Ginsenoside Rg1 attenuates ultraviolet B-induced glucocortisides resistance in keratinocytes via Nrf2/HDAC2 signalling

Oxidative stress, which occurs after ultraviolet (UV) radiation, usually results in Glucocorticoid (GC) resistance and the subsequent development of skin inflammation. One approach to protecting the skin against UV radiation is the use of antioxidants. The ginsenoside Rg1 is a novel natural antioxidant isolated from the medicinal plant Panax ginseng C.A. Mey. We demonstrated that UVB exposure exacerbated inflammation and reduced both the level of the glucocorticoid receptor (GR) and the efficacy of dexamethasone (Dex) in human keratinocytes (HaCaT cells). Pretreatment with Rg1 increased the expression of GR and restored Dex responsiveness to inflammation in UVB-irradiated HaCaT cells. Mechanistically, Rg1 rescued UVB-induced HDAC2 degradation. HDAC2 knockdown partially abolished the Rg1-induced up-regulation of GR and the enhancement of GC sensitivity. In addition, Rg1 reduced the production of reactive oxygen species (ROS), which preceded the up-regulation of HDAC2, and consequent sensitization of cells to Dex. Moreover, Rg1 treatment promoted the translocation and activation of Nrf2. Nrf2 knockdown partially abolished the Rg1-induced decrease of ROS production and increase of HDAC2. Rg1 also potentiated the anti-inflammatory effects of Dex in UVB-irradiated mouse skin. In conclusion, we demonstrated that Rg1 attenuated UVB-induced GC insensitivity. Notably, these effects were partially mediated by the Nrf2/HDAC2 pathway.


Rg1 up-regulated the transrepression and expression of the GR by reducing GR acetylation.
To understand the effects of Rg1 on UVB-induced Dex insensitivity, we next examined the GR, the GC receptor. The anti-inflammatory response mediated by the GC-GR complex occurs partially through repressing the expression of genes encoding NF-κ B-dependent cytokines 18 . Therefore, we assessed GR transrepression by detecting NF-κ B activation. Dex markedly reduced TNF-α -induced NF-κ B-dependent luciferase activity. UVB exposure did not further increase TNF-α -induced NF-κ B activation but it abolished the inhibitory effects of Dex. Pretreatment with Rg1 did not affect UVB+ TNF-α -induced NF-κ B activation, but it completely restored the inhibitory effects of Dex ( Fig. 2A). Interestingly, GR was decreased after treatment with Dex or UVB radiation (Fig. 2B, lanes [3][4][5][6], and pretreatment with Rg1 significantly restored the level of GR in HaCaT cells after UVB irradiation ( Fig. 2B; lanes 7, 8 and 9). Using realtime RT-PCR for analysis of GR mRNA, similar results were observed (Fig. 2C).
Since histone acetylation plays a significant role in the regulation of GR function 3 , we examined acetylated GR expression. As shown in Fig. 2D, UVB irradiation promoted GR acetylation, but Dex treatment had no effect. However, pretreatment with Rg1 markedly reduced GR acetylation (Fig. 2D, lanes 5 and 6).
HDAC2 was involved in Rg1-induced GR up-regulation. HDAC2 plays a critical role in the deacetylation of non-histone proteins, specifically transcription-related factors 16 . Therefore, we investigated whether the effects of Rg1 on GR acetylation and levels were dependent on HDAC2. As shown in Fig. 3A and Fig. 2S, Dex slightly decreased HDAC2 levels, and UVB exposure significantly reduced the HDAC2 level (Fig. 3A, lanes 3-6). However, pretreatment with Rg1 partially restored UVB-induced HDAC2 degradation (Fig. 3A, lanes 7-9).
Next, we examined whether shRNA knockdown of HDAC2 could abrogate the effects of Rg1 on UVB-induced GC insensitivity. In HaCaT cells transfected with HDAC2 shRNA, the HDAC2 level was markedly reduced (Fig. 3B, top lane) and Rg1-induced up-regulation of the GR was inhibited (Fig. 3B, middle lane). Moreover, the effect of Rg1 on Dex efficacy after UVB irradiation was effectively abolished in cells transfected with HDAC2 shRNA (Fig. 3C and D) compared with cells transfected with control shRNA.

Rg1 restored HDAC2 levels through ROS inhibition.
Oxidative stress significantly attenuates HDAC2 activity and levels, thereby limiting the GR-mediated recruitment of GCs to DNA 38 . Therefore, we evaluated the effect of Rg1 on ROS production. As shown in Fig. 4A, UVB irradiation significantly increased ROS production. However, pretreatment with 10 μ M Rg1 or 2 mM N-acetyl-L-cysteine (NAC, a ROS scavenger) markedly inhibited Scientific RepoRts | 6:39336 | DOI: 10.1038/srep39336 UVB-induced ROS production. Next, we determined whether the inhibition of ROS contributed to the effect of Rg1 on HDAC2 induction and GR up-regulation. As shown in Fig. 4B

Rg1 inhibited ROS production via Nrf2 activation.
A previous study demonstrated that Nrf2 increases the level of HDAC2 and plays key roles in protecting against UVB radiation and inflammation 19 . Therefore, we investigated whether Rg1 could enhance Nrf2 activity. As shown in Fig. 5A and Fig. 4S, the nuclear translocation and cytoplasmic expression of Nrf2 were reduced significantly after UVB irradiation (lanes 4 and 6). Dex treatment had no effects on this UVB-associated reduction in Nrf2 levels (lane 5), whereas Rg1 pretreatment markedly increased the nuclear translocation and cytoplasmic expression of Nrf2 after UVB irradiation (lanes 7-9). We also observed that TNF-α stimulated the nuclear translocation of Nrf2 (lane 2). RT-PCR demonstrated that pretreatment with Rg1 significantly increased the expression of Nrf2 target genes, including haem oxygenase-1 (HO-1), glutamate-cysteine ligase catalytic subunit (GCLC) and glutamate-cysteine ligase modifier subunit (GCLM)) ( Fig. 5B). Knock-down of Nrf2 by siRNA markedly inhibited the Rg1-induced up-regulation of HDAC2 and the GR (Fig. 5C) and abolished Rg1-mediated inhibition of ROS production (Fig. 5D). In addition, the ability of Rg1 to promote the effect of Dex on anti-inflammatory cytokines was abolished in cells transfected with Nrf2 siRNA (Fig. 5E top and bottom), compared with cells transfected with control siRNA.

Rg1 potentiated GC efficacy after UVB irradiation in vivo.
The morphology and histopathological changes of the mouse skin in the various groups are shown in Fig. 6. The red arrows in Fig. 6A(i-vi) indicate the epidermal thickness in shaved mice. The epidermal thickness in the UVB irradiation group (Fig. 6A-iv, red arrow) was clearly greater than those of the vehicle control and TNF-α treatment alone groups ( Fig. 6A-i,ii). Dex treatment slightly reduced the epidermal thickness ( Fig. 6A-v), and Rg1 treatment markedly decreased the epidermal thickness after UVB irradiation (Fig. 6A-vi). Furthermore, as shown in Fig. 6B, the epidermal thickness HaCaT cells were treated with dexamethasone (Dex, 1 μ M) and/or Rg1 (50 μ M) for 1 hour and subsequently exposed to UVB (60 mJ/cm 2 ). Then, the cells were treated with 10 ng/ml TNF-α for 24 hours. (A) IL-6 and IL-8 levels in the cell culture supernatant were analysed using ELISA. A minimum of two independent experiments using triplicate samples provided comparable results. a p < 0.05, aa p < 0.01 vs. vehicle. b p < 0.05, bb p < 0.01 vs. TNF-α only. c p < 0.05, cc p < 0.01 vs. TNF-α + UVB. (B) The localization of p65 was determined using western blot analysis of nuclear and cytoplasmic protein isolates. (C) MAPK phosphorylation in total protein isolates was evaluated using western blot assays with anti-pP38, anti-pJNK, and anti-pERK.
of the UVB-irradiated group (33.13 ± 2.41 μ m) increased approximately 3-fold compared with the control group (11.46 ± 1.17 μ m). The epidermal thickness of the Dex group (23.02 ± 0.92 μ m) and the Rg1 + Dex (12.55 ± 0.57 μ m) group was significantly lower than that of UVB radiation alone group.
Next, as shown in Fig. 6C, TNF-α treatment caused significant increases in IL-6 and IL-8 levels compared with the vehicle control group, and UVB irradiation enhanced this inflammatory response. Dex significantly inhibited the production of inflammatory cytokines, and the anti-inflammatory effects of Dex were reduced after UVB irradiation. Rg1 combined with Dex markedly reduced the IL-6 and IL-8 levels in skin homogenates compared with Dex treatment alone.
Oxidative stress in the form of ROS overproduction and lipid peroxidation causes skin injury in UVB irradiation-induced photodamage in shaved mice 39 . In our study, ROS levels ( Fig. 6D) and lipid peroxidation ( Fig. 6E) increased in the skin of UVB-irradiated mice compared with the control mice. Treatment with Rg1 in combination with Dex significantly reduced the ROS level and lipid peroxidation of the skin compared with Dex treatment alone.

Discussion
Therapeutic strategies to restore GR levels are urgently needed to overcome the challenge of GC resistance 2,40 .
In this study, we demonstrated that the ginsenoside Rg1 effectively reversed UVB-induced Dex insensitivity by up-regulating the GR and that these effects were partially mediated by the Nrf2/HDAC2 pathway.
Consistent with previous studies 41,42 , the cytotoxic effects of UVB radiation on HaCaT cells occurred in a dose-and time-dependent manner. In this study, 60 mJ/cm 2 UVB irradiation caused a significant increase in ROS production and induced minor cytotoxic effects (Fig. 1SA); therefore, this dose was used in the remaining experiments.
We demonstrated that Rg1 enhanced the anti-inflammatory effects of Dex on UVB-irradiated cells. Rg1 treatment alone partially inhibited TNF-α -induced IL-6 and IL-8 secretion after UVB exposure. This effect was likely due to the anti-inflammatory effect of Rg1 43 , as Rg1 inhibited MAPK phosphorylation and NF-kB translocation. However, the precise mechanism of the Rg1-mediated reduction of inflammation requires further investigation. It has been reported that a defect in GR deacetylation caused by reduced HDAC2 resulted in GC insensitivity in terms of NF-κ B-mediated gene expression 18 . Consistent with this finding, we found that UVB irradiation impaired Dex-mediated inhibition of NF-κ B activity and that pretreatment with Rg1 blocked this effect. This phenomenon may be explained by (i) increased Rg1-mediated GR transcriptional repression or (ii) reduced Rg1-mediated GR deacetylation via HDAC2 up-regulation. A reduction in GR levels is one of the mechanisms that contributes to GC resistance 2 . We also found that Rg1 up-regulated GR expression, which might be explained by the effects of deacetylation on ubiquitin and protein binding, resulting in the loss of selective GR degradation 12,44 .
HDAC2 is a critical component of the GR-corepressor complex that mediates the transrepression of NF-κ B transcriptional activity by deacetylating histones in the promoters of genes encoding pro-inflammatory factors 17 , and the GR 18 . Our results clearly demonstrated that HDAC2 silencing markedly inhibited, but did not fully block, the Rg1-induced restoration of GC efficacy. This phenomenon might be explained by (i) incomplete knockdown of HDAC2 at the protein level or (ii) another mechanism associated with the effect of GCs on inflammation. The level and transcriptional repression of the GR are not the only factors associated with GC sensitivity. For example, (i) GCs show non-genomic anti-inflammatory action by binding with or without the GR to activate annexin-1 45,46 , (ii) GCs induce the expression of many anti-inflammatory genes by transactivation of the GR 47 or (iii) GCs activate anti-inflammatory genes by recruiting coactivator molecules such as steroid receptor coactivator 1 (SRC1), SRC2, and SRC3 48 . In this respect, the effects of Rg1 on non-genomic effects of GC, GR transactivation, and the GR-corepressor complex should be considered and further investigated in the future.
ROS reduce the efficacy of GCs by modifying HDAC2 activity 14 . We found that Rg1 inhibited UVB-induced ROS production. In addition, Rg1 rescued H 2 O 2 -induced down-regulation of HDAC2 and GC insensitivity, indicating that Rg1 restored GC responsiveness by inhibiting the effects of oxidative stress. To the best of our knowledge, Nrf2-deficient cells are sensitive to oxidative stress 25,49 . In our study, Rg1 treatment activated Nrf2 by increasing the translocation of Nrf2 and expression of the major Nrf2 target genes HO-1, GCLC and GCLM, which exert potent antioxidant effects in multiple cell types, including keratinocytes 50,51 . More importantly, Nrf2 knockdown partially abolished the inhibitory effect of Rg1 on ROS production, indicating that the effect of Rg1 on ROS production was at least partially Nrf2-dependent. Additional mechanisms, including PI3K/AKT activation, also contributed to Rg1-mediated ROS inhibition (Fig. 5S). Yasuo et al. 52 reported that corticosteroid insensitivity in COPD could be reversed by inhibiting oxidative stress-dependent PI3K-δ activation. In this respect, a PI3K inhibitor or siRNA should be used to validate the contribution of PI3K/AKT to Rg1-mediated inhibition of ROS production in the future.
Furthermore, animal studies were performed to explore the combined effects of Rg1 and Dex, with the goal of identifying a safe and effective strategy to treat skin inflammation. No obvious liver or kidney damage was observed after treatment with Rg1 combined with Dex (Fig. 6S). In the present study, epidermal thickness and inflammation were used as the primary markers of UVB irradiation-induced damage. Our results indicated that Rg1 combined with Dex effectively decreased the inflammation and changes in epidermal thickness resulting from UVB irradiation. ROS overproduction and lipid peroxidation can cause skin injury in UVB-exposed mice 33 .

Figure 3. Rg1-induced HDAC2 up-regulation is essential for restoring GR levels and dexamethasone sensitivity in UVB-irradiated HaCaT cells.
HaCaT cells were treated with Dex (1 μ M) and/or Rg1 (50 μ M) for 1 hour and subsequently exposed to UVB (60 mJ/cm 2 ). Then, the cells were treated with 10 ng/ml TNFα for 24 hours. (A) HDAC2 levels in total protein isolates were evaluated using western blot. (B) Cells were transfected with HDAC2 shRNA, and GR levels in total protein isolates were evaluated using western blot. (C) IL-6 levels in cells transfected with HDAC2 shRNA were assessed using ELISA. (D) IL-8 levels in cells transfected with HDAC2 shRNA were assessed using ELISA. a p < 0.05, aa p < 0.01 vs. TNF-α + UVB. At least two independent experiments provided highly comparable results.
Our results demonstrated that Rg1 effectively reduced the amount of ROS and lipid peroxidation, suggesting that Rg1 enhances the anti-inflammatory effects of Dex in vivo by exerting antioxidant effects.
In conclusion, the ginsenoside Rg1 attenuated UVB-induced GC insensitivity. Notably, these effects were mediated, at least in part, via the Nrf2/HDAC2 pathway. However, it is still unknown if Rg1 affects endogenous cortisol production, and the question of whether other Rg1-mediated mechanisms are involved in GC sensitivity merits further investigation. Rg1 might be a potential candidate drug to attenuate UVB irradiation-induced GC resistance in chronic inflammatory or autoimmune diseases of the skin.

UVB-induced Dex resistance in vitro.
Cells were seeded in 6-or 96-wells plates overnight. Then, the cells were pre-treated with Dex (1 × 10 −6 M) and/or Rg1 (5 × 10 −5 M) for 1 hour, and the medium was discarded. The cells were irradiated using a Philips TL-D/08 UVB instrument (wavelength 290-340 nm, peak 311 nm, Philips, Amsterdam, the Netherlands) with a total energy dose of 60 mJ/cm 2 . After that, the medium was added back to the cells as described. One hour later, TNF-α (10 ng/ml) was added to the wells for 24 hours.

Measurement of reactive oxygen species.
The cells were plated in black 96-well plates and allowed to attach to the wells for 24 hours. The cells were pre-treated with NAC (2 mM) or Rg1 (5 × 10 −5 M) for 1 hour. Then, the medium was discarded and the cells were irradiated with a total energy dose of 60 mJ/cm 2 using a UVB instrument. The cells were subsequently stained with 5 μ M CellROX ™ Green Reagent (Thermo Fisher Scientific, MA, USA) and incubated at 37 °C for 30 min 53 . Western blot analysis. Whole cell protein and nuclear lysates were prepared, and the lysates were analysed using western blot assays as previously described 43 . Real-time PCR. The RT-PCR assays were conducted as previously described 54 .  39 . The mice were randomly divided into six groups of eight mice each. Group 1, which served as the control group, was treated with PBS and did not undergo UVB irradiation. Group 2 was treated with TNF-α and did not undergo UVB irradiation. Group 3 was treated with TNF-α and Dex, and did not undergo UVB irradiation. Group 4 was treated with TNF-α and underwent UVB irradiation. Group 5 was treated with TNF-α and Dex, and underwent UVB irradiation. Group 6 was treated with Rg1, TNF-α , and Dex, and underwent UVB irradiation.
TNF-α (2.5 mg/kg) was subcutaneously injected at a dose of 100 μ l per mouse after the final UVB irradiation. Dex (2.5 mg/kg) was intraperitoneally injected at a dose of 100 μ l per mouse before the TNF-α treatment. Rg1 (15 mg/kg) was intraperitoneally injected at a dose of 100 μ l per mouse daily for 3 weeks. The UVB source (311 nm) of simulated solar irradiation was provided by a UVB ultraviolet instrument (Sigma, SH4 B, Shanghai, China). The mice in the UVB-irradiated groups were exposed to UVB radiation three times per week for 3 weeks for a total energy dose of 600 mJ/cm 2 . A dose of 50 mJ/cm 2 was administered during the first week, and a dose of 75 mJ/cm 2 was administered in the remaining 2 weeks. The animals were sacrificed 24 hours after TNF-α treatment, and dorsal skin samples were immediately excised. The samples were washed with ice-cold PBS and stored at − 80 °C. The skin samples were assessed for histological, antioxidant and anti-inflammatory properties. The study was conducted in accordance with the European Communities Council Directive of November 24, 1986 (86/609 EEC) and was approved by the Ethics Committee of Changhai Hospital.

Evaluation of Lipid Peroxidation in vivo.
Lipid peroxidation in skin tissue was evaluated as previously described by Chiang-Wen Lee et al. 55 .

ROS levels in vivo.
ROS levels in the skin tissue of mice in each experimental group were measured as previously described in ref. 56. Histology. The skin tissue was dissected post-mortem, fixed in 10% formalin saline solution, decalcified, dehydrated, and embedded in paraffin. Sections of 5 μ m were prepared and stained with haematoxylin and eosin. The epidermal thickness of each sample was determined using a microscope.