Hmgb1 inhibits Klotho expression and malignant phenotype in melanoma cells by activating NF-κB

The molecular and cellular mechanisms behind the involvement of inflammation in melanoma have not been fully elucidated. In this study, knockdown of Hmgb1 expression increased apoptosis, reduced invasion and p-NF-κB expression, but increased Klotho protein level in melanoma tumor cells. The effect of Hmgb1 knockdown was overcome by LPS. Introduction of exogenous Hmgb1 significantly decreased apoptosis, increased invasion, elevated p-NF-κB, but lowered Klotho protein level in melanoma cells. The effect of exogenous Hmgb1 was agonized by NF-κB inhibitor CAPE. Hmgb1 knockdown activated, but exogenous Hmgb1 inactivated, p-IGF1R/p-PI3K p-85/p-Akt/p-mTOR signaling. Knockdown of Klotho gene expression significantly decreased apoptosis, increased invasion in melanoma cells, and inhibited xenograft A375 tumor growth. A significantly high percentage of cells stained positive for p-NF-κB, but negative for Klotho, in melanoma tissues compared to normal and benign skin tissues. The positive p-NF-κB and negative Klotho protein expression correlated with poor prognosis in melanoma patients. Multivariate analysis revealed an independent association between p-NF-κB / Klotho protein level and overall survival. In conclusion, Hmgb1 can inhibit Klotho gene expression and malignant phenotype in melanoma cells through activation of NF-κB signaling.


INTRODUCTION
Melanoma is a highly malignant tumor that is associated with the deaths of 50,000 people worldwide each year [1]. Currently, the incidence of melanoma is increasing worldwide [2]. However, the molecular mechanisms driving melanoma carcinogenesis and progression have not been fully elucidated.
Many tumors develop as a result of inflammation, and many biological characteristics of tumors exhibit a close relationship with the biological processes of inflammation [3]. NF-κB is an important nuclear transcription factor involved in tumor inflammation and development [4]. NF-κB can stimulate the expression of various cytokines and chemokines involved in epithelial cell growth and activity [5]. We hypothesize that Research Paper www.impactjournals.com/oncotarget NF-κB may be involved in the malignant transformation of epithelial cells.
Extracellular Hmgb1 is a member of the damageassociated molecular patterns and exerts an important role in the startup and promotion of inflammation as a danger signal [6]. Hmgb1 is involved in tumorigenesis and development as a late mediator of inflammation [7]. Our previous studies demonstrated that Hmgb1 is highly expressed in melanoma tissues, and is a predictive factor for the poor prognosis of melanoma patients [8]. Both Hmgb1 and LPS, an immune system activation factor, can activate NF-κB and subsequently promote tumor development [9]. A previous study demonstrated that TNF-α can suppress the expression of the anti-aging gene Klotho through activation of NF-κB [10]. In addition, the Klotho gene is progressively lost in melanoma under an unknown mechanism [11]. We therefore hypothesized that inflammation-activated NF-κB may activate Hmgb1, which subsequently suppresses Klotho gene expression.
This study investigated the effects of Hmgb1 and LPS on Klotho gene expression in melanoma cells and their relationship with NF-κB signaling and the biological significance of inflammation-Klotho in the malignant phenotype of melanoma.

Knockdown of Hmgb1 increased tumor cell apoptosis and decreased invasion in melanoma cells
In this study, 4 melanoma cell lines were used to screen Klotho and Hmgb1 protein expression. Western blot showed that low Klotho protein expression and high Hmgb1 protein expression were detected in WM35 and WM451 cells, whereas high Klotho protein expression and low Hmgb1 protein expression were detected in SK-28 and A375 cells ( Figure 1A). A375 and SK-28 cell lines with high Klotho protein expression were selected for further study. A pGFP-shHmgb1 vector was used to silence Hmgb1 gene expression in A375 ( Figure 1B) and SK-28 ( Figure 1C) cells. 24 hrs after transfection, Western blot showed significant decrease in Hmgb1 protein. The Transwell assay in A375 ( Figure 1D, 1E) and SK-28 ( Figure 1D, 1F) cells showed that shHmgb1 transfection significantly reduced invasion, whereas LPS treatment significantly increased cell invasion compared to NC and BC cells (p<0.001). Invasion in cells treated with shHmgb1 transfection and LPS was significantly higher than that in the NC and BC cells (p<0.001). However, no significant differences in the invasion of cells were observed between treatments with shHmgb1 + LPS and LPS alone (p>0.05) ( Figure 1D, 1E, 1F). Flow cytometry showed that shHmgb1 transfection significantly increased the percentage of sub G0/G1 in A375 ( The effects of exogenous Hmgb1 in cell invasion, cell cycle, and apoptosis in A375 and SK-28 cells The A375 and SK-28 melanoma cells were treated with exogenous Hmgb1 at 0.01 µg/ml, 0.05 µg/ml, 0.1 µg/ ml, and 0.5 µg/ml of Hmgb1 protein with or without 100 μM of NF-κB inhibitor CAPE for 24 and 48 hrs. 0.1 µg/ ml and 0.5 µg/ml of Hmgb1 was similarly effective (data not shown). 0.1 µg/ml of Hmgb1 was selected for further experiments. Invasion test showed that exogenous Hmgb1 (0.1 µg/ml) treatment for 48 hrs significantly increased cell invasion in A375 ( Figure 4A, 4B) and SK-28 cells ( Figure  4A, 4C) (p<0.05). CAPE not only inhibited cell invasion, but also reversed the effects of Hmgb1 on cell invasion in both cell lines (p<0.001). Flow cytometry showed that Hmgb1 treatment for 48 hrs significantly decreased the percentage of sub G0/G1 A375 ( Figure 5A-5E) and SK-28 cells ( Figure 5F-5J) (p<0.05). CAPE not only increased the percentage of sub G0/G1 (p<0.001) cells, but also reversed the effects of Hmgb1 on the cell cycle in both cell lines ( Figure 5E, 5J) (p<0.001). Exogenous Hmgb1 treatment significantly decreased the percentage of cell apoptosis in A375 ( Figure 6A-6E) and SK-28 cells ( Figure  6F-6J) (p<0.05). CAPE not only increased cell apoptosis (p<0.001), but also reversed the apoptotic effect of Hmgb1 in both cell lines ( Figure 6E, 6J) (p<0.001).

Hmgb1 / LPS can activate NF-κB signaling pathway and subsequently inhibit Klotho gene expression and its downstream signaling
The levels of p-NF-κB in shHmgb1 transfected A375 ( Figure 7A) and SK-28 ( Figure 7B) cells were significantly reduced compared to cells in the BC and NC control cells (p<0.001), suggesting that silencing of Hmgb1 gene expression reduced NF-κB phosphorylation. Although the p-NF-κB levels in LPS treated and shHmgb1 + LPS treated A375 and SK-28 cells were significantly increased (p<0.001) compared to cells in the control (BC and NC) cells, no significance in p-NF-κB levels was observed between cells treated with shHmgb1 + LPS and LPS alone (p>0.05), suggesting that both the Hmgb1 and LPS can activate NF-κB ( Figure 7A, 7B). We further found that Klotho protein expression in shHmgb1 treated A375 and SK-28 cells was significantly higher than that in control cells (p<0.001) ( Figure 7A, 7B), suggesting that silencing of Hmgb1 gene expression increased Klotho protein expression. In cells treated with LPS alone, no changes in Hmgb1 protein levels was observed, but the level of Klotho protein was significantly reduced compared to control cells (p<0.001). In A375 and SK-28 cells treated with shHmgb1 + LPS, Hmgb1 and Klotho protein expression was significantly reduced compared to control cells (p<0.001).
We further examined the expression of IGF-1R / PI3K / AKT / mTOR protein. In shHmgb1 treated A375 ( Figure 7A) and SK-28 ( Figure 7B) cells, the protein levels of p-IGF1-R, p-PI3K p85, p-Akt, and p-mTOR were significantly lower than that in control cells (p<0.001). In A375 ( Figure 7A) and SK-28 ( Figure 7B) cells treated with LPS, the levels of p-IGF1-R, p-PI3K p85, p-Akt, and p-mTOR were significantly higher than that in control cells (p <0.001). In cells treated with shHmgb1 + LPS or LPS alone, no significant differences in p-IGF1-R, p-PI3K p85, p-Akt, and p-mTOR levels were observed compared to control cells (p>0.05) ( Figure 7A, 7B). Both the Hmgb1 and LPS can inhibit Klotho protein expression, which correlated with the activity of NF-κB signaling pathway.
In exogenous Hmgb1 treated cells, significant increas in p-NF-κB level, but decrease in Klotho protein expression, was observed in A375 ( Figure 7C) and SK-28 ( Figure 7D) melanoma cells. After treatment with p-NF-κB inhibitor CAPE, p-NF-κB level was significantly    lowered, while the expression of Klotho protein was significantly increased ( Figure 7C, 7D). When Klotho was highly expressed, p-IGF-1R / p-AKT levels were significantly reduced. When Klotho was lowly expressed, p-IGF-1R / p-AKT was significantly increased ( Figure 7C, 7D). These findings suggest that Hmgb1 can activate NF-κB signaling pathway to decrease Klotho gene expression and the activity of downstream signaling.

Klotho knockdown stimulated xenograft tumor growth in mice
We further validated the effect of Klotho gene knockdown on the growth of xenograft A375 tumors in mice. The shKlotho or scrambled control-shRNA (NC) transfected A375 cells were injected subcutaneously in nude mice. After 42 days, the mice were sacrificed. The average volume of the tumors in the shKlotho group was significantly higher than that in the control (NC) group (1.264±0.642 vs. 0.408±0.207cm 3 , Figure 11A, p<0.05). The average tumor weight in the shKlotho group was 1.121±0.572g, which was significantly higher than that in the control group (0.335±0.159 g) ( Figure 11B, p<0.05). Kaplan-Meier analysis showed that the survival rate of mice in shKlotho group was lower than that in NC   Figure 11C, p<0.05). These results suggest that down-regulation of Klotho gene expression significantly enhanced xenograft melanoma tumor growth.

NF-κB activation and low Klotho protein expression correlated with poor prognosis of melanoma patients
Immunohistochemistry showed that p-NF-κB protein staining was mainly located in the nucleus, while Klotho protein staining was mainly located in the cytoplasm and cell membrane ( Figure 12A). Positive p-NF-κB protein staining was observed in 68% of melanoma tissues, which was significantly higher than that in the black nevus carcinoma and peritumoral tissues (23%). Positive Klotho protein staining was observed in 45% of melanoma tissues, which was significantly lower than that in black nevus carcinoma and peritumoral tissues (85%). p-NF-κB (Table 1) and Klotho (Table 2) protein expression in melanoma tissues significantly correlated with the clinical stage (P<0.05). The result of the spearman rank correlation analysis revealed that the expression of Klotho protein negatively correlated with p-NF-κB protein ( Figure 12B, r = 0.819, P < 0.001). Kaplan-Meier analysis showed that the prognosis of patients with high p-NF-κB protein levels (++, +++) was significantly poor than patients with low p-NF-κB levels (-, +) ( Figure 12C). In contrast, low Klotho protein levels (-, +) correlated with a poor prognosis ( Figure 12D). Multivariate analysis based on Cox models was also carried out to further explore the prognostic significance of p-NF-κB (Table 3) and Klotho (Table 4) protein. The results revealed that p-NF-κB and Klotho protein levels were independent predictors of poor overall survival, indicating that high p-NF-κB or low

DISCUSSION
Inflammation and inflammatory cytokines have been associated with tumor occurrence and development. This study showed that silencing Hmgb1 expression significantly increased the percentage of G0/G1 cells and apoptosis in melanoma cells, and reduced tumor cell invasion. In contrast, administration of exogenous Hmgb1 decreased apoptosis and the percentage of G0/G1 cells, and increased cell invasion and NF-κB phosphorylation in the melanoma cells. These findings suggest that Hmgb1 is involved in the development of melanoma. We also found that LPS can stimulate the invasion of melanoma cells, reduce apoptosis, and activate NF-κB. LPS and Hmgb1 activated NF-κB and enhanced tumor cell proliferation.
Our study suggests a pathway of inflammation-activated NF-κB-Hmgb1-Klotho in the cell survival, proliferation, and tumor growth of melanoma.
NF-κB is an important nuclear transcription factor associated with tumor inflammation [12]. After activation, NF-κB promotes the production of cytokines, chemokines, growth factors, and apoptotic proteins [13]. This study found that phospho-NF-κB levels were significantly higher in melanoma tissues than in black nevus and peritumoral tissues. Melanoma patients with high phospho-NF-κB levels had poor prognosis. Experiments with two melanoma cell lines showed that both Hmgb1 protein and LPS treatment increased the malignant phenotypes of melanoma cells via activation of NF-κB. Moreover, LPS can totally overcome the phenotypes and the decrease in Hmgb1 protein expression caused by Hmgb1 silencing. In contrast, treatment with NF-κB inhibitor CAPE can reverse the phenotypes caused by exogenous Hmgb1. These findings suggest that inflammation-activated NF-κB can stimulate Hmgb1 expression and its biological activities.   Previous reports implicate an association between inflammation and Klotho gene expression. For example, the inflammatory mediator TNF-α could significantly inhibit Klotho gene expression in renal epithelial cells [10] and fat cells [14]. This study found that Klotho protein expression was significantly lower in melanoma tissues than that in peritumoral tissues and black nevus tissues. The low Klotho protein expression correlated with high percentage of positive p-NF-κB staining in melanoma tissues. Low Klotho protein expression significantly correlated with poor prognosis of patients with melanoma. We also found that when NF-κB signaling was inhibited, Klotho protein expression was significantly increased. In contrast, when NF-κB signaling was activated, Klotho protein expression was significantly decreased. In short, we found that Klotho gene expression in melanoma cells is inhibited by inflammatory factors NF-κB.
Our previously studies found that Klotho protein inhibited the activation of IRS-1 / PI3K / Akt / mTOR and ERK / p70s6k and downstream signaling pathways in gastric cancer and liver cancer cells, mainly through the inhibition of IGF-1R phosphorylation [15,16]. This study found that after silencing Klotho gene, p-IGF-1R / p-AKT levels and cell invasion were significantly increased, whereas tumor cell apoptosis was significantly decreased. At the same time, the p-IGF-1R / p-PI3K / p-Akt / p-mTOR's levels were also increased or decreased with the activation or inhibitor of NF-κB. These findings indicate that Klotho may be an intermediate factor of NF-κB and insulin signaling pathways, and is involved in the malignant phenotypes of melanoma. We further validated that down-regulation of Klotho protein expression stimulated growth of xenograft A375 tumors in mice.
In conclusion, our study first proposed that inflammation inhibits Klotho gene expression in melanoma cells through activation of the inflammatory mediators Hmgb1 and NF-κB signal pathway.

MATERIALS AND METHODS
Samples 74 melanoma tissues were collected from the Second Xiangya Hospital and Third Xiangya Hospital, Central South University between 2002 and 2010 at the Department of Pathology. After passing the exclusion criteria including insufficient patient information and tumor material, 66 samples were finally collected for this study. 20 black nevi and peritumoral tissues were collected as control at the same stage. Melanoma was diagnosed by HE staining, HMB45 and S-100 protein immunohistochemical staining by three pathologists (double blind method). None of the 66 melanoma patients received chemotherapy and radiotherapy before surgery. Among the 66 melanoma patients, 30 were male and 36 were female with ages ranging from 28 to 65 years old. All specimens were fixed in 10% neutral formalin and embedded in paraffin. Tissues were sectioned at a thickness of 4 µm and stored at room temperature. The follow-up was performed by mail and telephone. The study was compliant with ethical standards.

Immunohistochemistry
Sections were deparaffinized with xylene followed by hydration in gradient ethanol and 3% hydrogen peroxide solution to eliminate endogenous peroxidase. After washing with 1xPBS buffer (pH7.4), antigen retrieval was performed by heating the sections in citric acid buffer (pH6.0) in a microwave oven at 750W for 12 min, cooling for 3 minutes, and heating at 750W for 12 minutes again, followed by cooling naturally to room temperature. After blocking at room temperature for 30 min, sections were incubated with mouse anti-human p-NF-κB and Klotho antibodies overnight at 4°C. After washing with 1xPBS buffer, sections were incubated Brown staining in the cytoplasm, membrane, and nucleus was justified as positive. 10 high-power fields were observed for each slice. The results are divided into: <5% (-), 6% to 30% (+), 3l % to 60% (++),> 60% (+++) according to the percentage of positive cells and the degree of staining intensity of cells. A case was defined as high expression when more than 31% (++ to +++) of cells were positively stained, while a case was defined as low expression when less than 30%, and the (-to +) of cells were positive.

Cell culture
Human melanoma cell lines WM35, WM451, SK-28, and A375 were purchased from the Cell Bank of Shanghai Institute. Cells were cultured in RPMI 1640 (GIBCO) containing 10% fetal bovine serum at 37 °C, 5% CO 2 .

Apoptosis and cell cycle analysis
To detect apoptosis, cells were digested with 0.25% EDTA-free trypsin and collected by centrifugation at 2,000 rpm for 5 min. After washing with 1 × PBS, 1 ~ 5 × 10 5 cells were suspended in 500 μL of binding buffer. 5 μL of annexin V-FITC was then added, followed by the addition of 5 μL of propidium iodide, and incubation at room temperature for 10 min in the dark room. Apoptotic cells were sorted by flow cytometry.
To perform cell cycle analysis, cells were collected as described above and fixed in pre-cooled 75% ethanol overnight at 4°C. After washing with 1×PBS (1000rpm, 5min), cells were resuspended in a solution of 800 µl of 1×PBS containing 1% BSA. After adding 100 µl of PI dye (3.8 × 10 -2 M sodium citrate, pH 7.0) and 100 µl of RNase A (10 mg / ml), cells were incubated at 37°C for 30 min in the dark. Cell cycles were analyzed by flow cytometry.

Western blot analysis
Cells were lysated and total protein was extracted. Western blot was performed as previously described [17]. Anti-Klotho, anti-p-IGF-IR, anti-IGF-IR, anti-p-AKT1, and anti-GAPDH antibodies were purchased from Santa Cruz biotechnology (Santa Cruz, CA, USA). Anti-AKT antibody, anti-p-PI3K, and anti-p-mTOR antibodies were purchased from Cell Signaling Technology (Danvers, MA, US).

Cell invasion assay
After digestion with 0.25% EDTA-free trypsin, cells were washed once with 1×PBS and suspended in RPMI 1640 medium containing 1% BSA at 5 × 10 4 /ml. 1 ml FBScontaining medium was added to 6-well plates and 2 ml cell suspension was added into the Transwell chamber. Cells were cultured at designated time points and membrane was removed from the lower chamber. Matrigel was wiped with a cotton swab and cells were fixed with 95% alcohol for 10 min followed by hematoxylin staining for 10 minutes. Cells were counted under inverted microscope.

Nude mouse xenograft studies
Sixteen male BALB/c-nu/nu (aged 4-6 weeks) mice were purchased from the animal laboratory of the Third Xiangya Hospital, Central South University. To explore the role of Klotho in melanoma cells, xenograft tumors were grown in nude mice. The shKlotho or scrambled control-shRNA transfected A375 cells (2×10 6 ) were injected subcutaneously in the ventral trunk of mice. Nude mice were sacrificed at 42 days after tumor implantation. Tumor volume, weight, and the survival rates of mice were measured. The experimental procedures were approved by the Ethics Committee of the Faculty of Experimental Animals, Central South University.

Statistic analysis
Statistical analysis was performed using SPSS13.0 statistical software. The count data were analyzed using