Subcutaneous Allergic Sensitization to Protease Allergen Is Dependent on Mast Cells but Not IL-33: Distinct Mechanisms between Subcutaneous and Intranasal Routes

Various factors found in the environment have been reported as triggers of allergic sensitization, which is characterized by Th2-dominant immune response and induction of allergen-specific IgE (1, 2). House dust mites (HDM), insects, pollen, fungi, bacterial components in house dust, animal dander, and enzymes used in the food industry are examples of allergenic environmental factors. One of the important properties of these allergenic environmental factors is protease activity (1–3). The protease activity of HDM allergens has been shown to cause barrier dysfunction, to induce the production of proinflammatory cytokines in epithelial cells and keratinocytes, to cleave various molecules, to modulate functions of various cell types, and to induce Th2 responses and IgE production (3–10). HDM group 1 allergens, Der f 1 and Der p 1, and papain, a papaya fruit–derived occupational allergen, belong to the same family of cysteine proteases, family C1 of clan CA (5, 11, 12). Protease activity of papain (10, 13) and recombinant Der f 1 (rDer f 1) (10) is essential to airway eosinophilia and serum IgE/IgG1 responses induced by inhalation of the protease allergens. In recent studies, papain has been used as a model protease allergen mimicking those contained in allergen sources (10, 13–16).

Protease allergens stimulate various cell types (3, 4, 17). Mast cell deficiency did not reduce the IgE/IgG1 responses and allergic airway inflammation induced by papain (10). Some murine models suggested contribution of basophils in papain- or HDM-induced allergic responses in vivo (15, 18, 19), although still controversial (10, 18, 20).

An epithelial cell–derived IL-1 family cytokine, IL-33, is implicated in allergic and type 2 immune responses in various models (10, 21–27). IL-33–mediated innate immunity is critical for the lung eosinophilia induced by high-dose, 3-day inhalation of papain (22). Serum Ab production and lung eosinophilia in another model of lower-dose papain inhalation for a longer period than a week are dependent on IL-33 and protease activity, in which the cooperation of IL-33–mediated innate responses and adaptive immune cells is required for robust eosinophilic lung inflammation (10). Halim et al. (28, 29) revealed that the IL-33–group 2 innate lymphoid cell (ILC2) axis critically contributes to the innate-type (22, 28) and acquired-type lung eosinophilia and initiation of Th2 differentiation and Ag-specific IgE/IgG1 production (10, 29). In contrast, the roles of IL-33 and IL-33–responsive ILC2s in response to cutaneous administration of protease allergen are unknown, although several lines of evidence suggest the importance of IL-33 and/or ILC2s in allergic skin diseases, such as atopic dermatitis (AD), and the corresponding animal models (27, 30–33).

“Atopic march” is a term defined as the natural progression of allergic diseases, in which development of atopic dermatitis predates development of food allergy, asthma, or allergic rhinitis, suggesting that skin, especially skin with defective barrier function or eczema, is the initial site of allergic sensitization in the natural exposure to environmental allergens (34–36). In the current study, we investigated whether mast cell deficiency and IL-33 deficiency affect skin inflammation and serum Ab responses induced by s.c. injection of papain into ear lobes. Furthermore, we examined the contribution of the s.c. presensitization to papain and IL-33 in another model, in which mice s.c. administered papain to induce skin inflammation were then exposed to a threshold dose of papain via airway to induce allergic airway inflammation.

Seven- to ten-week-old female C57BL/6N and BALB/c mice purchased from Charles River Japan (Yokohama, Japan), WBB6F1-^+/+ and WBB6F1-W/W^v mice purchased from Japan SLC (Shizuoka, Japan), and IL-33^−/− mice (C57BL/6N background) (22) were maintained in a specific pathogen-free animal facility at Juntendo University and used in accordance with the guidelines of the institutional committee on animal experiments.

Papain was purchased from Calbiochem (San Diego, CA). To prepare papain treated with E64 (Peptide Institute, Osaka, Japan), similar to our previous studies (10, 16), papain was reacted with an excess of E64 after the addition of L-cysteine and, after the reaction, L-cysteine and unbound free E64 were removed by dialysis (E64-papain). Reductions in the catalytic cysteine residue of papain-like cysteine proteases by the addition of a reducing reagent such as L-cysteine are necessary for E64 to bind covalently and irreversibly to papain. In some experiments (Figs. 2, 8), papain, which was incubated similarly to E64-treated papain (but without the addition of E64), dialyzed, and reoxidized, was prepared and used for comparisons with E64-papain. The Ags were administered to mice without being treated with reducing reagents just before the administration.

Mice were lightly anesthetized with an i.p. injection of pentobarbital (Somnopentyl; Kyoritsu Pharmaceutical, Tokyo, Japan), and the indicated amount of papain or E64-papain (10) dissolved in 10 μl PBS was s.c. injected into ear skin. Ear thickness was measured from day 0 until day 14 using dial thickness gauge (G-1A; Ozaki, Tokyo, Japan). Serum samples were collected on days 9–14. Specimens of ear skin tissue, six ear lobes from three mice at each time point, were collected (on day 14, unless otherwise described), fixed with 20% formalin neutral buffered solution, and embedded in paraffin. Paraffin-embedded sections were stained with H&E or Giemsa staining for histological analyses. Some specimens of ear skin tissue were used to extract total RNA for gene expression analyses.

Photographs of the papain-injected sites in the Giemsa-stained paraffin sections were taken under microscopy. Each photograph was separated into several 0.01-mm² (100 μm × 100 μm) areas, and eosinophils within each 0.01-mm² area were counted. The total eosinophil number in one or more photographs was divided by the total counted area (unit: mm²) in the same photograph(s) and indicated as eosinophils/mm². These histological analyses were performed by observers, including one who knows the experimental treatment and additional one or two who were blind to the experimental treatment.

Bone marrow–derived mast cells (BMMCs) were generated by culturing bone marrow cells from WBB6F1-^+/+ mice in RPMI 1640 media supplemented with 10% heat-inactivated FCS, 0.1 mM 2-ME, 0.01 mM MEM–nonessential amino acids, 100 U/ml penicillin, 0.1 mg/ml streptomycin, 10 ng/ml murine IL-3, and 10 ng/ml murine stem cell factor. After 5–8 wk of culture, cells suspended in saline were s.c. injected (1 × 10⁶ cells/20 μl saline/ear) to both ears of W/W^v mice. Then 7.5 wk after the s.c. injection of BMMCs, mice were treated s.c. with papain (10 μg/10 μl PBS/ear, day 0) in both ears.

Cervical lymph nodes were collected 10 d after the s.c. injection of PBS-, papain-, or E64-papain. Single-cell suspensions (2.5 × 10⁶ cells/ml) were prepared from the draining lymph nodes (DLNs) and restimulated with E64-papain (0, 30, or 100 μg/ml) in 96-well cell culture plates (200 μl/well) for 72 h, and culture supernatants were collected for the measurement of cytokines.

After restimulation for 72 h, culture supernatants were recovered by centrifugation at 310 × g for 5 min. Cytokine concentrations were measured with ELISA kits (Quantikine or DuoSet; R&D Systems, Minneapolis, MN). Lower detection limits were 31.25 pg/ml for IL-4 and IL-17A, 125 pg/ml for IL-13, and 62.5 pg/ml for IL-5 and IFN-γ.

Mice were lightly anesthetized with an i.p. injection of pentobarbital and allowed to inhale 40 μl papain following alternate application to both nares with a pipette. The intranasal (i.n.) administration of papain was performed twice (days 14 and 21). Four days after the last i.n. administration, sera and bronchial alveolar lavage (BAL) fluid cells were collected.

Four days after the last i.n. administration, mice were terminally anesthetized, the tracheas were cannulated, and internal airspaces were lavaged with 1000 μl PBS with 5% FCS. Fluids were centrifuged at 1200 × g, and the pellets were recovered for cellular analysis. Specimens were prepared on glass slides by Cytospin 4 (Thermo Shandon, Cheshire, U.K.), followed by Diff-Quick (Sysmex, Kobe, Japan) staining. Differential cell counts were performed with a minimum of 200 cells.

Serum total IgE was measured by a sandwich ELISA, as described previously (10). Papain-specific Abs were detected on plates coated with papain (30 μg/ml for Figs. 1, 2 and 5 μg/ml for Figs. 4–8) and blocked with BlockAce (Snow Bland, Sapporo, Japan), and the plates were developed with HRP-conjugated Abs specific to the murine IgE (biotin-conjugated anti-murine IgE plus HRP-conjugated streptavidin for Figs. 4–8) and IgG subclasses, as described previously (10, 37). Sera and detection Abs were diluted with Solution 1 and 2 of CanGetSignal (TOYOBO, Osaka, Japan), respectively. Serum dilutions were 1/200 (1/40 for Figs. 6, 7) and 1/5,000 (1/10,000 for Fig. 1) for detecting allergen-specific IgE and IgG1, respectively. Serum samples were serially diluted from 1/10 until 1/10⁷ for titration. For detecting papain-specific IgE, incubation with serum or a detection Ab was for 15 h at 4°C or for 2 h at room temperature, respectively. For detecting papain-specific IgGs, incubation with serum or a detection Ab was for 1.5 h at 37°C.

Ear lobes were excised and homogenized in QIAzol Lysis Reagent (Qiagen, Clifton Hill, Australia) by using TissueLyser II (Qiagen). Total RNA was extracted from the homogenized skin tissue specimens by using a RNeasy Plus Micro Kit (Qiagen). First-strand cDNA was synthesized from total RNA by using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). Real-time quantitative PCR was performed with the TaqMan method using an ABI 7500 system (Applied Biosystems). The mRNA levels of the target gene were normalized to the gene expression levels of GAPDH and are shown as relative expression levels to the control group.

One-way ANOVA with the Tukey post hoc test, unpaired t test, or Mann–Whitney U test was used as indicated in the figure legends. A p value <0.05 was regarded as statistically significant.

We previously reported the cysteine protease-dependent induction of serum IgE/IgG1 production and allergic airway inflammation by repeated inhalation of papain (10). In the current study, we tested the induction of skin inflammation by s.c. administered papain (Fig. 1). The s.c. injection of 10–50 μg/10 μl/ear papain resulted in ear swelling that started 1 h after the injection and lasted at least 14 d (Fig. 1A, 1C) and induced production of total IgE and papain-specific IgG1 in the serum (Fig. 1B, 1D). Injection of 50 μg papain resulted in the formation of scabs after the initial erythematous swelling subsided, whereas injection of 10 μg papain caused less severe or no scab formation (data not shown). The first histological manifestation was hemorrhage in the dermis, which was observed 1 h after 10 μg papain injection (Fig. 1E, 1 h). Moderate cellular infiltration including lymphocytes and eosinophils was observed on day 1 (Fig. 1E, day 1). Massive eosinophilic infiltration was found on day 10, after the initial moderate cellular infiltration subsided on day 3 (Fig. 1E, days 10 and 3).

To test the contribution of the cysteine protease activity of s.c. administered papain to the skin inflammation and Ab responses, we injected papain that had been treated with a cysteine protease-specific inhibitor, E64. The treatment of papain with E64 significantly reduced the papain-induced ear swelling (Fig. 2A) and production of total IgE and papain-specific IgE and IgG1 (Fig. 2B).

Tang et al. (18) reported that s.c. injection of papain (50 μg) leads to the expression of thymic stromal lymphopoietin (TSLP), a Th2-inducing cytokine (38, 39), in the ear skin. We confirmed the TSLP mRNA expression in the s.c. papain (10 μg/ear)-injected ear tissues, which was induced 4 h after the injection and lasted for at least 72 h (Fig. 2C). We analyzed gene expression of OX40L, which is induced in TSLP-stimulated dendritic cells to differentiate naive T cells to inflammatory Th2 cells (38), in the ear skin tissues. The elevated expression of OX40L mRNA was observed from 4 h and lasted for at least 7 d after the s.c. papain injection (Fig. 2C). The elevations of TSLP and OX40L mRNA expression were not observed in E64-papain–injected mice (Fig. 2C).

The results showed that the cysteine protease activity of papain induces the skin responses associated with upregulation of TSLP and OX40L gene expression in the skin and Ab responses.

We examined the cytokine production by DLN cells from papain-injected mice (Fig. 3). We restimulated DLN cells with E64-papain, but not intact papain to exclude the potential activation of T cells via the protease activity in vitro. When restimulated, DLN cells from papain-injected mice produced IL-4, IL-5, and IL-13 (Fig. 3A–C). A low concentration of IFN-γ was produced by DLN cells from papain-injected mice (Fig. 3D). The Th2/Th1 cytokine production in response to in vitro restimulation was absent or markedly reduced in the DLN cells from E64-papain–injected mice (Fig. 3). The results demonstrated that the protease activity of s.c. administered papain was necessary for the Th2-dominant cytokine production profile in the DLNs.

Next, we examined the contribution of mast cells to papain-induced skin inflammation and serum Ab responses (Fig. 4). Ear swelling was equally induced by s.c. injected papain both in mast cell–sufficient WBB6F1-^+/+ (wild-type) and mast cell–deficient WBB6F1-W/W^v (W/W^v) mice during the first week after injection (Fig. 4A). The swelling of papain-injected ears was less in W/W^v mice than wild-type mice from days 10 to 15 (Fig. 4A). Although serum levels of total IgE were not affected by the mast cell deficiency (Fig. 4B), reduced production of papain-specific IgE (Fig. 4C) and IgG1 (Fig. 4D) was observed in W/W^v mice. Upon restimulation with E64-papain in vitro, DLN cells from papain-injected W/W^v mice showed significantly lower IL-4 production and higher IFN-γ production compared with DLN cells from papain-injected wild-type mice (Fig. 4E). Reduced infiltration of eosinophils into dermis was also observed in W/W^v mice (Fig. 5F, 5G).

Then, we reconstituted mast cells in the dermal tissue of W/W^v mice by s.c. injection of BMMCs to the ear lobes and tested the effect of papain injection on the serum Ab responses and eosinophil infiltration (Fig. 5). Whereas serum levels of total IgE were similar among the papain-injected wild-type, W/W^v, and BMMC-reconstituted W/W^v mice (Fig. 5A), the reduction of papain-specific IgE (Fig. 5B, 5D), IgG1 (Fig. 5C, 5E), and eosinophil infiltration (Fig. 5F, 5G) in W/W^v mice was restored by the reconstitution of mast cells in the dermal tissue. These results showed that mast cells contribute profoundly to the induction of papain-specific Abs and eosinophil infiltration in the papain-induced skin inflammation model.

As precedent studies demonstrated the contribution of IL-33 to allergic airway inflammation (10, 22) and serum IgE/IgG1 responses induced by papain inhalation (10), we next examined the involvement of IL-33 in the skin inflammation and Ab responses induced by s.c. papain administration (Fig. 6). Total IgE and papain-specific IgE and IgG1 were induced equivalently between wild-type and IL-33^−/− mice on day 12 after the s.c. papain administration (Fig. 6A). There were no differences between wild-type and IL-33^−/− mice in the swelling of papain-injected ears, cellular infiltration to the dermis, or epidermal hyperplasia (Fig. 6B, 6C). Thus, IL-33 is not required for the skin inflammation or allergic sensitization to papain that penetrated the epidermal barrier.

The critical contribution of IL-33 to the inhaled papain-induced allergic airway inflammation models has been reported in a high-dose, 3-d inhalation model (100 μg/20 μl per animal; inhalations at days 0, 1, and 2; and BAL at day 3) (22) and a lower-dose, longer inhalation model (10–30 μg/40 μl per animal; inhalations at days 0 and 7; and BAL at day 10 or 11) (10). In the current study, we compared the requirement of IL-33 for the induction of serum IgE/IgG1 and lung eosinophilia by inhalation of a threshold dose of papain (10 μg/40 μl per animal) (10) between mice with and without allergic sensitization by s.c. papain administration prior to the i.n. challenges (Fig. 7).

In wild-type mice without s.c. sensitization to papain, inhalation (at days 14 and 21) of papain resulted in elevated serum levels of papain-specific IgE/IgG1 (Fig. 7A) and infiltration of eosinophils to the lung (Fig. 7B), similar to our previous data (10). In IL-33^−/− mice without s.c. sensitization, inhalation of the threshold dose of papain did not induce the responses. In the s.c. sensitized wild-type mice, airway exposure to papain induced more robust Ab responses and lung eosinophilia than in mice without s.c. papain injection. The s.c. sensitized IL-33^−/− mice showed lower, but still significant lung eosinophilia upon airway exposure to papain (Fig. 7B). The airway exposure to papain boosted total IgE and papain-specific IgE in the s.c. sensitized wild-type, but not IL-33^−/− mice (Fig. 7A compared with Fig. 6A). Papain-specific IgG1 was boosted in both of the s.c. sensitized wild-type and IL-33^−/− mice. Upon the airway exposure to PBS, no lung eosinophilia was observed in the s.c. sensitized mice (data not shown).

We examined dependency of the responses upon the i.n. challenges in s.c. sensitized wild-type mice on the protease activity of inhaled papain (Fig. 8). Prior to the i.n. challenges, levels of papain-specific IgE/IgG1 were measured to confirm that the s.c. sensitization was successful (Fig. 8A). Lung eosinophilia and boost of papain-specific IgE after the i.n. challenges were partially dependent on the protease activity of papain used for the i.n. challenges in the s.c. sensitized mice (Fig. 8B, 8C).

Upon i.n. challenge, mast cell–deficient mice showed weaker lung eosinophilia with lower incidence (Supplemental Fig. 1C). Ab responses after i.n. challenge were less affected by the mast cell deficiency compared with the lung eosinophilia (Supplemental Fig. 1B).

Environmental proteases induce disruption of epithelial and epidermal barrier functions (3, 40–42), production of proinflammatory cytokines and chemokines in epithelial cells (17, 43) and keratinocytes (44, 45), cleavage of cell surface molecules, and modulation of functions of various cell types (4). In the current study, we established and characterized a new murine model associated with skin inflammation and IgE/IgG1 responses induced by s.c. administration of a model protease allergen, papain, and another model with increased susceptibility of allergic airway inflammation to the allergen inhaled via airway subsequently after the s.c. presensitization. Single s.c. administration of papain (10 μg/ear) was considered to be a nearly minimum dose that can induce swelling of ear skin, which is accompanied by increased epidermal thickness, infiltration of eosinophils into the dermis, and elevation of serum IgE/IgG1 levels with a minimum tissue damage (Fig. 1). The protease activity of papain is indispensable for the allergic skin responses, Th2-dominant cytokine responses in DLNs, and Ab responses (Figs. 2, 3). Mast cell deficiency resulted in reduction of papain-specific IgE/IgG1 responses (Fig. 4), and topical reconstitution of tissue mast cells by s.c. BMMC injection recovered the Ab responses and eosinophil infiltration (Fig. 5). Unlike the allergic responses triggered by airway exposure to papain (10), dermal exposure to papain in IL-33^−/− mice resulted in skin inflammation and Ab responses similar to the wild-type mice (Fig. 6). Once the sensitization to papain was established by the dermal exposure, subsequent inhalation of a threshold dose of papain (10 μg/animal) induced robust lung eosinophilia and IgE boost in wild-type mice (Fig. 7) in a manner partially dependent on the protease activity of inhaled papain (Fig. 8) and lesser but significant lung eosinophilic inflammation in IL-33^−/− mice (Fig. 7).

Ab responses in i.p. immunization of mice with natural (7) or rDer p 1 plus Alum (8) or rDer f 1 with no adjuvants (9) are dependent on the protease activity of these HDM major protease allergens. Partial reduction (15, 46) or elimination (15) of Ab production has been reported for E64-treated or heat-denatured papain, respectively, when s.c. administered with no adjuvants. The cysteine protease inhibitor we used, E64, irreversibly binds to the catalytic site of papain-like cysteine proteases without altering their tertiary structure (8, 47). We showed that the loss of the protease activity but not structural change of the s.c. administered papain caused the loss or marked reduction of not only Ab production but also skin responses (Figs. 2, 3). This observation is similar to results from a model of allergic airway inflammation, in which the protease activity of i.n. administered papain and rDer f 1 crucially contribute to Ab responses and lung eosinophilia (10). Thus, the protease activity of protease allergens seems to be important regardless of the route of sensitization (s.c., i.n., or i.p.) toward Ab production and/or inflammation. We have recently demonstrated that Ab responses induced by epicutaneous administration of papain are also dependent on the protease activity (16).

Interestingly, in mice s.c. presensitized to papain, the protease activity of papain used for subsequent airway challenge contributed to the severity of lung eosinophilia and boost of Ag-specific IgE (Fig. 8).

Results obtained by using mast cell–deficient mice and reconstitution of tissue mast cells indicated contribution of mast cells to papain-specific Ab responses and eosinophil infiltration into dermis upon s.c. papain administration (10 μg/ear) (Figs. 4, 5). In contrast, our previous report suggested no significant contribution of mast cells to IgE/IgG1 responses and lung eosinophilia upon i.n. papain administration (10). These results suggest the contribution of mast cells to allergic sensitization to protease allergens administered via s.c. but not i.n. route. No decrease of serum total IgE by the mast cell deficiency (Figs. 4B, 5A) could be attributable to the fact that papain-specific IgE was induced in mast cell–deficient mice, although at lower levels than wild-type mice, and IgE harbored on the high-affinity IgE receptor expressed on the surface of mast cells in wild-type mice should exist as free IgE in serum in the mast cell–deficient mice. Upon subsequent i.n. challenge, mast cell–deficient mice showed weaker lung eosinophilia (Supplemental Fig. 1C). To our knowledge, this is the first demonstration that Ab responses against protein Ags were reduced in mast cell–deficient mice, although the mechanism is yet to be investigated. Recent reports showed contribution of mast cells to skin inflammation in mice (48, 49), but did not describe contribution of mast cells to Ag-specific Ab responses.

IL-33 was not necessary for the skin responses and Ab responses upon s.c. administration to ear skin (Fig. 6), in contrast to airway inflammation induced by exposure of airway mucosa to papain, in which the indispensable contribution of IL-33 (10, 22) and IL-33–activated ILC2s (28, 29) has been reported; these results also suggest the differential contribution of IL-33 to allergic reactions in the skin and airway. Similar to our results using papain without any adjuvant, Savinko et al. (50) reported that mice lacking a subunit of the receptor for IL-33, ST2, that were epicutaneously sensitized with OVA in the presence of a superantigen, staphylococcal enterotoxin B, as an adjuvant and subsequently i.n. challenged with OVA alone, showed decreased numbers of eosinophils in the airway, whereas dermal eosinophil numbers were equal to those in the epicutaneously sensitized wild-type mice. However, expression of IL-33 and ST2 (30, 31) and presence of ILC2s (32, 33) in the skin tissue of AD patients and association of single-nucleotide polymorphisms in the ST2 gene with AD (30) have been reported, and skin-specific overexpression of IL-33 in mice resulted in activation of ILC2s and spontaneous onset of AD-like inflammation (27), suggesting possible involvement of the IL-33–ST2 interaction and ILC2s in the pathogenesis of AD. Although a contribution of IL-33 to skin inflammation and Ab production was not observed in our model, suggesting that IL-33 does not contribute to the initial dermal sensitization phase, we cannot exclude the possibility of a contribution of IL-33 to the pathogenesis in different experimental settings such as those associated with more exacerbated or chronic symptoms.

The term “atopic march” refers to the natural history of allergic diseases, which is characterized by a typical sequence of IgE responses and clinical symptoms that appear early in life, persist over years, and often remit spontaneously with age. Generally, the earliest clinical sign in the atopic march is AD, and the onset of food allergy, atopic rhinitis, or asthma becomes prominent later as the initial signs of AD subside. Based on the sequence of the affected tissues, it is proposed that skin, especially skin with defective barrier function (51, 52) or eczema (36), is the initial site of allergic sensitization to allergens in the natural history of allergic diseases (34, 35, 53). In our murine model, in which mice s.c. administered papain were then exposed to a threshold dose of papain via airway (10 μg), mice with a history of allergic dermal sensitization have an increased susceptibility to allergic airway inflammation (Fig. 7). The protease activity of papain used for the i.n. challenges (Fig. 8) and IL-33 (Fig. 7) contributes to the atopic march responses. Two distinct pathways cooperatively contributed to the induction of maximum IgE response and lung eosinophilia in this s.c. sensitization–i.n. challenge model, that is, s.c. presensitization-dependent, IL-33–independent pathway and airway protease–induced, IL-33–mediated pathway. It is likely that the first pathway involves memory Th2 responses and Ab-mediated responses and that the second pathway is mediated by IL-33–responsive cells in the airway, which include ILC2s and/or other cell types (21, 54).

Ab production and eosinophil infiltration into dermis were dependent on mast cells, but not on IL-33 in the s.c. model (Figs. 4–6), whereas IL-33 contributed to Ab responses and lung eosinophilia (10) (Fig. 7) and mast cell did not significantly contribute to the responses (10) in i.n. models. The main role of IL-33 in the i.n. models is considered to stimulate ST2-expressing ILC2s to release abundant IL-5 and IL-13, both of which cooperating with Th2-derived ones cause lung eosinophilia, and the latter of which, IL-13, stimulates lung dendritic cells to initiate Th2 differentiation, leading to the IgE/IgG1 responses (10, 29). Hara et al. (55) suggested that uric acid is a key sensor molecule to alarm the protease inhalation or tissue damage by induction of IL-33 release. In the s.c. model, how mast cells contribute to the Ab and eosinophil responses, possible contribution of ILC2s activated with other cytokines than IL-33 (32), mediators inducing the responses, sensor molecules for the protease penetration, and so on are yet to be investigated in future studies.

In summary, we showed that s.c. administration of protease allergen causes skin inflammation, sensitization with IgE/IgG1 production, and Th2 differentiation in a manner dependent on the protease activity of the allergen (Figs. 1–3) similarly to an allergic airway inflammation model. However, the responsible mechanisms seem to be different between the s.c. and i.n. sensitizations, that is, IL-33 is dispensable and Ab production and eosinophil infiltration are dependent on mast cells in the s.c. model (Figs. 4–6), but not in the i.n. model (10). Furthermore, the dermal sensitization–airway challenge model with protease allergen enables the robust lung eosinophilia and IgE boost upon i.n. challenge with a threshold dose of the allergen in wild-type, but not IL-33^−/− mice (Fig. 7), and the protease activity of the inhaled Ag contributes to the responses (Fig. 8), suggesting that both the dermal sensitization prior to the airway exposure and protease-induced, IL-33–mediated responses upon the airway exposure contribute to the transition from skin inflammation to asthma and that the combination of both maximizes the responses. This study provides a concept of allergic sensitization that mammals can sense exogenous proteases as innate danger signals regardless of the sensitization route, but via distinct ways according to the sensitization route. Cooperation of initial sensitization via skin and re-exposure via airway to the protease allergen maximizes the magnitude of the atopic march, natural history of progression of allergic diseases. Based on clinical studies about the treatment of human atopic diseases, it is proposed that early treatment of children with AD will prevent the subsequent development of the atopic march (34, 35). Unknown mechanisms behind the mast cell–dependent, IL-33–independent skin inflammation and Ab responses in the s.c. model and investigations in epicutaneous sensitization models (16) should be addressed in future studies.

We thank Michiyo Matsumoto for secretarial assistance; Tomoko Tokura for technical assistance; Hideo Iida, Hiroko Ushio, and Nobuhiro Nakano for comments; Hiroshi Kawai, Toyoko Hidano, and Takatoshi Kuhara for animal care; and Noriyoshi Sueyoshi, Shinji Nakamura, Yuko Kojima, Atsushi Furuhata, Yasuko Toi, Katsumi Miyahara, and Mio Miyazawa for specimen preparation for histology.

The authors have no financial conflicts of interest.