The current study was approved by the Ethics Committee of Hunan Children’s Hospital and performed in strict accordance with the Declaration of Helsinki. All participants in this study signed informed consent documentation before sample collection. Animal experiments were approved by the Animal Ethics Committee of Hunan Children’s Hospital and strictly performed according to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health. Due efforts were made to limit animals’ pain.
Clinical sample collection
Normal subjects (n = 50) and patients with AD (n = 90) who were treated in the dermatology department of Hunan Children’s Hospital from June 2018 to October 2019 were selected for collection of blood samples and tissue samples (n = 10 for each group). Patients were included if they (1) aged between 6 and 12 years (including the boundary value) at the time of signing informed consent, regardless of gender; (2) had moderate or severe AD; (3) had condition not fully controlled by topical prescription drugs or not suitable for topical drug treatment. Patients were excluded if they had (1) cardiovascular, neurological, renal, liver, digestive tract, urogenital system, psychiatric, nervous system, musculoskeletal, skin, sensory, immune, endocrine (including uncontrolled diabetes or thyroid disease) or uncontrolled hematological abnormalities; (2) other active skin diseases (such as psoriasis or lupus erythematosus) or skin infections (bacteria, fungi or viruses) that might affect the evaluation of AD; (3) severe concomitant diseases (such as unstable chronic asthma) that may receive systemic hormone therapy or other interventions or need active and frequent monitoring.
Cell culture and transformation
Human immortalized keratinocytes (HaCaT) cell line (Cell Resource Center, Chinese Academy of Medical Sciences, 1101HUM-PUMC000373, Beijing, China) was cultured in Dulbecco’s modified Eagle medium (10569010, Gibco, Carlsbad, CA) appended to 10% fetal bovine serum (cat#10100147, Gibco) and penicillin mixture (working concentration of penicillin was 100 U/mL, streptomycin sulfate was 0.1 mg/mL, cat#P1400, Solarbio, Beijing, China) in an incubator with 5% CO2 at 37 °C. After culture for 1–2 passages, IL-32 (200 ng/mL; R&D Systems, Minneapolis, Minn.) was utilized for 48-h of cell treatment. The cells were then collected for the follow-up experiments.
Logarithmically growing cells were detached with 1 mL 0.25% trypsin (25200056, Gibco) for 3 min, and then the detachment was terminated utilizing the medium containing serum. Following cell concentration adjustment (1 × 105 cells/mL), the cells were seeded into a 6-well plate with a glass slide for 24-h of conventional culture. Under above 75% confluence, cell transfection was implemented by referring to Lipofectamine 2000 (Invitrogen, Carlsbad, CA) utilizing shRNA (sh)-negative control (NC), sh-JAK1 (JAK1 knockdown), vector (NC for JAK1 overexpression), or overexpression (oe)-JAK1. The plasmids were constructed by GenePharma (Shanghai, China) and the plasmid concentration was 50 ng/mL. The cells were collected after 48-h transfection for subsequent experiments.
Reverse transcription-quantitative polymerase chain reaction (RT-qPCR)
Total RNA extraction was processed by RNeasy Mini Kit (Qiagen, Valencia, CA). For mRNA determination, the complementary DNA (cDNA) was obtained by means of a RT Kit (RR047A, Takara, Otsu, Shiga, Japan). For miRNA determination, the miRNA First Strand cDNA SyntH&Esis (Tailing Reaction) kit (B532451-0020, Sangon, Shanghai, China) was adopted. SYBR Premix EX Taq kit (RR420A, Takara) was used to mix and load samples. The samples were subjected to RT-qPCR in a real-time fluorescence qPCR instrument (Bio-Rad CFX96, Bio-Rad Laboratories, Hercules, CA). The primers were synthesized by Sangon and displayed in Additional file 5: Tables S1 and S2. The relative expression of the product was calculated by 2−ΔΔCt method with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a normalizer for mRNA and U6 for miRNA.
Western blot analysis
Tissue and cell samples were lysed with enhanced radioimmunoprecipitation assay lysis containing protease inhibitor (1 mM, cat# 36978, Thermo Fisher Scientific, Rockford, IL), and then the protein concentration was determined by BCA protein quantitative Kit (Boster, Wuhan, China). Following electrophoresis separation, the protein was transferred to a polyvinylidene fluoride membrane which was sealed with 5% bovine serum albumin at ambient temperature for 2 h to block the nonspecific binding. Then, overnight incubation of membrane with diluted primary rabbit antibodies was performed: JAK1 (ab133666, 1: 1000, Abcam, Cambridge, UK), phosphorylated (p)-JAK1 (#74129, 1: 1000, Cell Signaling Technology [CST], Danvers, MA), p-STAT1 (#9167, 1: 1000, CST), STAT1 (ab230428, 1: 1000, Abcam), STAT3 (ab68153, 1: 1000, Abcam), p-STAT3 (ab76315, 1: 1000, Abcam), Dicer 1 (ab14601, 1: 2000, Abcam), DGCR8 (ab191875, 1: 1000, Abcam), Drosha (ab12286, 1: 10,000, Abcam), and GAPDH (ab8245, 1: 5000, Abcam, normalizer). The following day, the membrane was reacted with horseradish peroxidase-labeled goat anti-rabbit secondary antibody (ab205719; 1: 2000; Abcam) or rabbit anti-mouse secondary antibody (ab6728, 1: 1000, Abcam) at ambient temperature for 1 h and then detected with enhanced chemiluminescence solution (EMD Millipore, Billerica, MA). Image J analysis software was utilized for quantifying the gray level of each band.
Construction of mouse models
Seven-week-old male BALB/c mice (Hubei Provincial Center for Disease Control and Prevention,Wuhan, China) were fed adaptively at the temperature between 22 and 24 °C, under 12-h day/night cycles.
IL-32 transgenic (Tg) mice were generated  with the primer of: sense, 5ʹ-TGAGGAGCAGCACCCAGAGC-3ʹ, and antisense, 5ʹ-CCGTAGGACTGGAAAGAGGA-3ʹ. Genomic DNA samples were obtained from the tails of transgenic mice, followed by IL-32 gene expression determination utilizing RT-qPCR. IL-32 was not expressed in wild type (WT) mice. No overt phenotype was observed in IL-32-Tg mice compared with WT mice. The IL-32-Tg mice were viable and fertile, without tissue or organ abnormalities.
The construction process of phthalic anhydride (PA) model in mice was as follows: mice in the control group (n = 10) were treated with ddH2O, and mice in the PA group (n = 10) was treated with 100 μL (20 μL/cm2) of 5% PA. The mice were fed under this condition for 4 weeks, and then the next experiment was carried out.
MC903 model of AD in nude mice was constructed by using MC903 (Cayman, MI) according to the published experimental methods [15,16,17]. In brief, 2 nmol MC903 (20 μL dissolved in ethanol) was used on the back of ears of mice in the AD group (n = 10) for 14 days, and 20 μL ethanol was used for 14 days on mice in the control group (n = 10).
The ear thickness was evaluated utilizing a thickness gauge (Digimatic Indicator, Matusutoyo Co., Tokyo, Japan) to test the degree of skin inflammation caused by PA or MC903 treatment. To assess the severity of the PA-induced or MC903-induced AD, the clinical scores of mice in each group were scored as 0–6 points based on the average score of erythema (redness), scaly, itching and other symptoms, suggestive of the successful establishment of PA mouse model. At the end of the study, skin tissues and blood samples were collected. The skin-draining lymph nodes were obtained from the euthanized mice and weighed.
In vivo experiment
PA-IL32-AD modeled mice were randomized into three groups (control, sh-JAK1, miR-155 inhibitor), with 10 mice in each group. The control and inhibitors were special oligonucleotide sequences, which were synthesized by Beomic Biotechnology Co., Ltd. (Jiangsu, China). The constructed liposome-coated lentiviral vector was injected into mice via tail vein at 25 μg/mouse. After that, the mice in each group were routinely cultured for 7 days.
AD-reconstructed human epidermis (RHE) and corresponding treatment
RHE (0.33 cm2; 17 day) was purchased from Episkin (Lyon Cedex, France). This standard model was formed by the growth of human keratinocytes on an inert polycarbonate filter with chemical properties at the gas–liquid interface, with histological characteristics similar to that of real human epidermis. Briefly, an inflammatory AD cocktail constitutes 30 ng/mL of IL-4, 30 ng/mL of IL-13, and 3.5 ng/mL of tumor necrosis factor-α (TNF-α; PeproTech Inc, NJ) in the presence or absence of recombinant human IL-32 (100 ng/mL; YbdY Biotech, Seoul, Korea) was supplemented to the medium for 6-day culture. The culture medium was renewed every 48 h. After overnight incubation, the medium was renewed and the following treatments were started (all of them were treated with 6 independent reconstituted skin in vitro for 24 h):
control: The AD-RHE model was treated with PBS;
sh-JAK1: The AD-RHE model was treated with sh-JAK1;
miR-155-inhibitor: The AD-RHE model was treated with miR-155 inhibitor;
IL-32: The AD-RHE model was treated with IL-32 and the IL-32 gene fragment was transferred into the AD-RHE model with the same procedure as the construction of IL-32-Tg mice;
The sequence of sh-JAK1 was 5ʹ-CCAUCACUGUUGAUGACAAdTdT-3′, and the sequence of miR-155 inhibitor was 5′-ACCCCUAUCACGAUUAGCAUUAA-3′. The control and inhibitor were special oligonucleotide sequences (Biomics, Nantong, China). The constructed liposome-coated lentiviral vector was injected into mice through the tail vein at 25 μg/mouse (n = 10). In the end, the collected culture was frozen and stored at − 20 °C. RHE tissues were used for histological and immunohistochemical analyses.
Enzyme-linked immunosorbent assay (ELISA)
In this study, sandwich ELISA was used to detect IL-32 expression . The absorbance value was determined at 450 nm, and the concentration was calculated according to the standard.
Hematoxylin and eosin (HE) staining
For pathological observation, the ear skin of mice was embedded in paraffin for 24 h, and then fixed by a series of dehydration and rehydration. The paraffin-embedded tissues were cut into 4 mm sections, and then stained with HE. The sections were observed by microscopy (LAS; Leica Microsystems, Buffalo grove, IL) and five visual fields were randomly selected for evaluation.
EnVision (EnVision + , Dako, Carpentaria, CA) system was used for immunohistochemical detection of human and mouse skin tissues. In brief, the skin tissue samples were shaved and rehydrated in double distilled water. Next, 3% H2O2 was used to block endogenous peroxidase activity, and then sodium citrate solution was used for antigen repair. Then, antibodies to IL-32 (ab37158, 1: 100, Abcam), JAK1 (ab125051, 1: 100, Abcam), p-JAK1 (PA5-104554, 1: 100, Invitrogen), IL-32 (ab37158, 1: 100, Abcam), p-STAT1 (#9167, 1: 800, Cell Signaling Technology), STAT1 (ab230428, 1: 100, Abcam), STAT3 (ab68153, 1: 100, Abcam), and p-STAT3 (ab76315, 1: 100, Abcam) were used to incubate the samples at 37 °C for 3 h. Then, the sample was further reacted with secondary antibody obtained from the Zhongshan Biotechnology company (Beijing, China). After diaminobenzidine staining, routine staining was performed followed by microscopic examination.
Masson's trichrome staining
Skin tissue slices of mice in each group were dewaxed, stained with Weigert iron hematoxylin for 5–10 min, differentiated in acidic ethanol for 5–15 s, and treated with Masson for 3–5 min to return blue in color. Ponceau S Fuchsin staining solution was applied to slices for 5–10 min. Weak acidic working solution was prepared by mixing distilled water and weak acidic solution at a ratio of 2:1 and then used to wash slices for 1 min, which were rinsed with phosphomolybdic acid solution for another 1–2 min. Aniline blue staining solution was applied to slices for 1–2 min. Slices were then dehydrated with 95% ethanol and absolute ethanol, cleared with xylene, and sealed with neutral balsam. Nucleus and collagen fiber/protein were stained in blue while cytoplasm, muscle, and red blood cells were in red.
Data analysis was processed utilizing the SPSS 21.0 statistical software (IBM, Armonk, NY). Each experiment was repeated three times independently. The measurement data were summarized by mean ± standard deviation. Data between two groups were compared employing independent sample t test, and those among multiples utilizing one-way analysis of variance (ANOVA), combined with Tukey’s post hoc tests. p < 0.05 indicated statistically significant difference.