FcεRI and FcγRIII/CD16 Differentially Regulate Atopic Dermatitis in Mice¹

Atopic dematitis (AD)⁴ is a common chronic inflammatory skin disease that often begins in infancy and frequently occurs in subjects with a personal or family history of atopic disease (1). Skin lesions in AD display increased epidermal thickness and infiltration with inflammatory cells: activated memory CD4⁺ T cells, macrophages, mast cells, and eosinophils. Acute lesions are rather associated to a Th2 response, while a Th1 profile, with an accumulation of IFN-γ-producing cells, is predominant in chronic phase. Most AD patients show elevated serum IgE levels, with specific IgE directed against environmental allergens or microbial proteins as well as allergen-specific IgG (particularly IgG1 and IgG4) (2, 3).

The Ig Fc portion interacts with α subunits of the various FcRs, exerting pleiotropic effects within the immune system (4, 5). IgE binds to 2 FcεR, including the multimeric high-affinity FcεRI (4) and the low-affinity, lectin type FcεRII/CD23 (6). FcεRI is expressed on the surface of mast cells and basophils and triggers IgE-mediated degranulation and cytokine release (4). In humans, it is also expressed on professional APCs such as dendritic cells (including epidermal Langerhans cells and inflammatory dendritic epidermal cells), monocytes/macrophages, eosinophils, and platelets (4), where it plays a role in antigenic presentation and/or Ab-dependent cellular cytotoxicity reactions (4). Three FcRγ-associated, ITAM-containing activating receptors are found among FcγR family: the high-affinity FcγRI/CD64, the low-affinity FcγRIII/CD16 (in humans, the nonsignaling, glycolipid-anchored CD16b is expressed by neutrophils), and, in mice, the low-affinity FcγRIV. The FcγR family also comprises one monomeric, ITIM-bearing inhibitory receptor: the low-affinity FcγRII/CD32b (in humans, CD32a is also an activating receptor) (5). In mice, CD16, CD32, and FcγRIV also act as low-affinity IgE receptors (4, 7). Most FcγRs have a widespread cell distribution. Activating FcγRs trigger phagocytosis, cytokine release, Ag presentation, and Ab-dependent cellular cytotoxicity reactions (5).

Polymorphisms in FcεRI and CD16 were found to be correlated with atopy. Hasegawa et al. demonstrated the presence of a polymorphism in human FcεRIα gene promoter (chromosome 1q21) related to AD in a Japanese population (8). The α subunit of FcεRI plays an important role in IgE-mediated allergic reactions as an amplifier for cell surface expression and signal transduction of FcεRI (4). Polymorphisms in FcRβ gene located on human chromosome 11q12–13 have been associated to atopy. Indeed, different groups found single nucleotide polymorphisms in the FcRβ promoter in atopic patients (9). Those single nucleotide polymorphisms were causally linked with atopy via regulation of FcεRI expression (9). Likewise, Zeyrek et al. showed that a V158V genotype in CD16a gene polymorphism may represent a genetic risk factor for the development of atopic diseases (10).

IgE has also been shown to increase expression of both FcεRI and CD23 (4, 6), in the former case by preventing proteolytic degradation and receptor internalization. Indeed, FcεRI surface expression is increased in Langerhans cells, inflammatory dendritic epidermal cells, monocytes, and circulating dendritic cells (DCs) from patients with AD (4). In DCs, this increased expression has been associated to increased FcRγ expression, the limiting factor for receptor surface expression (11). In humans, FcεRI engagement has antiapoptotic effects on monocytes (12). An increased expression of CD64 and CD16 was also demonstrated in acutely and chronically inflamed skin, compared with healthy and nonlesional AD skin (13). However, the respective contributions of IgE and IgG receptors to the physiopathology of AD remain unclear. Thus, using a mouse model reproducing some features of the human disease and FcR-deficient animals, we have examined the role of FcγR-associated FcεRI and CD16 in AD.

BALB/c, C57BL/6, and OT-II (14) mice, FcRγ^−/− mice (BALB/c background) (15), and CD16^−/− mice (C57BL/6 background) (16) were from Charles River Laboratories, Taconic, and The Jackson Laboratory, respectively. DO11.10 mice (17) were a gift from Dr. C. Verwaerde (Institut Pasteur de Lille). FcεRI^−/− mice (BALB/c background) (18) were bred at the Institut Pasteur de Lille. Animals were kept under specific pathogen-free conditions. Eight- to 12-wk-old female mice were used for all the experiments, performed following approval by the Ethics Committee for Animal Experimentation from the Nord-Pas-de-Calais region. PBS-treated groups were composed of three to seven animals, while OVA-treated groups included six to nine animals. Equal numbers of gene-proficient and gene-deficient animals were used in every experiment.

AD was induced by epicutaneous OVA sensitization (19). Two paper discs (gift from Epitest) soaked with 25 μl of OVA solution (2 mg/ml in PBS) or PBS were applied on abdominal skin 24 h after shaving. Patches were secured with a bioocclusive dressing (Visulin; Hartmann), which was protected with an elastic bandage (Optiplaste; Smith & Nephew). Patches were left on for three 1-wk periods with a 2-wk interval between applications. Animals were sacrified by cervical dislocation after bronchoalveolar lavage. Skin and inguinal lymph nodes were collected in fixative for histology and/or directly frozen in liquid nitrogen for RNA analyses.

At the time of last patch removal (day 49) mice were challenged for 20 min by aerosol nebulization with OVA (1% in PBS) using an ultrasonic nebulizer (Ultramed; Medicalia) and serum was collected (19). On the next day, airway hyperreactivity (AHR) to increasing concentrations of methacholine was measured by whole-body plethysmography (Emka Technologies). Results were expressed as percentage of increase in enhanced pause over baseline value. Following pentobarbital anesthesia, lung samples were used for bronchoalveolar lavage (BAL). BAL fluid was analyzed on cytospin preparations following RAL 555 staining.

Tissue biopsies were fixed in ImmunoHistoFix (Intertiles) and embedded in ImmunoHistoWax (Intertiles) (20). Serial sections (5 μm) were stained with May-Grünwald-Giemsa for measurement of epidermal thickness as well as eosinophil and mast cell counts. Epidermal thickness was determined at a ×250 magnification with an ocular micrometer; an average of 10 measures were calculated for each sample. Cells were enumerated using an eyepiece equipped with a calibrated grid, by examining 20 random fields at ×1000 magnification; results were expressed as cell number per mm². For immunohistochemical analysis, 5-μm sections were dewaxed in acetone for 5 min and immunostained with anti-MHC class II (MHC-II), anti-CD4, anti-IL-10 mAbs (BD Pharmingen) and anti-Foxp3 (clone FJK-16s) (eBioscience), as previously described (20).

Identification of cell types expressing Foxp3 and IL-10 was performed by immunofluorescence on skin biopsies fixed in 4% paraformaldehyde in PBS and embedded in tissue freezing medium (Leica). Acetone/methanol-treated 5-μm cryosections were used in multiple staining with anti-Foxp3 revealed by Alexa 488-conjugated anti-rat IgG (Invitrogen) and biotin-conjugated anti-CD4 revealed by Alexa 555-conjugated streptavidin (Invitrogen). IL-10 was identified in T cells using anti-IL-10 revealed by Alexa 555-conjugated anti-rat IgG and biotin-conjugated anti-CD4 revealed by Alexa 488-conjugated streptavidin. Identification in DCs and mast cells was performed using anti-MHC-II revealed by Alexa 488-conjugated anti-rat IgG or FITC-conjugated anti-c-kit (BD Pharmingen), respectively, and biotin-conjugated anti-IL-10 revealed by Alexa 555-conjugated streptavidin. Nuclei were counterstained with Hoechst 33258 (Sigma-Aldrich).

Total IgE concentrations were measured in mouse serum by ELISA, using anti-IgE Abs (BD Pharmingen) (20). Total IgG1 and IgG2a were measured with the immunoassay kit SBA Clonotyping System/HRP (SouthernBiotech). Anti-OVA IgE was measured using anti-IgE (BD Pharmingen) as capture Ab and biotinylated OVA and HRP-conjugated streptavidin (Amersham Biosciences) for detection. Anti-OVA IgG1 and IgG2a were measured using OVA-coated plates and HRP-conjugated anti-mouse IgG1 and IgG2a (SouthernBiotech). Two-fold serial dilutions were prepared for each serum (starting dilution 1/25 for IgE, 1/5000 for IgG1, and 1/250 for IgG2a titrations). Ab titers were calculated as the dilution corresponding to twice the mean absorbance value obtained for nonsensitized mouse sera.

Tissue RNA extraction was performed as described (21). Reverse transcription was performed with 1 μg of RNA using SuperScript reverse transcriptase (Invitrogen). cDNAs were amplified by PCR in triplicate assays for 40 cycles in SYBR Green PCR Master Mix (Molecular Probes) on and ABI PRISM 7000 (Applied Biosystems). Oligonucleotide primers specific for mouse cytokines, chemokines, and their receptors and nuclear transcription factors are listed in Table I. Primers were used for amplification in triplicate assays. PCR amplification of GAPDH was performed to control for sample loading and to allow normalization between samples. Relative gene expression was calculated with the 2^−ΔΔCT method (22). Since expression levels were comparable in PBS-treated wild-type (WT) and corresponding FcR-deficient animals, results for OVA-sensitized animals were expressed as the mean fold induction compared with the mean expression level in PBS-treated mice.

Naive T cells purified from TCR_OVA transgenic mice (DO11.10 or OTII) were adoptively transferred into sex- and background-matched recipient mice. CD4⁺ cells were purified from spleens using a Dynal Mouse CD4 Negative Isolation kit (Invitrogen) (purity >95%). CD4⁺ cells were labeled with CFSE (Sigma-Alrdich) (23). Two days before the first sensitization, each recipient mouse received 2 × 10⁶ labeled cells via the lateral tail vein. At day 0 and day 3, mice were epicutaneously sensitized with 100 μl of OVA (2 mg/ml in PBS) or PBS. Two days after the second sensitization mice were sacrificed and inguinal lymph nodes were harvested and homogenized and processed for flow cytometry.

Cells were washed twice in ice-cold PBS, and after incubation with anti-mouse CD16/32 2.4G2 were incubated in 100 μl of PBS containing the appropriate mAb (2.5 μg/ml for 1 × 10⁶ cells) for 30 min on ice. At least 2 × 10⁶ cells were analyzed on a FACSCalibur flow cytometer. Unless otherwise mentioned, Abs were from BD Biosciences. For analysis of CFSE-labeled DO11.10 T cells, lymph node cells were stained with anti-CD4-PE and biotinylated KJ1-26 Ab (Caltag Laboratories) followed by incubation with streptavidin-allophycocyanin. For analysis of CFSE-labeled OT-II T cells, lymph node cells were stained with anti-CD4-PE/Cy7, anti-Vα2 TCR-PE, and biotinylated anti-Vβ5 TCR followed by incubation with streptavidin-allophycocyanin. Propidium iodide was added for dead cell exclusion, immediately before FACS analysis. CD4⁺CFSE⁺KJ1-26⁺ and CD4⁺CFSE⁺Vα2⁺Vβ5⁺ cells were gated and analyzed in mice on a BALB/c and C57BL/6 background, respectively.

Statistical significance was determined using Mann-Withney U test. Results were expressed as mean ± SD; p < 0.05 was considered significant.

Using a well-established model of AD based on Ag sensitization by repeated epicutaneous OVA applications, we assessed the effect of FcRγ, FcεRI, and CD16 deficiency on skin histology and inflammatory responses. In the absence of OVA sensitization, no inflammation was detected regardless of the mouse genotype (Fig. 1,A). Compared with OVA-sensitized FcRγ^+/+, skin inflammatory response was abolished in OVA-sensitized FcRγ^−/− mice, with an absence of hyperkeratosis, spongiosis, and dermal cellular infiltrates. Upon Ag sensitization, FcRγ^−/− mice exhibited no significant increase in either epidermal thickness (Fig. 1,B) or recruitment of inflammatory cells, mast cells (Fig. 1,C), eosinophils (Fig. 1,D), APCs (MHC-II⁺) (Fig. 1,E), and CD4⁺ T cells (Fig. 1,F), demonstrating that at least one FcRγ-associated receptor is essential for induction of AD. On the other hand, OVA-sensitized FcεRI^−/− and CD16^−/− mice showed significant but partial decrease of skin inflammatory response compared with their WT counterparts (Fig. 1,A). Compared with FcεRI^+/+ mice, OVA-sensitized FcεRI^−/− mice showed a 45% average decrease in epidermal thickening (Fig. 1,B) as well as 78%, 61%, 66%, and 62% average decrease in mast cell, eosinophil, APC (MHC-II⁺), and CD4⁺ T cell recruitment respectively (Fig. 1, C–F). Interestingly, OVA-sensitized CD16^−/− mice showed no increase in epidermal thickness or in eosinophil and APC infiltration compared with their PBS-treated counterparts (Fig. 1, B, D, and E). This represents a 63% and 44% average decrease of mast cell and CD4⁺ T cell recruitment compared with CD16^+/+ mice. Despite the different genetic background used for the experiments, these results nevertheless suggest that both FcεRI and CD16 contribute to AD skin lesions but differentially control dermal recruitment of inflammatory cells.

We next examined, by real-time PCR, mRNA expression for lesion-, inflammation-, or immune response-associated molecules in the skin from unsensitized or OVA-sensitized FcRγ^−/−, FcεRI^−/−, and CD16^−/− mice as well as from their WT counterparts. For each gene examined, skin expression was comparable in PBS-treated gene-deficient animals and in their WT counterparts (not shown). In WT animals, OVA sensitization increased mRNA expression of several genes associated to Th2, Th1, Th17, as well as to inflammatory response with the exception of IL-18, and thymic stromal lymphopoietin (TSLP), which have been associated to AD in other experimental models, and of IL-33, a Th2 cell chemoattractant and Th2 associated-cytokine inducer, as well as of immunoregulatory TGF-β (24, 25) (Fig. 2). In contrast, OVA sensitization of FcRγ^−/− mice did not significantly show increased expression of most of these genes (Fig. 2,A). Importantly, this inhibited response coincided with a 2 fold-increase in the expression of immunoregulatory IL-10 and Foxp3 (Fig. 2 A), while in FcRγ^+/+ mice, expression of these latter genes was not affected by OVA sensitization. This thus further indicates that at least one FcRγ-associated receptor is a key regulator of skin immune and inflammatory response in this model of AD.

In skin from OVA-sensitized FcεRI^−/− mice, IL-13, IL-6, and IL-17 expression was comparable to that found in corresponding PBS-treated animals, while a partial decrease of expression of several genes was observed compared with OVA-sensitized WT animals (Fig. 2 B): Th2-associated molecules (IL-4, IL-5, IL-31); CCR3, which contributes to eosinophil chemotaxis; CCL22 (macrophage-derived chemokine, MDC), a chemoattractant for Th2 lymphocytes into the inflammatory skin; Th1-associated molecules (IFN-γ and IP-10); CCL20, an APC-attracting chemokine; and proinflammatory IL-1β. As for FcRγ^−/− mice, FcεRI^−/− mice showed a 1.8-fold increase of IL-10 and Foxp3 mRNA expression following OVA sensitization.

Surprisingly, in CD16^−/− mice, OVA sensitization did not lead to a significant increase of Th2-related molecules and of CCL20 expression. IL-17, IL-1β, and IL-6 mRNA expression was also decreased compared with OVA-sensitized CD16^+/+ animals. In contrast, Th1 response was comparable to that of corresponding OVA-sensitized CD16^+/+ animals. Inhibition of Th2, Th17, and skin inflammatory response in OVA-sensitized CD16^−/− mice coincided with a moderate increase of IL-10 and Foxp3 mRNA expression (1.6- and 1.5-fold increase, respectively).

Since IL-10 can be produced by various immune and nonimmune cells types, in skin, we performed double immunohistochemistry to identify cellular source of IL-10. While keratinocytes were devoid of staining, IL-10-containing lymphocytes (CD4⁺IL-10⁺), DCs (MHC-II⁺IL-10⁺), and mast cells (c-kit⁺IL-10⁺) were identified (Fig. 3). A minor increase in IL-10⁺ skin cell density was found between OVA- and PBS-treated FcRγ^+/+, FcRγ^−/−, FcεRI^+/+, and FcεRI^−/− mice. However, no differences were observed between gene-deficient animals and their WT counterparts (data not shown).

As expected, only CD4⁺ cells expressed Foxp3 in skin (Fig. 4, A and B). No significant increase in Foxp3⁺ cell density was found between OVA- and PBS-treated FcRγ^+/+, FcεRI^+/+, CD16^+/+, and CD16^−/− mice. However, Foxp3⁺ cell density was increased, respectively, by 2.5- and 2.4-fold in OVA-treated FcRγ^−/− and FcεRI^−/− mice compared with their PBS-treated counterparts, and by 2.1- and 6-fold compared with OVA-treated FcRγ^+/+ and FcεRI^+/+ mice (Fig. 4 C).

Taken together, these data suggest that FcεRI and CD16 differentially regulate skin expression of DA-associated cytokines, with FcεRI controlling both Th1 and Th2 responses while CD16 only regulates Th2 response.

Serum levels of IgE and IgG1, both associated to a Th2 response, were decreased in OVA-sensitized FcRγ^−/− mice compared with OVA-sensitized FcRγ^+/+ animals, with a 64% and 55% average decrease for total IgE and OVA-specific IgE, respectively, and a 58.5% decrease for OVA-specific IgG1 (Fig. 5, A–C). IgG2a and IgG2b Abs, associated to a Th1 response, were undetectable in OVA-sensitized WT or FcR-deficient mice (data not shown). Interestingly, compared with their FcεRI^+/+ controls, only IgE response was affected in OVA-treated FcεRI^−/− mice: 67% and 54% average decrease for total IgE and OVA-specific IgE, respectively (Fig. 5, A–C). In contrast with the latter mice, in OVA-sensitized CD16^−/− mice, only OVA-specific IgG1 response was affected (63% decrease). Despite the different genetic background used for the experiments, these results nevertheless suggest that, in this model, FcεRI and CD16 specifically regulate the production of their own ligand.

To determine, at the molecular level, the factors that account for this differential regulation, we analyzed, by real-time PCR, cytokine expression in draining inguinal lymph nodes. In particular, we examined the expression or IL-21 and its receptor. Indeed, these latter are essential for IgG1 production (26). In OVA-sensitized FcRγ^−/− mice, expression of IL-4, IFN-γ, IL-21, and IL-21R were fully inhibited compared with FcRγ^+/+ animals (Fig. 5,D). IL-13 expression was comparable to that found in WT animals (data not shown). In OVA-sensitized FcεRI^−/− mice, IL-4 expression was reduced and IFN-γ completely inhibited compared with FcεRI^+/+ animals, while the IL-21/IL-21R axis remained intact (Fig. 5,E). In contrast, when compared with CD16^+/+ animals, IL-4 and IFN-γ expression was comparable in OVA-sensitized CD16^−/− mice, but expression of IL-21 and IL-21R was fully inhibited (Fig. 5 F).

We then investigated whether defective Ag presentation during early stages of epicutaneous sensitization was responsible for the reduced inflammatory skin response observed in OVA-sensitized FcRγ^−/−, FcεRI^−/−, and CD16^−/− mice. We thus examined Ag-induced T cell proliferation in vivo by transfer of CFSE-labeled CD4⁺ T cells from DO11.10 mice to recipient animals that were epicutaneously sensitized for 1 wk. In the absence of OVA sensitization, both percentage and total number of proliferating T cells were very low for each mouse strain (Fig. 6, A, left panel, and C). Compared with FcRγ^+/+ animals, OVA-sensitized FcRγ^−/− mice showed a 64% decrease in the percentage of proliferating T cells (Fig. 6,A, right panel). They exhibited no significant increase in the total number of proliferating cells, compared with corresponding PBS-treated animals (Fig. 6,C), indicating that FcRγ is essential for Ag presentation and thus T cell proliferation in this model. T cell proliferation following OVA sensitization was comparable in FcεRI^+/+ and FcεRI^−/− mice (Fig. 6, A, right panel, and C), in agreement with the absence of FcεRI expression on WT mouse DCs. In contrast, compared with CD16^+/+ animals, OVA-sensitized CD16^−/− mice showed 33% and 72% decrease in proliferating T cells by percentage and total cell number, respectively (Fig. 6, B, right panel, and C). These results suggest that, in this model of AD, CD16 plays a major role by controlling Ag presentation.

Since this mouse model is characterized (as for AD in humans) by Ag-induced lung inflammation, we examined AHR and recruitment of inflammatory cells in the BAL fluid following epicutaneous sensitization and a single nebulization. While OVA-sensitized and -challenged FcRγ^+/+ mice, upon nebulization with increasing doses of methacholine, displayed increased AHR compared with PBS-sensitized animals, OVA- and PBS-sensitized FcRγ^−/− mice responded similarly (Fig. 7,A), demonstrating the essential role of FcRγ in airway inflammation following epicutaneous sensitization. Along these lines, BAL inflammatory infiltrate in OVA-sensitized and -challenged FcRγ^−/− mice was virtually absent compared with FcRγ^+/+ mice (Fig. 7,B). Cellular inflammation in BAL was comparable in FcεRI^+/+ and FcεRI^−/− mice (Fig. 7,C). In contrast, lung inflammatory response was equally abolished in OVA-sensitized and -challenged CD16^−/− mice compared with CD16^+/+ animals (Fig. 7 D), indicating that CD16 is one key FcRγ-associated receptor for induction of lung inflammation in this model of AD.

Since FcεRI exerted a significant effect on skin inflammation without regulating APC function, we reasoned that mast cells would be the most likely cell type directly affected by FcεRI deficiency. We thus examined mast cell recruitment into draining lymph nodes following sensitization. In contrast to OVA-sensitized CD16^+/+ and CD16^−/− mice, which displayed a similar increase in inguinal lymph nodes compared with corresponding PBS-sensitized animals, FcRγ^−/− and FcεRI^−/− mice showed no significant Ag-induced increase in mast cell recruitment into lymph nodes (Fig. 8). This further suggests that FcεRI and CD16 differentially regulate distinct, but partially overlapping, sets of parameters from the AD-associated immune response.

In this paper, we provide evidence for the differential contribution of FcRγ-associated FcεRI and FcγRIII/CD16 in a murine model of AD. Absence of FcRγ-associated activating receptors led to a virtual absence of pathology with abolished skin and lung inflammation and drastic decrease of humoral response that were associated to an inhibition of Th2, Th1, Th17, and inflammatory cytokine expression. In contrast, a 2-fold increase of regulatory IL-10 and Foxp3 was observed. An increase in density of Foxp3⁺ cells was also evidenced. These later findings are likely linked to activation of the remaining inhibitory FcγRII/CD32b expressed by APC, which induces IL-10 production and Foxp3⁺ regulatory T cells (27). FcγRII/CD32b-independent mast cell activation might also lead to IL-10 production and/or induction of Foxp3 expression (28, 29, 30). Indeed, besides CD4⁺ lymphocytes, IL-10 expression was found in both DCs and mast cells but not in keratinocytes, known as a potential source of IL-10 (31). An increase in IL-10 and Foxp3 expression was also found in FcεRI- and, to a slightly lesser extent, in CD16-deficient animals. Foxp3⁺ cell density was also increased in skin from FcεRI-deficient animals. Defective Ag presentation during early sensitization phase, and thus abrogated Ag-induced T cell proliferation, accounts, at least in part, for the inhibition of pathology as previously observed in models of delayed hypersensitivity (32) and asthma (33). The lack of FcRγ expression on effector cells such as mast cells might further inhibit development of inflammation as shown in a mast cell-dependent asthma model (34).

Lack of FcεRI led to a partial decrease of epidermal thickening and cellular recruitment into dermis upon induction of AD but, surprisingly, had no effect on lung inflammation. FcεRI contribution to the pathology might be slightly underestimated since we have shown that CD16 expression on FcεRI-deficient mast cells, and probably basophils, is up-regulated and leads to more severe IgG-mediated anaphylaxis (35). In our AD model, the absence of FcεRI also led to impaired skin expression of Th1, Th2, Th17, and inflammatory cytokines but to an increased expression of Foxp3 and IL-10. As observed in some asthma models (36, 37), FcεRI involvement in this AD model results from its high expression levels on mast cells, which are very abundant in skin. However, FcεRI does not contribute the sensitization process at a distal inflammatory site such as lung.

Basophil contribution to the pathology cannot be ruled out, as this cell type contributes to IgE-induced chronic allergic dermatitis (38, 39), and since a basophil influx has been reported during patch tests in atopic patients (40). Histamine release by mast cells promotes Langerhans cell migration to draining lymph nodes through binding to H2 receptors (41). Thus, lack of FcεRI might indirectly explain the decreased skin APC number compared with sensitized WT animals. Following sensitization, mast cells also migrate to draining lymph nodes, where they induce T lymphocytes recruitment via MIP-1β production. In our model of AD, mast cell number in draining lymph nodes appears to be controlled by FcεRI. As mast cells undergo FcεRI-, IgE-, and Ag-dependent chemotaxis (42), an effect not reported for IgG isotypes, it is likely that the lack of FcεRI expression accounts for this defective migration and thus for indirect effects on other cell types besides Langerhans cells, particularly T and B cells and eosinophils. Indeed, mast cells also release Th2 and proinflammatory cytokines such as IL-4, IL-5, IL-13, and IL-6 upon FcεRI activation (43). These cytokines are known to regulate IgE production and eosinophilia, two hallmarks of AD. IL-6 has been shown to regulate IL-17 production (44), recently associated to AD (45). Thus, decreased IL-6 production by mast cells might indirectly account for the decreased IL-17 expression. Nevertheless, IL-17 expression in AD is low compared with psoriasis (46). Surprisingly, in FcεRI-deficient mice, Ag-specific IgG1 response was unaffected while Ag-specific IgE response was decreased (Fig. 3). We had previously reported a tendency to lower Ag-specific IgE concentrations but normal IgG levels following active anaphylaxis-inducing immunization (35). This regulation of IgE synthesis by FcεRI thus seems to complete a regulatory loop. Indeed, the ability for IgE to up-regulate expression of both its receptors has been previously documented (4, 6). IL-21 seems to be the only factor known to differentially regulate IgE and IgG1 synthesis (26). Expression of both IL-21 and its receptor were unaffected in draining lymph nodes from FcεRI-deficient animals, while expression of IL-4, the key cytokine for the regulation of IgE synthesis in this model, was reduced. Thus, the differential effect of FcεRI and CD16 on IgE and IgG1 production might be attributed to a distinct regulation of IL-4 and IL-21 (see below).

Compared with the pathology developed on the BALB/c background, AD is significantly less severe on a C57BL/6 background. Indeed, most of the histological, cellular, and biochemical parameters are quantitatively lower in the latter case. This renders direct comparisons between FcRγ- and FcεRI-deficient mice, on the one hand, and CD16-deficient mice, on the other hand, impossible. Thus, CD16 contribution to AD is only discussed by comparing CD16-deficient and -proficient mice on a C57BL/6 background.

CD16 deficiency led to a complete inhibition of epidermal thickening, dermal and lung recruitment of inflammatory cells following sensitization. Surprisingly, in our model of AD eliciting both a Th1 and a Th2 response (1), skin Th1-associated cytokine response was unaffected in CD16-deficient mice while Th2 response was completely abolished. These results on CD16 selective contribution to the Th2 response, and to AD, are in agreement with those obtained in a (strongly Th2-polarized) asthma model, in which CD64 does not play a role (47). Both IL-17 and IL-6 expression was inhibited while IL-10 and Foxp3 expression was enhanced in CD16-deficient mice, as it was the case for FcεRI-deficient animals. In contrast, CD16-deficient mice showed a major decrease in Ag-induced T cell proliferation, indicating that CD16 plays an essential role in antigenic presentation and initiation of immune response in this model of AD. Indeed, CD16 is highly expressed on DC and is known to preferentially bind IgG1 (immune complexes) (48), the major Th2-associated isotype in mouse, whose levels are very high following OVA sensitization in this AD protocol (Fig. 3). Besides its expression on APC, CD16 is also expressed on mast cells (29) and basophils (49). It is thus possible that CD16 deficiency directly contributed to the decreased mast cell number and to a dampened mast cell activation in skin following Ag sensitization. However, mast cell number in draining lymph nodes from CD16-deficient animals was not affected compared with WT animals, suggesting that, in contrast with FcεRI, CD16 is not involved in chemotactic activity. Finally, as observed for FcεRI and IgE, CD16 deficiency led to a decreased specific IgG1 response, while IgE production remained similar. Consistent with this result, expression of both IL-21 and its receptor was abrogated in draining lymph nodes from CD16-deficient animals, while IL-4 expression was comparable to that found in WT animals. This contrasts with an asthma model where production of both isotypes was decreased (47). This discrepancy might be due to our less Th2-prone experimental setting, to the different pathology (AD vs asthma), and/or to a different genetic background (C57BL/6 vs BALB/c).

Among FcRγ-associated receptors, FcεRI and CD16 were a priori the likeliest to be involved in the regulation of AD, hence our study. Indeed, the only other two FcRγ-associated receptors are FcγRI/CD64 and FcγRIV, which are mainly expressed on macrophages and preferentially bind IgG2a and IgG2b (48), two Th1-associated isotypes. In our model, OVA-specific IgG2a and IgG2b were not detected. FcγRIV has also been reported to bind IgE^b, found in C57BL/6 but not in BALB/c mice, with an even lower affinity than CD16 (7). Due to this cell distribution and isotype specificity, it is thus unlikely, but cannot be completely ruled out, that either of these receptors plays a major role in AD physiopathology.

In conclusion, the present study delineates for the first time a differential but partially overlapping role of FcεRI and CD16 in a mouse model reproducing some features of AD physiopathology, where the contribution of IgG immune complexes was documented but poorly understood (50). Even if considerable differences exist between mice and humans regarding identity, structure, cell distribution, and/or isotype specificity of FcγR and FcεRI, we might nevertheless speculate that the partial inhibition obtained by targeting only IgE/FcεRI interactions, using anti-IgE, in asthma (51) or atopic skin reactions (52) might be due to a significant contribution of CD16 or other FcγR to the pathology. Thus, a therapy preventing both IgE/FcεRI and IgG/FcγR interactions might represent a further improvement over the present ones.

The authors have no financial conflicts of interest.