Brazilian Green Propolis Rescues Oxidative Stress-Induced Mislocalization of Claudin-1 in Human Keratinocyte-Derived HaCaT Cells

Claudin-1 (CLDN1) is expressed in the tight junction (TJ) of the skin granular layer and acts as a physiological barrier for the paracellular transport of ions and nonionic molecules. Ultraviolet (UV) and oxidative stress may disrupt the TJ barrier, but the mechanism of and protective agents against this effect have not been clarified. We found that UVB and hydrogen peroxide (H2O2) caused the internalization of CLDN1 and increased the paracellular permeability of lucifer yellow, a fluorescent marker, in human keratinocyte-derived HaCaT cells. Therefore, the mechanism of mislocalization of CLDN1 and the protective effect of an ethanol extract of Brazilian green propolis (EBGP) were investigated. The UVB- and H2O2-induced decreases in CLDN1 localization were rescued by EBGP. H2O2 decreased the phosphorylation level of CLDN1, which was also rescued by EBGP. Wild-type CLDN1 was distributed in the cytosol after treatment with H2O2, whereas T191E, its H2O2-insensitive phosphorylation-mimicking mutant, was localized at the TJ. Both protein kinase C activator and protein phosphatase 2A inhibitor rescued the H2O2-induced decrease in CLDN1 localization. The tight junctional localization of CLDN1 and paracellular permeability showed a negative correlation. Our results indicate that UVB and H2O2 could induce the elevation of paracellular permeability mediated by the dephosphorylation and mislocalization of CLDN1 in HaCaT cells, which was rescued by EBGP. EBGP and its components may be useful in preventing the destruction of the TJ barrier through UV and oxidative stress.


Introduction
The skin forms a barrier between the body and its external environment in order to prevent the intrusion of pathogens and the uncontrolled loss of internal water and solutes. The epidermis is the outermost layer of skin and provides the first line of defense against ultraviolet (UV) radiation and other environmental factors. UV is divided into three ranges based on wavelength: UVA (320-400 nm), UVB (280-320 nm), and UVC (100-290 nm). Among these, UVC is unlikely to be present in terrestrial sunlight because it is blocked by the ozone layer, whereas UVA and UVB can come into contact with the skin. The skin can be separated by the basement membrane into two layers, the dermis and epidermis. The mammalian epidermis is composed of four layers, including the basal, spinous, granular, and cornified layers [1]. UVA can penetrate the dermal layer of the skin and reach the capillaries, whereas UVB is blocked in the upper dermis. According to wavelength-dependent studies, UVB radiation has more cytotoxic and mutagenic effects than UVA does [2,3]. Long-term exposure to UVB induces damage to both the dermal and epidermal skin [4]. UVB-induced cell death is Effects of UVB and H2O2 on reactive oxygen species (ROS) production and cell toxicity in HaCaT cells. (A,B) Cells exposed to UVB were cultured for 3 h and then incubated with 2',7'dichlorodihydrofluorescein diacetate (H2DCFDA) (ROS) and Hydrop (H2O2) for 30 min. The florescence intensities of each dye were measured using an Infinite F200 Pro microplate reader and are shown relative to the values (0 mJ/cm 2 ). (C,D) Cells were exposed to UVB or transiently exposed to H2O2 for 3 h, and then they were cultured for 24 h. Cell viability was measured using a WST-1 assay. (E) Cells were exposed to H2O2 for 3 h, and then they were incubated with H2DCFDA. The florescence intensities of H2DCF are shown as relative to the values in 0 μM H2O2. n = 4-6; *** p < 0.001 and ** p < 0.01 significantly different from 0 mJ/cm 2 UVB or 0 μM H2O2.

Effect of H2O2 on the Cell Localization of TJ Protein
Immunofluorescence measurements showed that CLDN1, CLDN4, and ZO-1 were localized at the TJ under control conditions (Figure 2A,B). The fluorescence signal of CLDN1 at the TJ was not changed by treatment with 200 μM H2O2 for 3 h, but it became weaker at 6 h. In contrast, those of CLDN4 and ZO-1 were unchanged. CLDN1 was relocalized at the TJ after 48 h. The amount of CLDN1 protein was unchanged by H2O2 treatment, but that of CLDN4 transiently increased at 6 h ( Figure 2C). The florescence intensities of each dye were measured using an Infinite F200 Pro microplate reader and are shown relative to the values (0 mJ/cm 2 ). (C,D) Cells were exposed to UVB or transiently exposed to H 2 O 2 for 3 h, and then they were cultured for 24 h. Cell viability was measured using a WST-1 assay. (E) Cells were exposed to H 2 O 2 for 3 h, and then they were incubated with H 2 DCFDA. The florescence intensities of H 2 DCF are shown as relative to the values in 0 µM H 2 O 2 . n = 4-6; *** p < 0.001 and ** p < 0.01 significantly different from 0 mJ/cm 2 UVB or 0 µM H 2 O 2 .

Effect of H 2 O 2 on the Cell Localization of TJ Protein
Immunofluorescence measurements showed that CLDN1, CLDN4, and ZO-1 were localized at the TJ under control conditions (Figure 2A,B). The fluorescence signal of CLDN1 at the TJ was not changed by treatment with 200 µM H 2 O 2 for 3 h, but it became weaker at 6 h. In contrast, those of CLDN4 and ZO-1 were unchanged. CLDN1 was relocalized at the TJ after 48 h. The amount of CLDN1 protein was unchanged by H 2 O 2 treatment, but that of CLDN4 transiently increased at 6 h ( Figure 2C).  The expression levels of CLDN1 and CLDN4 are represented relative to the values at 0 h. n = 3-8; ** p < 0.01 and * p < 0.05 significantly different from 0 h; ## p < 0.01 significantly different from 6 h.

Effects of EBGP on the UVB-and H 2 O 2 -Induced Destruction of the TJ Barrier
The function of the TJ barrier was evaluated using transepithelial electrical resistance (TER) and lucifer yellow (LY) flux with HaCaT cells cultured on Transwell inserts. TER decreased after 6 h of treatment with UVB or H 2 O 2 , then returned to the control level after 48 h ( Figure 3A). EBGP-induced cell toxicity was examined using a 2-(4-Iodophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium (WST-1) assay. Cell viability was not significantly changed by EBGP at concentrations less than 20 µg/mL, but decreased at 50 µg/mL ( Figure 3B). The H 2 O 2 -induced mislocalization of CLDN1 was significantly rescued by 10 µg/mL EBGP ( Figure 3C), but not by 50 µg/mL EBGP. Similarly, the TJ localization of CLDN1 was decreased by UVB, which was rescued by 10 µg/mL EBGP. Both H 2 O 2 and UVB caused a decrease in TER and an increase in LY flux, indicating that H 2 O 2 and UVB induced the destruction of the TJ barrier. These effects were rescued by EBGP ( Figure 3D). These results indicated that UVB and H 2 O 2 may have induced the destruction of the TJ barrier mediated by the loss of CLDN1 in the TJ, which was rescued by EBGP.

Effects of EBGP on the UVB-and H2O2-Induced Destruction of the TJ Barrier
The function of the TJ barrier was evaluated using transepithelial electrical resistance (TER) and lucifer yellow (LY) flux with HaCaT cells cultured on Transwell inserts. TER decreased after 6 h of treatment with UVB or H2O2, then returned to the control level after 48 h ( Figure 3A). EBGP-induced cell toxicity was examined using a 2-(4-Iodophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2Htetrazolium (WST-1) assay. Cell viability was not significantly changed by EBGP at concentrations less than 20 μg/mL, but decreased at 50 μg/mL ( Figure 3B). The H2O2-induced mislocalization of CLDN1 was significantly rescued by 10 μg/mL EBGP ( Figure 3C), but not by 50 μg/mL EBGP. Similarly, the TJ localization of CLDN1 was decreased by UVB, which was rescued by 10 μg/mL EBGP. Both H2O2 and UVB caused a decrease in TER and an increase in LY flux, indicating that H2O2 and UVB induced the destruction of the TJ barrier. These effects were rescued by EBGP ( Figure 3D). These results indicated that UVB and H2O2 may have induced the destruction of the TJ barrier mediated by the loss of CLDN1 in the TJ, which was rescued by EBGP.  Cell viability was measured using a WST-1 assay. (C) The cells were pre-incubated in the absence (Veh) or presence of 10 µg/mL EBGP for 30 min, followed by exposure to UVB or H 2 O 2 for 6 h. The cells were immunostained with anti-CLDN1 antibodies. Scale bars indicate 10 µm. The TJ localization of CLDN1 is represented as the percentage of total amount. (D) The cells were pre-incubated in the absence (Veh) or presence of 10 µg/mL EBGP for 30 min, followed by exposure to UVB or H 2 O 2 for 6 h. Control cells (Cont) were not treated with UVB or H 2 O 2 . TER was measured using a volt-ohmmeter and is represented as a percentage of the values in control cells. Paracellular lucifer yellow (LY) flux was analyzed using fluorescence spectrometry. n = 4; ** p < 0.01 significantly different from 0 h, 0 µg/mL, or Cont; ## p < 0.01 significantly different from 6 h of UVB and H 2 O 2 or Veh of UVB and H 2 O 2 .

Effect of CLDN1 Expression on the TJ Barrier
To confirm the involvement of CLDN1, we examined the effect of knockdown of CLDN1 expression. The mRNA expression of CLDN1 was decreased by the introduction of CLDN1 small interfering RNAs (siRNAs) ( Figure 4A). The knockdown of CLDN1 induced a decrease in TER and an increase in LY flux ( Figure 4B), indicating that CLDN1 had an important role in the regulation of the TJ barrier. H 2 O 2 significantly changed neither TER nor LY flux in CLDN1 knockdown cells ( Figure 4C). In addition, EBGP showed no recovery effects on the TJ barrier. These results indicated that CLDN1 was necessary for the EBGP-induced rescue effect of the TJ barrier.
represented as a percentage of the values at 0 h. (B) Cells were incubated in the presence or absence of EBGP for 24 h. Cell viability was measured using a WST-1 assay. (C) The cells were pre-incubated in the absence (Veh) or presence of 10 μg/mL EBGP for 30 min, followed by exposure to UVB or H2O2 for 6 h. The cells were immunostained with anti-CLDN1 antibodies. Scale bars indicate 10 μm. The TJ localization of CLDN1 is represented as the percentage of total amount. (D) The cells were preincubated in the absence (Veh) or presence of 10 μg/mL EBGP for 30 min, followed by exposure to UVB or H2O2 for 6 h. Control cells (Cont) were not treated with UVB or H2O2. TER was measured using a volt-ohmmeter and is represented as a percentage of the values in control cells. Paracellular lucifer yellow (LY) flux was analyzed using fluorescence spectrometry. n = 4; ** p < 0.01 significantly different from 0 h, 0 μg/mL, or Cont; ## p < 0.01 significantly different from 6 h of UVB and H2O2 or Veh of UVB and H2O2.

Effect of CLDN1 Expression on the TJ Barrier
To confirm the involvement of CLDN1, we examined the effect of knockdown of CLDN1 expression. The mRNA expression of CLDN1 was decreased by the introduction of CLDN1 small interfering RNAs (siRNAs) ( Figure 4A). The knockdown of CLDN1 induced a decrease in TER and an increase in LY flux ( Figure 4B), indicating that CLDN1 had an important role in the regulation of the TJ barrier. H2O2 significantly changed neither TER nor LY flux in CLDN1 knockdown cells ( Figure  4C). In addition, EBGP showed no recovery effects on the TJ barrier. These results indicated that CLDN1 was necessary for the EBGP-induced rescue effect of the TJ barrier.

Effects of Endocytosis Inhibitors on the Localization of CLDN1
The localization of plasma membrane protein is regulated by endocytotic processes, including clathrin-and caveolae-dependent pathways. The H 2 O 2 -induced mislocalization of CLDN1 was inhibited by monodansylcadaverine (MDC), a clathrin-dependent endocytosis inhibitor, but not by methyl-β-cyclodextrin (MβCD), a caveolae-dependent endocytosis inhibitor ( Figure 5A), indicating that H 2 O 2 may increase the internalization of CLDN1 mediated by the clathrin-dependent endocytosis pathway. An increase in the paracellular permeability caused by H 2 O 2 was rescued by MDC, but not by MβCD ( Figure 5B). These results coincided with the effect of MDC on the H 2 O 2 -induced mislocalization of CLDN1 in the immunofluorescence measurements. The recovery effect of MDC was blocked by the knockdown of CLDN1 expression. These results support that the disappearance of CLDN1 in TJ may induce disruption of the TJ barrier. The localization of some CLDNs has been reported to change according to their phosphorylation status [19][20][21]. Therefore, we decided to investigate the effect of H 2 O 2 on the phosphorylation of CLDN1.

Effects of Endocytosis Inhibitors on the Localization of CLDN1
The localization of plasma membrane protein is regulated by endocytotic processes, including clathrin-and caveolae-dependent pathways. The H2O2-induced mislocalization of CLDN1 was inhibited by monodansylcadaverine (MDC), a clathrin-dependent endocytosis inhibitor, but not by methyl-β-cyclodextrin (MβCD), a caveolae-dependent endocytosis inhibitor ( Figure 5A), indicating that H2O2 may increase the internalization of CLDN1 mediated by the clathrin-dependent endocytosis pathway. An increase in the paracellular permeability caused by H2O2 was rescued by MDC, but not by MβCD ( Figure 5B). These results coincided with the effect of MDC on the H2O2induced mislocalization of CLDN1 in the immunofluorescence measurements. The recovery effect of MDC was blocked by the knockdown of CLDN1 expression. These results support that the disappearance of CLDN1 in TJ may induce disruption of the TJ barrier. The localization of some CLDNs has been reported to change according to their phosphorylation status [19][20][21]. Therefore, we decided to investigate the effect of H2O2 on the phosphorylation of CLDN1.

Effects of Phosphorylation Mimic Mutants of CLDN1 on the TJ Barrier
H2O2 transiently decreased the phosphothreonine (p-Thr) levels of CLDN1 after 3-6 h ( Figure  6A). In contrast, phosphoserine (p-Ser) levels were unchanged by H2O2 (data not shown). The H2O2- 2.6. Effects of Phosphorylation Mimic Mutants of CLDN1 on the TJ Barrier H 2 O 2 transiently decreased the phosphothreonine (p-Thr) levels of CLDN1 after 3-6 h ( Figure 6A). In contrast, phosphoserine (p-Ser) levels were unchanged by H 2 O 2 (data not shown). The H 2 O 2 -induced dephosphorylation of CLDN1 was inhibited by EBGP ( Figure 6B). It has been reported previously that CLDN1 may be phosphorylated at T191 and T195 [22]. To clarify the dephosphorylation site of CLDN1 associated with H 2 O 2 treatment, we investigated the effects of phosphorylation mimic mutants of CLDN1, T191E, and T195E. In the absence of H 2 O 2 , the wild-type (WT), T191E, and T195E proteins were localized in the TJ area. H 2 O 2 increased the intracellular distribution of WT and T195E, but T191E was still mainly localized in the TJ area ( Figure 6C). Compared to WT in the absence of H 2 O 2 , T191E showed a similar TJ barrier function, but T195E did not ( Figure 6D). These results indicated that the phosphorylation of T191 may be necessary for the TJ localization of CLDN1.
induced dephosphorylation of CLDN1 was inhibited by EBGP ( Figure 6B). It has been reported previously that CLDN1 may be phosphorylated at T191 and T195 [22]. To clarify the dephosphorylation site of CLDN1 associated with H2O2 treatment, we investigated the effects of phosphorylation mimic mutants of CLDN1, T191E, and T195E. In the absence of H2O2, the wild-type (WT), T191E, and T195E proteins were localized in the TJ area. H2O2 increased the intracellular distribution of WT and T195E, but T191E was still mainly localized in the TJ area ( Figure 6C). Compared to WT in the absence of H2O2, T191E showed a similar TJ barrier function, but T195E did not ( Figure 6D). These results indicated that the phosphorylation of T191 may be necessary for the TJ localization of CLDN1.  were prepared from the cells exposed to 200 µM H 2 O 2 in the absence (Veh) or presence of 10 µg/mL EBGP for 6 h. After immunoprecipitation with anti-CLDN1 antibodies, the immunoprecipitants were applied to 12.5% SDS-PAGE and blotted with antiphosphothreonine (p-Thr) and anti-CLDN1 antibodies. The levels of p-Thr are represented relative to the values at 0 h or the control (Cont). (C,D) Cells were transfected with FLAG-tagged wild-type (WT), T191E, and T195E CLDN1. At 48 h after transfection, the cells were exposed to 200 µM H 2 O 2 for 6 h. (C) The cells were immunostained with anti-FLAG antibodies (green) and DAPI (blue), a nuclear marker. Scale bars indicate 10 µm. (D) TER and paracellular LY flux were analyzed using a volt-ohmmeter and fluorescence spectrometry, respectively. n = 3-4; ** p < 0.01 and * p < 0.05 significantly different from 0 h, Cont, or WT alone; ## p < 0.01 and # p < 0.05 significantly different from Veh with H 2 O 2 or WT with H 2 O 2 .

Phosphorylation of CLDN1 by PKC
The phosphorylation level of CLDN1 is regulated by atypical PKC and protein phosphatases (PPs) [23]. Go6976, a selective PKCα and β isoform inhibitor, decreased p-Thr levels and the TJ localization of CLDN1 ( Figure 7A,B). H 2 O 2 decreased PKC activity and increased PP activities, which were significantly inhibited by EBGP ( Figure 7C). To support the involvement of PKC, we examined the effects of phorbol 12-myristate 13-acetate (PMA), a PKC activator, and cantharidin, a PP2A inhibitor, on the TJ barrier. The H 2 O 2 -induced mislocalization of CLDN1 was inhibited by both PMA and cantharidin ( Figure 7D). Similarly, the reduction in the TJ barrier function by H 2 O 2 was significantly rescued by both PMA and cantharidin ( Figure 7E,F). The recovery effects of PMA and cantharidin were blocked by the knockdown of CLDN1 expression. These results indicated that PKC activator and PP inhibitor could rescue the H 2 O 2 -induced mislocalization of CLDN1. were prepared from the cells exposed to 200 μM H2O2 in the absence (Veh) or presence of 10 μg/mL EBGP for 6 h. After immunoprecipitation with anti-CLDN1 antibodies, the immunoprecipitants were applied to 12.5% SDS-PAGE and blotted with antiphosphothreonine (p-Thr) and anti-CLDN1 antibodies. The levels of p-Thr are represented relative to the values at 0 h or the control (Cont). (C,D) Cells were transfected with FLAG-tagged wild-type (WT), T191E, and T195E CLDN1. At 48 h after transfection, the cells were exposed to 200 μM H2O2 for 6 h. (C) The cells were immunostained with anti-FLAG antibodies (green) and DAPI (blue), a nuclear marker. Scale bars indicate 10 μm. (D) TER and paracellular LY flux were analyzed using a volt-ohmmeter and fluorescence spectrometry, respectively. n = 3-4; ** p < 0.01 and * p < 0.05 significantly different from 0 h, Cont, or WT alone; ## p < 0.01 and # p < 0.05 significantly different from Veh with H2O2 or WT with H2O2.

Phosphorylation of CLDN1 by PKC
The phosphorylation level of CLDN1 is regulated by atypical PKC and protein phosphatases (PPs) [23]. Go6976, a selective PKCα and β isoform inhibitor, decreased p-Thr levels and the TJ localization of CLDN1 ( Figure 7A,B). H2O2 decreased PKC activity and increased PP activities, which were significantly inhibited by EBGP ( Figure 7C). To support the involvement of PKC, we examined the effects of phorbol 12-myristate 13-acetate (PMA), a PKC activator, and cantharidin, a PP2A inhibitor, on the TJ barrier. The H2O2-induced mislocalization of CLDN1 was inhibited by both PMA and cantharidin ( Figure 7D). Similarly, the reduction in the TJ barrier function by H2O2 was significantly rescued by both PMA and cantharidin ( Figure 7E,F). The recovery effects of PMA and cantharidin were blocked by the knockdown of CLDN1 expression. These results indicated that PKC activator and PP inhibitor could rescue the H2O2-induced mislocalization of CLDN1.

Discussion
Disruption of the TJ barrier in the skin is caused by UV exposure and oxidative stress. In previous research, TJ integrity has been examined using high doses of UV, which can induce noticeable cell damage [24]. The production of ROS and H 2 O 2 was increased by UVB in a dose-dependent manner ( Figure 1A,B). ROS production has been reported to be increased by UVB irradiation in a dose-and time-dependent manner [25]. Park et al. [26] reported that ROS production was increased 1.3-fold by UVB irradiation (10 mJ/cm 2 ) and incubation for 8 h in HaCaT cells, which is similar to our data. Transient H 2 O 2 treatment increased the production of ROS and decreased cell viability in a dose-dependent manner ( Figure 1D,E). The percentage of damaged cells was about 10%-20% in the present experimental conditions of transient H 2 O 2 treatment. Transient H 2 O 2 (200 µM) treatment induced the mislocalization of CLDN1 without affecting the amount of CLDN1 protein (Figure 2). There was no apparent change in CLDN4 localization, but the amount of CLDN4 protein increased transiently after 6 h of H 2 O 2 treatment. Recently, El-Chami et al. [27] reported that H 2 O 2 (1 mM, a higher concentration than in our study) induced the mislocalization of CLDN1, CLDN4, and occludin in a continuous cell line of rat epidermal keratinocytes. We suggest that oxidative stress selectively induces the mislocalization of CLDN1 in low-level cell toxic conditions.
The protein level of CLDN1 is decreased in atopic dermatitis, whereas that of CLDN4 is increased [28]. The loss of CLDN1 localization in the TJ may induce a compensatory elevation of CLDN4 expression. Nevertheless, the barrier function of TJ was decreased after 6 h of transient H 2 O 2 treatment ( Figure 3A). The TJ barrier function was reduced by the knockdown of CLDN1 by siRNA ( Figure 4A,B). Furthermore, the rescue effects of EBGP on the decrease in TER and increase in LY flux by H 2 O 2 were blocked by the knockdown of CLDN1 ( Figure 4C). There results indicate that CLDN1 has an important role in the maintenance of the TJ barrier. Previously, Furuse et al. [29] have reported that a continuous TJ was observed in the stratum granulosum of the epidermis in CLDN1-deficient mice and in WT mice using ultrathin section electron microscopy. However, a small molecule tracer (~600 D) passes through the TJ of the epidermis in CLDN1-deficient mice, whereas the diffusion is prevented at the TJ in WT mice. We suggest that CLDN1 cannot be replaced by CLDN4 in the skin.
The H 2 O 2 -induced mislocalization of CLDN1 was inhibited by a clathrin-dependent endocytosis inhibitor ( Figure 5A), suggesting that H 2 O 2 enhances the endocytosis of CLDN1 from the TJ to intracellular compartments. Although the protein levels of CLDN1 did not change, we did not detect the subcellular localization of CLDN1 in the organelle (Figure 2). We have to clarify the subcellular localization of CLDN1 using organelle markers in further studies. The p-Thr level of CLDN1 was decreased by H 2 O 2 ( Figure 6A). The phosphorylation status of proteins is regulated by various protein kinases and PPs. Go6976 induced the dephosphorylation and mislocalization of CLDN1 (Figure 7). On the contrary, the H 2 O 2 -induced mislocalization of CLDN1 was rescued by the PKC activator PMA. These results suggest that the TJ localization of CLDN1 is upregulated by PKC-dependent phosphorylation. Dephosphorylation of CLDN1 is caused by the activation of PP2A [23]. H 2 O 2 increased PP activities, and the mislocalization of CLDN1 was rescued by a PP inhibitor, cantharidin, indicating that PPs may also be involved in the dephosphorylation of CLDN1 by H 2 O 2 . CLDN1 is phosphorylated in both renal tubular MDCK I and colonic HT29 cells, but PKC-induced changes in the phosphorylation state were detected only in MDCK I cells [21]. The regulatory mechanism for the phosphorylation of CLDN1 may be different in each tissue.
Threonine phosphorylation sites of CLDN1 by PKC were predicted at T191 and T195 using the NetPhos 2.0 and Disphos 1.3 servers [22]. A phosphorylation mimic T191E mutant was localized to the TJ and maintained the TJ barrier in cells treated with H 2 O 2 , whereas WT and a T195E mutant lost their localization and barrier function in the TJ ( Figure 6C,D). These results indicate that phosphorylation at T191 may be necessary for CLDN1 to localize to the TJ in HaCaT cells. The necessity of phosphorylation at T191 has been reported using human embryonic kidney 293 cells [30] and MDCK cells [31].
H 2 O 2 caused a decrease in CLDN1 localization in TJ and an increase in paracellular permeability, which were rescued by EBGP ( Figure 3C,D). Similar effects were observed in the UVB-treated cells.
Although the effect of EBGP may be due to antioxidant activity, some cinnamic acid derivatives and flavonoids, which are contained in propolis, have beneficial effects on PKC. Artepillin C enhances adipocyte differentiation and glucose uptake mediated by the activation of PKC [32]. Chlorogenic acid prevents α-amino-hydroxy-5-methyl-isoxazole-4-propionate-mediated excitotoxicity in optic nerve oligodendrocytes through a PKC-dependent pathway [33]. Caffeic acid phenethyl ester inhibits the expression and activity of PP2A [34]. Our preliminary data indicate that kaempferide, which is abundant in EBGP [35], had lower antioxidant activity compared to EBGP, but it rescued the H 2 O 2 -induced mislocalization of CLDN1 and the reduction in the TJ barrier function. The components of propolis, including kaempferide, vary depending on the area [14]. A comparison of the effects of propolis produced in various places may be good for the identification of active components. Further studies are needed on which components of EBGP can rescue the H 2 O 2 -induced mislocalization of CLDN1 using human keratinocytes and what doses are effective. The identification of active components could lead to expanding the range of raw substances beyond green propolis, which could prove useful for the same functions.
In conclusion, we found that UVB irradiation increased ROS production, including H 2 O 2 , and both UVB and H 2 O 2 caused the mislocalization of CLDN1 in HaCaT cells. The H 2 O 2 -induced mislocalization of CLDN1 was rescued by EBGP, PMA, and cantharidin. H 2 O 2 decreased PKC activity and increased PP activities, which were inhibited by EBGP. Go6976 decreased p-Thr levels and the TJ localization of CLDN1. These results suggest that the TJ localization of CLDN1 is controlled by PKC. H 2 O 2 decreased the p-Thr level of CLDN1, but this effect was blocked by EBGP. The T191E CLDN1 mutant was localized to the TJ after treatment with H 2 O 2 . We suggest that the phosphorylation of CLDN1 at T191 is necessary for its localization in the TJ. EBGP and its components may be useful in preventing the destruction of the TJ barrier by UVB and oxidative stress.

Cell Culture
HaCaT cells, an immortalized nontumorigenic human keratinocyte-derived cell line [36], were grown in Dulbecco's Modified Eagle's Medium (Sigma-Aldrich) supplemented with 5% fetal bovine serum (FBS, Sigma-Aldrich), 0.07 mg/mL penicillin-G potassium, and 0.14 mg/mL streptomycin sulfate in a 5% CO 2 atmosphere at 37 • C. One day before experiments, cells were transferred to FBS-free medium. Cell viability was examined using a WST-1 assay.

UVB Irradiation
UVB irradiation was carried out using a UV Crosslinker CL-1000M (Analytik Jena, Upland, CA, USA), which emits most of its energy within the UVB range (peaking at 302 nm). HaCaT cells were irradiated at a dose of 5-50 mJ/cm 2 in Hank's balanced salt solution. After UVB radiation, cells were incubated in fresh medium until analysis.

Production of Reactive Oxygen Species
H 2 DCFDA can detect several ROS, including H 2 O 2 , ·OH, and peroxy radical, whereas OxiOrange and Hydrop selectively detect ·OH and H 2 O 2 , respectively. HaCaT cells were incubated with these ROS-sensitive fluorescent probes for 30 min. The fluorescence intensity of each probe was detected using an Infinite F200 Pro microplate reader (Tecan, Mannedorf, Switzerland).

Confocal Microscopy
Cells were plated on cover glasses. After forming a confluent monolayer, the cells were incubated with cold methanol for 10 min at −30 • C and then permeabilized with 0.2% Triton X-100 for 10 min. Following permeabilization, the cells were blocked with 4% Block Ace (Dainippon Sumitomo Pharma, Osaka, Japan) for 30 min and incubated with anti-CLDN1, anti-CLDN4, anti-FLAG, or anti-ZO-1 antibodies (1:100 dilution) for 16 h at 4 • C, followed by incubation with Alexa Fluor 488-or 555-conjugated secondary antibodies (1:100 dilution). The fluorescence images were observed using an LSM 700 confocal microscope (Carl Zeiss, Jena, Germany).

SDS-Polyacrylamide Gel Electrophoresis and Immunoblotting
Cells were scraped into cold phosphate-buffered saline and precipitated by centrifugation. They were lysed in a RIPA buffer containing 150 mM NaCl, 0.5 mM EDTA, 1% Triton X-100, 0.1% SDS, 50 mM Tris-HCl (pH 8.0), and a protease inhibitor cocktail (Sigma-Aldrich) and were sonicated for 20 s. After centrifugation at 6000× g for 5 min, the supernatants were collected and used as cell lysates, which included membrane and cytoplasmic proteins. Samples were applied to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and blotted onto a polyvinylidene fluoride membrane. The membrane was then incubated with the respective primary antibody (1:1000 dilution) at 4 • C for 16 h, followed by a peroxidase-conjugated secondary antibody (1:3000 dilution) at room temperature for 1.5 h. Finally, the blots were incubated in EzWestLumi Plus (Atto, Tokyo, Japan) or ImmunoStar Basic (Wako Pure Chemical) and scanned with a C-DiGit Blot Scanner (LI-COR Biotechnology, Lincoln, NE, USA). The blots were stripped and reprobed with an anti-β-actin antibody. Band density was quantified using ImageJ software (National Institute of Health software). The signals were normalized using a β-actin loading control.

Measurement of Paracellular Permeability
Cells were plated on Transwell plates (0.4 µm pore size, Corning Inc., Corning, NY, USA). After forming a confluent monolayer, TER was measured using a Millicell-ERS epithelial volt-ohmmeter (Millipore, Billerica, MA, USA). TER values (ohms × cm 2 ) were normalized by the area of the monolayer and were calculated by subtracting the blank values from the filter and the bathing medium. The paracellular permeability to LY, a fluorescent paracellular flux marker, for 1 h from upper to lower compartments was measured with an Infinite F200 Pro microplate reader.

PKC and Serine/Threonine Protein Phosphatase Activity Assays
Cells were harvested in 1× Passive Buffer (Promega, Madison, WI, USA). PKC activity was measured using a CycLex PKC Super Family Kinase Assay Kit (Medical & Biological Laboratories, Nagoya, Japan) in accordance with the manufacturer's protocol. Serine/threonine PP activities were investigated using paranitrophenylphosphate (pNPP) as a substrate at pH 7.5 (neutral condition) and pH 8.4 (alkaline condition). The hydrolysis of pNPP was measured by monitoring the absorbance at 405 nm.

Vector Construction and Transfection
A vector containing human CLDN1 cDNA was prepared as described previously [31] and was called CLDN1/pCMV, which encoded a FLAG tag at the amino terminus. Mutants of T191E and T195E were generated as described previously [31]: siRNAs against CLDN1 (SASI_Hs01_00211216) and a negative control (SIC-001) were purchased from Sigma-Aldrich. Plasmid vector and siRNAs were transfected into cells using Lipofectamine 2000, as recommended by the manufacturer.

Statistics
Results are presented as means ± S.E.M. Differences between groups were analyzed using one-way analysis of variance, and corrections for multiple comparison were made using Tukey's multiple comparison test. Comparisons between two groups were made using Student's t-test. Statistical analyses were performed using KaleidaGraph version 4.5.1 software (Synergy Software, Reading, PA, USA). Significant differences were assumed at p < 0.05.
Author Contributions: K.M., M.K., and S.S. performed the experiments and analyzed the data. T.M. contributed to the experiment plan and to the discussion of the manuscript. A.I. contributed to supervision of the project, interpretation of the data, and writing the paper.
Funding: This work was supported in part by grants from the Yamada Research Grant from Yamada Bee Company, Inc., and the Koyanagi-foundation.