An Atopic Dermatitis-Like Skin Disease with Hyper-IgE-emia Develops in Mice Carrying a Spontaneous Recessive Point Mutation in the Traf3ip2 (Act1/CIKS) Gene

The term atopic dermatitis (AD) was first coined by Wise and Sulzberger (1) to describe “confusing types of localized and generalized lichenification, generalized neurodermatitis or a manifestation of atopy.” Today, AD (or atopic eczema) is recognized as a strongly heritable disease characterized by complex symptoms, including chronically relapsing, extreme pruritus and eczematous skin disease, both of which are frequently associated with IgE hyperresponsiveness to environmental allergens (2–4). The rapid increase in the prevalence of AD during the past three decades has sparked an intense effort to elucidate the underlying pathogenesis of the disease and has led to the use of radical treatments for the disorder (3, 5).

The causative factors for AD generally fall into two categories: environmental and genetic. In the environmental category, the involvement of allergens, such as house dust, mites, and air pollution, has been suggested strongly by epidemiological studies (6). Conversely, linkage studies of atopic and nonatopic phenotypes also strongly suggest that genetic factors linked to several different candidate regions on chromosomes are involved (see Ref. 7 and references therein). Uncertainty as to the specific genetic factors that contribute to AD exists due to the difficulty of performing linkage analyses on multigenic diseases in humans, largely because of the effects of the different genetic backgrounds. Furthermore, appropriate animal models for human AD are lacking.

Recently, two promising mouse models for human AD were described by Matsuda et al. (8) and Natori et al. (9). These models are represented by the inbred strains NC/Nga (NC) and Naruto Research Institute Otsuka Atrichia (NOA). The NC strain was established in 1957 by Kondo et al. (10, 11). NC mice spontaneously suffer severe dermatitis in the presence of nonspecific allergens. Morbid NC mice show AD symptoms, including itching, erythema, hemorrhage, edema, crust, drying, and excoriation/erosion hyperplasia of the epidermis of the face, neck, and/or back, which are exacerbated by aging. Furthermore, NC mice display some of the characteristic histopathological features of AD, such as macrophage and eosinophil invasion of the dermis, increased numbers and activation of mast cells and lymphocytes, reduction of ceramide (12), the appearance of activated mast cells, and CD4⁺ T cells in the lesion. These lines of evidence suggest that the symptoms displayed by NC mice are clinically, pathologically, and immunologically similar to those of human AD. Quantitative traits linkage analysis suggests that NC mice carry several genetic determinants of dermatitis, and the major determinant is located on murine chromosome 9 (13).

Like the NC mouse, the NOA mouse also shows ulcerative skin lesions with an accumulation of mast cells and increased serum IgE levels. Linkage analysis has demonstrated that the major gene responsible for dermatitis in the NOA mouse is located in the middle of chromosome 14. The incidence of disease in these animals clearly differs according to parental strain, and the mode of inheritance is autosomal recessive with incomplete penetrance. Statistical analyses have shown that the critical region is in the vicinity of D14Mit236 and D14Mit160 (9). Furthermore, two modifier genes have been reported as candidate loci, one in the middle of chromosome 7 and the other in the telomeric region of chromosome 13. These loci correspond to regions of synteny in human chromosomes where linkages to asthma, atopy, or related phenotypes (e.g., serum IgE levels) have been documented (14). However, no mouse strains have been shown to develop dermatitis due to alterations in a single gene thus far.

We have identified a new mouse model for human AD, the phenotype of which is controlled by a recessive mutation. This mutation was successively isolated by positional cloning and found to be a nonsense mutation in the TNFR-associated factor 3-interacting protein 2 (Traf3ip2) locus on mouse chromosome 10. The TRAF3IP2 protein, an adapter molecule, was cloned using a functional genetic screen based on its ability to activate NF-κB (15). The molecule also was cloned simultaneously using a yeast two-hybrid screen. In that report, the molecule was named connection to IκB kinase and stress-activated protein kinase/Jun kinase (CIKS) based on its interaction with IκB kinase-γ (16). The TRAF3IP2 protein does not possess any identified enzymatic domains but does contain a helix-loop-helix at its N terminus, a coiled-coil at its C terminus (15), and two putative TNFR-associated factor (TRAF) binding sites (17), suggesting that it functions through protein–protein interactions in signaling pathways. Qian et al. (18) showed that TRAF3IP2 plays a critical role in the CD40-mediated signaling pathway in epithelial cells in vitro. Additionally, targeted disruption of the Traf3ip2 locus in mice causes a dramatic increase in the number of peripheral B cells, culminating in lymphadenopathy, splenomegaly, hypergammaglobulinemia, and autoantibody production. Detailed analyses have indicated that TRAF3IP2 negatively regulates CD40- and B cell-activating factor (BAFF)-mediated signaling and suggest that the TRAF3IP2-mediated attenuation of this pathway plays an important role in the homeostasis of B cells (19). Conversely, TRAF3IP2 functions as a positive regulator in the IL-17–dependent signaling pathway associated with autoimmunity and inflammatory diseases (20). This antipathetic nature of TRAF3IP2 in signaling pathways has been suggested to be a result of the differential usage of protein–protein interacting domains in the TRAF3IP2 molecule (21).

In this report, we show that the AD-like phenotype in AD from Japanese mice (adjm) is caused by the presence of a truncated form of the TRAF3IP2 molecule and that the severity of the phenotype is controlled by the genetic background of the mouse strains used. The use of these mouse strains (e.g., KOR, C57BL/6, BALB/c, AKR/J, and A/J; see 1Materials and Methods) to study the dual functions of TRAF3IP2 should help to elucidate the mechanisms underlying the two major symptoms of human AD, especially hyper-IgE-emia, and should facilitate exploration of the question of why human AD is multigenic.

KOR mice were derived from Japanese wild mice (Mus musculus molossinus) (11). KOR-adjm/adjm and C57BL/6-, BALB/c-, AKR/J-, and A/J-adjm congenic strains were maintained in-house at the Saitama Cancer Center. All of the mice were housed under specific pathogen-free conditions (22 ± 2°C, 50 ± 10% controlled humidity, and a 12 h/12 h light/dark schedule). A regular laboratory diet (F-2; Funabashi Farm, Chiba, Japan) and water were provided ad libitum. All of the animal experiments were approved by the Committees for Guidelines and Regulation of Animal Experiments of the Saitama Cancer Center.

As a prefixative, tissues were immersed twice in 4% paraformaldehyde and irradiated using a microwave oven for 30 s at 550 W. Tissues were then postfixed in the same fixative for 4 h on ice. Fixed tissues were dehydrated in methanol, embedded in polyester wax (Sekisui Medical, Tokyo, Japan), and sectioned at 3 μm. Sections were dewaxed, stained with H&E (Wako Pure Chemicals, Tokyo, Japan) or with Masson's trichrome solution (Muto Pure Chemicals, Tokyo, Japan), and mounted on a coverslip for photomicrography.

Blood was collected from the retro-orbital plexus of mice under ether anesthesia and immediately heparinized. Plasma samples were obtained by centrifugation and stored at −20°C until use. The total serum IgE level was measured using a mouse IgE measurement kit (Mouse IgE EIA Kit) according to the manufacturer’s instructions (Yamasa Shoyu, Tokyo, Japan). Briefly, the test serum (100 μl of a 1:24 dilution) was incubated in a microplate precoated with anti-mouse IgE mAbs for 30 min at room temperature. After being washed, the wells were further incubated with HRP-labeled anti-mouse IgE for 30 min at room temperature, followed by washing and incubation with the colorimetric substrate for 30 min. The reaction was stopped by the addition of 100 μl per well of a 2.0 N HCl solution, and the OD was measured at 450 nm. Mouse IgE isotype standard solutions were used to construct a standard curve (10–500 ng/ml). Total serum IgE levels were calculated from the standard curve.

Genome-wide simple sequence length polymorphism tests were performed using Mit microsatellite marker pairs (ResGen, Huntsville, AL) and a PCR amplification kit (Takara Bio, Shiga, Japan). Approximately 20 ng genomic DNA was added to the final 20-μl reaction mixture. The thermocycling program consisted of one initial denaturation at 94°C for 1 min and 30 cycles of 94°C, 55°C, and 72°C for 1 min each. PCR-amplified products were separated in a 4% agarose gel (NuSieve:Agarose, 1:3; Cambrex Bio Science Rockland, Rockland, ME) in Tris-borate-EDTA buffer and visualized by UV light after being stained with ethidium bromide.

The nucleotide sequences of the genes (i.e., Traf3ip2, Fyn, ENSMUSG00000064311, Q9D6T1, and Lama4) were obtained from the Ensembl Genome Browser (www.ensembl.org), and PCR primers were designed for the amplification of each exon. To screen for mutations between KOR and KOR-adjm, total RNA was isolated from 5-wk-old mouse brain and spleen using TRIzol (Invitrogen, Carlsbad, CA) following the manufacturer’s protocol. cDNA was generated with the Omniscript RT Kit (Qiagen, Hilden, Germany) using 1 μg of DNase-pretreated total RNA. The entire genomic region of the Traf3ip2 gene was amplified by long and accurate PCR (Takara Bio) to produce overlapping PCR products from both strains as well as from other inbred strains (C57BL/6J, A/J, and BALB/c) as controls. The sequences of the primers used, including the sequencing primers, are shown in Table II. PCR products were gel-purified, sequenced using a BigDye Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, CA), and analyzed on a 3100 Genetic Analyzer (Applied Biosystems). Genomic DNA from KOR, KOR-adjm/+, KOR-adjm/adjm, C57BL/6-adjm/+, C57BL/6-adjm/adjm, BALB/c, BALB/c-adjm/+, and BALB/c-adjm/adjm animals also was amplified (forward primer, AI429613int1F; reverse primer, AI429613int2R; see Table II) by PCR. PCR products were digested with PstI, separated in a 1% agarose gel, and stained with ethidium bromide.

Total RNA was isolated from 5-week-old BALB/c-+/+ and BALB/c-adjm/adjm mouse spleen using TRIzol (Invitrogen) and a TRIzol Plus Purification Kit (Invitrogen) following the manufacturer’s protocol. cDNA was generated with the ThermoScript RT-PCR System (Invitrogen) using 0.1 μg of DNase-pretreated total RNA. The cDNA was amplified for 30 cycles (98°C for 20 s and 68°C for 4 min) using KOD FX (TOYOBO, Osaka, Japan). Primer sets for amplification of a 272-bp product used primers (AI429613ex1F and AI429613ex2R; Table II) located at exons 1 and 2. The products were subjected to agarose gel electrophoresis. cDNA integrity was confirmed by Gapdh (22).

Total RNA was isolated from 5-wk-old BALB/c-+/+ and BALB/c-adjm/adjm mouse skin, spleen, and brain using TRIzol and a TRIzol Plus Purification Kit following the manufacturer’s protocol. Reverse transcription and quantitative PCR were carried out using a SuperScript VILO cDNA Synthesis Kit (Invitrogen) and a QuantiTect SYBR Green PCR Kit (Qiagen) according to the manufacturer’s protocol. Gapdh was used as an endogenous control. Primers for mouse Traf3ip2 (Mm_AI429613_1) and Gapdh (Mm_Gapdh_3) were purchased from Qiagen. All of the samples were analyzed in triplicate using the ABI Prism 7500 Fast Real-Time PCR System (Applied Biosystems). Data were analyzed using the Relative Quantification Software of the Applied Biosystems 7500 Fast, with expression levels in C57BL/6-+/+ spleen assigned an arbitrary value of 1.

Proteins were isolated from murine spleen using T-PER Tissue Protein Extraction Reagent (Pierce, Rockford, IL) according to the manufacturer’s protocol. Proteins (5 μg) were fractionated by 10% SDS-PAGE and transferred onto a Hybond-P polyvinylidene difluoride membrane (GE Healthcare, Buckinghamshire, U.K.). A rabbit polyclonal Ab raised against TRAF3IP2 (anti-Act1 [H-300]) was obtained commercially (Santa Cruz Biotechnology, Santa Cruz, CA) and has been characterized previously (23, 24). The Ab was used at a dilution of 1:400. The membrane was subsequently stripped and blotted with an anti–β-actin mAb (1:10,000; Sigma-Aldrich, St. Louis, MO) to control for protein loading. HRP-conjugated secondary Abs were used at a dilution of 1:40,000; blots were developed using an ECL Advance Western blotting Detection Kit (GE Healthcare).

The major cytokines present in Th1, Th2, and Th17 cells were analyzed semiquantitatively at the in-house microarray technology office of the Tokyo Metropolitan Institute of Medical Science using cytokine Ab arrays for 97 cytokines (Mouse Cytokine Antibody Array G Series 6; RayBiotech, Norcross, GA) following the manufacturer’s protocol.

Affected and nonaffected skin surrounding the eyes was removed, minced into pieces ~3 mm in width, incubated with DMEM/10% FBS/2 mg/ml dispase II at 4°C for 18 h, and then washed with PBS. Epidermis was separated from dermis and minced as finely as possible; it was then incubated with DMEM/10% FBS/1 mg/ml collagenase type III at 37°C for 1 h and filtered through a 70-μm nylon mesh sheet to yield a single-cell suspension.

Spleen cells were resuspended in DMEM/10% FBS and filtered through a 70-μm nylon mesh sheet to prepare single-cell suspensions.

The single-cell suspensions were treated with hypotonic buffer to induce hemolysis; they were then treated with anti-CD16/32 mAb (BD Pharmingen, San Diego, CA) and normal rat serum to prevent nonspecific binding and stained with FITC-conjugated anti-CD4 (L3T4) mAb (BD Pharmingen). CD4⁺ cells were trapped on anti-FITC microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) by MACS. The enriched CD4⁺ cells were treated with BD Cytofix/Cytoperm solution (BD Biosciences, San Jose, CA) and stained with PE-conjugated anti–IL-17A (TC11-18H10) mAb (BD Pharmingen); 10⁵ cells in each CD4⁺ T cell suspension were analyzed using a FACSCalibur (BD Biosciences, San Jose, CA).

As stated above, KOR is an inbred strain established from Japanese wild mice (M. musculus molossinus) (11). During the maintenance of the strain, we discovered two littermates (marked by an arrow in Fig. 1A) that developed severe dermatitis. This dermatitis became apparent at 5 wk of age, was localized mainly to the face (including the ears), and became progressively more severe. When two pairs of nonaffected mice from the same litter were crossed, 4 of the 12 offspring in the next generation developed dermatitis (Fig. 1A), indicating that the mutation was recessive. Using the founder pair (marked with arrows in Fig. 1A), we established a mutant strain that naturally develops dermatitis. This strain has been designated KOR-adjm/adjm. To maintain the KOR-adjm/adjm strain, nonaffected littermates are crossed randomly (e.g., pairs shown by closed arrowheads in Fig. 1A); heterozygosity is confirmed by the appearance of affected individual(s) in the next generation. The heterozygous pairs are used for strain maintenance.

Histological examination of skin specimens from KOR-adjm/adjm mice revealed that the affected skin had a thickened epidermis and showed a massive infiltration of inflammatory cells, including eosinophils (Fig. 1B–G). Scratching of the face and ears with the hind legs, most likely due to itching, was observed frequently in all of the mice analyzed (data not shown). Furthermore, the levels of serum IgE in these animals spontaneously increased with age and reached >10 ng/ml by 11 wk of age (Fig. 2A). The hyper-IgE-emia showed a sexual disparity, with female mice displaying serum IgE levels two times higher than those of male mice until 10 wk of age (Fig. 2B). Importantly, except for the short life span of these animals, all of the phenotypes observed in the mutant mice resemble those seen in patients with AD. Indeed, treatment of the affected skin with an ointment containing Tacrolimus (FK506; Astellas Pharma, Tokyo, Japan) ameliorated the dermatitis in the mutant mice, similar to what has been observed in AD patients (Y. Matsushima, unpublished observations).

The life span of KOR-adjm/adjm mice was much shorter than that of KOR-+/+ or KOR-adjm/+ mice. The t_1/2 spans were estimated to be 15.1 wk in homozygous male mutant mice and 11.6 wk in homozygous female mutant mice, whereas there was no difference in the t_1/2 spans of wild-type and heterozygous mice (Fig. 3). Female sterility also was found in all of the homozygous mutant mice (Fig. 3). The underlying mechanism of early death and female sterility in KOR-adjm/adjm mice remains unclear, although infection is suggested to be a major cause of death because ulcerative yellow pustules were sometimes seen in the forelimbs of severely affected individuals (data not shown). The pustules may have been caused by infection with Staphylococcus aureus, which is an indigenous bacterium in the lesional skin of AD patients.

A preliminary linkage analysis using (BALB/c × KOR-adjm/adjm) N₂ mice suggested that the mutation was controlled by a single recessive locus on mouse chromosome 10 (data not shown). No segregation was observed between the two prominent phenotypes, dermatitis and hyper-IgE-emia. We established four adjm congenic strains in different genetic backgrounds (C57BL/6, BALB/c, AKR/J, and A/J). Both dermatitis and hyper-IgE-emia were transmitted stably to the progeny of the congenic strains, although the severity of the dermatitis and the levels of serum IgE varied depending on the genetic background (Fig. 4). The original KOR-adjm/adjm strain had the most severe phenotypes, the mutation on congenic strains with a propensity for Th2 responses (i.e., BALB/c, AKR/J, and A/J) showed less severe phenotypes, and the mutation backcrossed into a Th1-skewed strain (C57BL/6J) showed the least severe phenotype. Interestingly, the female sterility that was observed in the original KOR strain was not observed in any of the congenic strains (data not shown). These results suggest that the adjm locus is responsible for the dermatitis and hyper-IgE-emia in the mutant mice and that additional loci may influence the adjm locus as modifiers.

To identify the gene responsible for the adjm phenotype, 1170 (BALB/c × KOR-adjm/adjm) N₂ segregants were generated. Of these, 140 developed dermatitis by 8 wk of age. The frequency of the N₂ segregants with dermatitis was lower than expected, likely because we did not account for animals with a late onset of dermatitis. Linkage analysis mapped the adjm locus near the D10mit53 locus. Further analysis with newly identified microsatellite markers (Table I) in 86 other mice obtained during the development of the adjm congenic strain in the BALB/c background further restricted the locus to the 940 kb between D10Stm216 and D10Stm238 (Fig. 5A). In this region, five genes were identified in the European Molecular Biology Laboratory database (Fig. 5A). Two of the five genes appear to be the most likely candidates for the adjm mutation, because both have been shown to be involved in signal transduction in the immune system; one gene encodes for a Src family kinase (Fyn), and the other gene (Traf3ip2, formerly AI429613) encodes a homologue of human CIKS (16)/NF-κB activator 1 (Act1) (15).

A mutation survey comparing mutants and their isogenic littermates revealed that the mutant mice (but not their control littermates) carried a C → T transition in the second exon of the murine Traf3ip2 gene (DNA Data Bank of Japan/European Molecular Biology Laboratory/GenBank accession numbers AB238206–AB238207; www.ncbi.nlm.nih.gov; Fig. 5B), whereas the sequence of the Fyn gene in both the mutant mice and the controls was identical. The presence of this alteration was confirmed by PstI digestion of PCR-amplified genomic DNA from adjm homozygotes and heterozygotes (Fig. 5C). The transition changed amino acid 214 from glutamine (CAG) to a stop codon (TAG) (Fig. 5B), and this results in a C-terminal truncation of the TRAF3IP2 protein. The truncation removes the second TRAF binding site (EEERPA) and the N-terminal coiled-coil domain (21) (Fig. 5D). These lines of evidence suggest that the Traf3ip2 gene is a candidate gene for the adjm locus.

To examine the effect of the adjm mutation on Traf3ip2 RNA expression, a series of primer sets were used to amplify the 5′ and 3′ regions of cDNA fragments (Table II). RT-PCR analysis of splenic RNA resulted in a single band from wild-type, adjm/+, and adjm/adjm mice. However, the Traf3ip2 transcript level appeared to be reduced in adjm mutants relative to that of GAPDH controls (Fig. 6A). From a quantitative analysis of band intensities, we estimated that Traf3ip2 transcripts in adjm/+ and adjm/adjm mice are ~40 and 10% as abundant as those in wild-type mice, respectively.

To confirm the results described above, we carried out quantitative RT-PCR using total RNA isolated from 5-wk-old BALB/c-+/+ and BALB/c-adjm/adjm mouse skin, spleen, and brain. The relative amounts of Traf3ip2 transcripts in skin, spleen, and brain from BALB/c-adjm/adjm mice were ~40% as abundant as those in wild-type mice (Fig. 6B). Therefore, the relative amounts of Traf3ip2 transcripts found in BALB/c-adjm/adjm by RT-PCR, as mentioned above, are greatly underestimated.

The adjm mice showed prominent defects in TRAF3IP2 protein expression in multiple tissues. Western blot analysis of protein extracts from spleens of wild-type mice using a TRAF3IP2-specific Ab yielded bands of the expected size (~72 kDa). However, protein extracts from spleens of adjm/adjm mice showed no specific Ab labeling in the positions expected for TRAF3IP2 (Fig. 6C).

Recently, a new lineage of CD4 Th cells that produced IL-17, the Th17 lineage, was identified (20, 21). IL-17 is a proinflammatory cytokine that upregulates the expression of inflammatory genes in fibroblasts, epithelial cells, and some other cell types. IL-17 levels are elevated in patients with allergic and autoimmune diseases (see Ref. 21 and references therein). To examine the relationship between IL-17 expression and seriousness of dermatitis, we estimated the relative amounts of eight major cytokines in the serum of C57BL/6-+/+ and C57BL/6-adjm/adjm mice with mild or severe dermatitis by Ab microarray. Of these eight cytokines, we found that IL-17 was expressed strongly in the serum of affected female mice (C57BL/6-adjm/adjm) but not in that of affected male mice (Fig. 7A).

We then attempted to examine the numbers of Th17 cells in the lesional skin of severely affected mice. However, no Th17 cells were found (data not shown), suggesting that the population of Th17 cells in the skin is very low (<0.01% in CD4⁺ T cells). When the spleen was examined, twice as many Th17 cells were found in the spleens of severely affected homozygous female mice compared with unaffected female mice, although little difference was found in Th17 cells of affected male mice (Fig. 7B). This observation is consistent with the results concerning the relative amounts of IL-17 in affected and nonaffected mice (Fig. 7A). We do not have conclusive evidence that shows whether the differences in Th17 cell number and IL-17 levels in female mice are specific to the C57BL/6 strain or common to other genetic backgrounds. However, our preliminary results suggest that the differences are strain specific, because the relative levels of IL-17 in C57BL/6 were 1.2 times those of BALB/c mice (data not shown). This point will be addressed in future experiments.

Two mouse strains, NC and NOA, have been reported as possible models of human AD. Although the phenotypes of both strains are controlled by polygenic traits, the major genetic determinant is located on chromosome 9 (derm1 locus) in NC (13) and on chromosome 14 in NOA (9). NOA also possesses two additional modifier genes on chromosomes 7 and 13 (14). The phenotype in KOR-adjm/adjm, dermatitis and hyper-IgE-emia, is controlled by a single locus on chromosome 10 (Fig. 5). Because the mutant gene in the KOR-adjm/adjm strain differs from those in the strains mentioned above, it is evident that KOR-adjm/adjm is a new model for human AD. This conclusion also is supported by the results of linkage analysis of NC, which revealed no close genetic association between dermatitis and hyper-IgE-emia. Conversely, a close genetic association was found in the congenic KOR-adjm/adjm strains described in this paper (Fig. 4). Animals homozygous for the adjm mutation showed distinct sex differences in hyper-IgE-emia (Fig. 2), life span, the amount of serum IL-17, and the splenic Th17 cell population (Fig. 7). These lines of experimental evidence suggest the existence of modifier gene(s) for the adjm locus that are controlled by sex hormones. We consider this issue very interesting and important with respect to discovery of the molecular regulatory mechanisms of the adjm gene and signaling pathways (e.g., the IL-17/IL-17R signaling pathway [see Ref. 21 and references therein]) by hormone(s).

We propose that the mutation detected in Traf3ip2 (Act1/CIKS) is responsible for the phenotype observed in KOR-adjm/adjm. TRAF3IP2 is associated with and activates IκB kinase and stimulates both the NF-κB and the JNK signaling pathways (15). TRAF3IP2 does not possess any enzymatic domains, but it contains a helix-loop-helix at its N terminus and a coiled-coil at its C terminus, along with two putative TRAF binding sites, EEESE (residues 38–42) and EERPA (residues 333–337). Importantly, the adjm mutation deletes the C-terminal TRAF binding site and the coiled-coil region (17, 21). Previous studies demonstrate that deletion of the C-terminal region (residues 316–573) abolishes TRAF3IP2-induced NF-κB activation (15). Therefore, it is likely that the adjm mutation leads to the loss of TRAF3IP2 function, although the truncated protein is produced (Fig. 5D).

Together with recent findings examining the dual function of TRAF3IP2 (21), the results of this study also explain why a tight genetic association exists between hyper-IgE-emia and dermatitis. The forced expression of TRAF3IP2 in 293T cells and HeLa cells has been reported to activate both NF-κB and JNK (18). Endogenous TRAF3IP2 was recruited to CD40, a member of the TNFR superfamily, in human epithelial cells upon stimulation with CD40 ligand (CD40L). Furthermore, transfection of Traf3ip2 into TRAF3IP2-negative epithelial cells rendered them sensitive to CD40L-induced NF-κB activation (17, 18, 25, 26). Despite these results, the recent generation of TRAF3IP2-null mice shows that TRAF3IP2 functions as a negative regulator rather than an activator in CD40- and BAFF-mediated signaling (18, 19, 21). Therefore, the truncation of TRAF3IP2 may cause a malfunction of this adaptor protein, which results in the hyper-IgE-emia observed in our mutant mouse.

IL-17 plays an important role in promoting tissue inflammation in host defenses against infection and in autoimmune diseases. IL-17 was first discovered as the founding member of the IL-17 cytokine family and has been reported to regulate the expression of proinflammatory cytokines, chemokines, and matrix metalloproteinases (27, 28). Previous work has demonstrated that IL-17 activates the NF-κB and MAPK pathways and requires TRAF6 to induce IL-6 (20, 26, 27). Additionally, Chang et al. (27) have demonstrated that the intracellular region of the IL-17R family proteins shares sequence homology with the Toll/IL-1R domain and with Traf3ip2. Furthermore, TRAF3IP2 and IL-17R directly associate, likely via a homotypic interaction. In fibroblasts, TRAF3IP2 deficiency abrogates IL-17–induced cytokine and chemokine expression as well as the induction of C/EBPβ, C/EBPδ, and IκBζ. Yamamoto et al. (29) reported that mice deficient for IκBζ, a member of the IκB family involved in the regulation of NF-κB signaling, develop AD-like skin lesions with acanthosis and lichenoid changes by 10 wk of age. Together with the results of the current study and those of the study on Traf3ip-null mice (19), this observation suggests that dysregulation of the NF-κB signaling pathway leads to the development of AD-like skin disease in mice. The KOR-adjm/adjm and congenic mice established in the current study therefore will be useful animal models of human AD.

The absence of TRAF3IP2 results in a selective defect in IL-17–induced activation of the NF-κB pathway. These results indicate that TRAF3IP2 is a membrane-proximal adaptor of the IL-17R and that it has an essential role in the induction of inflammatory genes. Thus, this study not only reveals an immediate signaling mechanism downstream of an IL-17 family receptor for the first time but also has implications for the therapeutic treatment of various immune diseases.

Th17 cells are a novel subset of CD4⁺ T cells that are regulated by TGF-β, IL-6, and IL-23. Th17 cells have been shown to be important in inflammation and in the control of certain bacteria. After stimulation with IL-17, the recruitment of TRAF3IP2 to IL-17R requires the IL-17R conserved cytoplasmic SEFIR (similar expression to fibroblast growth factor genes and IL-17Rs) domain and is followed by the recruitment of the TGF-β–activated kinase 1 and an E3 ubiquitin ligase (TRAF6), both of which mediate the downstream activation of NF-κB. IL-17–induced expression of inflammation-related genes was abolished in TRAF3IP2-deficient primary astroglial and gut epithelial cells, and this reduction was associated with less severe inflammatory disease in vivo in both autoimmune encephalomyelitis and dextran sodium sulfate-induced colitis (20). Therefore, our mutant mice provide an excellent tool for detailed analysis of the function of the N-terminal TRAF3IP2 protein in these pathways because they express a truncated protein.

The adapter molecule TRAF3IP2 regulates autoimmunity through both T and B cell-mediated immune responses. The coordinated regulation of T and B cell-mediated immune responses plays a critical role in the control and modulation of autoimmune diseases. Whereas the TRAF3IP2 molecule is an important negative regulator of B cell-mediated humoral immune responses through its role in CD40L and BAFF signaling (19), recent studies have shown that TRAF3IP2 is also a key positive regulator of the IL-17 signaling pathway and that it is critical for Th17-mediated autoimmune and inflammatory responses. The dual functions of TRAF3IP2 are evident in TRAF3IP2-deficient mice that display B cell-mediated autoimmune phenotypes, including a dramatic increase in peripheral B cells, lymphadenopathy and splenomegaly, hypergammaglobulinemia, and Sjögren’s disease in association with lupus nephritis (19, 21), but show resistance to Th17-dependent experimental autoimmune encephalomyelitis and colitis (20, 21, 30). Such seemingly opposite functions of TRAF3IP2 in the CD40 or BAFFR and the IL-17R signaling pathways are orchestrated by the different domains of TRAF3IP2. Whereas TRAF3IP2 interacts with the IL-17R through the C-terminal SEFIR domain in TRAF3IP2, TRAF3IP2 is recruited to CD40 and BAFFR indirectly by TRAF3 binding to the TRAF binding site in TRAF3IP2 (17, 21). Such a delicate regulatory mechanism may provide a common vehicle that promotes the balance between host defense to pathogens and tolerance to self (21).

Conflicting evidence has been reported regarding the association between IL-17 production and AD exacerbation. Two reports suggest that Th17 cells exacerbate dermatitis because increased numbers of Th17 cells were found in the peripheral blood and acute lesional skin of AD patients (31). Kawakami et al. (32) also support this idea with their study of NK cell function in AD patients (23). In contrast, two reports suggest that low IL-17 expression or reduced responses to IL-17 in the skin cause AD-like symptoms (24, 33). To resolve these conflicting lines of evidence, we determined the number of Th17 cells in affected skin regions of the faces of homozygous adjm mice. No Th17 cells were observed in the skin (data not shown), although these cells were found in the spleens of severely affected homozygous female mice at twice the number as in nonaffected female mice (Fig. 7B). Similarly, IL-17 protein levels in the peripheral blood of affected female mice were 3.3- to 6.5-fold higher than those of unaffected mice. Conversely, few differences were observed between affected and unaffected male mice (Fig. 7). This is consistent with the finding that female mice were affected more severely than male mice by hyper-IgE-emia and dermatitis (Figs. 2–4). On the basis of our results with female mice, it is suggested that reduced IL-17 responses cause AD. However, it remains unknown why no correlation was found between IL-17 expression and features of AD in male mice.

Act1/TRAF3IP2-deficient mice showed a drastic increase of peripheral B cells and plasma cells, together with lymphadenopathy and splenomegaly. Serum levels of IgG and IgE in these animals were >10-fold higher than those of normal mice (19). These phenotypes occur mainly as a result of TRAF3IP2 deficiency in B cells, because B cell-specific TRAF3IP2-deficient mice showed similar, although less severe, phenotypes. It is therefore probable that the hyper-IgE-emia observed in KOR-adjm/adjm mice and congenic adjm/adjm mice is due to increased CD40- and BAFFR-mediated B cell stimulation and Ig class switching in the absence of functional TRAF3IP2. Importantly, lymphadenopathy and splenomegaly were observed in all of the mutant mice generated in this study.

Hyper-IgE syndrome (HIES; i.e., Job’s syndrome) is an autosomal dominant syndrome that is often associated with dermatitis. Recently, close association of this syndrome with mutations at the STAT3 locus (34–36) has been demonstrated. These reports showed that HIES phenotypes can cause a specific deficit in the generation of Th17 cells and IL-17 production (35, 36). Although it is interesting that separate mutations affecting IL-17 signaling result in a nearly identical AD phenotype and hyper-IgE-emia, the two mutants show opposite phenotypes with respect to the generation of Th17 cells and IL-17 production. With respect to IL-17R signaling (21), it is reasonable that both mutants show nearly identical phenotypes, because both mutations attenuate downstream IL-17/IL-17R signaling. HIES mutants suffer an attenuation due to the insufficiency of IL-17, whereas the adjm mutation also causes attenuation through the dysfunction of the IL-17R signaling complex brought about by the presence of deficient TRAF3IP2 protein.

Although we assume that TRAF3IP2-null mice in the BALB/c background develop less severe dermatitis than KOR-adjm/adjm mice (TRAF3IP2-null mice have been shown to develop skin inflammation with epidermal hyperplasia and T cell infiltration), detailed studies of dermatitis development in this mouse strain have not been reported. However, we found that the AD-like skin disease that developed in BALB/c-adjm/adjm mice was less severe than that which occurred in KOR-adjm/adjm mice. Although, on the basis of the phenotypes of adjm mutant and TRAF3IP2-null mice, there is no doubt that a deficiency in functional TRAF3IP2 is responsible for the development of dermatitis in these strains, the underlying mechanism remains to be clarified.

Because, as observed in BAFF transgenic mice (37–39), TRAF3IP2-null mice have been shown to produce autoantibodies against DNA and histones (19), it is possible that autoantibodies against some components of the skin trigger dermatitis. Expression of TRAF3IP2 is not restricted to B cells but is ubiquitous. It is therefore also possible that the deficiency of functional TRAF3IP2 leads to dysregulated homeostasis of skin cells, such as epithelial cells and fibroblasts. Indeed, TRAF3IP2-deficient embryonic fibroblasts demonstrated enhanced NF-κB activation when stimulated via ectopically expressed CD40 or BAFFR. It has been suggested that TRAF3IP2 in non-B cells may play a role in the signaling events mediated by other members of the TNFR superfamily, especially the subsets of signaling proteins that use TRAFs. If this is the case, then B cell-specific TRAF3IP2-deficient mice may not develop dermatitis. However, studies of the development of dermatitis in such mice have not been reported.

We thank Dr. Y. Nishito for the cytokine Ab array analysis and S. Yamada, C. Matsuo, Y. Seino, M. Iyobe, T. Mizuno, and K. Mukai for technical support.

Disclosures The authors have no financial conflicts of interest.