Eosinophils are major effector cells in type 2 inflammatory responses and become activated in response to IL-4 and IL-33, yet the molecular mechanisms and cooperative interaction between these cytokines remain unclear. Our objective was to investigate the molecular mechanism and cooperation of IL-4 and IL-33 in eosinophil activation. Eosinophils derived from bone marrow or isolated from Il5-transgenic mice were activated in the presence of IL-4 or IL-33 for 1 or 4 h, and the transcriptome was analyzed by RNA sequencing. The candidate genes were validated by quantitative PCR and ELISA. We demonstrated that murine-cultured eosinophils respond to IL-4 and IL-33 by phosphorylation of STAT-6 and NF-κB, respectively. RNA sequence analysis of murine-cultured eosinophils indicated that IL-33 induced 519 genes, whereas IL-4 induced only 28 genes, including 19 IL-33–regulated genes. Interestingly, IL-33 induced eosinophil activation via two distinct mechanisms, IL-4 independent and IL-4 secretion/autostimulation dependent. Anti–IL-4 or anti–IL-4Rα Ab-treated cultured and mature eosinophils, as well as Il4- or Stat6-deficient cultured eosinophils, had attenuated protein secretion of a subset of IL-33–induced genes, including Retnla and Ccl17. Additionally, IL-33 induced the rapid release of preformed IL-4 protein from eosinophils by a NF-κB–dependent mechanism. However, the induction of most IL-33–regulated transcripts (e.g., Il6 and Il13) was IL-4 independent and blocked by NF-κB inhibition. In conclusion, we have identified a novel activation pathway in murine eosinophils that is induced by IL-33 and differentially dependent upon an IL-4 auto-amplification loop.
Interleukin-33 is a member of the IL-1 family, which includes IL-1 and IL-18, and binds to its heterodimeric receptor, IL-33R, consisting of the ST2 molecule and IL-1R accessory protein. Binding of IL-33 to its receptor typically triggers NF-κB phosphorylation and translocation to the nucleus to activate transcription of target genes. However, a recent study in fibroblasts overexpressing ST2 has demonstrated that IL-33 can signal through phospho-ERK, independently of NF-κB, indicating two distinct pathways of activation (1). Whether this pathway is operational in other cells remains unclear. IL-33 potently polarizes Th2 cells (2) and activates mast cells (3), basophils (4), and alternatively activated M2 macrophages (5). Recently, a pathogenic role of IL-33 in inflammatory airway disease has been reported [for review (6)]. Notably, asthmatic patients exhibit high levels of IL-33 in bronchoalveolar lavage fluid and in bronchial epithelial and airway smooth muscle cells (7, 8). Interestingly, intranasal administration of IL-33 in naive mice induces inflammatory airway responses (5) and exacerbates eosinophil-mediated airway inflammation (9), suggesting a key role of IL-33 in the development of allergic disease pathogenesis. Although eosinophils have been shown to respond to IL-33 (9), little is known about the molecular mechanism involved.
During immune responses, eosinophils are recruited from the circulation to inflammatory sites, rich in cytokines (e.g., IL-4, IL-13, and IL-33) and chemokines (e.g., CCL11); become activated; and modulate the response through diverse mechanisms. Upon activation, eosinophils undergo degranulation by releasing cationic proteins, which consequently induce cell damage and dysfunction (10). In vitro studies have shown that eosinophil granule constituents are toxic to a variety of tissues such as the intestine, skin, and trachea (10). In addition, activated eosinophils secrete an array of cytokines (e.g., IL-6, IL-10, and IL-13) and chemokines (e.g., CCL3, CCL5, and CCL17) capable of activating T cells (9, 10). Moreover, eosinophils have a profibrogenic role by producing TGF-αβ, resistin-like molecule α (RELM-α), metalloproteinase 9, and fibroblast growth factor 2 (11–13). Notably, eosinophils have the capacity to auto-activate themselves as they express and respond to GM-CSF, which promotes their survival (14, 15). It has not yet been determined whether similar auto-activation loops exist for other eosinophil-derived cytokines, especially IL-4, as eosinophils are often a chief source of IL-4 (16–18).
In allergic airway inflammation, the overexpression of IL-4 and IL-13 has been demonstrated to have a role in the development of pulmonary eosinophilia (19, 20). IL-4 mediates its effects through either the type I IL-4R (composed of IL-4Rα and the common γ-chain) or the type II IL-4R (composed of IL-4Rα and IL-13Rα1), which can also mediate IL-13 signaling (21–23). Engagement of both receptors induces the phosphorylation of STAT-6, which subsequently dimerizes and translocates to the nucleus to induce transcription of specific genes (22, 23). Despite the similar intracellular cascades of IL-4 and IL-13, IL-4 can mediate specific signals independently of IL-13 (24). In asthma, IL-13Rα1 has been highlighted as having a critical role in the pathogenesis of allergic lung responses, regulating different subsets of genes according to the stimulus (IL-4, IL-13, or allergen) (25).
In this study, we demonstrated that murine eosinophils directly respond to IL-4 and IL-33, but not IL-13, by the phosphorylation of STAT-6 and NF-κB, respectively. Using transcriptome analysis, we identified the genes associated with the IL-33 and IL-4 pathways. We found that IL-33 upregulated 519 genes, whereas IL-4 only induced 28 genes, including 19 genes that were also IL-33 induced. Interestingly, IL-33 induced eosinophil activation via two distinct mechanisms depending upon IL-4 secretion and autostimulation. Anti–IL-4 or anti–IL-4Rα Ab-treated cultured and mature eosinophils, as well as Il4- or Stat6-deficient cultured eosinophils, had attenuated protein secretion of a subset of IL-33–induced genes, including Retnla and Ccl17. However, the induction of most IL-33–regulated transcripts (e.g., Il6 and Il13) was independent of IL-4 and blocked by an inhibitor of NF-κB. Our study provides evidence for a novel eosinophil activation pathway that is triggered by IL-33 and involves an IL-4/STAT-6 auto-amplification loop or an IL-4–independent/NF-κB–dependent pathway.
Materials and Methods
BALB/c wild-type, Il13ra1−/− (IL-13Rα1), Il4−/− (IL-4), Il5 (IL-5) CD2-transgenic, or Stat6−/− (STAT-6) mice were analyzed at 4–8 wk of age. All mice were housed under specific pathogen-free conditions and treated according to institutional guidelines.
Bone marrow–derived eosinophil culture
Bone marrow–derived eosinophil culture was modified from Dyer et al. (26). Briefly, total bone marrow cells were isolated, and erythrocytes were lysed by RBC lysis buffer (Sigma-Aldrich). After a density gradient of Histopaque 1083 (Sigma-Aldrich), the low-density bone marrow cells were collected and plated at 1 × 106 cells/ml in IMDM (Life Technologies) supplemented with 10% FBS (HyClone), penicillin-streptomycin (Life Technologies), 200 mM l-glutamine (Life Technologies), and 55 μM 2-ME (Sigma-Aldrich). During the first 4 d, the medium also contained stem cell factor (PeproTech) and Fms-like tyrosine kinase 3 ligand (PeproTech) at 100 ng/ml each. From day 4 to day 14, the cells were cultured in medium containing 10 ng/ml IL-5 (PeproTech). The medium was changed every 2 d until day 14.
For eosinophil activation, cells were collected, pooled, and plated for at least 1 h in a tissue culture dish, to remove any contaminating cells, such as stromal cells or macrophages. Then the nonadherent cells were collected, washed, counted, and incubated with different treatment, according to the experiments. Murine rIL-4 and rIL-13 were purchased from PeproTech, and IL-33 was purchased from R&D Systems. The NF-κB inhibitor BAY 11-7082 was purchased from Santa Cruz Biotechnology and was administered at 5 μM for all of the experiments in which it was used.
Eosinophil purification from Il5-transgenic mice
Briefly, spleen from Il5-transgenic mice was collected and crushed onto a 40-μm cell strainer (BD Falcon). After RBC lysis, the T and B cells were depleted by using microbeads anti-CD90.2 and anti-CD19 (MACS; Miltenyi Biotec). The cell suspension was incubated for 1 h in a tissue culture dish to deplete the adherent cells (macrophages). Then the cells were collected, plated at a density of 2.5 × 106 cells/ml, and activated for 24 h in complete IMDM (see details in bone marrow–derived eosinophil culture section) supplemented with 10 ng/ml IL-5.
Extraction of mRNA and quantitative RT-PCR analysis
Total RNA was isolated with the RNeasy mini kit (Qiagen), according to the manufacturer’s protocol. cDNA was synthetized from 1 or 0.5 μg RNA using the iScript synthesis kit (Bio-Rad). Quantitative RT-PCR (qRT-PCR) was performed using a 7900HT Fast Real-Time PCR system (Applied Biosystems) with FastStart Universal SYBR Green Master mix (Roche). Primers will be provided upon request.
RNA sequencing analysis
FASTQ files from Illumina Pipeline were aligned by TopHat (version 1.4.1) (27), with -T and -G parameters and iGenome annotation from Illumina (08/30/11) for the mm9 genome. The -T parameter is used to align reads to the mouse transcriptome, and the -G parameter is used to provide transcriptome annotation. Produced .bam files were fed to cuffdiff (version 1.3.0) (27) annotated with the same GTF file. Only genes for which at least one isoform was significantly changed and had fragment per kb per million mapped reads >2 in at least one condition were analyzed. Experimental conditions were compared with appropriate controls, and a total set of 1593 genes was generated. To generate heat maps, genes were clustered using Gene Cluster 3.0 (28) with the following parameters: clustering average linkage and correlation (centered) similarity. For visualizing of clustered data, Java Tree View (29) software was applied: http://jtreeview.sourceforge.net/. The database is available following this link: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=nfelrwciwwesida&acc=GSE43660
Eosinophil markers were assessed by FACS staining for CCR3 and Siglec-F (BD Biosciences) at day 14 of culture. For the detection of phospho-p38 or phospho–STAT-6, the cells were activated with IL-33 (100 ng/ml) for 5 min and IL-4 or IL-13 (100 ng/ml) for 30 min, washed with PBS, then processed for fixation (1% formaldehyde) and permeabilization (90% methanol), and stained with 1 μg/ml anti-mouse phospho-p38 (Cell Signaling Technology) or phospho–STAT-6 Ab (BD Biosciences) per 1 × 106 cells. For the intracellular staining, bone marrow–derived cultured eosinophils, activated overnight by IL-4 or IL-33, were incubated for 4 h with 5 μg/ml brefeldin A (Sigma-Aldrich) to block the protein secretion. Cells were washed with PBS and then processed for the fixation, permeabilization, and staining with anti-mouse Siglec-F PE (BD Biosciences), anti-mouse RELM-α (PeproTech), anti-ST2 (R&D Systems), and anti–IL-4Rα (BD Biosciences) Abs and secondary Ab: goat anti-rabbit Alexa Fluor 488 (Life Technologies, Invitrogen). All Abs were used at 1 μg/ml per 106 cells. Flow cytometry was performed on FACSCalibur, and data were analyzed with FlowJo software (Tree Star).
For cell culture, samples were incubated with 10 μg/ml anti-mouse IL-4 (eBioscience) or anti-mouse IL-4Rα (BD Biosciences) Ab for 4 or 24 h, depending on the experiment. Control IgG1 and IgG2a Abs were purchased from BD Biosciences.
Proteins extracted with radioimmunoprecipitation assay buffer were loaded in equal amounts on 4–12% polyacrylamide gradient gels (Invitrogen) and subjected to Western blot with anti-phospho NF-κB p65 and anti–NF-κB p65 Abs (Cell Signaling Technology).
Murine IL-4 and IL-6 levels were analyzed by ELISA using a kit from BioLegend. Murine IL-13 and CCL17 were measured by ELISA using the DuoSet ELISA Development Systems from R&D Systems. For detection of murine RELM-α, purified anti–RELM-α and biotinylated anti–RELM-α (PeproTech) Abs were used according to a protocol provided by the manufacturer. The lower detection limits for IL-4 and IL-13 were 1.95 and 62.5 pg/ml, respectively, and for IL-6, CCL17, and RELM-α were 15.6 pg/ml.
Statistics were done by using the Student t test or one-way ANOVA with Bonferroni’s correction. The p values <0.05 were considered as significant.
Gene expression profiles in murine eosinophils activated by IL-4 or IL-33
We first generated murine bone marrow–derived eosinophils, primarily as per Dyer et al. (26). After 14 d of culture, we confirmed that the cells were mature eosinophils, characterized by multilobed nuclei (Fig. 1A) and the expression of Siglec-F and CCR3 (Fig. 1B) and the expression of eosinophil granule proteins by staining for eosinophil peroxidase activity (data not shown). To minimize the possibility that noneosinophils could be contaminating cells and responsible for the signaling induction by IL-4 or IL-33, adherent cells were removed prior to cellular activation, and eosinophils represented ≥99% of the cells. IL-33 signaling triggered the phosphorylation of p65, a subunit of NF-κB, within 5 min; total NF-κB (p65) did not change over time, demonstrating that eosinophils respond to IL-33 (Fig. 1C). We also confirmed that IL-33 induced the phosphorylation of p38 within 5 min of stimulation (Fig. 1D). Similarly, IL-4 induced STAT-6 phosphorylation in eosinophils within 30 min, whereas IL-13 had no effect on STAT-6 phosphorylation (Fig. 1E).
We generated genome-wide transcriptome data based on RNA sequence analysis from eosinophils activated with IL-4 or IL-33. We analyzed two time points, 1 and 4 h, to determine early and late gene induction. We found that multiple genes were differentially expressed after 1 and 4 h of IL-4 or IL-33 exposure (Fig. 2A). Each treatment induced a specific pattern of gene expression, with certain genes being regulated (≥2-fold change) by IL-33, but not by IL-4 and vice versa (Fig. 2B). IL-33 exhibited a more profound effect compared with IL-4, inducing 519 genes. IL-4 induced 28 genes, with only 9 being specifically induced by IL-4. Notably, 19 genes were induced by both IL-33 and IL-4. Functional analysis revealed that the IL-33–induced genes were involved in cytokine production, immune response, and NF-κB signaling, whereas IL-4–induced genes were involved in cellular homeostasis and Jak/STAT signaling (data not shown) (http://toppgene.cchmc.org). Validating genes by qRT-PCR, we confirmed that Cxcl2, Cxcl10, Clec4e, Adora2b, Il6, and Il13 were specifically upregulated by IL-33, but not IL-4, after 1 and 4 h of activation (Fig. 2C). We also confirmed that eosinophils upregulated Retnla mRNA after 1 and 4 h of IL-4 exposure, but only after 4 h of IL-33 exposure, and upregulated Tfec after 1 and 4 h of IL-4 or IL-33 exposure.
IL-33 is a potent activator of eosinophils
Because IL-33 induced Il6 and Il13 expression, we evaluated whether these two cytokines were released in response to 24 h of exposure to different concentrations of IL-33. Eosinophil secretion of IL-6 and IL-13 increased in response to IL-33 in a dose-dependent manner (Fig. 3A). This response was specific to IL-33, as IL-4 had no effect. Similarly, CCL17 release increased in a dose-dependent manner in response to different doses of IL-33, but not IL-4 (Fig. 3B).
Eosinophils also secreted IL-4 in response to IL-33 in a dose-dependent manner (Fig. 3C). Notably, CCL17 and IL-4 release increased following IL-33 exposure, but IL-33 did not regulate Ccl17 or Il4 expression as determined by RNA sequencing. Indeed, Il4 mRNA was not induced by IL-33 at 2, 6, or 24 h of exposure, suggesting that the protein is preformed in the cells and released when the eosinophils were activated (Fig. 3D). By ELISA, we detected IL-4 in a total protein lysate of unstimulated eosinophils compared with the negative control, Il4-deficient eosinophils (data not shown). In contrast, Ccl17 was increased after 6 h of IL-33 treatment and decreased at 24 h (Fig. 3D). For comparison, Il6 was induced at 2 and 6 h and diminished at 24 h. Finally, we observed that 24 h of exposure to different concentrations of IL-4 or IL-33 induced significant, dose-dependent production of RELM-α by eosinophils (Fig. 3E). Additionally, we confirmed that IL-4– and IL-33–treated eosinophils did indeed express RELM-α by flow cytometry analysis (Fig. 3F).
IL-33 induces RELM-α and CCL17 expression through an IL-4 autocrine mechanism
Because eosinophils secreted RELM-α in response to IL-4 or IL-33 and released IL-4 after IL-33 exposure, we hypothesized that an IL-4 autocrine loop is involved in IL-33–induced RELM-α. We first generated Il4−/− and wild-type bone marrow–derived eosinophils. We monitored the cultures and did not observe any differences in terms of proliferation, differentiation, and morphology between the IL-4–deficient and wild-type eosinophils. In addition, IL-4–deficient eosinophils expressed CCR3, Siglec-F, IL-4Rα, and ST2 at similar levels compared with wild-type eosinophils (data not shown). We then activated Il4−/− and wild-type bone marrow–derived eosinophils (obtained at day 14 of culture) for 24 h with IL-4 or IL-33 and compared RELM-α as well as IL-6, IL-13, and CCL17 expression at the mRNA and protein levels. Notably, RELM-α was not induced in Il4−/− eosinophils stimulated with IL-33 compared with wild-type eosinophils; however, Il4−/− eosinophils responded to IL-4 by releasing RELM-α (Fig. 4A). Il4−/− eosinophils stimulated with IL-33 secreted IL-6 and IL-13 at similar levels to IL-33–stimulated wild-type eosinophils (Fig. 4B). Interestingly, the release of CCL17 was lower in IL-33–stimulated Il4−/− eosinophils in comparison with IL-33–stimulated wild-type eosinophils, suggesting a role for an IL-4 autocrine loop in CCL17 expression (Fig. 4C). However, in Fig. 3B, the treatment by IL-4 did not induce CCL17 expression. Thus, we tested whether the addition of IL-4 with IL-33 could induce the expression of CCL17. Il4-deficient eosinophils treated with IL-33 and IL-4 produced CCL17 at similar levels compared with wild-type eosinophils (data not shown). However, the cotreatment did not increase the expression of CCL17 in wild-type eosinophils, suggesting that the IL-4 has an additional, but not synergistic effect. By qRT-PCR analysis, we confirmed that, compared with IL-33–treated wild-type eosinophils, IL-33–treated Il4−/− eosinophils displayed decreased Retnla and a trend for decreased Ccl17 mRNA expression, but unchanged Il6 and Il13 mRNA expression (Fig. 4D). Collectively, these data suggest the existence of an IL-4 autocrine loop that is responsible for IL-33–induced RELM-α and, to a lesser extent, CCL17. In contrast, IL-6 and IL-13 are independent of the IL-4 autocrine loop and may be directly induced by IL-33.
IL-33 induces eosinophil activation through IL-4–dependent and –independent pathways
To further investigate the role of IL-4 in mediating activation of eosinophils by IL-33, we examined the effect of IL-4–specific neutralizing Ab. Eosinophils were incubated for 24 h with anti–IL-4 Ab in the presence of IL-33; IL-4 was not detected (by ELISA) following anti–IL-4 treatment (Fig. 4E). Anti–IL-4 Ab decreased the IL-33–induced secretion and transcription of RELM-α and CCL17 (Fig. 4F, 4G). Notably, cells incubated with IL-33 and IgG1 control Ab produced RELM-α at similar levels to cells incubated with IL-33 alone, indicating the specificity of the IL-4 neutralizing Ab in inhibiting IL-33–induced RELM-α expression. The expression and secretion of IL-6 and IL-13 were not affected by anti–IL-4 treatment (Fig. 4F, 4H). In addition, the expression of Cxcl2, Cxcl10, Clec4e, and Adora2b was not decreased with anti–IL-4 treatment, indicating that, similar to IL-6 and IL-13, these genes are not dependent on the IL-4 pathway (Fig. 4F). However, we observed a trend for Tfec mRNA to be decreased in the presence of anti–IL-4 Ab (Fig. 4F), suggesting that the IL-4 autocrine loop is involved in the regulation of Tfec expression. We validated our system with an ex vivo culture of mature eosinophils from Il5-transgenic mice. Fig. 5 shows that mature eosinophils respond to IL-33 by releasing IL-4, RELM-α, CCL17, IL-6, and IL-13. However, in the presence of the anti–IL-4 Ab, the RELM-α and CCL17 secretion (unlike IL-6 and IL-13) were significantly decreased.
IL-33 requires type I IL-4R to activate eosinophils
Next, we aimed to determine whether IL-4R type I and/or II are required for eosinophil activation. We first treated eosinophils with IL-33 for 24 h in the presence of anti–IL-4Rα neutralizing Ab or IgG2a, an isotype control. Neutralizing IL-4Rα inhibited the IL-33–induced expression and production of RELM-α by eosinophils (Fig. 6A, 6B). Although the IL-33–induced expression of Ccl17 mRNA was not significantly decreased by anti–IL-4Rα treatment (Fig. 6A), the production of CCL17 protein after 24 h of IL-33 and anti–IL-4Rα treatment was dramatically decreased in comparison with IL-33 and IgG2a treatment (Fig. 6B). Treatment with anti–IL-4Rα Ab did not change the production of IL-4 induced by IL-33 (Fig. 6B). In addition, IL-33–induced IL-6 and IL-13 expression and production were not affected by the anti–IL-4Rα or IgG2a Ab treatments (Fig. 6A, 6B). Similarly, the IL-33–induced expression of Cxcl2, Cxcl10, Clec4e, and Adora2b was not affected by the treatment with the anti–IL-4Rα Ab, indicating that these genes are directly regulated by IL-33 (Fig. 6A). IL-33–induced Tfec expression tended to be decreased in the presence of the IL-33 and anti–IL-4Rα Ab compared with IL-33 and IgG2a (Fig. 6A), indicating that this gene may be dependent on the IL-4 pathway. In summary, we found that IL-33 requires IL-4Rα to induce RELM-α and CCL17 secretion.
We then tested whether IL-4R type II was required for eosinophil activation. Because anti-mouse IL-13Rα1 neutralizing Ab was not commercially available, we generated eosinophils from Il13ra1-deficient mice. Wild-type and Il13ra1-deficient eosinophils were treated for 24 h with IL-33. IL-33 induced RELM-α, CCL17, IL-4, and IL-13 secretion in Il13ra1−/− eosinophils at similar levels compared with wild-type eosinophils (Fig. 6C). Overall, these results demonstrated that IL-4Rα, but not IL-13Rα1, is required for the IL-33–mediated release of RELM-α and CCL17.
Role of STAT-6 in eosinophil activation
To further elucidate whether IL-4 signaling is involved in the expression of genes induced by IL-33, we generated Stat6-deficient bone marrow–derived eosinophils and examined their response to IL-33 or IL-4. We found that RELM-α production was not induced by either IL-4 or IL-33 in Stat6-deficient eosinophils compared with wild-type eosinophils (Fig. 7A), which is consistent with our previous observations that RELM-α expression is IL-4/IL-4Rα dependent. In addition, IL-33 did not induce CCL17 production in Stat6−/− eosinophils (Fig. 7B). IL-4, IL-6, and IL-13 were secreted by wild-type and Stat6-deficient eosinophils after IL-33 exposure (Fig. 7C). Although Stat6-deficient eosinophils responded to IL-33 and expressed ST2 (data not shown), they secreted significantly less IL-4, IL-6, and IL-13 compared with wild-type eosinophils, suggesting a dysfunction in these cells, probably due to the loss of STAT-6.
NF-κB is involved in the expression of genes induced by IL-33
To determine whether the induction of genes by IL-33 occurs via NF-κB signaling, eosinophils were pretreated for 1 h with a NF-κB inhibitor (BAY11-7082) and then activated for 4 h with IL-33 or IL-4. Cxcl2, Cxcl10, Clec4e, Il6, Il13, Adora2b, Ccl17, Tfec, and Retnla displayed a decrease in their IL-33–induced expression when treated with the NF-κB inhibitor, indicating that induction by IL-33 is regulated directly or indirectly by NF-κB signaling (Fig. 8A). Notably, Cxcl2, Cxcl10, Clec4e, Il6, Il13, and Adora2b were not induced by IL-4, suggesting that these genes may be directly regulated by the binding of NF-κB to the promoter region. This notion is supported by the presence of NF-κB binding/consensus sites in the promoter regions of human IL6 and IL13 and of murine Cxcl2 and Cxcl10 (see link: http://bioinfo.lifl.fr/NF-KB/). We previously showed that Ccl17, Tfec, and Retnla were induced by IL-33 through an IL-4 autocrine loop. Consistent with this indirect mechanism, IL-4–induced Ccl17, Tfec, and Retlna expression were not affected by the NF-κB inhibitor, and the decrease in their IL-33–induced expression by NF-κB inhibitor correlated to a decrease in IL-4 release (Fig. 8B). Indeed, our results indicate that the IL-4 autocrine loop requires IL-33–associated NF-κB signaling (see Fig. 9 for the overall proposed model).
Although eosinophils have been shown to respond to IL-33 or IL-4 (9, 30–33), the regulation and role of both of these cytokines, as well as IL-13, in directly activating eosinophils have not been well characterized. In this study, we report RNA sequence analysis of murine eosinophils activated by IL-4 and IL-33. We demonstrate that IL-33 is a powerful activator of eosinophils, whereas IL-4 and IL-13 induce modest and no activation of eosinophils, respectively. We demonstrate that IL-33 mediates its effects by two pathways, distinguished by their dependence on IL-4. IL-33 directly stimulates eosinophils to release preformed IL-4, which then auto-activates eosinophils via IL-4Rα and STAT-6; additionally, IL-33 directly induces transcription of a myriad of gene products via NF-κB. Interestingly, functional analysis revealed that IL-33 mediates the expression of cytokine and chemokine gene products, such as IL-6, IL-13, CXCL2, and CXCL10, whereas IL-4 induces genes enriched in the STAT-6 signaling pathway, such as RELM-α and TFEC, which are known to be STAT-6 dependent (34, 35). Fig. 9 summarizes our findings, which support the ability of IL-33 to profoundly induce eosinophil activation, as measured by transcriptional induction and release of pleiotropic immunomodulatory mediators, and the induction of an IL-4–driven autoinflammatory loop that is likely to contribute to a variety of Th2 and/or innate immune responses. Whereas mature eosinophils are classically known as primarily granule protein-secreting cells, there are emerging data that these cells remain transcriptionally active and secrete a number of gene products, including IL-4 and RELM-α mRNA and protein (18, 36–38).
We found that IL-33 induces IL-4 release by eosinophils. However, IL-4 mRNA did not increase after IL-33 treatment. Interestingly, we found that IL-4 was prestored in murine eosinophils under homeostatic conditions (at day 14 of culture) (data not shown). Similarly, human eosinophils have been demonstrated to contain prestored IL-4 in vesicles and to rapidly release IL-4 upon stimulation (39, 40). Previous studies have reported that IL-33 can induce IL-4 production by human basophils (4, 41) and murine splenic eosinophils when these cells are cotreated with IL-33 and IL-5 or GM-CSF (30). Moreover, a recent study showed that, in a dextran sulfate sodium–induced colitis model, IL-33 exacerbates the disease via IL-4 (42). In addition, IL-33 has been demonstrated to enhance B cell activation through IL-4 secreted by mast cells and eosinophils (43). Notably, IL-4–producing eosinophils have been demonstrated to have a role in the initiation of type 2 immune responses. In a helminth infection model, eosinophils were identified as one of the major innate IL-4–producing cells in the lung of infected mice (16, 17), whereas IL-33 is known to initiate the immune response (44). Interestingly, in adipose tissue, eosinophil-derived IL-4 has been demonstrated to have a role in glucose metabolism through the maintenance of adipose M2 macrophages, indicating an anti-inflammatory role of IL-4–secreting eosinophils (18). Moreover, two recent studies have demonstrated that, after muscle or liver injury, eosinophil-derived IL-4 is required for tissue regeneration, through the regulation of the resident cells (45, 46). In addition, IL-33 protects mice from adipose tissue inflammation by increasing Th2 cells and macrophages (47). On the basis of this collective dataset as well as our data that IL-33 potently induces eosinophil production of a plethora of potent cytokines, we speculate that IL-33–activated eosinophils could be a key driving factor in the regulation of eosinophilic inflammatory responses. Our results also shed light on the potential function of eosinophil-derived IL-4; rather than strictly stimulating other cell types, eosinophil-derived IL-4 may have an essential role in auto-activating eosinophils, consistent with most studies that have shown that eosinophil-derived IL-4 is not essential for Th2 cell polarization (48–50). Recently, an in vitro study demonstrated that the activation of the IL-4/IL-4Rα axis in murine eosinophils from IL-5–transgenic mice enhances eosinophil migration toward CCL11 (33), suggesting a role of IL-4 as a potent coactivator of eosinophils in allergic reactions.
By dissecting the mechanism of IL-4 action on eosinophils, we discovered that IL-33 requires IL-4 signaling to induce RELM-α and CCL17. First, we found that IL-4 induces RELM-α production by eosinophils through the type I IL-4R. To further examine the specific role of IL-4 in IL-33–induced eosinophil responses, we generated eosinophils from Il4-deficient mice, which demonstrated that eosinophils required IL-4 to induce RELM-α and CCL17, but not IL-6 and IL-13. In complementary experiments using anti–IL-4 or anti–IL-4Rα treatment, we validated the IL-4/IL-4Rα dependency of IL-33–induced RELM-α and CCL17 expression. Interestingly, although single treatment by IL-4 did not induce the expression of CCL17 (Fig. 3B), treatment with IL-4 and IL-33 increased CCL17 expression in Il-4–deficient eosinophils, suggesting that STAT-6 and NF-κB signaling are both required for CCL17 production (data not shown). However, in wild-type eosinophils, CCL17 expression was not increased by the cotreatment with IL-4 and IL-33 compared with IL-33 alone. These data indicate that CCL17 reached a maximal expression after IL-33 treatment or cotreatment with IL-4 and suggest that the signal cannot be amplified due to a saturation of the IL-4R by the IL-4. By using in vivo eosinophils from Il5-transgenic mice, we confirmed the in vitro findings showing that IL-33 elicits two activation pathways. Although in vivo eosinophils respond to IL-33 and IL-4, we found that the levels of cytokines released are lower than bone marrow–derived eosinophils. In Il5-transgenic mice, the overexpression of IL-5 may preactivate directly or indirectly the eosinophils. It is known that activated eosinophils are more sensitive to apoptosis. In addition, the eosinophil preparation process could also weaken the cells and alter their viability, and thus their potential to respond to a stimulus.
Moreover, we showed that the IL-33–induced expression of Tfec was dependent on the IL-4 autocrine loop, whereas the IL-33–mediated induction of Cxcl2, Cxcl10, Clec4e, and Adora2b transcripts was not affected by neutralizing the IL-4/IL-4Rα axis. Although IL-4 displays a lower effect than IL-33 on eosinophil activation, these findings showed that IL-33 induces and amplifies eosinophil activation partially via IL-4. Overall, we suggest that IL-4 contributes to the IL-33–induced eosinophil activation by the release of CCL17, which enhances the recruitment of TCD4+ cells, and by producing RELM-α, which promotes fibrogenic responses and eosinophil chemoattraction (51–53).
In agreement with a recent study (9), we showed that Il6 and Il13 were rapidly expressed in eosinophils and released within 24 h of treatment with IL-33. Moreover, our study showed that IL-6 and IL-13 transcripts and proteins were not induced by IL-4 treatment alone. Using a NF-κB inhibitor, we demonstrated that IL-33–induced Il6 and Il13 expression were dependent on NF-κB signaling, suggesting that NF-κB binds to Il6 and Il13 promoters and consequently induces their transcription. It is important to note that the IL-33–induced expression of some genes (e.g., Il6 or Il13) was not entirely abolished by NF-κB inhibition, indicating that NF-κB is not likely the only signaling molecule involved in the activity of IL-33; other mediators such as ERK, p38, or JNK could be involved as well.
In conclusion, we have demonstrated the ability of IL-33 to directly and profoundly induce eosinophil activation by a NF-κB–dependent mechanism. Additionally, we have determined that IL-33 induces NF-κB–dependent eosinophil release of IL-4. Furthermore, we have identified an IL-33–induced IL-4–driven eosinophil autoinflammatory loop that most likely contributes to a variety of Th2 and/or innate immune responses, involving both immature developing eosinophils and mature eosinophils. To date, few papers have demonstrated the effect of IL-33 on human eosinophils (4, 54, 55), and the transcriptome or gene expression profile has not been shown. In this study, we demonstrated that IL-6 is induced by IL-33 in murine eosinophils, similarly to reported induction in human eosinophils (55). We also found that Ccl2 and Icam1 are expressed in murine eosinophils after IL-33 treatment, which is consistent with protein expression in IL-33–activated human eosinophils (55).
We thank Dr. Yui-Hsi Wang (Cincinnati Children’s Hospital Medical Center) for the kind gift of the Stat6−/− and Il4−/− mice, Melissa Mingler for animal colony maintenance, and Shawna Hottinger (medical writer, Cincinnati Children's Hospital Medical Center) for editorial assistance.
This work was supported by National Institutes of Health/National Institute of Allergy and Infectious Diseases Grants R01AI083450 and R37AI045898, the Campaign Urging Research for Eosinophilic Disease Foundation, Food Allergy Research & Education, and the Buckeye Foundation.
The authors have no financial conflicts of interest.