Two tissue microarray (TMA) sections of melanoma samples were purchased from Pantomics, Inc. (Richmond, CA, USA) and Super BioChips (Seoul, Korea), respectively, and used for immunohistochemistry. The specimens included 70 primary melanoma and 25 metastatic melanoma tissue samples. Three normal skin tissues and 10 primary melanoma tissue samples were obtained from Yanbian University Hospital and used for quantitative real-time PCR (qRT-PCR) and methylation-specific PCR (MSP) analysis. The clinicopathological characteristics of all patients are shown in Table I. This study was approved by the Institutional Review Board of Yanbian University Hospital.

The human melanoma cell lines (SK-MEL-2 and SK-MEL-24), and adult human epidermal melanocytes (HEMs) were purchased from ATCC (Rockville, MD, USA). Melanoma cell lines were cultured in microvascular endothelial cell medium (MEM) supplemented with 10% fetal bovine serum (FBS), 100 U/ml ampicillin, 100 µg/ml streptomycin, and 2 mM glutamine (Gibco BRL, Gaithersburg, MD, USA). HEMs were cultured in MGM™-4 melanocyte cell basal medium (Lonza Group AG, Basel, Switzerland) supplemented with MBM-4 plus SingleQuots™. Cells were grown in a humidified atmosphere of 5% CO₂ at 37°C. SK-MEL-24 cells were positive for GPX3 on both the mRNA and protein levels. GPX3-depleted stable SK-MEL-24 cell lines were created using GPX3 shRNA-encoding plasmid (Origin Pharmaceutical Services Ltd., Abingdon, Oxfordshire, UK). Retroviruses were generated in 293T cells and the transduced cells were selected with 1 µg/ml puromycin. Silencing of GPX3 in SK-MEL-24 cell lines was confirmed by reverse transcription polymerase chain reaction (RT-PCR) analysis (Fig. 1A).

Melanoma tissue sections were deparaffinized in xylene and rehydrated in graded alcohol. Antigen retrieval was performed by a 2 min autoclave treatment using antigen retrieval buffer (Dako North America, Inc., Carpinteria, CA, USA). Sections were treated with primary antibody against GPX3 (working dilution: 1:100; Abcam plc, Cambridge, UK) at room temperature (RT) for 1 h, and the Donkey Anti-Mouse IgG H&L (Alexa Fluor^® 488) (Abcam) was used as the secondary antibody. Counterstaining was performed using 4′,6-diamidino-2-phenylindole (DAPI). In this study, sections of cell blocks from HeLa cells were used as a positive control for staining of GPX3. Rabbit IgG (R&D Systems, Wiesbaden-Nordenstadt, Germany) was used as a negative control in this study.

GPX3 expression was determined in both HEMs and melanoma cell lines by immunocytochemistry. Cell lines were plated on glass coverslips containing 6-well plates on the day before staining. After fixation with 95% ethanol, the cells were blocked with 5% bovine serum antigen (BSA) at RT for 1 h. Endogenous peroxidase activity was inactivated with a mixture of H₂O₂ and methanol, in a ratio of 1:40. The cells were then treated with primary antibody against GPX3 (working dilution: 1:200; Abcam) at RT for 1 h, and the Real™EnVision™HRP Rabbit/Mouse detection system (Dako North America, Inc.) was used as the secondary antibody. After visualization with 3,3′-diaminobenzidine, counterstaining was performed using hematoxylin. HeLa cells were used as a positive control for GPX3.

Total RNA was isolated with TRIzol reagent (for cell line samples; Invitrogen, Carlsbad, CA, USA) and the RNeasy FFPE kit (from paraffin tissue samples; Qiagen GmbH, Hilden, Germany), and cDNA was synthesized using total RNA and AccuPower^® RT PreMix (Bioneer, Seoul, Korea) according to the manufacturer's protocol. qRT-PCR was performed as previously described using 1X SYBR-Green Master Mix (Applied Biosystems, Foster City, CA, USA) and synthesized cDNA in an ABI7500 Real-time PCR system (Applied Biosystems) under the following conditions: initial denaturation for 10 min at 95°C, followed by 40 cycles of 95°C for 20 sec, 50°C for 30 sec, and 72°C for 45 sec. Conventional RT-PCR analysis was performed using AccuPower PCRPreMix (Bioneer) and the synthesized cDNA with an annealing temperature of 58°C. β-actin was used as the housekeeping gene. Oligonucleotide primers used for the PCR were: 5′-CAACCAATTTGGAAAACAGG-3′ and 5′-GTGGGAGGACAGGAGTTCTT-3′ for GPX3; and 5′-ATAGCACAGCCTGGATAGCAACGTAC-3′ and 5′-CACCTTCTACAATGAGCTGCGTGTG-3′ for β-actin (29).

Two melanoma cell lines (SK-MEL-24 and SK-MEL-2) were treated with 5-Aza (Sigma-Aldrich, Inc., St. Louis, MO, USA), 1 or 10 µM, for 72 h, with drug replacement every 24 h. Cells were then harvested and mRNA expression of GPX3 was detected in each group of the cells.

Genomic DNA extraction was performed in HEMs and melanoma cell lines, three normal skin samples, and 10 melanoma tissue samples using the QIAamp DNA mini kit (Qiagen GmbH). Modification of DNA was performed with the EZ DNA Methylation kit (Zymo Research Corp., Irvine, CA, USA), and the methylation-specific or unmethylation-specific amplification was performed with an annealing temperature of 59°C using the following primers: 5′-TATGTTATTGTCGTTTCGGGAC-3′ and 5′-GTCCGTCTAAAATATCCGACG-3′ for methylation-specific amplification; and 5′-TTTATGTTATTGTTTTGGGATG-3′ and 5′-ATCCATCTAAAATATCCAACACTCC-3′ for unmethylation-specific amplification. The amplified DNA products were loaded on 2% agarose gels for electrophoresis and stained with ethidium bromide.

To determine the influence of GPX3 on the proliferative ability of the melanoma cells, the trypan blue exclusion assay was performed in GPX3-depleted SK-MEL-24 (GPX3^Δ) cells and cells transfected with the control vector (Mock). Cells were seeded in 6-well plates and the number of cells in each group was counted at 36 and 72 h using a hemocytometer.

To evaluate the influence of GPX3 on the migration ability of the melanoma cells, a wound healing assay was performed in GPX3^Δ- and Mock-SK-MEL-24 cells. The cells were seeded in 24-well culture plates in triplicate. After 24 h of culturing, linear scrape injuries were made with a yellow pipette tip on the cells that displayed growth to a confluent monolayer. Wound closing in each group was photographed at the indicated time points and analyzed with the Image J analysis software package (National Institutes of Health, Bethesda, MD, USA).

To evaluate the influence of GPX3 on the invasiveness of the melanoma cells, an invasion assay was performed using Transwell filters (pore size, 8 µm; BD Biosciences, Bedford, MA, USA) in GPX3^Δ- and Mock-SK-MEL-24 cells. Both the lower and top sides of Transwell filters were coated with 8 µg/µl Matrigel (BD Biosciences, San Jose, CA, USA) and the filters were placed on 24-well culture plates with a culture medium with 15% BSA. The cells were suspended in a medium containing 3% BSA and then seeded in the coated Transwell chambers. After 36 h of culturing, the bottom side of the membrane with the invaded cells was fixed with 4% paraformaldehyde and then stained with crystal violet. The number of invaded cells was counted with a microscope and comparatively investigated between GPX3^Δ- and Mock-SK-MEL-24 cells.

In this study, we used commercially available software (SPSS version 15.0 for Windows; SPSS, Inc., Chicago, IL, USA) for statistical analysis. The Mann-Whitney U test was used to analyze differences between groups in the in vitro studies. The χ² test and Fisher's exact test were used to investigate differences between groups of clinical samples. Survival analysis for melanoma patients was performed using Kaplan-Meier analysis, and the differences were compared using the log-rank test. A value of P<0.05 was considered to indicate a statistically significant difference.

In the current study, both mRNA and protein expression of GPX3 were comparatively investigated in HEMs and melanoma cell lines. Our results showed that GPX3 expression was downregulated in melanoma cell lines to a greater extent than in HEMs on both the mRNA and protein levels (Fig. 1B and C). The methylation status of GPX3 was evaluated in both HEMs and melanoma cell lines using MSP primers against the GPX3 promoter. Full methylation (presence of only methylated CpGs) was found in the SK-MEL-2 cell line, which was negative for GPX3 on both the RNA and protein levels. Moreover, HEMs and SK-MEL-24 cells, with expression of GPX3 on both the RNA and protein levels, showed unmethylation (presence of only unmethylated CpGs) and partial methlylation (presence of both methylated and unmethylated CpGs) for GPX3, respectively (Fig. 1E). To investigate whether methylation status was related to repression of GPX3 expression, melanoma cell lines were treated with the DNA demethylation agent 5-Aza, and our results showed that demethylation of the GPX3 gene by 5-Aza in both SK-MEL-2 and SK-MEL-24 cell lines restored GPX3 expression at mRNA and protein levels (Fig. 1D and F). No change was found in GPX3 expression of HEMs according to the presence or absence of 5-Aza treatment (Fig. 1D).

The methylation status of GPX3 was also determined in 3 normal skin and 10 primary melanoma tissue samples. Unmethylation for GPX3 was found in the 3 normal skin samples. By contrast, full methylation and partial methylation were detected in three and seven melanoma tissues samples, respectively (Fig. 2A). Moreover, RT-PCR analysis showed that GPX3 mRNA expression was significantly decreased in primary melanoma tissues compared to normal skin tissues (Fig. 2B).

SK-MEL-2 showed increased invasion ability than SK-MEL-24 cells, which was found to be negative for GPX3 on both RNA and protein levels (data not shown). However, SK-MEL-2 and SK-MEL-24 cells may have a different molecular basis. Therefore, the differences in invasive ability between the cells cannot exactly represent the role of GPX3. We also found that invasive activities were suppressed to a greater extent in SK-MEL-2 and SK-MEL-24 cells with 5-Aza treatment than without 5-Aza treatment (data not shown). However, no change was found in HEMs according to the presence or absence of 5-Aza treatment. Other tumor suppressor genes in addition to GPX3 have been found to be methylated in melanoma (30). Consequently, 5-Aza treatment can restore the RNA and protein expression of various genes that are methylated in melanoma. Therefore, we constructed GPX3-depleted cell lines to investigate the influence of GPX3 on the biological behavior of melanoma cells.

The proliferative, migration, and invasive ability of cells were comparatively investigated in the GPX3^Δ- and Mock-SK-MEL-24cells. Compared to Mock-SK-MEL-24 cells, GPX3^Δ-SK-MEL-24 cells demonstrated 1.43- and 2.07-fold higher proliferative ability in each time point, respectively (P<0.001, and P<0.001, respectively). Moreover, GPX3^Δ-SK-MEL-24 cells showed 3.06-fold (P<0.001), and 6.47-fold (P<0.001) higher migration and invasive ability, respectively (Fig. 3A–D).

GPX3 expression was detected in the cytoplasm of cancer cells in melanoma tissue samples (Fig. 4A). The expression frequency of GPX3 was significantly increased in primary sites (33.8%) compared to metastatic sites (10.8%) of melanoma tissues (P=0.036) (Fig. 4B). Kaplan-Meier analysis was performed in 46 primary malignant melanoma patients. We found that patients with GPX3 expression showed increased survival rates (median survival duration 71 months) compared to patients without GPX3 expression (median survival duration 18 months) (P=0.013) (Fig. 4C). No significant association was found between GPX3 expression and age or gender of melanoma patients.