The p38-MK2/3 Module Is Critical for IL-33–Induced Signaling and Cytokine Production in Dendritic Cells

The novel member of the IL-1 cytokine superfamily, IL-33, is considered an “alarmin” that is released upon necrosis from endothelial and epithelial cells or from fibroblasts (1, 2). The release of IL-33 induces the recruitment and activation of IL-33R–expressing cells, such as mast cells and dendritic cells (DCs). In DCs, IL-33 induces the production of IL-6, IL-13, and TNF-α, but not of IL-4, and it mediates the upregulation of MHC class II and CD86 (3, 4). Thereby, IL-33–activated DCs mediate an atypical Th2 response in naive T cells that is characterized by the production of IL-5 and IL-13 (3, 4). However, despite the importance of IL-33–activated DCs for inducing a Th2 response, the mechanisms of IL-33–induced signaling in DCs that regulate their effector functions are unknown.

In general, binding of IL-33 to IL-33R results in IL-1R accessory protein– and MyD88-dependent activation of TAK1 and IKK2, which mediates IκBα degradation and, thus, NF-κB activation (5, 6). IL-33 also induces the activation of MAPKs, such as p38 (6–8). Consequently, NF-κB and MAPK activation result in cytokine production and survival (6). In mast cells, the p38 targets, Mapk-activated protein kinases 2/3 (MK2/3) (9, 10), play an important role in IL-33–induced signaling and cytokine production in vitro and in vivo (11).

MK2/3 mediate the stabilization of cytokine mRNAs (12) and participate in the activation of the PI3K/protein kinase B (PKB)/Akt signaling and ribosomal S6 kinases (11, 13–15). Recently, we found that the stem cell factor (SCF) and IL-3 potentiate IL-33–induced mast cell effector functions (16–18). Furthermore, we demonstrated that epidermal growth factor receptor (EGFR) is critically involved in IL-33–induced and Erk1/2-dependent IL-13 production in Th2 cells (19). These data show that IL-33R–mediated effector functions are controlled by activated growth factor receptors. In this article, we show that GM-CSF, a differentiation and growth factor for bone marrow–derived DCs (BMDCs), is required to mediate an effective IL-33–induced and RelA- and p38-MK2/3-mTOR–dependent cytokine response. This article proposes GM-CSF and the p38–MK2/3–mTOR signaling module as potential targets in manipulating IL-33R–meditated effector functions in DCs.

Wild-type (wt) littermates, Mapkapk2^tm1Mgl (mk2^−/−)/Mapkapk3^tm1Mgl (mk3^−/−) mice (9), and myd88^−/− mice (20) were maintained at the Animal Research Facility of the Medical School Hannover and in the Animal Research Facility of Jena University Hospital. St2^−/− (il-33r^−/−) (21) mice were provided by Prof. Dr. Löhning (Deutsches Rheuma-Forschungszentrum, Berlin, Germany). rela^fl/fl (22), relb^fl/fl (23), and cd11c-cre (24) mice were provided by Prof. Dr. Weih (Fritz-Lipmann-Institut, Jena, Germany). rela^fl/fl and relb^fl/fl mice were crossed with cd11c-cre mice to obtain a conditional deletion of rela or relb in DCs. The obtained mice are rela^fl/fl;tg/+, relb^fl/fl;tg/+, and rela^fl/fl;+/+, rela^fl/fl;+/+ (as control littermate mice). We used sex- and age-matched knockout and wt littermates.

For generation of BMDCs, bone marrow was obtained from the femurs and tibias of mice. After erythrocyte lysis, cells were seeded (2 × 10⁵ cells per milliliter) in RPMI 1640 culture medium (Sigma-Aldrich) supplemented with 10% FCS (Sigma-Aldrich), 1% antibiotics (Jena Bioscience), HEPES (Serva), 2-ME (Thermo Fisher Scientific), and GM-CSF (20 ng/ml) (conditioned medium from X63AG–GM-CSF cells). Medium was refreshed on days 3, 6, and 8. On days 8–10, nonadherent cells were harvested and measured for expression of CD11c and CD11b by flow cytometry. BMDCs with purity > 90% were used for experiments.

After 8 d in culture, BMDCs were labeled with Abs in PBS containing 0.25% BSA and 0.02% sodium azide. Dead cells were excluded by propidium iodide (PI). Nonspecific binding was blocked with anti-CD16/CD32 (clone 2.4G2) and rat-IgG (Jackson ImmunoResearch). Cells were stained with anti-murine CD11c Ab and anti-murine CD11b Ab (eBioscience) for identification of BMDCs (see gating strategy in Supplemental Fig. 1A). For other experiments, we also stained cells with the anti-murine IL-33R Ab (clone DJ8; MD Biosciences). For determination of dead BMDCs, samples were treated with the respective inhibitor (as indicated in the figure legends) and PI. We performed these cell death analyses for ELISA experiments in which BMDCs were incubated for 24 h with the respective inhibitor. Samples were analyzed by flow cytometry (LSR II or FACSCanto II flow cytometer [BD] and FlowJo v9 [TreeStar, Ashland, OR].

BMDCs were GM-CSF starved for 1 h, incubated with inhibitors (30 min) as indicated (all from Merck Millipore, with the exception of SB239063 [Santa Cruz Biotechnology]), and stimulated with IL-33 (PeproTech) or LPS (Enzo Life Sciences) or with IL-33 together with GM-CSF (both from PeproTech). Cells were lysed with buffer containing 20 mM HEPES (Serva) (pH 7.5), 10 mM EDTA (Roth), 40 mM β-glycerophosphate (AppliChem), 2.5 mM MgCl₂ (Roth), 2 mM orthovanadate (Sigma-Aldrich), 1 mM DTT (AppliChem), 20 μg/ml aprotinin (Sigma-Aldrich), 20 μg/ml leupeptin (Sigma-Aldrich), and 1% Triton (Sigma-Aldrich), and protein concentration was determined using a BCA Assay (Pierce). Subsequently, protein samples were boiled in 6× Laemmli buffer.

Samples were separated on 10% SDS-Laemmli gels and transferred by Western blotting onto nitrocellulose membranes (biostep). Membranes were blocked with 0.1% Tween/TBS buffer supplemented with 5% dry milk and incubated (overnight) with Abs detecting phosphorylated/nonphosphorylated proteins. We used anti-pS176/177IKK1/2/anti-IKK1/2, anti-pT180/Y182-p38/anti-p38, anti-pY694-STAT5/anti-STAT5, anti-MK2, anti-MK3, anti-pS32-IκBα/anti-IκBα, anti-pS468-p65/anti-p65, anti-RelB, anti-pT389-p70S6K/anti-p70S6K, anti-pS473-PKB/Akt/anti-PKB/Akt, anti-ubiquitin, and anti-tubulin (all from Cell Signaling Technology, with the exception of anti-IKK1/2 and anti-ubiquitin [Santa Cruz Biotechnology]). Membranes were washed in 0.1% Tween/TBS and incubated with HRP-conjugated secondary Abs: anti-rabbit Ig, anti-goat Ig (both from Santa Cruz Biotechnology) and anti-mouse Ig (Thermo Fisher Scientific). Detection was performed using ECL reagent (Pierce).

BMDCs were seeded at 10⁶ cells per milliliter in GM-CSF–free medium. Cells were preincubated with vehicle (DMSO) or inhibitors for 30 min. BMDCs were stimulated with IL-33, GM-CSF, or both (all PeproTech). For ELISA experiments, supernatants were collected after 24 h and analyzed for IL-6, IL-13, or TNF-α using matched-pair Abs (eBioscience). For IL-33 analysis, a Mouse IL-33 DuoSet ELISA (R&D Systems) was used. For proliferation assays, cells were cultured for 54 h, and [³H]thymidine (1 μCi) in 25 μl of complete IMDM (PAA) (without GM-CSF) was added to each well for an additional 18 h. Incorporated radioactivity was measured using a beta scintillation counter (PerkinElmer, Rodgau-Jügesheim, Germany).

Western blots were quantified with ImageJ (National Institutes of Health). To determine the stability of proteins, the intensities of protein bands were determined and normalized to the respective internal controls (tubulin). The control (unstimulated cells) of wt cells was set as 1. To quantify protein phosphorylation, the intensities of the phosphorylation bands were determined and normalized to the respective total protein bands. The control (unstimulated cells) of wt cells was set as 1. We calculated the fold increase in phosphorylation compared with the unstimulated wt control.

The number of performed independent experiments is given in the figure legends. For all experiments, bone marrow of several mice was pooled and cultured in GM-CSF–containing media to generate BMDCs. For all experiments, the BMDCs derived from pooled bone marrow were seeded. For every condition we used ≥6-fold determination (shown is one representative experiment). Cytokine concentration is indicated as the mean ± SE. Statistical analyses were performed with IBM SPSS Statistics version 20.0 (IBM, Ehningen, Germany). Statistical significance was assessed by the Mann–Whitney U test, unless otherwise stated. Statistical significance was accepted for p ≤ 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001).

As shown by other investigators, BMDCs express IL-33R (4, 25–28). Consequently, stimulation of wt, but not of t1/st2^−/− (il-33r^−/−), BMDCs with IL-33 induced the production of IL-6 and IL-13 (Supplemental Fig. 1B). However, and in accordance with the literature (3, 4), we could not detect IL-4 in the supernatants of IL-33–activated BMDCs (data not shown). In contrast to IL-33, LPS induced the production of IL-12 (Supplemental Fig. 1C). These data demonstrate the functionality of the IL-33R on BMDCs. Next, we investigated the IL-33–induced signaling in BMDCs. Similar to LPS, IL-33 also induces the activation of IKK2, resulting in phosphorylation of IκBα and of S468 on p65, which correlates with NF-κB activation (29) (Fig. 1A). However, despite the phosphorylation of IκBα and p65, IL-33 did not induce degradation of IκBα (Fig. 1A). Also, higher concentrations of IL-33 failed to induce IκBα degradation (Supplemental Fig. 1D). Furthermore, IL-33 and LPS induced the activation of p38, PKB/Akt, and p70S6K (Fig. 1B). Due to the lack of IκBα degradation, we hypothesized that, in the presence of a strong ongoing protein biosynthesis, IκBα degradation is quantitatively not sufficient to result in a net loss of IκBα. To test this hypothesis, we used cycloheximide to block the protein biosynthesis. Treatment with cycloheximide slightly reduced the basal content of IκBα and resulted in IL-33–induced IκBα degradation (Fig. 1C). Next, we tested whether IL-33 induces IκBα ubiquitinylation. As shown in Fig. 1D (left side; DMSO treatment), IL-33 did not induce a detectable IκBα ubiquitinylation. Thus, we hypothesized that IκBα ubiquitinylation is too transient and/or too weak for its detection. Therefore, we used the proteasome inhibitor MG132 to stabilize IκBα. Interestingly, proteasome inhibition by MG132 leads to accumulation of ubiquitinylated IκBα and, surprisingly, to its degradation in response to IL-33 (Fig. 1D), indicating that proteasome inhibition might induce an alternative pathway [e.g., autophagy (30)] for IκBα degradation. Nevertheless, these results demonstrate that, under physiological conditions, the IL-33–induced ubiquitinylation is too weak to induce a strong signal, which results in a net loss of IκBα.

Because we could not detect IκBα degradation, we tested whether IL-33 induces MyD88-dependent canonical NF-κB signaling in BMDCs. MyD88 deficiency blocked the IL-33–induced NF-κB activation and cytokine response (Supplemental Fig. 1E, 1F). Furthermore, inhibition of TAK1 with 5Z-7-oxozeanol (Supplemental Fig. 2A, 2B) or IKKs with the IKK inhibitor VII (Supplemental Fig. 2C, 2D) also blocked the IL-33–induced NF-κB (p65) activation and cytokine production without inducing cell death (Supplemental Fig. 2E, 2F). These data show that IL-33 induces MyD88-TAK1-IKK2–dependent NF-κB (p65) activation and cytokine production. Next, we wanted to determine the role of p65 (RelA) in the IL-33–induced production of IL-6 and IL-13. Therefore, we used BMDCs generated from rela ^fl/fl;tg/+ and rela ^fl/fl;+/+ mice as controls. rela ^fl/fl;tg/+ mice are characterized by a conditional deletion of exons 7–10 of the rela gene in DCs. Exons 7–10 encode for the Rel homology domain, which is important for DNA binding and dimerization of RelA (22). Therefore, deletion of exon 7–10 leads to expression of a shorter and, thus, inactive RelA in DCs (22). Unexpectedly, the IL-33–induced IL-6 response, but not the IL-13 response, was reduced in rela ^fl/fl;tg/+ BMDCs (Fig. 1E, 1F). To exclude that IL-33 alternatively induces the activation of RelB in BMDCs, we generated BMDCs from relb ^fl/fl;tg/+ mice, as well as from relb ^fl/fl;+/+ mice as controls. relb ^fl/fl;tg/+ mice are characterized by deletion of exon 4 of the relb gene, which encodes for the Rel homology domain (31). Compared with relb ^fl/fl;+/+ BMDCs, stimulation of relb ^fl/fl;tg/+ BMDCs with IL-33 leads to an enhanced phosphorylation of RelA and an increased release of IL-6 and IL-13 (Supplemental Fig. 2G, 2H). In contrast, the LPS-induced IL-12 production was blocked in relb ^fl/fl;tg/+ BMDCs (Supplemental Fig. 2I), demonstrating that LPS-induced IL-12 production depends on RelB. These data show that IL-33 induces activation of the MyD88-TAK1-IKK2–dependent signaling, resulting in p65-dependent IL-6 production.

We found that the IL-33–induced IL-6 production depends on the MyD88–TAK1–IKK2–p65 signaling pathway. However, IL-33–induced IL-13 production only depends on MyD88, TAK1, and IKK, and not on p65, indicating an alternative pathway that mediates the production of IL-13. In mast cells, p38 and its targets, the MAPKAPs MK2/3, mediate activation of the PI3K/PKB–p70S6K signaling pathway and, thus, the IL-33–induced cytokine production (11). We and other investigators found that IL-33 induces activation of p38 in BMDCs (28). Therefore, we investigated the roles of p38 and MK2/3 in the IL-33–induced signaling and cytokine production in BMDCs. Activation of p38-MK2/3 signaling and PKB/Akt is mediated via the MyD88–TAK–IKK signaling pathway (Supplemental Fig. 3). Furthermore, the p38 inhibitor SB239063 reduced activation of the PKB/Akt–p70S6K signaling module (Fig. 2A) and decreased the production of IL-13 (Fig. 2B) but not of IL-6 (Fig. 2C). Thereby, SB239063 did not induce cell death (Fig. 2D). Next, we determined the role of the p38 targets MK2/3. As shown for p38 inhibition, MK2/3 deficiency also reduced activation of the PKB/Akt–p70S6K signaling module (Fig. 3A) and decreased the production of IL-6 but not IL-13 (Fig. 3B, 3C). These data show that the p38-MK2/3 signaling module mediates activation of PKB/Akt–p70S6K signaling and is critical for the production of IL-13 but not IL-6.

We found that the p38-MK2/3 signaling module mediates activation of the mTOR target, p70S6K, in BMDCs. Therefore, we hypothesized that the p38–MK2/3–PKB/Akt–mTOR–p70S6K signaling pathway mediates the production of IL-13, but not IL-6. To test this hypothesis, we used the mTOR inhibitor, rapamycin. Unexpectedly, treatment of BMDCs with rapamycin reduced the production of IL-6 and IL-13 (Fig. 3D, 3E) without inducing cell death (Fig. 3F). This result was contradictory to our hypothesis in which we speculated that the MK2/3–PKB/Akt–mTOR–p70S6K signaling pathway mediates the production of IL-13 but not of IL-6. Thus, we hypothesized that inhibition of the p38-MK2/3 signaling module leads to a compensatory upregulation of an IL-33–induced signaling pathway, leading to equal IL-6 production in wt and mk2^−/−/3^−/− BMDCs. MKs are known to control the activation of NF-κB (32). Thus, we speculated that the p38-MK2/3 signaling module negatively regulates the activation of p65. Indeed, SB239063 treatment and MK2/3 deficiency increased the phosphorylation of RelA on S468, indicating increased RelA activation (29) (Fig. 4A, 4B). This indicates that reduced activation of the PKB/Akt-p70S6K signaling module in SB239063-treated and mk2^−/−/3^−/− BMDCs might be compensated for by an increased RelA activation. Therefore, IL-33 induces a similar IL-6 response in wt and mk2^−/−/3^−/− BMDCs. In summary, we speculate that the p38-MK2/3 module mediates a balanced activation of the transcriptional and translational response in IL-33–stimulated BMDCs.

Next, we wanted to identify the molecular mechanism leading to enhanced phosphorylation of p65 in MK2/3-deficient BMDCs. We hypothesized that, in unstimulated BMDCs, MK2/3 bind to and, thus, mask the phosphorylation site on p65. Stimulation with IL-33 leads to the release of MKs and, thus, to improved accessibility of p65 phosphorylation sites. However, immunoprecipitation of MK2 and p65 did not show complex formation of p65 with MK2/3 (Fig. 4C, 4D), but it revealed a stable association of p65 with IκBα (Fig. 4D). These data show that MKs do not control the phosphorylation of p65 by complex formation.

We demonstrated that SCF and IL-3 potentiate IL-33–induced and NF-κB–dependent cytokine responses in mast cells (16–18). Furthermore, we found that expression of EGFR on Th2 cells is critical for IL-33R to mediate TCR-independent Th2 responses (19). In addition, GM-CSF is pivotal for IL-33–induced DC activation and, thus, Th2 polarization in vivo (33), Therefore, we tested the influence of GM-CSF on the IL-33–induced cytokine response in BMDCs. GM-CSF alone did not induce the production of IL-6 and IL-13, but it strongly potentiated the weak IL-33–induced production of IL-6 and IL-13 (Supplemental Fig. 4A, 4B).

In mast cells, growth factors (e.g., SCF and IL-3), together with IL-33, are critical to induce the production of TNF-α (11). In BMDCs, single stimulation with GM-CSF or IL-33 induced the production of very low levels of TNF-α. In contrast, costimulation with GM-CSF and IL-33 leads to a potentiated TNF-α production (Supplemental Fig. 4C). These data confirm that GM-CSF is critical for IL-33 to induce the activation of DCs (33).

In bone marrow–derived mast cells (BMMCs), the sequential stimulation with growth factors (e.g., SCF or IL-3) and IL-33 is important to induce optimal potentiated cytokine production (16–18). Consequently, we tested several prestimulation times for GM-CSF. We did not find a profound difference between different prestimulation times with regard to IL-33–induced IL-6 and IL-13 production (Supplemental Fig. 4D, 4E). Thus, we stimulated BMDCs with GM-CSF and IL-33 simultaneously in further experiments.

Next, we tested whether GM-CSF stimulation upregulates the expression of IL-33R or leads to production of IL-33. As shown in Supplemental Fig. 4F and 4G, stimulation with GM-CSF did not upregulate the expression of IL-33R or induce production of IL-33. Furthermore, we investigated whether costimulation with GM-CSF (10 ng/ml) and IL-33 also induces potentiated proliferation, which might explain an increased cytokine response. Single stimulation with GM-CSF (10 ng/ml) or IL-33 did induce a 2-fold increase in proliferation. As a positive control, we used GM-CSF (20 ng/ml) and found a strong induction of BMDC proliferation (Supplemental Fig. 4H). Costimulation with GM-CSF (10 ng/ml) and IL-33 did not enhance the proliferation compared with single stimulation (Supplemental Fig. 4H). This demonstrates that the potentiated cytokine response is not mediated as a result of upregulated expression of IL-33R and IL-33 or by increased cell numbers.

We identified the p38-MK2/3 signaling module and p65 as critical for the IL-33–induced IL-6 and IL-13 production. Moreover, GM-CSF potentiated the IL-33–induced cytokine production in BMDCs. GM-CSF is known to induce the activation of p38, PKB/Akt, and STAT5. Furthermore, STAT5 is important for IL-33 to induce an effective cytokine production in T cells and mast cells (34, 35). We determined whether costimulation with GM-CSF and IL-33 increases the activation of key signaling molecules, such as p38, p65, PKB/Akt, and STAT5. Stimulation with IL-33 or GM-CSF alone induced the activation of p38 (Fig. 5A), PKB/Akt (Fig. 5B), and p65 (Fig. 5C). In contrast to IL-33, GM-CSF induced the activation of STAT5 (Fig. 5D). Costimulation with GM-CSF and IL-33 increases the activation of p38 (Fig. 5A) and PKB/Akt (Fig. 5B). However, the IL-33–induced activation of p65 (Fig. 5C) and the GM-CSF–induced activation STAT5 (Fig. 5D) were not influenced by costimulation. This indicates that the potentiated cytokine production is primarily mediated by an increased activation of the p38–MK2/3–PKB/Akt signaling module.

Next, we determined the roles of p65 and the p38-MK2/3 signaling module in the potentiated cytokine response induced by costimulation. First, we investigated the role of the canonical NF-κB signaling induced by IL-33. MyD88 deficiency (Fig. 5E–G) and the IKK-inhibitor VII (Fig. 5H–J) reduced the potentiated production of IL-6, IL-13, and TNF-α. This demonstrates that the MyD88-IKK2 signaling module is critical for the potentiated production of IL-6, IL-13, and TNF-α. Next, we investigated the role of the downstream targets of the MyD88-IKK2 signaling complex, p65 and the p38-MK2/3 signaling module. In rela ^fl/fl;tg/+ BMDCs, we found strongly reduced production of IL-6 but not of IL-13 or TNF-α (Fig. 6A–C). In contrast, the p38 inhibitor SB239063 (Fig. 6D–F) and MK2/3 deficiency (Fig. 6G–I) strongly reduced the production of IL-13 and blocked the production of TNF-α but did not influence IL-6 production.

We found that mTOR, a downstream target of PI3K/PKB/Akt signaling, is critical for IL-33–induced cytokine production. Therefore, we also tested the influence of mTOR in BMDCs costimulated with GM-CSF and IL-33. Treatment of BMDCs with the mTOR inhibitor rapamycin reduced the production of IL-6, IL-13, and TNF-α (Figs. 6J–L).

Together, these data show that p65 is important for potentiated IL-6 production, whereas the p38-MK2/3 signaling module mediates the production of IL-13 and TNF-α. In contrast, mTOR-dependent signaling is important for the production of IL-6, IL-13, and TNF-α (Fig. 7).

In this article, we show that, in BMDCs, IL-33 induces a MyD88-TAK1-IKK–dependent phosphorylation of IκBα and p65. However, IL-33 did not induce a detectable destabilization of IκBα. This indicates either that IL-33 does not induce IκBα degradation or that our BMDCs are defective for proteasomal degradation induced by TLR/IL-1R family members. However, stimulation with LPS induced degradation of IκBα and the protein biosynthesis inhibitor cycloheximide enables IL-33 to induce IκBα degradation. This demonstrates that the proteasomal degradation is intact and that, in the presence of ongoing IκBα resynthesis, the IL-33–induced IκBα degradation is quantitatively insufficient to result in an effective and, thus, detectable net loss of IκBα. In contrast to BMDCs, IL-33R is strongly expressed in BMMCs, and IL-33 induces IκBα degradation (16, 17). Therefore, we speculate that the low expression level of IL-33R on BMDCs is the reason for the missing strong impact of IL-33 on IκBα stability. Despite the lack of detectable IL-33–induced IκBα degradation, we think that IL-33 induces IκBα degradation and, thus, p65 activation in BMDCs. This is in line with the result that IL-33–induced IL-6 production depends on p65. In contrast, the IL-33–induced IL-13 production does not depend on p65. This indicated that, in addition to p65, alternative signaling pathways are important to mediate IL-13 production in IL-33–stimulated BMDCs. In BMMCs, we found that the p38-MK2/3 signaling module is critical for IL-33–induced cytokine production (11). Furthermore, the p38-MK2/3 signaling module is also involved in the activation of PKB/Akt and p70S6K (11, 14, 36). In BMDCs, we also found that the p38 inhibitor and MK2/3 deficiency strongly reduced the IL-33–induced activation of the PKB/Akt-p70S6K signaling axis. Interestingly, the IL-33–induced production of IL-13, but not IL-6, depends on the p38-MK2/3 signaling module. However, the production of both cytokines depends on mTOR. Therefore, we hypothesize a mechanism that compensates for the reduced activation of the PKB/Akt–mTOR–p70S6K signaling in MK2/3-deficient BMDCs. Indeed, we found that MK2/3 negatively controls IL-33–induced p65 phosphorylation on S468, which correlates with increased p65 activation (29). This might indicate that the p38-MK2/3 signaling module negatively regulates the activation of p65.

However, the mechanism behind this regulatory function of the p38-MK2/3 signaling module is unknown. We could not detect a direct association of MKs with p65, excluding that MKs control the phosphorylation of p65 by complex formation. Gorska et al. (32) showed that MK2 deficiency leads to accumulation of p38 in the nucleus, which results in enhanced phosphorylation of p65. Thereby, MK2 associates with p38 and, thus, mediates translocation of p38 from the nucleus into the cytoplasm. These data show that localization of p38 and MK2 regulates the magnitude of p65 activation. Furthermore, the p38-MK2/3 signaling module mediates the expression of phosphatases, leading to inhibition of MAPKs and p65 (37, 38). We are currently investigating which of these possibilities regulates IL-33–induced p65 phosphorylation in BMDCs.

In summary, the p38-MK2/3 signaling module mediates activation of the PKB/Akt-mTOR-p70S6K signaling pathway and simultaneously controls the activation of p65. Therefore, we hypothesize that the p38-MK2/3 signaling module balances the transcriptional and translational response in IL-33–stimulated DCs (Fig. 7).

In addition to the p38-MK2/3 signaling module, RelB negatively regulates IL-33–induced p65 activation and the resulting cytokine production. Thereby, the basal level of RelB might determine the strength of the IL-33–induced and p65-dependent cytokine response. Interestingly, Marienfeld et al. (39) reported that expression of RelB decreases the activity of p65 by sequestering it in the RelB/p65 complex. These data are in line with our results and indicate that RelA is inhibited by IκBα and is additionally regulated by RelB in BMDCs. Taken together, these data show that the p38-MK2/3 signaling module and RelB are central regulators of IL-33–induced cytokine responses in BMDCs.

We found that SCF and IL-3 strongly increase IL-33–induced mast cell effector functions (16–18). Furthermore, we showed that, on Th2 cells, expression of EGFR is critical for the IL-33–induced Erk1/2 activation resulting in IL-13 production (19). These data demonstrated that growth factor receptors are critical for IL-33 to induce its full biological effector functions in mast cells and Th2 cells. We found a similar situation in BMDCs: GM-CSF increases the IL-33–induced and RelA-dependent IL-6 and p38-MK2/3–dependent IL-13 production. Moreover, effective TNF-α production is only induced in the presence of GM-CSF and IL-33 in a p38-MK2/3–dependent manner. Therefore, we conclude that GM-CSF increases the activation of p38–MK2/3–PKB/Akt–mTOR–p70S6K signaling, resulting in increased production of IL-6, IL-13, and TNF-α.

GM-CSF induces the activation of STAT5. Importantly, STAT5 is critical for IL-33 to induce its full biological effector functions in mast cells and T cells (34, 35). In BMDCs, GM-CSF, but not IL-33, induces the activation of STAT5. Furthermore, costimulation with GM-CSF and IL-33 increased the activation of p38, but not p65 or STAT5, indicating a preferential role for p38-MK2/3 signaling in the potentiated cytokine response in BMDCs. The fact that IL-6, IL-13, and TNF-α production strongly depends on p65 or the p38-MK2/3-signaling module indicates a negligible role for STAT5. Together, these data show that the IL-33–induced signaling and the resulting effector functions are regulated by GM-CSF, which determines the strength of IL-33–induced IL-6, IL-13, and TNF-α production.

IL-33–activated BMDCs mediate T cell polarization, including an atypical Th2 response in naive T cells that is characterized by strong production of IL-5 and IL-13 and low production of IL-17 and IL-10 (3, 4). Therefore, it is speculated that such naive T cells maintain their capacity to fully differentiate into Th2 cells, Th17 cells, or regulatory T cells (4). The link between IL-33–activated DCs and Th2 cell–mediated effector functions in vivo has also been shown recently. Thereby, IL-33–induced and MyD88-dependent signaling in DCs are critically involved in Th2-mediated skin inflammation (40). Moreover, IL-33–activated DCs are dependent on released GM-CSF to mediate allergic sensitization (33).

Therefore, we hypothesize that the release of IL-33 and GM-CSF from epithelial cells induces the activation of p65 and the p38-MK2/3 signaling module in DCs, resulting in Th2 polarization and, finally, allergic inflammation. These data, together with the fact that MK2/3 are also critical for the IL-33–induced signaling in mast cells, highlight the role of MK2/3 (11) in IL-33–induced signaling in innate cells. Thus, inhibition of MK2/3 might be an attractive therapeutic intervention to treat IL-33–driven diseases.

We thank Claudia Küchler, Karin Müller, and Freya Rost for excellent technical support.

The authors have no financial conflicts of interest.