IL-33 Induces IL-13–Dependent Cutaneous Fibrosis

Fibrosis is characterized by the excess accumulation of extracellular matrix (ECM) components, including collagen (1, 2). The progressive replacement of parenchymal tissues with ECM is observed in fibrotic diseases, such as systemic sclerosis, idiopathic pulmonary fibrosis, and liver cirrhosis, leading to impaired organ function. Fibrosis is estimated to contribute to nearly 45% of deaths in the developed world. Although fibrosis is postulated to result from chronic irritation or damage to the affected organs, the cellular and molecular factors that sustain the fibrotic cascade remain poorly understood (1, 3–5).

IL-33 (IL-1F11, NF-HEV) is the most recently discovered member of the IL-1 cytokine family (6). IL-33 is constitutively expressed in barrier tissues, such as skin, where it is found preferentially localized to the nucleus of epithelial and endothelial cells (7–10). The receptor for IL-33 is composed of two subunits, IL-1RAcP (IL-1R3) and ST2 (IL-1R4) (6, 11). IL-1RAcP is widely expressed, whereas ST2 expression is restricted to cell types that include Th2 cells, eosinophils, basophils, invariant NKT cells, and mast cells (6, 12–16). Consistent with the expression of ST2 by Th2-associated cell types, in vivo systemic administration of recombinant IL-33 induces Th2 cytokine production, eosinophilia, and mucous hypersecretion in the lung and gut (6, 14). IL-33 is released by cells undergoing necrotic cell death and in this respect IL-33 is thought to function as damage-associated molecular pattern (DAMP) (17–19).

Recent studies have revealed an association between IL-33/ST2 and the development of fibrotic disorders, such as scleroderma and progressive systemic sclerosis (20, 21). However, the consequences of dysregulated IL-33 activity in skin have not been reported. To model chronic IL-33 release caused by sustained tissue damage, we repeatedly administered IL-33 s.c. and show that it induces ST2-dependent cutaneous fibrosis and inflammation. IL-33–induced fibrosis is associated with altered expression of ECM-modifying genes. Importantly, we demonstrate that IL-13 is a critical downstream mediator of IL-33–induced cutaneous fibrosis. In addition, we show that complete development of IL-33–induced fibrosis requires eosinophils and RAG-dependent lymphocytes. For the first time, our data reveal profibrotic properties of IL-33 and collectively suggest that IL-33 may function as an endogenous DAMP involved in the initiation of fibrotic disease.

B6/129.IL-1RAcP^−/−, B6.IL-1R1^−/−, B6.RAG^−/−, and 129S6.IL-13^−/− mice were housed in specific pathogen-free conditions at Schering-Plough Biopharma according to Institutional Animal Care and Use Committee guidelines (22–25). B6.ST2^−/− mice were a kind gift from the Neurath laboratory, University of Mainz, Mainz, Germany (26). BALB/c.∆dblGATA, WB/B6F1.cKit^w/v, B6.IL-4^−/−, and appropriate control mice were obtained from The Jackson Laboratory (Bar Harbor, ME) (27, 28). B6/129 mice were obtained from The Jackson Laboratory. C57BL/6 and 129S6 mice were obtained from Taconic Farms (Germantown, NY). Age- and sex-matched mice, 8–16 wk of age, were used for the experiments described.

Mice were injected s.c. daily for 7 d with 5 μg mouse serum albumin (MSA) or recombinant mIL-33. Recombinant IL-33 (aa 112–270) was produced in Escherichia coli by Aragen Biosciences (Morgan Hill, CA) and contained <0.5 U/ml endotoxin. Twenty-four hours after the final injection, the injection site was harvested for analysis. Experimental protocols were approved by the Schering-Plough Biopharma Institutional Animal Care and Use Committee.

RNA isolation was performed by standard techniques and gene expression was calculated using the Δ-ΔCt method (using the mean cycle threshold [Ct] value for ubiquitin and the gene of interest for each sample). Primers were obtained commercially from Applied Biosystems (Foster City, CA) or designed using Primer Express (PE Biosystems, Foster City, CA). The equation 1.8e (Ct ubiquitin − Ct gene of interest) × 10⁴ was used to obtain the normalized values.

Skin samples from the injection site were harvested and formalin fixed for 24 h. Samples were then paraffin embedded and 5 μm sections cut. Sections were stained with H&E, modified Masson’s trichrome, or astra blue and violet red according to the manufacturer’s directions (American MasterTech, Lodi, CA). To quantify mast cell and eosinophil skin infiltration, astra blue and violet red-stained sections were scanned using a Mirax Midi slide scanner (Zeiss, Jena, Germany). Mast cells were identified based on the characteristic astra blue staining of mast cell granules and the presence of a singular nucleus, whereas eosinophils were identified by violet red cytoplasmic staining colocalized with a bilobed nucleus. A minimum cross-sectional area of 0.5 mm² skin per mouse was randomly selected and surveyed to obtain manual cell counts. Images were obtained using a bright-field microscope (Olympus Model BX51, Olympus, Center Valley, PA) with attached digital camera (QImaging Model Retiga 2000R, QImaging, Surrey, British Columbia, Canada) and were white balanced using Adobe Photoshop elements (v.2.0, Adobe Systems, San Jose, CA).

Immunohistochemical studies were performed on formalin-fixed, paraffin-embedded tissue sections using a cross-reactive rabbit polyclonal Ab against anti-human CD3 (A0452; DakoCytomation, Carpenteria, CA, dilution 1:200) and a rat mAb against anti-mouse F4/80 (BM8; eBioscience, San Diego, CA, dilution 1:400). Paraffin-embedded tissues were sectioned at 5 μm thickness, deparaffinized, and quenched with 3% hydrogen peroxide for 10 min. Slides were heat retrieved with citrate buffer at pH 6.1 (Catalog S1699, DakoCytomation) for 4 min at 123°C using a Biocare Decloaker chamber, then cooled for 15 min, followed by a running tap water rinse. Slides were mounted on a Dako Autostainer and covered with fresh TBS to prevent drying of sections. Sections were then incubated with anti-CD3 or anti-F4/80 Abs at room temperature for 60 min and rinsed with TBS. CD3-stained sections were incubated for 30 min in Rabbit Envision-Plus (Catalog K4011, DakoCytomation). F4/80 stained sections were incubated for 30 min in Rat Probe and then Rat Polymer-HRP (Catalog RT517L, Biocare Medical, Concord, CA). Slides were rinsed with TBS and developed with DAB-Plus (DakoCytomation), counterstained in modified Mayer’s hematoxylin and blued in 0.3% ammonia water, followed by a tap water rinse. CD3⁺ and F4/80⁺ cells were counted as described previously.

Total soluble collagen levels in the skin were quantified using the sircol assay according to the manufacturer’s directions (Biocolor, Carrickfergus, U.K.). Briefly, a 6 mm dermal punch biopsy was isolated from the injection site and s.c. fat removed. Samples were homogenized in 3 ml 0.5 M acetic acid, supplemented with complete protease inhibitors (Roche, Indianapolis, IN). After overnight extraction at 4°C, the samples were spun down and 0.2 ml extract was assayed to determine collagen content. Results are reported as total micrograms of collagen per punch biopsy.

Bone marrow-derived eosinophils (bmEos) were generated as described (29). Briefly, bone marrow was isolated from C57BL/6 and ST2^−/− mice and single-cell suspensions prepared. Cells were plated at 10⁶/ml in RPMI 1640 containing 20% FBS (Irvine Scientific, Santa Ana, CA), 2 mM glutamine, 25 mM HEPES, 1× MEM nonessential amino acids, 1 mM sodium pyruvate (Life Technologies BRL, Rockville, MD), 50 μM β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO), 100 ng/ml stem cell factor, and 100 ng/ml FLT3-L (Peprotech). After 4 d, cells were replated in RPMI 1640 containing 20% FBS, 2 mM glutamine, 25 mM HEPES, 1× MEM nonessential amino acids, 1 mM sodium pyruvate, 50 μM β-mercaptoethanol, and 10 ng/ml rmIL-5. Cells were harvested 12 d after the start of the culture. For in vitro stimulations, 5 × 10⁵ cells/ml were either left untreated or stimulated with 10 ng/ml rmIL-33. Supernatants were harvested after 24 h. IL-13 (PeproTech, Rocky Hill, NJ) ELISAs were performed according to the manufacturer’s directions.

A total of 1 × 10⁵ bmEos were Fc-blocked with 5 μg/ml anti-CD16/32 Ab (2.4G2; BD Biosciences, San Jose, CA) for 15 min. Cells were subsequently washed and coincubated with anti–mSiglec-F-PE (E50-2440; BD Biosciences) and anti–mST2-FITC (DJ8; MD Biosciences, Zurich, Switzerland) or control Abs (BD Biosciences) for 30 min in PBS containing 2% BSA (Sigma-Aldrich). Samples were washed and data collected on a BD LSRII flow cytometer (BD Biosciences). Data were analyzed using FlowJo software version 7.2.4 (Tree Star, Ashland, OR).

Statistical analyses to compare collagen content in skin and cell numbers between experimental and control groups were performed using Student t tests, with p values ≤ 0.05 considered significant. Studies using TAQman analysis to compare gene expression levels between control and experimental groups were analyzed using the Mann-Whitney U test. p values ≤ 0.05 were considered significant. Statistical analyses were performed using GraphPad Prism version 4.02 for Windows (GraphPad Software, San Diego, CA).

To assess the consequences of dysregulated IL-33 release in skin, IL-33 or control protein (MSA) were injected s.c. everyday for 7 d. IL-33– and MSA-injected mice did not exhibit any discernible changes in behavior over the course of the injection series (e.g., pruritus) nor was any overt skin pathology observed. Nevertheless, histological analysis of skin from IL-33– but not MSA-injected mice revealed prominent inflammation and edema at the injection site and development of subcuticular fibrosis as evidenced by Masson’s trichrome staining (Fig. 1A). Soluble collagen levels were increased 4-fold in skin punch biopsies obtained from IL-33–injected mice (Fig. 1B). The majority of infiltrating leukocytes consisted of granulocytic leukocytes with dense clusters of these cells occasionally observed in s.c. adipose tissue (Fig. 1A, 1C). Histochemical staining revealed that the majority of infiltrating leukocytes were eosinophils and that s.c. injection of IL-33 caused significant accumulation of these cells compared with MSA-injected skin (Fig. 1C, 1D). Expression levels of eosinophil major basic protein in skin from IL-33– and control protein-treated mice yielded a similar pattern of results (data not shown). CD3⁺ mononuclear cells were also significantly increased in skin obtained from IL-33–injected mice, whereas mast cells and B220⁺ cells were present at similar numbers in skin sections obtained from mice injected with IL-33 or MSA (Fig. 1E, 1F and data not shown). No pathological changes were observed in skin sections adjacent to the injection site (data not shown).

To determine whether IL-33–induced inflammation is dependent on signaling through the IL-33Rs, ST2, and IL-1RAcP, we next injected ST2^−/− and IL-1RAcP^−/− mice s.c. with IL-33. Neither IL-1RAcP^−/− nor ST2^−/− mice exhibited histological signs of inflammation or subcuticular fibrosis after receiving IL-33 injections (Fig. 2A, 2B), and skin sections from IL-33– and MSA control-injected IL-1RAcP^−/− or ST2^−/− mice were indistinguishable by histology (data not shown). IL-1RAcP is a shared subunit of the IL-33R that also pairs with IL-1R1 to form the heterodimeric receptor for IL-1α and -1β (22). To determine whether signaling through IL-1R1 is required for the development of IL-33–induced cutaneous fibrosis and inflammation we treated IL-1R1^−/− mice s.c. with IL-33. IL-1R1^−/− mice injected with IL-33 developed inflammation and subcuticular fibrosis in a manner that was indistinguishable from control mice indicating that IL-1α and -1β signaling is not required for IL-33–induced skin pathology (Fig. 2C). Collectively, these results demonstrate that repeated s.c. administration of IL-33 induces an ST2/IL-1RAcP–dependent skin inflammation and fibrosis.

We next examined gene expression in skin obtained from IL-33–injected mice. Consistent with the ability of IL-33 to induce expression of Th2 cytokines, skin samples obtained from IL-33–injected mice had significantly elevated expression of IL-4, IL-5, and IL-13, whereas expression of the Th1- and Th17-associated cytokines, IFN-γ and IL-17, respectively, remained unchanged (Fig. 3A and data not shown). As repeated injection of IL-33–induced the development of subcuticular fibrosis locally, we also examined expression of a panel of ECM-associated genes (Supplemental Table I). IL-33–injected skin exhibited substantially increased expression of various components of the ECM, such as collagen VIa, collagen IIIa, and fibronectin 1 (Fig. 3B, Supplemental Table I). In addition, expression of ECM-modifying components, such as tissue inhibitor of metalloprotease (TIMP)-1, matrix metalloprotease (MMP)-12, and MMP-13 were also significantly elevated (Fig. 3B). Interestingly, IL-33 treatment also resulted in significantly reduced expression of several collagen isoforms, including collagens II, IV, XI, XIII, XVI, and XVIII (Supplemental Table I). Thus, these results highlight the novel observation that IL-33 can modulate expression of ECM-associated genes in the skin. TGF-β and IL-13 are two cytokines known to play a critical role in promoting fibrotic disease in multiple animal models (1). Skin isolated from IL-33–injected mice expressed levels of TGF-β that were similar to those observed in MSA-injected skin samples (Fig. 3C). Because expression of IL-13, but not TGF-β, was significantly increased in IL-33–injected skin samples these results suggested that IL-33–induced fibrosis may develop in an IL-13–dependent manner.

To determine whether IL-33–induced skin pathology requires IL-13, we injected IL-13^−/− mice with IL-33 s.c. for 7 d and then examined skin samples histologically and quantified collagen content. Skin sections prepared from IL-13^−/− mice injected with IL-33 revealed only minor trichrome staining in the subcutis (Fig. 4A) and the collagen content in skin obtained from IL-33–injected IL-13^−/− mice was found to be significantly reduced compared with wild-type (WT) animals treated with IL-33 (Fig. 4B). Strikingly, the soluble collagen content in skin obtained from IL-33–injected IL-13^−/− mice was similar to levels observed in control protein-treated animals suggesting that IL-13 acts as a critical mediator of IL-33–induced skin fibrosis. Expression levels of collagen VIa, collagen IIIa, MMP-12, MMP-13, and TIMP-1 were reduced in skin obtained from IL-33–injected IL-13^−/− mice compared with WT mice (Fig. 4C, 4D), whereas expression of fibronectin was similarly induced by IL-33 in WT and IL-13^−/− mice indicating that IL-33–dependent expression of fibronectin occurs via an IL-13–independent mechanism (Fig. 4E). Although local collagen deposition was reduced in skin from IL-33–injected IL-13^−/− mice, inflammatory infiltrates were still clearly present (Fig 4A). Comparing IL-33–injected WT and IL-13^−/− mice, a nonsignificant trend toward reduced eosinophil infiltrates was observed in the IL-13^−/− mice (Fig. 4F). IL-5 is known as a critical cytokine promoting eosinophil maturation and survival, and the chemokine RANTES (CCL5) is known as a key factor for recruitment of eosinophils via CCR3 signaling (30, 31). We therefore investigated mRNA levels of IL-5 and RANTES in skin of IL-33–injected WT and IL-13^−/− mice: no differential expression for IL-5 and RANTES was detected (Fig. 4G), suggesting that neither IL-5 nor RANTES expression was IL-13 dependent. In contrast, expression of other CCR3-dependent chemokines, such as CCL7 (MCP-3), CCL11 (eotaxin), and CCL24 (eotaxin-2), were reduced in IL-33–injected IL-13^−/− mice compared with WT mice (Supplemental Fig. 1). These data suggest that expression of CCL7, CCL11, and CCL24, but not CCL5 and IL-5 are regulated via IL-13–dependent pathways in this model.

As expected, IL-13 expression in skin samples from IL-13^−/− mice was undetectable, but we also noted that IL-4 expression was greatly reduced (Fig. 5A, 5B). The coding sequences for IL-4 and IL-13 are tightly linked in an 11 kb region of mouse chromosome 11 and previous studies have indicated that targeted deletion of IL-13 may inadvertently impact expression of IL-4 by disrupting regulation of IL-4 transcription (32). As it has also been reported that systemic IL-4 overexpression via transgene can promote collagen accumulation in skin, we assessed the requirement for IL-4 in the development of IL-33–induced skin pathology (33). H&E-stained skin sections prepared from IL-4^−/− mice injected with IL-33, exhibited inflammation that was grossly indistinguishable from WT mice injected with IL-33 (Fig. 5C). In addition, the collagen content in skin samples obtained from IL-4^−/− and WT mice injected with IL-33 was similar, indicating that IL-33–induced inflammation and fibrosis are not dependent on IL-4 (Fig. 5D). Collectively, these results indicate that IL-33–induced fibrosis develops via an IL-13–dependent and IL-4–independent mechanism.

The accumulation of eosinophils at the site of IL-33 injection suggested that this cell type may promote IL-33–induced fibrosis. To determine whether IL-33–induced fibrosis requires eosinophils, we injected ∆dblGATA mice with IL-33 (27). The eosinophil- deficient status of ∆dblGATA mice was confirmed after completing the IL-33 treatment regimen by measuring blood eosinophils using a hematologic analyzer (data not shown). Notably, the collagen content in skin obtained from ∆dblGATA mice injected with IL-33 was significantly reduced compared with WT mice (Fig. 6A). Interestingly, the collagen content in skin from IL-33–treated ∆dblGATA mice was also significantly increased compared with skin samples obtained from control (MSA)-injected mice (Fig. 6A). Thus, these results indicate that IL-33–induced fibrosis requires eosinophils for complete development and that additional cell types also contribute. H&E-stained skin sections prepared from ∆dblGATA and WT mice injected with IL-33 revealed that both strains had developed inflammation in subcuticular tissue (Fig. 6B). The predominant cell type present in skin obtained from IL-33–injected ∆dblGATA mice were mononuclear cells morphologically resembling monocytes. In fact, many of the infiltrating mononuclear cells in skin samples obtained from IL-33–treated ∆dblGATA mice expressed F4/80 (Fig. 6C). F4/80⁺ mononuclear cells were present in significantly increased numbers in skin samples obtained from IL-33–treated ∆dblGATA mice compared with MSA-treated mice (Fig. 6D). We also observed some F4/80⁺ monocyte-like cells in WT skin injected with IL-33; however, we were unable to quantify these cells because of abundant eosinophil infiltrates that also express F4/80 [data not shown and (34)]. Nevertheless, these data indicate that a monocyte-like cell type can accumulate in IL-33–induced cutaneous lesions.

To further dissect the mechanism of IL-13–dependent cutaneous fibrosis in this model, we next sought to determine whether eosinophils could be a source of IL-13 after stimulation with IL-33. To these ends, we prepared bmEos from WT and ST2^−/− mice (29). bmEos cultures prepared from WT and ST2^−/− mice had comparable growth kinetics and viability and yielded >94% Siglec-F⁺ cells in resultant cultures (Fig. 6E and data not shown). ST2 expression was consistently low but detectable on the surface of Siglec-F⁺ bmEos from WT mice and as expected ST2 expression was undetectable on the surface of bmEos prepared from ST2^−/− mice (Fig. 6E). Notably, WT bmEos secreted abundant IL-13 in response to stimulation with IL-33, whereas IL-13 production by IL-33–stimulated ST2^−/− bmEos was undetectable (Fig. 6F). Thus, bmEos produce IL-13 in response to stimulation with IL-33 in an ST2-dependent fashion. Consistent with the reduced fibrosis exhibited by IL-33–treated eosinophil-deficient mice, IL-13 expression levels were also decreased in IL-33–treated ∆dblGATA mice compared with WT mice treated with IL-33 (Fig. 6G). These results suggest that IL-13 production by IL-33–stimulated eosinophils may promote IL-33–induced cutaneous fibrosis.

CD3⁺ lymphocytes also accumulated at the site of IL-33 injection. To determine whether lymphocytes are required for the development of fibrosis, we injected IL-33 into RAG^−/− mice, which lack IL-33–responsive Th2 and iNKT cell populations as well as B cells and γδ T cells (24). Skin samples obtained from IL-33–treated RAG^−/− mice contained collagen levels intermediate between control protein-treated mice and IL-33–treated WT mice, indicating that RAG-dependent lymphocytes contribute to the development of IL-33–induced fibrosis (Fig. 7A). Consistent with the reduced fibrosis exhibited by IL-33–treated RAG-deficient mice, IL-13 expression levels were also decreased in IL-33  treated RAG^−/− mice compared with WT mice treated with IL-33 (Fig. 7C). Notably, eosinophil infiltrates were found in H&E-stained skin specimens from IL-33–injected RAG^−/− mice lending further support to the suggestion that eosinophils may serve as a source of IL-33–induced IL-13 in vivo (Fig. 7B).

Several reports have suggested the potential for IL-33 to stimulate mast cells (6, 35–39). To investigate the possible contributions of skin resident mast cells to IL-33–mediated subcuticular fibrosis, we injected mast cell-deficient cKit^w/v mice with IL-33 (40). No histological differences with regard to infiltrates were noted between H&E-stained skin sections prepared from cKit^w/v and WT control mice injected with IL-33, indicating that mast cells or mast cell activation are not required for IL-33–induced skin pathology (Supplemental Fig. 2A). Similarly, no overt difference in collagen deposits was revealed by Masson’s trichrome staining (Supplemental Fig. 2A). This result is consistent with the observation that s.c. IL-33 injection of WT mice does not result in a local increase in mast cell numbers (Supplemental Fig. 2B).

Fibrosis is thought to be a consequence of chronic tissue irritation or damage (1, 3–5). However, the cellular and molecular factors that promote the accumulation of ECM in fibrotic disease remain poorly defined. IL-33 is a recently described IL-1 family member that is constitutively expressed in barrier tissues, such as skin. IL-33 is released on necrotic cell death and therefore has been suggested to function as DAMP (17–19). As IL-33 has been associated with fibrotic disease, we examined the consequences of dysregulated IL-33 signaling in the skin by repeated s.c. administration of IL-33 (7–9, 20, 41, 42). We show that IL-33 induces cutaneous fibrosis and intense inflammation that is associated with large numbers of infiltrating eosinophils, CD3⁺ cells and F4/80⁺ myeloid cells. IL-33–induced fibrosis develops through an IL-13–dependent mechanism that requires both eosinophils and RAG-dependent lymphocytes. We also show that IL-33– stimulated bmEos produce IL-13, thus highlighting a novel cellular source of IL-33–induced IL-13 production. Collectively, these studies identify IL-33 as a novel factor sufficient to induce cutaneous fibrosis.

Injection of IL-33 s.c. resulted in the development of an ST2-dependent fibrotic lesion, which was evidenced by trichrome staining in subcuticular tissue and increased collagen content in dermal punches. The pathologic changes induced by IL-33 were associated with altered expression of ECM genes in the collagen, MMP, TIMP, and bone morphogenetic protein families. In particular, the expression levels of collagen VIa and collagen IIIa were significantly increased in skin samples obtained from IL-33–injected mice suggesting that these collagen subtypes may contribute to the increased collagen deposition in IL-33–treated skin. TIMP-1 and TIMP-2 expression were also increased by IL-33 treatment. TIMP family members can inhibit MMP-mediated collagenase activity, which may promote IL-33–induced collagen accumulation by decreasing collagen turnover (43). Our findings indicate that IL-33 regulates the expression of these genes through an IL-13–dependent pathway. IL-13 has been shown to directly modulate expression of collagen VIa, collagen III, and TIMP-1 in previous studies (44, 45). Interestingly, fibronectin-1 expression was increased independently of IL-13 in IL-33–injected skin samples, suggesting that expression of some ECM components may be regulated directly through IL-33– or other IL-13–independent downstream mediators. Overall, our data highlight IL-33–mediated alterations in the expression levels of various collagen subtypes as well as enzymes known to affect ECM remodeling. It should be noted that our studies use a recombinant truncated form of IL-33 (aa 112–270) and we cannot formally exclude the possibility that full length IL-33 (aa 1–270) may have differential effects in vivo. However, after production of both IL-33 forms via in vitro transcription and translation, we have been unable to detect differences in the in vitro assays (unpublished observations).

IL-13 is a profibrotic cytokine that is sufficient for the induction of fibrosis in skin and lung (46, 47). In addition, IL-13 is required for the development of fibrosis in some animal models (48–51). The factors that regulate IL-13 expression in fibrotic disease remain poorly understood. In this study, we show that IL-33–induced fibrosis requires IL-13 but not IL-4. IL-4 can signal through the type I IL-4 receptor, whereas both IL-4 and IL-13 can signal through the type II IL-4 receptor (52). Our results suggest that the type I IL-4 receptor is not required for IL-33–induced fibrosis. IL-13 can signal through the type II IL-4 receptor and IL-13Rα2, which both can contribute to the development of fibrosis (53–55). Future studies will examine the requirement for specific IL-13 receptors in the development of IL-33–induced fibrosis.

In addition to IL-13, TGF-β is another cytokine with profibrotic properties in several organs, including skin (56, 57). Although our data show that mRNA expression of TGF-β is not modulated in IL-33–injected skin (Fig. 3C), our results do not exclude a role for its involvement in IL-33–induced cutaneous fibrosis as TGF-β activity is regulated at multiple posttranscriptional levels and our assays have not exhaustively examined TGF-β activity (58, 59). Interestingly, in mice that overexpress IL-13 in lung tissue, blocking TGF-β activity partially inhibited the development of fibrosis indicating that TGF-β expression can be induced downstream of IL-13 (60). Because we observe no evidence of fibrosis in IL-13–deficient mice injected with IL-33, our results argue that if TGF-β contributes to IL-33–induced cutaneous fibrosis then its activity is likely restricted to effects downstream of IL-13.

The full development of IL-33–induced cutaneous fibrosis requires both eosinophils and RAG-dependent lymphocytes. Eosinophils have been shown to contribute to the development of fibrosis in transgenic mice that overexpress IL-13 in the lung (61). Importantly, human eosinophils secrete IL-13 in response to stimulation with IL-5 and produce IL-8 after stimulation with IL-33 in vitro (15, 62). We demonstrate that murine bmEos express ST2 and secrete IL-13 in response to stimulation with IL-33. Whether murine eosinophils respond to IL-33 in vivo remains to be examined, nevertheless these results suggest that eosinophils may promote the development of IL-33–induced fibrosis by producing IL-13. Transgenic overexpression of IL-13 by keratinocytes is sufficient to induce the accumulation of eosinophils in skin (46), suggesting that IL-33–induced production of IL-13 by eosinophils may contribute to a positive feedback loop for eosinophil recruitment to sites of IL-33 release through IL-13–dependent expression of CCL7, CCL11, and CCL24 (Supplemental Fig. 1).

In addition to eosinophils, RAG-dependent lymphocytes contribute to driving IL-33–induced fibrosis. Th2 cells and iNKT cells are both CD3⁺ populations that express ST2 and produce IL-13 in response to stimulation with IL-33 (6, 12, 16). Staining for CD3 in fibrotic lesions revealed significant accumulation of these cells in IL-33–treated skin. By contrast, B220⁺ B cells, which are another RAG-dependent cell type, were observed infrequently in IL-33–induced fibrotic lesions and their numbers were not modulated in response to injection with IL-33. We also have not detected surface expression of ST2 on B cells by flow cytometry (data not shown). γδ T cells are a CD3⁺ RAG-dependent cell type that comprises a small percentage of total lymphocytes. Vγ3 Vδ1 T cells are a specialized subset of γδ T cells, termed “dendritic epidermal T cells,” that are skin-resident and have been shown to contribute to wound repair (63, 64). Expression of ST2 by γδ T cell subsets has not been reported, thus we cannot formally exclude a potential contribution of these cells to IL-33–induced fibrosis. These data suggest that Th2, iNKT, or possibly γδ T cells may also contribute to the development of IL-33–induced fibrosis.

F4/80⁺ monocyte-like cells also infiltrated IL-33–induced fibrotic lesions. Macrophages have been reported to express ST2 and to respond to IL-33 in vitro (65, 66). In addition, monocytes can be differentiated in the presence of IL-4/IL-13 in vitro to become alternatively activated M2 macrophages, which contribute to the tissue remodeling. In vivo, macrophage infiltration is observed in different fibrotic disease models (67–72). Future work will characterize the F4/80⁺ infiltrates in IL-33–induced fibrotic lesions to determine whether these cells are directly responsive to IL-33.

Clinical studies have identified a positive correlation with IL-33 and/or IL-33R and several fibrotic diseases. Sera obtained from systemic sclerosis patients and idiopathic pulmonary fibrosis patients exhibiting acute exacerbation contained elevated levels of sST2, the endogenous soluble receptor antagonist of IL-33 (20, 41). Moreover, abundant staining for ST2 is observed in sclerotic lesions, whereas IL-33 protein exhibits marked downregulation in early sclerotic skin lesions (21). In addition, both IL-33 and ST2 were expressed at significantly elevated levels in hepatic fibrotic lesions obtained from patients (42). Collectively, the data implicate the IL-33/IL-33R pathway as a potential mediator of pathogenesis in fibrotic diseases. Future studies will examine the impact of neutralizing endogenous IL-33 on the development of fibrosis.

We thank S. Jungers and F. Shen for technical assistance with flow cytometry and hematologic analysis. The many helpful discussions with E. Bowman, D. Cua, C. Tato and M. Kleinschek were much appreciated throughout these studies.

Disclosures Schering-Plough Biopharma (formerly DNAX) is fully funded by the Schering-Plough Corporation. The authors have no further conflicting financial interests.