Spontaneous atopic dermatitis in mice with a defective skin barrier is independent of ILC2 and mediated by IL‐1β

Abstract Background Atopic dermatitis (AD) is one of the most common skin diseases with a multifactorial etiology. Mutations leading to loss of skin barrier function are associated with the development of AD with group 2 innate lymphoid cells (ILC2) promoting acute skin inflammation. Filaggrin‐mutant (Flgft/ft) mice develop spontaneous skin inflammation accompanied by an increase in skin ILC2 numbers, IL‐1β production, and other cytokines recapitulating human AD. Here, we investigated the role of ILC2, effector cytokines, inflammasome activation, and mast cell function on the development of chronic AD‐like inflammation in mice. Methods Mice with a frameshift mutation in the filaggrin gene develop spontaneous dermatitis. Flgft/ft mice were crossed to cell‐ or cytokine‐deficient mouse strains, or bred under germ‐free conditions. Skin inflammation was scored, and microbiome composition was analyzed. Skin protein expression was measured by multiplex immunoassay. Infiltrating cells were analyzed by flow cytometry. Results Wild‐type and Flgft/ft mice significantly differ in their microbiome composition. Furthermore, mutant mice do not develop skin inflammation under germ‐free conditions. ILC2 deficiency did not ameliorate chronic dermatitis in Flgft/ft mice, which was also independent of IL‐4, IL‐5, IL‐9, IL‐13, IL‐17A, and IL‐22. Inflammation was independent of NLRP3 inflammasome activation but required IL‐1β and IL‐1R1‐signaling. Mechanistically, IL‐1β promoted hyperactivation of IL‐1R1‐expressing mast cells. Treatment with anti‐IL‐1β‐antibody alleviated dermatitis exacerbation, while antibiotic intervention ameliorated dermatitis in neonatal mice but not in adults with established inflammation. Conclusions In summary, we identified a critical role for the microbiome and IL‐1β mediating chronic inflammation in mice with an impaired skin barrier.


| INTRODUC TI ON
Atopic dermatitis (AD) is a common eczematous pruritic disease with an onset at an early age, affecting up to 30% of children in the Western world. Etiology of AD is multifactorial including genetic predisposition, environmental factors, and immune status, leading to a high complexity in clinical presentation. 1-3 Predisposing genetic factors for the development of AD include mutations in genes affecting the integrity of the skin barrier, such as mattrin (TMEM79) and filaggrin (filament aggregation protein, FLG). [4][5][6] Filaggrin mutations were found in 30% of AD patients in Poland, 7 China (26.0%), 8 and Korea (15.7%), 9 while healthy individuals had none. Importantly, filaggrin expression is downregulated in AD patients independent of their FLG genotype as a consequence of increased type 2 cytokines contributing to the aggravation of disease. 10 We have previously separated and described the two mutated genes-Tmem79/mattrin and filaggrin-leading to the allergic skin phenotype of flaky tail mice. 5,11 Single mutant mice both have a defective skin barrier, and both spontaneously develop AD-like inflammation.
Pathogenesis in Tmem79 ft/ft mice is dependent on adaptive immunity, while Flg ft/ft mice develop dermatitis through innate immune cells. 11 However, the mechanisms underlying inflammation are unclear.
Results: Wild-type and Flg ft/ft mice significantly differ in their microbiome composition. Furthermore, mutant mice do not develop skin inflammation under germ-free conditions. ILC2 deficiency did not ameliorate chronic dermatitis in Flg ft/ft mice, which was also independent of IL-4, IL-5, IL-9, IL-13, IL-17A, and IL-22. Inflammation was independent of NLRP3 inflammasome activation but required IL-1β and IL-1R1-signaling. Mechanistically, IL-1β promoted hyperactivation of IL-1R1-expressing mast cells.
Treatment with anti-IL-1β-antibody alleviated dermatitis exacerbation, while antibiotic intervention ameliorated dermatitis in neonatal mice but not in adults with established inflammation.
Conclusions: In summary, we identified a critical role for the microbiome and IL-1β mediating chronic inflammation in mice with an impaired skin barrier.

K E Y W O R D S
atopic dermatitis, filaggrin, IL-1β, innate lymphoid cells, microbiome

G R A P H I C A L A B S T R A C T
Filaggrin deficiency leads to skin dysbiosis early after birth altering adult immune responses, while mice raised under germ-free conditions remain disease-free. NLR Family Pyrin Domain Containing 3-independent processing of IL-1 in the skin promotes atopic dermatitis (AD)-like ILC2-independent inflammation. IL-1 deficiency or targeting IL-1 by monoclonal antibodies ameliorates dermatitis. IL-1R1-expressing dermal mast cells are key responders to IL-1, acquire a hyperactive phenotype, and promote AD-like inflammation.
As an atopic disorder, AD is classically considered a type 2-driven immunopathology involving type 2 T helper (Th2) cells, interleukin (IL)-4, IL-5, IL-9, and IL-13, as well as IgE, mast cells, basophils, and eosinophils-with more recent data expanding this view to include Th17 and IL-22 cellular responses in the genesis of AD. 12 While T cells are promoting inflammation in certain instances, [13][14][15] they are largely dispensable in the Flg ft/ft model. Mast cells have long been associated with AD, and increased numbers are found in the skin of atopic patients. 16 Upon activation by cytokines, FcεRI-bound IgE, or pathogen-and danger-associated molecular patterns, mast cells can release large amounts of pro-inflammatory mediators, such as tumor necrosis factor (TNF). 17 suggested that the microbiota promotes upregulation of IL-17A and the infiltration of neutrophils and eosinophils into the skin. 33 Polymorphisms in members of the IL-1 family of cytokines and their receptors are associated with skin disorders, such as cutaneous lupus erythematosus, psoriasis, and atopic dermatitis. 34,35 We have previously reported increased IL-1α, IL-1β, and IL-1R1 expression in skin of Flg ft/ft mice and AD patients with mutations in FLG. 36 In the present study, we set out to investigate the mechanisms underlying AD-like inflammation in Flg ft/ft mice. We discovered that ILC2-while required for acute MC903-induced dermatitis-were dispensable for spontaneous AD-like inflammation in Flg ft/ft mice with an impaired skin barrier. Instead, the development of skin inflammation was dependent on an interplay between microbiota, IL-1β, and mast cells.

| Mice
The following mice were backcrossed onto the

| Scoring of skin inflammation
Severity of skin inflammation was clinically scored (total range: 0-12) by macroscopic diagnostic criteria as previously described. 5 The total score is the sum of individual scores ranging from 0 to 3 (0, none; 1, mild; 2, moderate; 3, severe) that were applied to edema, erythema, scaling, and erosion.

| Preparation of bone marrow-derived mast cells (BMMC) and stimulation
Bone marrow was isolated from the femur and tibia of donor mice and cultured in media (RPMI + 10% FBS + L-glutamine + penicillin/streptomycin + HEPES + non-essential amino acids) containing 10 ng/mL SCF and 10 ng/mL IL-3 (R&D systems) for a total of 4 weeks with media changes twice a week. 54

| Microbiome analysis
Skin microbiome samples were acquired by exposing sterile swabs to the ear skin of Flg ft/ft and wild-type mice, as previously published. 56 Mice were kept in the same or adjacent cages, looked after by the same person using the same products. Same surface area was sampled for all age-and sex-matched mice. To avoid cross-contamination, sterile gloves were changed between each sample collected. Samples were instantly frozen in liquid nitrogen, and 16S rRNA gene sequencing and microbiome analysis was performed by Second Genome (San Francisco, CA), as previously described. 57

| Axenic mouse model generation
Male and female Flg ft/ft mice were shipped from Trinity College Dublin to the Instituto Gulbenkian De Ciência in Portugal and re-derived by embryo transfer from a quarantine facility into SPF housing. Subsequent litters were generated by timed-pregnancies. Fetuses were transferred to GF isolators and fostered by GF C3H mothers, as described in the relevant EMMA protocol (http://strains.emmanet.org/protocols/GermFree_0902.pdf). Germ-free and age-matched SPF control litters were raised and maintained under strictly identical conditions (food, water, humidity, temperature), except the microbiological status.

| Statistics
GraphPad Prism (version 7) was used to generate graphs and for statistical analysis. Area under curve (AUC), Student's t test, and ANOVA were used to determine statistical significance. P-values < 0.05 were considered statistically significant.
Please refer to the Supporting Information for additional materials and methods.

| Flg ft/ft mice develop spontaneous atopic dermatitis-like skin inflammation
We sought to investigate the innate mechanisms that elicit in- Expression of genes associated with AD (Il33, Il1a, Il1b) was significantly altered in the absence of microbiota ( Figure 2I). Importantly,   Clinical Severity Score

| Atopic dermatitis is independent of group 2 innate lymphoid cells
The appearance and severity of dermatitis in Flg ft/ft mice is dependent on cells of the innate immune system ( Figure 3A) 11 with a significant increase in dermal ILC2 numbers in Flg ft/ft mice ( Figure 1F). Therefore, we tested whether ILC-deficient mice on  Figure 3C). These data from three separate and distinct models of ILC2 deficiency led us to conclude that ILC2 were dispensable for skin inflammation in this spontaneous and chronic model of AD-like inflammation in mice with a defective skin barrier.

| Group 2 innate lymphoid cells promote acute skin inflammation
Daily application of MC903 elicits acute AD-like skin inflammation 28 ( Figure 3D). We, and others, have previously shown using antibody-mediated depletion models as well as bone marrow chimeric mice that development of MC903-elicited skin inflammation is dependent on ILC2. 25,27 Indeed, using ILC2-deficient Rora fl/sg Il7ra Cre/+ mice, we could confirm that ear swelling was ameliorated in acute dermatitis ( Figure 3D). As expected, ILC2 were not present in the inflamed skin of Rora fl/sg Il7ra Cre/+ mice ( Figure 3E). Moreover, cellular infiltration was blunted in ILC2-deficient mice including the recruitment of granulocytes, Th2 and Th17 cells ( Figure 3E,F). Furthermore, type 2-associated cytokine production in draining LN was impaired ( Figure 3G). Because of the divergent roles ILC2 play in the initiation processes of acute chronic dermatitis, we sought to determine which other factors promote AD in the clinically more relevant Flg ft/ft model.   . Graph shows the mean ± SEM from 6 to 7 mice per group. C, Macroscopic clinical scoring of wild-type (dashed black), Rora fl/fl Flg ft/ft (red), and Rora fl/sg Il7r Cre/+ Flg ft/ft (blue) mice. Graph shows the mean + SEM from 9 mice per group. D, Skin inflammation was induced in Rora fl/sg Il7r Cre/+ (cko; red line) and control (WT; black line) mice by daily topical application of 4 nmol MC903 in 100% ethanol onto the right ear. The left ear was treated with ethanol and served as internal control. Ear thickness was measured daily. Mean ± SEM from at least 6 mice per group from two independent experiments is depicted. *P < 0.05, t test of AUC. E, Frequency of ILC2, mast cells, neutrophils, and eosinophils isolated from the ears of MC903-(blue) and ethanol-treated (Vehicle, gray) Rora fl/sg Il7r Cre/+ (cko) and control (WT) mice. F, Frequency of T helper cell subsets isolated from the ears of MC903-(blue) and ethanoltreated (gray) Rora fl/sg Il7r Cre/+ (cko) and control (WT) mice. Bar graphs show the mean ± SEM from six mice of two independent experiments. *P < 0.05. G, Draining (MC903, blue) and non-draining (Vehicle, gray) lymph node cells were restimulated with anti-CD3/anti-CD28 for 72 h and indicated cytokines in the supernatant were measured by ELISA. Bar graphs show the mean + SEM from 6 to 8 mice per group from two independent experiments. ns, not significant, t test of AUC  IL-1β requires processing by the NLRP3 inflammasome to be cleaved from pro-IL-1β to become bioactive IL-1β. We generated   Figure 6D). Indeed, when we induced inflammation in these mice using MC903, IL-1β-responsive mast cells were sufficient to aggravate inflammation ( Figure 6D). Interestingly, germ-free Flg ft/ft mice had significantly decreased serum levels of IgE compared to their SPF counterparts ( Figure 2H) and-lacking the IgE-mediated activation of mast cells-also showed decreased levels of MCP-1 ( Figure 6F). Importantly, GF mice showed reduced expression of IL-1β further reducing activation of mast cells ( Figure 2I).

| Dermal mast cells promote inflammation in mice with impaired skin barrier
These data indicate that IL-1β-mediated hyperactivation of mast cells contributes significantly to dermatitis in Flg ft/ft mice.

| D ISCUSS I ON
In the present study, we have shown that the development of skin inflammation in Flg ft/ft mice is independent of group 2 innate lymphoid cells but requires an interplay between the microbiome, IL-1β, Isotype Anti-IL-1α Anti-IL-1β  shown to decrease expression of filaggrin and tight junction proteins. 78 Similarly, IL-22 is able to regulate keratinocyte function. 63,79 Despite these cytokines being upregulated in the skin, the single knockout of each cytokine did not alleviate disease. This highlights that single target therapy may not be useful to therapeutically deplete cytokines during chronic disease. In this context, dupilumab, targeting the IL-4Rα-chain and thereby the actions of IL-4 and IL-13, has recently been approved by the FDA for the treatment of moderate-to-severe AD patients (reviewed in 80 ) and is proving to be efficacious. 81  WT expressed increased amounts of IL-1α in the epidermis but 21 days after acute mechanical skin injury released less IL-1α and more IL-1β. 86 These results indicate important, but divergent, roles for IL-1 family cytokines in acute and chronic dermatitis. One of our next steps will therefore include the analysis what impact IL-1-family members have on expression of key skin barrier proteins.
In Flg ft/ft mice, we observed that IL-1β enhances FcεRI-mediated signaling in mast cells confirming previous reports. 87 Recently, it was shown that mast cells with a mutated NLRP3 inflammasome and higher caspase-1 activity produced IL-1β in response to TNFα or LPS stimulation. 66 Using kit w-sh Flg ft/ft mice as mast cell-recipient mice, we show that IL-1β-responsive mast cells promote inflammation via IL-1R1-signaling. Because kit w-sh mice certainly have phenotypes independent of mast cell deficiency, 88 further studies using more sophisticated models of mast cell deficiency are required to analyze their in vivo function in mice with skin barrier defects.
Treatment with the NLRP3-inflammasome inhibitor MCC950 did not ameliorate inflammation in our multifactorial model of AD when we triggered an acute inflammatory response with MC903.
In a recent study, NLRC4 was found in the scale extracts of psoriasis patients; however, no inflammasome components (including NLRC4) were found in atopic dermatitis patients. 89 Inflammasome-independent processing of IL-1β has been observed in a variety of settings (reviewed in 90 ). In particular, neutrophil-and mast cell-derived proteases play prominent roles in the extracellular processing of IL-1family cytokines, 91,92 with human mast cell protease 1 shown to process pro-IL-1. 93 Increased mast cell chymase activity may therefore contribute to IL-1β maturation in the Flg ft/ft model; however, further investigations are required to fully understand the role of mast cell proteases in AD.
Skin microbiome sequencing revealed a shift toward pathogenic staphylococci, which has been reported in AD patients. 30 While the microbiota in flaky tail mice induced IL-17A-mediated neutrophilia, we could show that in Flg ft/ft mice IL-17A was redundant. 33 Indeed, IL-17A-mediated inflammation is observed in Our results indicate that the pathogenic microbiome present in neonates imprinted a hyperresponsive phenotype in mast cells.
This phenotype is maintained when adult mice were treated with antibiotics. When we treated neonate Flg ft/ft mice with antibiotics before inflammation is established, or when mice are raised in GF conditions, the subsequent adult dermatitis phenotype is significantly ameliorated. A recent study in a cohort of newborn children suggests that early colonization with commensal staphylococci genera protects from AD and that microbiotic changes only occur after the onset of AD. 94 Therefore, the Flg ft/ft mouse model provides a basis for future investigations into the extrinsic factors and intrinsic mechanisms of neonatal inflammation before AD-like inflammation develops. While the broad application of state-of-the-art techniques gives a comprehensive analysis of the AD-like inflammation in mutant mice on the Flg ft/ft background, future approaches need to target the separate pathways leading to disease and untangle the relative contribution to pathogenesis.
In summary, we revealed a critical role of IL-1β in the initiation and maintenance of skin inflammation in a mutant mouse model of defective skin barrier. Further investigations will be required to assess the contribution of IL-1β-responsive mast cells, as well as roles of the integrity of the skin barrier and interplay of the microbiome, in the generation of inflammation in AD patients. These endeavors will lead to the development of novel management options for children suffering from AD.