mainly IL-5 and IL-13 but not IL-4. The IL-33-induced IL-5-producing T cells differ from the classical Th2 cells in several aspects. Th2 cells are driven by IL-4 via IL-4R and produce principally IL-4 (43). In contrast, the IL-5-producing T cells reported here are polarized by IL-33 via ST2 and secrete mainly IL-5 and IL-13 but not IL-4. Moreover, differentiation of Th2 cells requires signaling via STAT6 (34) and GATA3 (38, 39), whereas IL-33 failed to induce GATA3 or T-bet and induced IL-5⁺ T cell differentiation through the classical Toll/IL-1R pathway via MyD88, MAPKs, and NF-κB, independent of IL-4 and STAT6. It should be noted that naive CD4⁺ cells can be polarized into IL-5-producing T cells only in the presence of TCR activation. This is consistent with a recent report that, in contrast to IL-18, IL-33 does not stimulate CD4⁺ cells in the absence of TCR engagement (21).

The role of ST2 in the asthma model has been controversial (10, 11, 13, 14, 15, 42). Data presented here clearly demonstrate that IL-33 is important for Ag-specific asthma. Consistent with the mRNA data reported previously (44), we show here that IL-33 protein is abundantly expressed in the lungs of mice with OVA-induced airway inflammation and alveolar macrophages are one of the cell types that express IL-33. Importantly, we also found that ST2 deficiency resulted in a significantly attenuated OVA-induced IL-5 but not IL-4 and IL-13 production, eosinophilia in the BAL and lung inflammation. Our results are in agreement with a previous report which demonstrated that treatment with soluble ST2 ameliorated OVA-induced airway inflammation (11) and with the finding that ST2 is not required for IL-4 production (14). However, our results are in contrast to the reports that ST2^−/− mice developed unchanged or enhanced eosinophilia in the airway inflammation (14, 42). This discrepancy can be explained by the difference in the mouse strain used and experimental protocols. Our data presented here are based on ST2^−/− of the BALB/c background and a short model of asthma, using only one sensitization and three challenges over a period of 12 days (30) In agreement with the report of Hoshino et al. (14), we failed to detect any significant difference between WT and ST2^−/− mice by using the long protocol with two sensitizations and four challenges over a period of 28 days. All together, it appears that the preferential contribution of IL-33/ST2 signaling to allergic airway inflammation may depend on the time and dose of Ag, with IL-33/ST2 more involved in acute response than chronic immune response. Consistent with this notion, we demonstrate here that administration of IL-33 along with a low dose of Ag (OVA, 10 μg) selectively promoted allergen-specific IL-5-producing T cells but not the classical IL-4-producing Th2 cells and exacerbated airway inflammation. However, in the presence of the standard dose of OVA (100 μg), IL-33 enhanced the induction of both classical Th2 and IL-5⁺IL-4⁻ T cells (data not shown). This finding suggests that the preferential induction of the IL-5-producing T cells over the classical Th2 cells might be largely determined by the dose of allergen and the availability of IL-33 during sensitization. Importantly, we demonstrated that IL-33 could drive the development of Ag-specific IL-5-producing T cells in vivo and exacerbate OVA-induced airway inflammation in IL-4^−/− mice to a similar extent as in WT mice. Furthermore, adoptive transfer of OVA plus IL-33-polarized T cells from IL-4^−/− asthmatic mice into naive IL-4^−/− mice induced airway inflammation. These data clearly demonstrate that IL-33 is capable of driving a population of IL-5-producing T cells independent of IL-4 and may play an important role in allergic airway inflammation.

Several cytokines, including IL-25, IL-1, and IL-18, can induce IL-5 and IL-13 in conjunction with IL-4 (45, 46, 47, 48, 49, 50). To our knowledge, IL-33 is the only cytokine described to date which polarizes IL-4-independent IL-5⁺ T cells. These results also suggest that type 2 responses may result from at least two pathways: the IL-4-dependent and the IL-33-dependent pathways. Both routes might be required for an optimal type 2 response, since these responses are reduced in IL-4^−/− and ST2^−/− mice (2, 13). Our finding may also explain the long-held conundrum that ST2 played an important role in type 2 responses, but was not required for the Th2 cell development in ST2^−/− mice (10, 11, 13, 14, 15).

Interestingly, an increase in the percentage of IFN-γ⁺CD4⁺ cells in mice given IL-33 without Ag indicates that IL-33 may amplify existing type 1 responses in these mice. This is in agreement with recent findings that IL-33 can affect both type 2 and type 1 immune responses depending on local environment (51).

Schmitz et al. (16) observed a systemic innate type 2 response in conjunction with lung eosinophilia in naive mice when 2 μg of IL-33 was administrated i.p. daily for 7 days (16). We evaluated a contribution of IL-33-induced innate immunity to our experimental model by administration of IL-33 but not Ag at the sensitization phase followed by Ag challenge. We found that without Ag at sensitization, a low dose of IL-33 (2 μg) and a short time of administration (3 days) failed to induce inflammation in the lungs. It thus appears that in our experimental model, IL-33- driven exacerbation of OVA-induced airway inflammation is mainly attributed to the direct effect of IL-33 on adaptive rather than innate immunity such as mast cells, NK cells, or basophils. Along this line, we have found that IL-33 did not affect the total cell number or eosinophilia in the BAL of mast cell-deficient (Kit^Wsh/Kit^Wsh) mice treated with IL-33 (N. Pitman, G. Murphy, D. Xu, and F. Y. Liew, unpublished data), which was recently confirmed by Kondo et al. (21).

Our earlier report (26) shows that ST2 down-regulated the LPS-induced inflammatory response. Thus, ST2^−/− mice produced more IL-1, TNF-α, and IL-12 in response to LPS stimulation and were more susceptible to LPS-induced shock compared with WT mice. In the present study, the levels of IL-1, TNF-α, and IL-12 production could not be detected in WT and ST2^−/− mice in the experimental asthma setting. This result suggests that even if the OVA used was contaminated with trace amounts of LPS (<0.18 endotoxin units/μg by Limulus test), the effect observed was unlikely to be due to LPS-induced shock. Therefore, ST2 may play different roles in allergic response and LPS tolerance.

Although the molecular mechanism by which IL-33 preferentially induces IL-5/IL-13 is currently unknown, IL-5 and IL-13 are likely to be responsible for IL-33-induced exacerbation of OVA-specific airway inflammation in WT and IL-4^−/− mice. IL-5 is critical for eosinophil differentiation, recruitment from bone marrow, and migration to the lungs (52). IL-13 stimulates mononuclear cell influx to the lungs by triggering a variety of chemokines from resident lung cells (53, 54). It is also well documented that IL-13 is indispensable for epithelial cell hyperplasia, mucus production, and IgE synthesis (16, 21, 53, 54). Although IL-33/ST2 may not be able to drive the classical Th2 cell differentiation directly, it may still be able to increase IL-4 synthesis by Th2 cells indirectly (16, 23, 24, 55), possibly via the production of IL-13, vascular endothelial growth factor, and GM-CSF which have been reported to enhance Th2 cell polarization (55, 56, 57). We have detected an increase in vascular endothelial growth factor and GM-CSF production by IL-33-stimulated CD4⁺ T cells (data not shown). In addition, IL-33 enhanced the expression of ST2 on both IL-5⁺CD4⁺ and classical Th2 cells and stimulated IL-5 and IL-13 production by mature Th2 cells (16, 21). Therefore, it is likely that IL-33 is capable of activating both IL-4-dependent (indirectly) and IL-4-independent (directly) pathways.

IL-33 is clearly detected in clinical disease (58). Moreover IL-33 mRNA (44) and protein are present in the lungs of mice with OVA-induced airway inflammation. We propose here that under pathological conditions such as the presence of allergens, locally expressed IL-33 may polarize IL-5-producing T cells even in the absence of IL-4. In addition, our study provides a possible mechanism explaining the existence and activation of IL-5-producing T cells in IL-4-deficient mice in different disease models (3, 6, 7, 8, 9). Importantly, there is an increasing interest in different pathological phenotypes of asthma and how they relate to clinical syndromes and respond to treatment (59). Therefore, further clinical studies into the relative contribution of classical Th2 cells and IL-33-triggered IL-5-producing T cells to airway inflammation and changes in airway physiology may provide important insights into the etiology of different phenotypes of asthma, which may provide a novel therapeutic target.

We thank Roderick Ferrier and Dr. Ashley Miller for experimental assistance. The cells from Jα18^−/− mice were provided by Prof. Vincenzo Cerundolo (Weatherall Institute of Molecular Medicine, University of Oxford, U.K.). Cells from MyD88^−/− mice were provided by Prof. Richard K. Grencis (University of Manchester, Manchester, U.K.). Cells from TRIF^−/− mice were provided by Dr Clare Bryant (University of Cambridge, Cambridge, U.K.). IL-4^−/− and IL-4R^−/− mice were generously provided by Prof. James Alexander (University of Strathclyde).

The authors have no financial conflict of interest.