IL-33 causes selective mast cell tolerance to bacterial cell wall products by inducing IRAK1 degradation


Full correspondence: Professor Silvia Bulfone-Paus, Faculty of Human and Medical Sciences, 3.50 Stopford Building, Oxford Road, University of Manchester, M13 9RD, UK

Fax: +44-161-275-1813



Mast cells are important cellular constituents of epithelial–mesenchymal interactions, densely located at sites of microbial entry into the host where they are continuously exposed to products from commensals. In order to avoid excessive activation and the associated pathology, mast cell responses to TLR agonists must be tightly regulated. Here, we show that exposure in vitro to subactivating levels of the epithelial cell product, IL-33, renders mast cells insensitive to bacterial cell wall products. Mast cell responsiveness to Ag, cytoplasmic dsDNA, and TLR7/8 agonists is unaffected or enhanced by IL-33. The IL-33–induced mast cell selective tolerance requires the IL-33 receptor ST2 and peritoneal mast cells from St2−/− mice display a constitutively activated phenotype, demonstrated by increased expression of activation markers including CD11b and CD28. IL-33 exposure neither affects the levels of TLR4, MyD88, TIRAP, IL-1R associated kinase 2 (IRAK2), or IRAK4, nor induces persistent A20 or Tollip expression, but potently causes ST2-dependent IRAK1 degradation. We show that while IRAK2 is redundant for TLR4 signaling, IRAK1 is essential for TLR4 signaling in mast cells. We suggest that IL-33 produced during homeostasis retains mast cells in an unresponsive state to bacterial cell wall products via IRAK1 degradation, thus preventing chronic inflammation and tissue destruction.


Mast cells are closely associated with the epithelium [1], contributing to barrier function [2]. Therefore, responses to commensal bacteria must be tightly regulated to prevent the pathology associated with unnecessary immune activation. The crucial role of the tissues in regulating immunity is increasingly being recognized [3] and we hypothesized that epithelial or endothelial cell products may orchestrate this regulation.

IL-33 and its receptor ST2 have complex roles in the LPS response [4-8]. IL-33 is expressed constitutively by epithelial cells [9, 10] and its expression is upregulated during inflammation [11]. Mast cells express ST2 and are activated by IL-33 to release cytokines and chemokines [12-15] and, furthermore, the fact that LPS and IL-33 share the TLR/IL1R signaling pathway raises the possibility that cross-talk or cross-tolerance could occur between them. While the role of IL-33 during inflammation has been widely investigated, little is known about the function of constitutively expressed IL-33.

Typically, TLR4 signaling proceeds via MyD88-dependent and independent pathways but in mast cells TLR4 ligation does not engage the MyD88-independent pathway [16]. The molecular details of the MyD88-dependent pathway in mast cells have not been fully investigated, but conventionally MyD88 is recruited via the adaptor TIRAP and associates with IL-1R–associated kinase 4 (IRAK4). IRAK4 activates IRAK1 and IRAK2, which then dissociate from the complex and associate with TRAF6. This ultimately leads to the activation of several signaling pathways and transcription factors, including MAPK and NFκB, resulting in cell activation and the expression of proinflammatory cytokines [17, 18].

In this study we demonstrate that IL-33 at concentrations insufficient for mast cell activation causes mast cell insensitivity to LPS and peptidoglycan (PGN), while leaving intact or enhancing the response to other stimuli. IL-33 causes IRAK1 degradation in mast cells and while IRAK1 is not required for TLR4 signaling in the macrophage [19], it is shown here that Irak1−/− bone marrow-derived mast cells (BMMCs) are unable to respond to LPS. We suggest that IL-33 acts in a homeostatic manner to maintain mast cells in a selectively unresponsive state and, in support of this hypothesis, find that mast cells from St2−/− animals are hyperactivated in vivo.


IL-33 exposure renders mast cells selectively unresponsive

The ability of IL-33 to regulate the response of mast cells to LPS was investigated. Peritoneal cell-derived mast cells treated with IL-33 for 24 h, thoroughly washed, then challenged with LPS produced reduced amounts of IL-6 (Fig. 1A), suggesting that this cytokine is able to induce endotoxin tolerance (ET) in mast cells.

Figure 1.

IL-33 treatment renders mast cells selectively unresponsive. Supernatant IL-6 levels are shown. (A) Peritoneal cell-derived mast cells were stimulated for 24 h with 100 ng/mL LPS or 1 ng/mL IL-33 (first treatment) then stimulated with 100 ng/mL LPS for an additional 24 h (second stimulation). (B) BMMCs were treated for 24 h with 100 ng/mL LPS (gray bars) or 1 ng/mL IL-33 (black bars) then treated with 100 ng/mL LPS, 1 ng/mL IL-33, or media alone as indicated. (C and D) BMMCs were treated for 24 h with 1 ng/mL IL-33 or media alone then stimulated with (C) 100 ng/mL LPS, 10 μg/mL PGN, 1 μg/mL poly(dA:dT), 5 μg/mL R837, or 5 μg/mL CL075 or (D) 100 ng/mL LPS or 10 ng/mL DNP for 24 h. (E) BMMCs were treated for 24 h with the indicated IL-33 concentrations then stimulated with 100 ng/mL LPS or media alone. (F) BMMCs were treated for 24 h with the indicated IL-33 concentrations. Data are shown as mean ± SEM of results pooled from 7 (B), 6 (C, D), and 5 (E, F) experiments, or are from one experiment representative of two experiments (A), each performed in duplicate. Significance compared with the first treatment with media alone, or to media alone (F). *p < 0.05, Wilcoxon test.

In order to further investigate the role of IL-33 in regulating the mast cell response we used BMMCs since these cells are easily obtained in large numbers. BMMCs were treated for 24 h with IL-33 or LPS and restimulated with LPS. We used IL-6 expression and secretion as an indication of cell activation together with the production of TNF-α and CCL2. IL-33 exposure rendered the cells significantly less responsive to LPS treatment, and the reduction in IL-6 production was comparable with that seen in endotoxin-tolerant BMMCs, which had been pretreated for 24 h with LPS (Fig. 1B). Although pretreatment with either LPS or IL-33 significantly reduced the response of the cells to IL-33, the cells still produced substantial amounts of IL-6 (Fig. 1B). IL-33 had a similar inhibitory effect on the secretion of TNF-α and CCL2 after LPS treatment (Supporting Information Fig. 1A and B) and the reduction in IL-6, CCL2, and TNF-α secretion correlated with an inhibition in mRNA expression (Supporting Information Fig. 1C and E). Thus, IL-33 regulates mast cell responsiveness.

To investigate the specificity of the IL-33 inhibition on mast cell activation, a panel of PAMPs was used to stimulate mast cells. No detectable IL-6 or CCL2 was released from BMMCs treated with CpG, flagellin, or poly(I:C) (data not shown). IL-6 and CCL2 secretion were induced by PGN, poly(dA:dT) (conjugated to a transfection reagent), the TLR7 agonist, R837, and CL075, a dual TLR7/TLR8 agonist (Fig. 1C and Supporting Information Fig. 1F). IL-33 pretreatment of BMMCs rendered the cells less responsive to PGN, had no effect on the response to poly(dA:dT), and enhanced the response to the TLR7/8 agonists (Fig. 1C and Supporting Information Fig. 1F). IL-33 pretreatment also enhanced the BMMCs response to stimulation via engagement of the FcεRI (Fig. 1D).

In conclusion, IL-33 potently rendered mast cells selectively tolerant to further stimulation with the bacterial cell wall products, LPS and PGN, but not cytoplasmic dsDNA, and enhanced the ability of the cells to respond to Ag via FcεRI and to viral RNA via TLR7 and TLR8.

Suboptimal levels of IL-33 induce LPS tolerance

Since IL-33 is known to stimulate mast cells [12-15], our finding that the cytokine also inhibits the activation of these cells upon stimulation with LPS and PGN is unexpected. Therefore we tested the IL-33 concentrations required to induce LPS tolerance in mast cells. The inhibition caused by IL-33 preexposure was concentration dependent, with significant effects observed with as little as 1 pg/mL (Fig. 1E). With a higher IL-33 concentration of 10 ng/mL, the pretreatment alone was sufficient to cause IL-6 secretion when the cells were incubated in media alone for the second 24 h treatment (Fig. 1E). It is interesting to note that the low concentrations of IL-33 able to induce LPS tolerance are subactivating and do not induce BMMCs to produce IL-6 (Fig. 1F).

ST2 is required for IL-33–induced LPS insensitivity

IL-33 signals via a receptor complex consisting of the IL-1 receptor accessory protein (IL1-RAcP) and ST2 [20, 21], and BMMCs were differentiated from St2−/− mice [22] to establish whether IL-33–induced LPS tolerance via this receptor complex. Treatment with IL-33–induced LPS unresponsiveness in WT BMMC controls but not in the St2−/− cells as measured by IL-6 production (Fig. 2A and B), suggesting that ST2 is required for IL-33 to exert this effect.

Figure 2.

ST2 is required for the IL-33–induced tolerance and mast cells from St2−/− animals are hyperactivated. (A) WT (B) St2−/− BMMCs were treated for 24 h with media alone, 100 ng/mL LPS or 1 ng/mL IL-33 then stimulated with 100 ng/mL LPS or media alone for 24 h. Data are shown as mean ± SEM and are pooled from two experiments each performed in triplicate on three independent BMMC cultures. (C) Intracellular IL-6 levels in mast cells from n = 13 WT or n = 14 St2−/− animals untreated or stimulated with 100 ng/mL LPS for 4 h. (D) CD11b expression on media-treated peritoneal mast cells from n = 15 WT or n = 16 St2−/− animals. (E) CD11b expression on peritoneal mast cells (n = 10) or from animals treated i.p. 10 μg LPS (n = 9). (F, G, and H). Surface expression of (F) CD28, (G) MHC class II, and (H) CD80 on peritoneal mast cells from WT or St2−/– animals (n = 5). Each symbol represents an individual animal and data shown are pooled from three experiments (C and D), two experiments (E) or one experiment (F to H). (A and B) significance compared with the first treatment with media alone (C to H) significance assessed by Mann–Whitney test *p < 0.05, **p < 0.01.

Since the SCF receptor, c-kit, has been shown to coregulate IL-33 signaling in mast cells [23] it may be involved in the IL-33–induced LPS unresponsiveness. BMMCs were differentiated from KitW-sh/W-sh mice to investigate this possibility. As described [23], KitW-sh/W-sh BMMCs did not express c-kit (Supporting Information Fig. 2A). IL-33 treatment of KitW-sh/W-sh BMMCs resulted in LPS unresponsiveness comparable with that observed in WT cells (Supporting Information Fig. 2B and C) demonstrating that c-kit is not required for this effect. IL-33–mediated tolerance is, therefore, dependent on ST2 yet independent of c-kit.

Mast cells from St2−/− animals are hyperactivated

The finding that subactivating levels of IL-33 induce LPS tolerance in mast cells raises the interesting possibility that IL-33 constitutively produced in vivo, in concentrations too low to activate mast cells, may act to prevent excessive mast cell activation. Since IL-33 acts via its receptor, ST2, to exert this effect (Fig. 2A and B), we analyzed peritoneal mast cells from ST2-deficient animals in order to investigate the relevance of homeostatic IL-33 in vivo.

We hypothesized that mast cells from St2−/− animals, deprived of the inhibitory IL-33/ST2 signaling pathway, would be hyperresponsive. Peritoneal cells were stained for intracellular IL-6 both in resting conditions and after LPS exposure in vitro. Mast cells were identified by their high expression of c-kit (Supporting Information Fig. 3). WT and St2−/− mast cells expressed similar levels of IL-6 in unstimulated conditions, however, and responded to LPS stimulation similarly (Fig. 2C).

Interestingly, a significantly higher level of CD11b was detected on the surface of mast cells freshly isolated from St2−/− animals (Fig. 2D). Since CD11b is upregulated on activated neutrophils and eosinophils, the increased CD11b expression on St2−/− mast cells may be indicative of mast cell activation. In order to determine if CD11b is a marker of mast cell activation, mice were treated i.p. with LPS and, indeed, CD11b was upregulated on the surface of mast cells (Fig. 2E). In addition, CD28, which has previously been shown to be upregulated on mast cells upon treatment with LPS [24], was significantly higher on the St2−/‒ mast cells than WT cells (Fig. 2F). Similarly, the activation markers MHC class II and CD80 were expressed at higher levels on ST2-deficient mast cells, although these differences did not reach statistical significance presumably due to the low numbers of animals studied (Fig. 2G and H).

Taken together these data indicate that although mast cells freshly isolated from St2−/− animals express similar levels of IL-6 and respond in a similar manner to LPS stimulation in vitro, they are hyperactivated, supporting the hypothesis that IL-33 encountered in vivo inhibits mast cell activation during immune-homeostasis.

IL-33–induced tolerance is not explained by known mechanisms of ET

Several mechanisms have been described to mediate ET [18, 25] and we investigated whether known pathways contribute to the IL-33–induced mast cell ET. Recently, TNF-α was shown to render macrophages unresponsive to LPS [26]. To examine the possibility that TNF-α released upon IL-33 stimulation in BMMCs could be the cause of LPS tolerance, BMMCs were treated with TNF-α for 24 h prior to stimulation with LPS. In contrast to macrophages however, TNF-α did not affect the responsiveness of mast cells to LPS (Supporting Information Fig. 4), suggesting that IL-33-elicited TNF-α has no role in the IL-33–induced tolerance.

The deubiquitinating enzyme, A20, has been implicated as a mediator of ET [27] but in BMMCs, IL-33–induced only transient A20 expression (Supporting Information Fig. 5A and B), suggesting that it may not be playing a role here. Tollip is another inhibitor of IL-1R signaling which has been implicated in ET [28, 29] but neither IL-33 nor LPS treatment affected Tollip expression in BMMCs (Supporting Information Fig. 5C).

In macrophages, reduced TLR4 surface levels and intracellular IRAK4 downregulation upon LPS treatment have been proposed to account for ET [18, 30]. In BMMCs, however, levels of TLR4 were unaffected by 24 h treatment with LPS or IL-33 (Fig. 3A), and the levels of MyD88, TIRAP, IRAK2, and IRAK4 were similarly unaltered by treatment with either agonist (Fig. 3B and E) although IL-33 stimulation caused IRAK2 modification (Fig. 3D), resulting in an apparently larger band as previously observed in LPS-treated macrophages [31]. The inhibitory form of MyD88, MyD88s [32], was not detected in any of the mast cell lysates (data not shown).

Figure 3.

The levels of TLR4, MyD88, TIRAP, IRAK2, and IRAK4 in BMMCs are unaffected by IL-33 or LPS treatment. BMMCs were treated for (A) 24 h or (B to E) the indicated times with 100 ng/mL LPS, 1 ng/mL IL-33, or media alone. (A) TLR4 levels were assessed by flow cytometry on intact (top) and permeabilized (bottom) cells. Shaded histogram represents isotype control staining, open histogram represents TLR4 staining. (B to E) Immunoblot analysis of whole cell lysates with antibodies against (B) MyD88, (C) TIRAP, (D) IRAK2, or (E) IRAK4. β-actin served as a loading control. Data shown are from one experiment representative of at least two experiments.

IRAK1 is degraded upon IL-33 stimulation of mast cells

Treatment with either IL-33 or LPS induced a significant reduction in IRAK1 protein levels in BMMCs that was observed after 2 h and was sustained for at least 24 h (Fig. 4A and B and Supporting Information Fig. 6A). The levels of IRAK1 appear to correlate with the responsiveness of the BMMCs to LPS (Supporting Information Fig. 6). IRAK1 is approximately 78 kDa and after activation becomes phosphorylated and ubiquitinated and may appear as a band of about 100 kDa [33]. After phosphorylation, IRAK1 is degraded via the proteasome [34], and, in agreement with published data [30], LPS treatment of BM-derived macrophages (BMMs) resulted in loss of IRAK1 protein in a similar manner to that observed in BMMCs (Fig. 4C). IL-33, however, had no effect on IRAK1 protein levels in BMMs (Fig. 4C).

Figure 4.

IRAK1 is degraded upon IL-33 stimulation. Immunoblot analysis of cell lysates with antibodies against IRAK1 or β-actin. (A and B) BMMCs or (C) BMMs were treated for the times indicated with 100 ng/ml LPS, 1 ng/mL IL-33, or media alone. (D) WT and St2−/− BMMCs were treated for 24 h with 100 ng/mL LPS or 1 ng/mL IL-33 as indicated. (E and F) BMMCs were stimulated for 24 h with the indicated IL-33 concentrations. (G and H) BMMCs were treated where indicated with 10 μM MG115 for 10 min then incubated for a further 10 min with media alone or 1 ng/mL IL-33. (A, C, D, E, and G). Representative blots from one experiment representative of at least three experiments are shown. (B, F, and H) The quantification relative to β-actin is shown as mean ± SEM of data pooled from six (B, H) and five (F) experiments (smear included in quantification of H). Significance compared with the samples treated with media alone for the relevant time, *p < 0.05, Wilcoxon test.

The downregulation of IRAK1 protein induced by IL-33 in BMMCs was dependent on ST2 expression (Fig. 4D), demonstrating that IL-33 is acting via its receptor to induce IRAK1 degradation. Loss of IRAK1 was observed upon treatment with pg/mL concentrations of IL-33 (Fig. 4E and F), the same concentrations that were effective at inducing ET (Fig. 1E). When cells were pretreated with the proteasome inhibitor MG115, the IL-33–induced loss of IRAK1 protein was significantly inhibited (Fig. 4G and H), suggesting that active degradation is responsible for the loss of IRAK1.

IRAK1 is required for LPS signaling in the mast cell

The molecular details of TLR4 signaling in mast cells are incompletely understood and in order to establish the role of IRAK1 and IRAK2, Irak1−/− and Irak2−/ BMMCs were generated. Irak1−/ and Irak2−/− BMMCs expressed similar levels of FcεRI, TLR4, ST2, and c-kit to WT cells (Supporting Information Fig. 7A and B) and were not defective in their expression of TIRAP or MyD88 (Supporting Information Fig. 7C and D). Irak1−/ BMMCs expressed WT levels of IRAK2, while Irak2−/ BMMCs expressed WT levels of IRAK1 (Supporting Information Fig. 7D and E). Irak1−/ and Irak2−/ BMMCs responded similarly to WT cells to stimulation through FcεRI or with PMA/I (Supporting Information Fig. 7F and G). Taken together, these results suggest that there is no fundamental defect in the Irak1−/ and Irak2−/ BMMCs.

The response of Irak1−/ BMMCs to LPS and PGN was dramatically inhibited (Fig. 5A and B). When Irak1−/ and Irak2−/ BMMCs were stimulated with the other PAMPs included in this study, Irak2−/ cells were less responsive to the TLR7 ligand, R837, than the WT cells (Fig. 5B). Irak2−/ cells were more sensitive to poly(dA:dT) and Irak1−/– BMMCs were more sensitive to the TLR7 agonist (Fig. 5B).

Figure 5.

IRAK1 is required for cell responsiveness to LPS and PGN. IL-6 levels in cell supernatants of WT (black bars), Irak1−/− (grey bars) or Irak2−/− (unfilled bars) BMMCs stimulated with (A) the indicated concentration of LPS, (B) 10 μg/mL PGN, 1 μg/mL poly(dA:dT), 5 μg/mL R837, 5 μg/mL CL075, (C) the indicated concentrations of IL-33 or media alone for 24 h are shown. (D) Irak1−/− BMMCs were treated for 24 h with 1 ng/mL IL-33 (filled bars) then stimulated with 100 ng/mL LPS as indicated. (E and F) IRAK1 expression in peritoneal mast cells isolated from WT or St2−/‒ animals. (E) A representative histogram of one WT (black line) and one St2−/‒ (grey line) animal is shown, and (F) data from n = 5 animals. Data are shown as mean ± SEM of data pooled from eight (A to C) and five (D) experiments. Statistical significance compared with the WT cells; **p < 0.01 Wilcoxon test.

Although the response of the Irak2/− BMMCs to LPS and PGN was comparable with that of the WT cells (Fig. 5A and B), they responded less well to IL-33 stimulation, particularly at higher concentrations (Fig. 5C). This was in contrast to the Irak1−/− cells, which seemed only to be defective in IL-33 signaling at lower cytokine concentrations (Fig. 5C). Therefore, in BMMCs IRAK2 is part of the ST2 signaling pathway but is redundant in TLR4 signaling.

The Irak1−/ BMMCs were defective in their response to LPS (Fig. 5A) and pretreatment of the Irak1−/ BMMCs with IL-33 did not further reduce the response of the cells to LPS (Fig. 5D) although the levels of IL-6 produced with or without IL-33 pretreatment were very low.

This finding that IRAK1 is absolutely required for TLR4 signaling in the mast cell may explain the mechanism behind the IL-33–induced tolerance observed here, since IL-33 potently induced IRAK1 degradation.

To further investigate the hypothesis that IL-33 produced under homeostatic conditions renders mast cells selectively responsive by causing the degradation of IRAK1, freshly isolated WT and St2−/− peritoneal mast cells were used. The IRAK1 levels detected in St2−/ mast cells were significantly higher than those detected in WT cells (Fig. 5E and F) suggesting that in the absence of ST2, IRAK1 is constitutively expressed at higher levels in mast cells in vivo.


Our results show that treatment of mast cells with subactivating levels of IL-33 renders the cells selectively insensitive to bacterial cell wall products. This IL-33–mediated tolerance is dependent on the IL-33 receptor ST2 and is mediated by a potent, rapid, and sustained IRAK1 degradation.

IL-33 at concentrations insufficient to cause mast cell activation causes significant LPS tolerance in mast cells and we propose that low concentrations of IL-33 produced constitutively in vivo during immune-homeostasis may act to prevent mast cells from reacting to commensal bacterial cell wall products (Fig. 6). It has been demonstrated that in addition to its potential role as an alarmin, IL-33 can be actively secreted by viable cells [35, 36] and pg/mL concentrations of IL-33 have been detected in the peritoneal cavity of healthy women [37], suggesting that the cytokine is present extracellularly during immune homeostasis. In physiological conditions, low levels of IL-33 may retain the mast cell in a selectively tolerant state: Insensitive to bacterial cell wall products yet able to respond to Ag and viral PAMP; thus preventing unwanted cell activation by extracellular commensal bacteria. In the event of pathogen encounter, the increased levels of IL-33 produced by damaged epithelial and endothelial cells [9, 10], and by activated inflammatory cells [11] would overcome the tolerant state and cause mast cell activation, allowing the cell to respond robustly when the host is challenged (Fig. 6).

Figure 6.

IL-33 acts during homeostasis to inhibit mast cell responses to LPS. We propose that during immune homeostasis low concentrations of IL-33 (red circles) are produced and act on mast cells to cause IRAK1 degradation. In the absence of IRAK1, mast cells are unable to respond to LPS, preventing them from responding to commensal bacteria (left). In the event of infection or damage, greater concentrations of IL-33 would be produced, activating the mast cells to secrete cytokines and chemokines (right).

In support this hypothesis, mast cells from St2−/− mice have a more activated phenotype as evidenced by increased CD11b, CD28, MHC class II, and CD80 expression. We demonstrate here that CD11b is upregulated upon treatment of animals with LPS, and it has been previously shown that CD28 is increased upon LPS treatment of mast cells [24]. In addition, mast cells from St2−/− mice display increased levels of IRAK1 expression, suggesting that in the absence of IL-33 signaling, the kinase accumulates in cells.

We had expected that mast cells freshly isolated from St2−/− animals would respond more than WT to LPS but this was not the case. Presumably, other regulatory mechanisms are at work which compensate for the lack of ST2 in these animals, preventing overreaction of mast cells to commensal bacteria.

A similar mechanism has been described to regulate the response of macrophages to LPS. The proinflammatory cytokine TNF-α inhibits the response of macrophages to LPS both in vitro and in vivo [26]. It appears that TNF-α may be important in inducing macrophage, but not mast cell, ET [26] while IL-33 acts to prevent mast cell responses to LPS, suggesting homeostatic roles for both cytokines.

One mechanism implicated in ET is the downregulation of a key component of the TLR signaling pathway [18] and IL-33 potently induces sustained IRAK1 degradation via the action of the proteasome. IL-33–induced IRAK1 degradation within 2 h and this correlated with the onset of LPS tolerance, implying that the loss of IRAK1 is responsible for the LPS tolerance observed.

The relative importance of IRAK1 and IRAK2 in the TLR4 signaling pathway seems to depend on the cell type: Irak2−/− macrophages are less able to respond to a variety of TLR ligands than Irak1−/ macrophages; but Irak1−/ and Irak2−/ mouse embryonic fibroblasts are equally defective in their response to LPS stimulation [19]. The roles of IRAK1 and IRAK2 in IL-1R/TLR signaling in the mast cell have not been previously investigated and it is shown here that the response of Irak1−/ BMMCs to LPS is dramatically reduced. This key difference in the requirement for IRAK1 in the TLR4 signaling pathway in mast cells in contrast to IRAK2 in macrophages is very interesting. Taken together with previous findings [16], these data suggest that some aspects of TLR/IL-1R signaling are mast cell specific as recently discussed [38].

Since BMMCs are dependent on IRAK1 for TLR4 signaling, LPS-mediated IRAK1 degradation may contribute the ET observed here and elsewhere [39] in mast cells. The degree of IRAK1 degradation induced by LPS was less than that observed with IL-33, however, while the ET was comparable, suggesting that other mechanisms may mediate ET in mast cells. One example of an alternative mechanism could be the upregulation of SOCS proteins, as has been previously suggested [39].

Irak1−/ BMMCs responded significantly less to stimulation with PGN which may explain why IL-33 treatment causes insensitivity to this PAMP. Although we cannot exclude the possibility that in addition to acting via IRAK1 degradation IL-33 may induce tolerance via other as yet undefined mechanisms, the fact that IL-33 was unable to further reduce the response of Irak1−/ BMMCs to LPS supports this hypothesis.

IRAK2 is expressed and functional in BMMCs as evidenced by the reduced response of Irak2−/ BMMCs to higher concentrations of IL-33. It seems that IRAK1 and IRAK2 are both involved in IL-33 signaling in mast cells, with IRAK1 playing a role when lower concentrations of the cytokine are encountered, while IRAK2 is central to the response to higher concentrations.

Mast cells express a constitutively high level of ST2 whereas the levels on the surface of macrophages are difficult to detect [5]. This difference in receptor expression could explain why mast cells are sensitive to low concentrations of IL-33 in the manner described here. It is interesting to note that IL-33 treatment of macrophages did not induce IRAK1 degradation, in contrast to its effects in mast cells, suggesting that the signaling pathways induced by this cytokine may be different in these two cell types.

IL-33 is a cytokine with wide-ranging effects [40]. Although generally considered to be proinflammatory [11], it has been shown to promote healing in the dextran sodium sulfate model of colitis [41]. In addition, in sepsis, administration of IL-33 results in a significant decrease in the systemic proinflammatory cytokines and promotes the survival of the animals by increasing neutrophil recruitment [42]. Furthermore, intranuclear IL-33 is known to inhibit NF-κB signaling [43], demonstrating a novel antiinflammatory role for the cytokine. Taken together with the data presented here, these findings should caution against the development of future therapeutics aimed at inhibiting IL-33 signaling, as there may be unwanted effects on immune-homeostasis.

Researchers are uncovering important roles for the tissues in regulating and directing the immune response [3]. The mechanism described here would enable the epithelium or endothelium to regulate the ability of mast cells to respond to bacteria by releasing IL-33. Epithelial and endothelial cells could educate the mast cell at distinct tissues and under different physiological conditions to respond appropriately to bacteria (Fig. 6), avoiding unnecessary immune activation but maintaining the ability to respond to pathogens. We therefore propose a hitherto unidentified role for IL-33 in maintaining immune homeostasis by retaining mast cells in a selectively unresponsive state.

Materials and methods


C57BL/6 mice were housed in full accordance with the Animal Scientific Procedures Act 1986 under Home Office approval at the University of Manchester and St2−/− animals [22] at the MRC laboratories in Cambridge. KitW-sh/W-sh and C57BL/6 controls were housed at the Research Centre Borstel and animal experiments were performed in accordance with institutional guidelines and were approved by the local authorities (Ministry of Agriculture, the Environment, and Rural Areas, Schleswig-Holstein). Irak1−/−, Irak2−/−, and WT control BM were obtained from Osaka University [19]. Animals were generated and housed with the approval of the Animal Research Committee of the Research Institute for Microbial Diseases (Osaka University).

Cell culture

BMMCs were generated by culturing BM cells in Iscove's modified Dulbecco medium (IMDM; PAA) with 10% heat-inactivated FBS (Biochrom), penicillin, streptomycin, L-glutamine (PAA), vitamins, sodium pyruvate, nonessential amino acids, and β-mercaptoethanol (Invitrogen) for 4 weeks with 10 ng/mL SCF and 5 ng/mL IL-3 (R&D Systems). Cells from C57BL/6 mice were >90% c-kit+ (CD117) ST2+ FcεRI+ as analyzed by flow cytometry. Cells from BALB/c mice were >80% c-kit+ FcεRI+. Cells were subsequently cultured at 0.5–1 × 106 cells/mL with media changed weekly, for a further 4 weeks.

BMMs were generated from C57BL/6 BM. Cells were cultured in RPMI supplemented with 10% FBS (Biochrom), nonessential amino acids, sodium pyruvate (Invitrogen), penicillin, streptomycin, L-glutamine (PAA) for 6 days with 200 ng/mL M-CSF (R&D Systems). Cultures were >90% F4/80+ CD11b+ as assessed by flow cytometry. Adherent macrophages were removed in ice-cold Versine (Invitrogen) with a cell scraper.

Peritoneal cell–derived mast cells were cultured as described [44]. Cells were collected from C57BL/6 mice after i.p. injection of 8 mL of 0.89% NaCl solution. Total peritoneal cells were cultured at a density of 1 × 106 cells/mL in RPMI 1640 (Invitrogen) supplemented with 10% FBS (Biochrom), 10 μg/mL gentamicin (PAA), 10 ng/mL IL-3, and 30 ng/mL SCF (R&D Systems). After 48 h, media containing nonadherent cells was removed and replaced by fresh culture medium. Cells were used in assays after 9 days of culture.

Peritoneal cell isolation

Peritoneal lavage was collected in PBS or 0.89% NaCl, with FBS. For experiments in which cells were activated, cells were resuspended at 1 × 106/mL in media and stimulated with LPS for 4 h: 3 μg/mL brefeldin A (eBioscience) was added to cell cultures 1 h into the stimulation.

Cell stimulation and ELISA

BMMCs were stimulated in media containing 1 ng/mL IL-3 with LPS from Escherichia coli 0127:B8, 100 ng/mL PMA with 1 μM ionomycin (PMA/I; Sigma-Aldrich), recombinant IL-33, TNF-α (R&D Systems), poly(I:C), flagellin, poly(dA:dT)/LyoVec, R837 and CL075 (InvivoGen). Phosphorothioate-modified CpG and control CpG sequences were taken from Ref. [45] and were as follows, CpG, 5′-GAG AAC GCT CGA CCT TCG AT-3′, CpG control, 5′-GAG AAG CCT GCA CCT TGC AT-3′ (Eurofins). Prior to treatment with human-DNP albumin, cells were incubated overnight with 200 ng/mL anti-dinitrophenyl IgE (SPE-7; Sigma-Aldrich). ELISA was performed on supernatants to determine IL-6, TNF-α, and CCL2 concentrations using antibodies and protein standards from R&D Systems according to their protocols.


Total RNA was isolated from cell pellets using the RNeasy systems (QIAGEN). cDNA was synthesized using the Transcriptor High Fidelity cDNA Synthesis Kit (Roche). Levels of TNF-α, CCL2, and β-actin were determined using the RealTime ready Universal ProbeLibrary and primers (Roche). For IL-6 the forward (5′-TCC TAC CCC AAC TTC CAA TCC TC-3′) and reverse (5′-TTG GAT GGT CTT GGT CCT TAG CC-3′) primers were used (Eurofins) with universal probe #6 (Roche). The FastStart Universal Probe Master (with ROX; Roche) was used with a StepOnePlus real-time PCR system (Applied Biosystems).

Western blotting

Cells were lysed in 1% NP-40 (Roche), 50 mM Tris-HCl buffer pH 8.0 (Fisher), 30 mM NaCl, 3x Halt protease and phosphatase inhibitor, and 5 mM EDTA (Pierce). Insoluble materials were removed by centrifugation. Primary antibodies against β-actin, IRAK1, IRAK4, MyD88, Tollip, A20 (Cell Signaling), IRAK2 (ProSci), and TIRAP (Abcam) were used. Anti-rabbit and anti-mouse secondary antibodies conjugated to HRP (Cell Signaling) were detected with ECL Prime, ELC Plus (GE Healthcare), or ECL (Pierce).

Flow cytometry

Cells were stained in 2% newborn calf serum (Invitrogen), 0.1% sodium azide (Fisons), 0.2 mM EDTA, in PBS (Invitrogen). Fc receptors were blocked with anti-CD16/CD32 (eBioscience). Antibodies against c-kit (CD117, clone 2B8; BD Biosciences and eBioscience), T1/ST2 (MD Biosciences), IRAK1 (Cell signaling), FcεRI, LAMP1 (CD107a), F4/80, TLR4/MD-2 complex (clone MTS410), CD11b, CD80, MHC class II, IL-6 and the 7-AAD viability stain (all from eBioscience) were used. Fluorescence was detected using an LSRII flow cytometer (BD Biosciences) and data analyzed using FlowJo software (version 7.6.1; TreeStar). For intracellular staining, fixation and permeabilization buffers (eBioscience) were used according to the manufacturer's protocols.

In vivo LPS challenge

Animals were treated i.p. with 10 μg LPS and peritoneal cells harvested after 3 h.

Statistical analysis and data presentation

Statistical analysis was performed by using SPSS (Version 16.0, IBM) unless otherwise stated.


We gratefully acknowledge the technical assistance of Bill Moser and Helen Jolin, and thank Prof. Richard Grencis, Prof. Werner Muller, Dr. Andrew McKenzie, and Dr. Anne Krug for kindly proving animals. We wish to thank Prof. Richard Grencis and Prof. Werner Muller for helpful discussions and are grateful to Prof. Ian Sabroe for critical reading of the manuscript. We are grateful to the University of Manchester for funding this study.

Conflict of interests

The authors declare no financial or commercial conflict of interest.


BM-derived macrophage


BM-derived mast cell


endotoxin tolerance


IL-1R associated kinase




phorbol myristate acetate