Signs of atopic dermatitis and contact dermatitis affected by distinct H2-haplotype in the NC/Nga genetic background

We recently advocated in favour of naming a novel H2-haplotype consisting of Kd, D/Ldm7, I-Ak and I-Ek in the atopic dermatitis (AD) mouse model NC/Nga as “H-2nc.” The role of the H2-haplotype in AD development was investigated in H2b-congenic NC/Nga mice (NC.h2b/b and NC.h2b/nc) established by backcrossing. A severe 2,4-dinitrofluorobenzene (DNFB)-induced dermatitis in NC/Nga was alleviated partially in NC.h2b/nc and significantly in NC.h2b/b. The AD phenotype was correlated with thymic stromal lymphopoietin (TSLP)-epidermal expression levels and serum levels of total IgE and IL-18/IL-33. Histologically, allergic contact dermatitis (ACD) was accompanied by lymphocytes and plasma cells-infiltrating perivasculitis in NC.h2b/nc and NC.h2b/b and clearly differed from AD accompanied by neutrophils, eosinophils and macrophages-infiltrating diffuse suppurative dermatitis in NC/Nga. Interestingly, IFN-γ/IL-17 production from autoreactive CD4+ T-cells remarkably increased in DNFB-sensitised NC.h2b/b but not in NC/Nga. Our findings suggest that AD or ACD may depend on haplotype H-2nc or H-2b, respectively, in addition to the NC/Nga genetic background.

proteases, and adhesion molecules. In humans, the deficiency of the skin barrier molecule filaggrin leads to alkalisation of the skin, favouring bacterial infections and increased metal ion-protein hapten complexes 12 . In NC/ Nga mice having developed AD, the expression of filaggrin is low, but increases with treatment of an opioid analgesic drug JTC-801 13 . Our group recently found that the clonal deletion of T cell repertoires with specific T cell receptor Vβ chains is induced by the expression of two endogenous superantigens (Mls-1 a , MMTV(SHN)) in the genetic background of NC/Nga mice 14 . This finding was in line with a previous report showing that under conventional condition, the Th1 dominant state mediated by IL-12 or IL-18 after Staphylococcal enterotoxin B (SEB) exposure was inhibited by the absence of Vβ8 + T-cells in NC/Nga mice, resulting in the induction of the Th2 dominant state 15 .
Genome-wide expression analysis indicates that besides inflammatory factors or the skin barrier system, the human genotype of AD sensitivity is also associated with MHC regions 16,17 . We recently proposed the name "H-2 nc " for the novel H2-haplotype consisting of K d , D dm7 /L dm7 -hybrid mutant (D/L dm7 ), I-A k and I-E k in NC/Nga mice 18 . Here, H-2 b congenic NC/Nga mice (NC.h2 b/b or NC.h2 b/nc ) established by backcrossing (8 generations) were investigated for their sensitivity to DNFB-induced dermatitis. The role of the H2-haplotype in AD development in the NC/Nga strain is discussed.

Results
Establishment of H-2 congenic NC/Nga mice (NC/Nga). To investigate whether the H2-haplotype (H-2 nc ) in NC/Nga mice (NC/Nga) is a cause for high sensitivity to AD-like disease, we established H-2 congenic mice (NC.h2 b/b , NC.h2 b/nc and NC.h2 nc/nc ) by backcrossing (8 generations) the NC/Nga strain (Fig. 1A). Because the haplotype H-2 nc in NC/Nga strain shows the phenotype (K d , I-A k , I-E k ) 18 , we tested peripheral blood mononuclear cells (PBMCs) from filial generation after backcrossing with NC/Nga strain by flow cytometry on the expression of K d and K b . K d -negative and K b -positive populations, K d -positive and K b -positive populations and K d -positive and K b -negative populations in PBMCs indicate the haplotypes H-2 b , H-2 b/nc and H-2 nc , respectively (Fig. 1B). It was confirmed that the H2-phenotype of NC.h2 b/b shows K b , D b , and I-A b (Fig. 1C).
In DNFB-induced dermatitis model, the degrees of inflammation and dermatitis scores in NC.h2 b/nc and NC.h2 b/b are alleviated compared with that in NC/Nga. DNFB-induced dermatitis in NC/Nga (or NC.h2 nc/nc ), NC.h2 b/b and NC.h2 b/nc was evaluated by both a scoring index of AD 19 and the myeloperoxidase (MPO) activity using in vivo Imaging System. The degree of dermatitis score in DNFB-applied NC.h2 b/b was significantly alleviated compared with that in NC/Nga at days 12 and 14 (p < 0.05, ANOVA), whereas DNFB-applied NC.h2 b/c was partially alleviated ( Fig. 2A,C). Particularly, dryness and erosion were strongly reduced in NC.h2 b/b and NC.h2 b/c mice compared with the effects seen in NC/Nga (Fig. 2D). The result of in vivo Imaging System at day 14 were in line with the degree of dermatitis scores, reflecting the degree of inflammation with MPO production due to the accumulation of activated neutrophils and macrophages (Fig. 2B).  (Fig. 3A). Thickening of the epidermis and dermis in the skin or ear were strongest developed in the following order: NC/Nga < NC.h2 b/nc < NC.h2 b/b (Fig. 3A). A sign of AD-like lesion accompanied by band-like infiltration of neutrophils, eosinophils and macrophages in the dermis was observed in NC/Nga (Fig. 3B,C). Additionally, signs of ACD-like lesions 20 accompanied by fibroblast proliferation in dermis (Fig. 3E) and distinctive perivasculitis with infiltrating lymphocytes and plasma cells (Fig. 3F) were mainly observed in both NC.h2 b/nc and NC.h2 b/b . Although high expression levels of thymic stromal lymphopoietin (TSLP) was remarkably detected in the NC/Nga epidermis (Fig. 3D), TSLP expression was observed in perivascular tissues rather than in the epidermis in NC.h2 b/nc and NC.h2 b/b (Fig. 3G). Although it was at a low level, serum TSLP levels also significantly increased in DNFB-induced dermatitis of NC/Nga than in that of NC.h2 b/b (p < 0.05, ANOVA; Supplemental Figure S1A,B). Furthermore, serum TNF-α and CCL2 levels in DNFB-applied NC/Nga also partially increased compared with those in DNFB-applied NC.h2 b/b . However, there is no difference in serum IL-1β levels in both mice (Supplemental Figure S1A, B).

In axillary lymphadenopathy with DNFB-induced dermatitis, the production of IFN-γ and IL-17 from autoreactive CD4 + T-cells in NC.h2 b/b were enhanced compared with that in NC/Nga.
Mice with DNFB-induced dermatitis exhibit an axillary lymphadenopathy phenotype (data not shown) that we further investigated in this study. As IL-4, IFN-γ and IL-17 are hardly detected in serum, we assessed their productions from CD4 + T-cells in axillary lymph nodes (LNs) by co-culture of these CD4 + T-cells with naïve CD11b + myeloid cells (as antigen-presenting cells) whose growth response was stopped by X-ray irradiation. The productions of IFN-γ and IL-17, but not of IL-4, from proliferating autoreactive CD4 + T-cells in NC.h2 b/b were remarkably detected compared with NC/Nga (p < 0.05, ANOVA) ( Fig. 4A-C). This result indicates that the total IgE production in DNFB-induced dermatitis of the NC/Nga strain is not dependent on IL-4-based mechanisms because IL-4 was hardly detected.

Discussion
Here, we investigated the role of the H2-haplotype in the NC/Nga genomic background using H2-congenic mice (NC.h2 b/b , NC.h2 b/nc and NC.h2 nc/nc ) established by backcrossing the NC/Nga strain. Severe dermatitis induced by repeated DNFB-application under SPF condition in NC/Nga was alleviated partially in NC.h2 b/nc and significantly in NC.h2 b/b as shown by decreased epidermal expression levels of TSLP and serum levels of total IgE, TSLP, IL-18 and IL-33. Histologically, ACD-like lesion of NC.h2 b/nc and NC.h2 b/b differed from AD-like lesions observed in NC/Nga. With the poor expression of IL4, the total IgE production in DNFB-induced dermatitis of NC/Nga is likely stimulated via an IL-4-independent mechanism.
The IL-1 family members IL-18 and IL-33 are highly inflammatory cytokines constitutively expressed in barrier cell types 21 , acting as regulators of innate and acquired immune responses by amplifying both Th1 and Th2 responses with or without TCR activation 22 . IL-18 and IL-33 signal their biologically activities through the heterodimeric receptors IL-18R and IL-33R in mast cells. IL-18 serum levels are elevated in AD patients or AD-induced NC/Nga mice 23,24 . IL-33 is an important mediator of allergy through its ability to induce Th2 cytokines and has been associated with development of severe AD 25 . IL-33 initiates allergic inflammation by activating innate lymphoid cells type 2 (ILC2s) to produce large amounts of Th2 cytokines IL-5 and IL-13 responsible for eosinophilia and IgE-class switching, respectively 7,26-28 . Our results indicate that autoreactive CD4 + T-cells in axillary lymphadenopathy of DNFB-sensitised mice produce increased levels of IFN-γ and IL-17 in NC.h2 b/b compared with NC/Nga. Autoreactive CD4 + T-cells were previously reported to be induced in the DNFB-induced ACD model 29 . We confirm in this study that autoreactive CD4 + T-cells lead to ACD-like lesions in both DNFB-induced NC.h2 b/b and BALB/c mice. Although there are no reports on autoallergens specific to autoreactive CD4 + T-cells in ACD-developed mice, it is conceivable that autoreactive CD4 + T cells recognizing a carrier-protein of DNFB (a hapten) as self-antigens may be generated. In humans, the autoallergens Hom s 1-5 had been identified by screening of a human epithelial complementary DNA library for IgE binding of sera from AD patients [30][31][32] . Hom s 2 corresponds to the alpha-chain of the nascent polypeptide-associated complex (α-NAC) and α-NAC-specific autoreactive CD8 + T-cells were identified as secreting both IL-4 and IFN-γ in AD patients 33 . However, autoreactive CD4 + T-cells from PBMCs are not generated by α-NAC peptides in AD patients 34 . Recently, we detected IFN-γ-producing autoreactive CD8 + T-cells from NC/Nga mice with DNFB-induced AD that were expanded by co-culturing with dendritic cells treated with proteasome inhibitor MG132 (data not shown). Although an MHC class I mutant D/L dm7 has been downregulated by degradation with proteasome activation, the increase in D/L dm7 by MG132 addition was associated with the generation of autoreactive CD8 + T-cells, but not of autoreactive CD4 + T-cells (data not shown). Thus, autoreactive CD8 + T-cells or CD4 + T-cells are induced in AD-like lesions or in ACD-like lesions, respectively. Although it is a matter of speculation, the autoreactive Th1/Th17-like cells detected in ACD-like lesions may be inhibited in vivo by immunosuppressive cells other than CD4 + T-cells as these lesions display a weak inflammation phenotype compared with AD-like lesions and no sign of autoimmune disease and may play a role of inhibiting the generation of autoreactive CD8 + T-cells in changing AD-like lesions into ACD-like lesions.

Materials and Methods
Mice. NC/Nga (H-2 nc ) (male and female) and C57BL/6 (H-2 b ) mice (male) (Japan SLC, Hamamatsu, Japan) were maintained in SPF conditions, and used at 18-20 weeks of age. H2-congenic NC/Nga strain (NC.h2 nc/nc , NC.h2 b/nc and NC.h2 b/b ) were generated by backcrossing (8 generations) of NC/Nga mice. Mice were maintained by breeding in SPF air conditions with microbiological monitoring tests (twice/year). All mice were maintained in our full-barrier animal facility under controlled temperature, humidity and 12 hour light/dark regimen. All experiments were approved by the Institutional Review Board for animal studies of NVLU, and performed following the guidelines provided by the Committee.

2,4-dinitrofluorobenzene (DNFB)-induced dermatitis model.
The methods in this study have been described previously 2 . AD-like skin lesions were induced by the repeated application of 25 µl of 0.15% DNFB (Wako Pure Chemicals, Tokyo, Japan) in acetone/olive oil (3:1) to the skin of the ears, calva, and neck on days 0, 3, 5, 7, 9, 11, and 13. The control mice were applied the acetone/olive oil (3:1) alone as the vehicle of DNFB. Dermatitis was evaluated by assigning an inflammation score 19 . Briefly, inflammation of the face, ears, and the anterior part of the body was scored as follows: 0, none; 1, mild; 2, moderate; and 3, severe. This scoring was based on the severity of erythema/hemorrhage (e/h), edema (ed), excoriation/erosion (e/e), and scaling/dryness (s/d), and total points were evaluated as the severity of dermatitis.
In vivo Imaging for MPO activity. A chemiluminescent in vivo reagent for monitoring inflammation (XenoLight RediJect Inflammation probe, ParkinElmer) was administrated by intraperitoneal (i.p.) injection at 120 μL /mouse. MPO activity was analyzed by using IVIS in vivo imaging (IVIS Lumina II, ParkinElmer) at 10 minutes post i.p. injection of the probe under Pentobarbital anesthesia (Somnopentyl, Kyoritsu seiyaku).
ELISA. Blood samples were collected from the orbital sinus in mice by using hematocrit tube, and serum samples were obtained by centrifugation. The total serum IgE was measured by a mouse IgE ELISA kit (Shibayagi Co., Ltd., Gunma, Japan). Serum IL-18 was measured by mouse IL-18 ELISA kit (MBL). Serum IL-33 was measured by mouse IL-33 Quantikine ELISA kit (R&D system). Culture supernatants were collected from each assay. The IFN-γ and IL-4 were measured by mouse Th1/Th2 ELISA Ready-SET-Go! (eBioscience). The IL-17 was measured by mouse IL-17 Quantikine ELISA kit (R&D system). All assays were following the manufacturer's instructions.
Histopathology and immunohistochemistry. Histopathological analyses were performed on a minimum of forth animals per experimental group. Tissues were immersion-fixed in 10% buffered formalin and processed by routine methods for paraffin sectioning. Paraffin sections, 5 μm thickness, were stained with hematoxylin and eosin (H&E), and examined by light microscopy.
Statistical analysis. Statistical analysis was performed with ANOVA using Excel (Microsoft) and StatPlus (AnalystSoft, Alexandria, VA). A p-value < 0.05 was considered significant. ANOVA test was performed after normal distribution test (Shapiro-Wilk test).