Calycosin alleviates allergic contact dermatitis by repairing epithelial tight junctions via down‐regulating HIF‐1α

Abstract Calycosin, a bioactive component derived from Astragali Radix (AR; Huang Qi), has been shown to have an effect of anti‐allergic dermatitis with unknown mechanism. This study aims to investigate the mechanism of calycosin related to tight junctions (TJs) and HIF‐1α both in FITC‐induced mice allergic contact dermatitis and in IL‐1β stimulated HaCaT keratinocytes. Th2 cytokines (IL‐4, IL‐5 and IL‐13) were detected by ELISA. The epithelial TJ proteins (occludin, CLDN1 and ZO‐1), initiative key cytokines (TSLP and IL‐33) and HIF‐1α were assessed by Western blot, real‐time PCR, immunohistochemistry or immunofluorescence. Herein, we have demonstrated that allergic inflammation and the Th2 cytokines in ACD mice were reduced significantly by calycosin treatment. Meanwhile, calycosin obviously decreased the expression of HIF‐1α and repaired TJs both in vivo and in vitro. In HaCaT keratinocytes, we noted that IL‐1β induced the deterioration of TJs, as well as the increased levels of TSLP and IL‐33, which could be reversed by silencing HIF‐1α. In addition, administration of 2‐methoxyestradiolin (2‐ME), a HIF‐1α inhibitor,significantly repaired the TJs and alleviated the allergic inflammation in vivo. Furthermore, TJs were destroyed by DMOG or by overexpressing HIF‐1α in HaCaT keratinocytes, and simultaneously, calycosin down‐regulated the expression of HIF‐1α and repaired the TJs in this process. These results revealed that calycosin may act as a potential anti‐allergy and barrier‐repair agent via regulating HIF‐1α in AD and suggested that HIF‐1α and TJs might be possible therapy targets for allergic dermatitis.

Calycosin, an isoflavonoid phytoestrogen isolated from Astragali Radix, was reported to possess antitumour and anti-inflammation properties. 7 The dried roots of Astragali Radix, called "Huang Qi" in Chinese, are well known as a basic traditional Chinese medicine for thousands of years. "Huang Qi" has powerful immunomodulatory function and is widely used for treating dermatitis, chronic rhinitis, asthma and glomerulonephritis. [8][9][10] Calycosin-7-O-β-D-glucoside and calycosin are the main active flavone of "Huang Qi," and the former would transform into the latter in vivo. 11 Our previous study showed that administration of calycosin in the initial stage of sensitization when the ear has not yet produced inflammation significantly reduced the ear inflammation in mice ACD model. 12 However, the impact of calycosin treated throughout the ACD model when a large number of inflammatory factors appear is unknown, and the mechanism is also worth exploring.
The epidermis of allergic dermatitis patients could have significant barrier disruption. 13 Epidermal barrier dysfunction is a prerequisite for the penetration of high molecular weight allergens, such as pollens, house-dust mite products, microbes and food. 14,15 A disturbed barrier alone could drive dendritic cells to enhance Th2 polarization in patients with AD through elevated antigen permeation. 16 In recent years, more and more researches have demonstrated that the dysregulation of the epithelial barrier might enhance susceptibility and result in the allergen uptake, which initiates an allergic immune response by promoting the release of interleukin (IL)-25, IL-33 and thymic stromal lymphopoietin (TSLP), 17 which are key initiators of allergic diseases leading to the induction and maintenance of Th2 responses. 18,19 The impairment of epidermal tight junctions can induce immune dysregulation in allergic dermatitis, 20 and disruption of tight junctions was demonstrated by down-regulation of a variety of proteins including CLDN1, occludin and ZO-1. CLDN1 regulates the pathogenesis, severity and natural course of human AD, and reduced levels of CLDN1 in AD skin have been inversely correlated with expression of Th2 markers and a propensity to infection. [21][22][23] It is noteworthy that restoring the expression of CLDN1 alleviates atopic dermatitis. 24 So regulating tight junctions might involve in the mechanism of attenuation of AD.
AD is characterized by an increase in the release of local inflammatory factors, such as IL-1β, TNF-α and Th2 cytokines. 25,26 Many studies have shown that inflammatory cytokines play a vital role in the dysregulation of the epithelial barrier function. 27,28 HIF-1α can be activated by these inflammatory mediators and can activate inflammation-modulating proteins itself. 29 Therefore, HIF-1α may relate to the regulation of the epithelial barrier function. Furthermore, the role of HIF-1α in inflammatory reactions associated with dermatitis has recently become evident. 30 It is reported that UVB exposure may increase TSLP expression in keratinocytes through a HIF-1α-dependent mechanism. 31 Moreover, a murine study has demonstrated that HIF-1α protein is an early responder to UV radiation in the skin. 32,33 HIF-1α may be an important pathway involved in processes such as delivery of "second signals" in contact hypersensitivity reactions to allergen. 34 Although some plant isoflavones were reported to exert inhibition effect on HIF-1α, the effect of calycosin on it in the context of allergic inflammation remains to be illustrated. Therefore, HIF-1α is likely to play a role in AD and the anti-allergy mechanism of calycosin may relate to it.

| Animal
BALB/c mice were purchased from Shanghai Slack Laboratory Animal Co., Ltd. All animals were housed in a specific pathogen-free facility at 18-25°C and 40%-70% humidity and were used at 6-8 weeks of age. All procedures involving animals were approved by the Animal Care and Use Committee of Nanjing University of Chinese Medicine and performed strictly according to the Guide for the Care and Use of Laboratory.

| Mouse Th2-mediated ACD model
After acclimatization for 3 days, the abdomens of BALB/c mice were shaved with a razor over an area of about 3 × 3 cm 2 . The abdominal skin of the mice was treated with 1.5% FITC (Sigma, St. Louis, MO, USA) in 80 μL of acetone and dibutylphthalate (1:1; vehicle) on days 1 and 2, and ears were treated with 20 μL of 0.6% FITC solution on day 6. On day 7 (24 hours after elicitation), ear thickness was measured with a thickness gauge (7301; Mitutoyo, Kawasaki, Japan), and ear swelling was calculated (ear thickness in each group minus the average ear thickness in the control group). The mice were then killed, and ear tissues were collected. The levels of IL-4, IL-5 and IL-13 in ear tissue homogenate were detected by ELISA. Histopathological changes of the ears were determined by HE staining.

| Drug administration
A: The dose-response study of calycosin in ACD model The establishment of FITC-induced ACD model is as described above. In ACD model, mice were treated daily with calycosin (2, 10 or 50 mg/kg, intraperitoneally), DEX (0.67 mg/kg, intraperitoneally) or normal saline (model group) for 7 days during the whole course of the experiment. In this experiment, the ID50 was calculated by combining the results of ear swelling, lymphocyte infiltration and cytokines. The calculation process is as follows: calculate the inhibition rate (IR) of each indicator for each dose (2, 10 and 50 mg/kg) separately, and then, the IR was multiplied by their respective weight coefficients (ie IR of ear swelling*0.4, IR of lymphocyte infiltration*0.3 and IR of cytokines*0.3) and sum the 3 indicators to get the total inhibition rate and, finally, calculate ID50 by ID50 calculator software based on the total inhibition rate of each dose.
B: The effect investigation of 2-ME on ACD.
The establishment of FITC-induced ACD model is as described above. Mice were treated once daily with 2-ME (50 mg/kg, intraperitoneally) in the whole course of ACD model.

| Western blotting
Western blotting was performed to assess the changes in the protein expression levels of HIF-1α, CLDN1, occludin, ZO-1, TSLP and IL-33.

| Histological and immunohistochemical analysis
Ears were fixed in 10% neutral formalin, embedded in paraffin and sectioned. Dry tissue sections of 5 μm thickness at 60°C constant temperature box were baked for 30 minutes. Slides were undergone dewaxing, and hydration was performed with sequential dimethylbenzene soak for 20 minutes. Sequential ethanol soaks were carried out for 3 minutes, each starting 100% ethanol for twice, followed by 95%, 80% and 75% ethanol, and finishing with a 50% ethanol. Antigen was retrieved by citric acid buffer water bath heating by microwave oven at high fire for 12 minutes, followed by adding the buffer twice at high fire for 5 minutes every time and then restored at room temperature.
Block non-specific binding sites with 10% normal goat serum for

| Immunofluorescence analysis
The changes in the expressions and distribution of CLDN1, occludin and ZO-1 were analysed by immunofluorescence analysis.

| Transfection with siRNA for HIF-1α in HaCaT
In all experiments, 150 pmol siRNA (the target sequence of HIF-1α-

| Enzyme-linked immunosorbent assay (ELISA)
Ears were ground into homogenates with ice-phosphate-buffered saline (PBS), and the homogenates were centrifuged at 3500 g at 4°C for 10 minutes. ELISAs were performed to relatively quantify protein in the ear tissue homogenate. All of the procedures followed the manufacturer's instructions. Mouse TNF-α, IL-1β, IL-4,IL-5 and IL-13 ELISA kit were obtained from eBioscience Co., Ltd (San Diego, CA, USA).

| Statistical analysis
Data are expressed as mean values + SD. Unpaired Student's t test was used when comparing 2 groups. One-way ANOVA was used to compare multiple groups and 2 groups with Dunnett's test. Statistical analysis was performed by Prism 5.00 software (GraphPad, San Diego, CA, USA). The differences were considered significant for P < .05.

FITC-induced mice ACD model
To evaluate the activity of calycosin against ACD, a dose-response study was carried out. In ACD model, mice were treated daily with

| Calycosin reduced the expression of HIF-1α both in vivo and in vitro
To address the effect of calycosin on the expression of HIF-1α, Western blot,immunohistochemistry and real-time PCR were performed. The results showed that HIF-1α increased significantly in the ACD model compared with that in control, and calycosin (10 mg/ kg) reduced the expression of HIF-1α obviously ( Figure 4A). Immunohistochemistry results also showed that the expression of HIF-1α was reduced by calycosin (10 mg/kg) ( Figure 4B). Furthermore, calycosin (10 μmol/L) could significantly reduce the protein and mRNA expression of HIF-1α induced by IL-1β in HaCaT cells ( Figure 4C,D).
These results suggested that calycosin down-regulated the HIF-1α expression both in vivo and in vitro.

| Knocking-down HIF-1α reversed the down-regulation of occludin and the redistribution of CLDN1 and ZO-1 induced by IL-1β in HaCaT cells
HaCaT cells were stimulated with LPS, TNF-α or IL-1β, respectively, for 24 hours, and the results showed that IL-1β was the most significant factor in increasing the expression of HIF-1α ( Figure 5A) in a dose-dependent manner ( Figure 5B). However, increasing concentrations of FITC (0.1, 1, 10 and 100 μmol/L) exposure had no detectable change in the protein expressions of HIF-1α ( Figure 5C). The nuclear translocation of HIF-1α was higher in IL-1β-stimulated cells than that in the controls ( Figure 5D), which indicates that IL-1β induces the stabilization of HIF-1α, leading to its nuclear translocation. Meanwhile, the mRNA levels of HIF-1α, TNF-α and IL-1β were significantly F I G U R E 2 Calycosin improved epithelial tight junctions of ACD mice. On day 7 in the ACD model, mice were killed, and then, the ears were removed and fixed by 10% formalin. A-C, IHC was performed to detect the expression of CLDN1, occludin and ZO-1 (n = 5, magnification: ×200). Data of IOD (integral optical density) were counted by IPP (Image-Pro Plus) software (mean + SD, n = 5, ***P < .001, **P < .01) Meanwhile, the expressions of TSLP and IL-33 were down-regulated by knocking-down HIF-1α ( Figure 5I). Immunofluorescence assays further showed that HIF-1α silencing could reverse the deterioration of occludin, ZO-1 and CLDN1 induced by IL-1β ( Figure 5J). Taken together, these findings suggested that HIF-1α plays a critical role in the damage of tight junctions.

| HIF-1α inhibitor administered in ACD mice attenuated allergic inflammation and repaired the tight junctions
2-ME, an inhibitor of HIF-1α, 40 was administered daily to further determine the effect of HIF-1α on epithelial tight junction and the follow-up allergic inflammation. The ear swelling and inflammatory cell infiltration were significantly suppressed by 2-ME ( Figure 6A,B).
The levels of TNF-α, IL-1β and IL-13, as well as HIF-1α, could be remarkably reduced by 2-ME ( Figure 6C,D). 2-ME could also repair the tight junctions by means of up-regulating the expression of CLDN1, occludin and ZO-1 ( Figure 6E).

| Calycosin repaired the tight junctions by down-regulating the HIF-1α expression in HaCaT cells
To further testify the role of HIF-1α in calycosin-regulated tight junctions, we evaluated the effect of pharmacologic HIF activation by DMOG on the expressions of tight junction proteins. The results showed that DMOG promoted protein accumulation of HIF-1α and reduced the expressions of CLDN1 and occludin. Calycosin reversed the effect of DMOG on the expressions of HIF-1α, CLDN1 and occludin ( Figure 7A). Furthermore, we overexpressed HIF-1α in HaCaT cells by transfecting pLenti-h-HIF-1α plasmid ( Figure 7B), and transfection efficiency was confirmed by Western blot and qPCR ( Figure 7C,D). We found that cell morphology changed after HIF-1α overexpressed ( Figure 7E) and the expressions of occludin and CLDN1 were significantly reduced ( Figure 7F). After transfection with HIF-1α, the expression of HIF-1α itself was reduced by calycosin significantly. Moreover, the reduced expressions of occludin and CLDN1 could be reversed by calycosin ( Figure 7G,H). These results suggested that calycosin repaired the tight junctions by down-regulating the HIF-1α.

| DISCUSSION
When allergic dermatitis occurs, it was often accompanied by itch- Considering the mechanism of calycosin repairing the tight junctions, we investigated HIF-1α. It has been reported that calycosin reduced NF-κB activation in LPS-stimulated human keratinocytes 44 and LPS induced HIF-1α protein accumulation by activating NF-κB. 45 Moreover, the role of HIF-1α in allergic diseases has been increasingly recognized. 46,47 A previous study showed that HIF-1α increases epithelial permeability in human nasal epithelium. 48 Moreover, tight junction proteins are indicated as MMP (matrix metalloproteinases) F I G U R E 6 2-ME inhibited ACD-associated inflammation and repaired the tight junctions. A, Ear swelling was calculated on day 7 in the ACD mice model (means + SD, n = 6, ### P < .001 vs control; **P < .01 vs model). B, Haematoxylin and eosin (H&E)-stained ear skin sections from FITC-induced ACD model mice (n = 5; magnification: ×100). C, The expressions of HIF-1α in the total protein were detected by Western blot. The data are expressed as the mean + SD of the ratios of indicated protein to GAPDH. **P < .01, *P < .05. D, The levels of cytokines IL-4, IL-13, TNF-α and IL-1β in the ear tissue homogenates were detected by ELISA (mean + SD, n = 6, ### P < .001 vs control; **P < .01 vs model). E, IHC was performed to detect the expression and distribution of CLDN1, occludin and ZO-1 (n = 5, magnification: ×200). Data of IOD (integral optical density) were counted by IPP (Image-Pro Plus) software (mean + SD, n = 5, ***P < .001) substrates. Both occludin and the scaffolding protein ZO-1 are the substrates for gelatinases, and HIF-1α can trigger MMP activation. 49 We considered that HIF-1α regulated tight junction proteins. In our study, by silencing or overexpressing HIF-1α in vitro,we demonstrated that HIF-1α plays a critical role in regulating the inflammation microenvironment and tight junctions. Also, we found that HIF-1α To investigate the mechanisms of how calycosin regulates the expression of HIF-1α, we first considered the proteasome pathway.
When oxygen is available, PHDs (prolyl hydroxylases) are active and hydroxylate HIF-α, marking it for proteasomal degradation in a process mediated by von Hippel-Lindau tumour suppressor protein (VHL)-dependent ubiquitination. If oxygen concentration drops, PHDs become inactive, resulting in HIF-α accumulation. 55 In our study, we found that DMOG, a small-molecule dimethyloxalylglycine which inhibited all PHD isoforms, 56  stabilization in immune cells can occur in an oxygen-independent manner by activating NF-κB. 29 UVB induces HIF-1α expression via the JNK, ERK and P38 pathways in human keratinocytes. 31 Calycosin reduced NF-κB activation in TNF-α-stimulated 16HBE cells or LPS-stimulated human keratinocytes. 12,44 Calycosin inhibits the phosphorylation of ERK 1/2 and NF-κB. 7,57 Hence, calycosin may also regulate the expression of HIF-1α by transcriptional regulation.
Our results in vitro also showed that the knockdown of HIF-1α in HaCaT cells can reverse the levels of TSLP and IL-33 induced by IL-1β. Some researches have shown that HIF-1α is associated with TSLP and IL-33. In UVB-mediated immune response in keratinocytes, UVB exposure may increase TSLP expression through a HIF-1αdependent mechanism. 31 It has been reported that epithelial barrier dysfunction can lead to loss of adhesion between cells and cause the release of allergy-related cytokines, such as TSLP, in skin keratinocytes. 58 Recent research by our group showed that CLDN1 down-regulation in HaCaT cells could exacerbate the production of TSLP. 59 Calycosin may reduce the production of TSLP and IL-33 by repairing the tight junctions.
To conclude, calycosin may be a potential drug for the treatment of allergic dermatitis by regulating tight junctions via inhibiting the expression of HIF-1α, and as a consequence, HIF-1α and TJs might be possible therapy targets for allergic dermatitis. In addition, this study also suggests the theoretical possibility that the combined treatment of dexamethasone and calycosin might be a useful modality for allergic dermatitis treatment, and further studies involving the combined treatment of steroid and barrier repair therapies are needed.

CONFLI CT OF INTEREST
The authors declare no conflict of interest.