Total RNAs were extracted from 4 cervical cancer cell lines, 6 normal cervical tissues and 10 cervical cancer tissues using TRIzol (Invitrogen) and RNeasy Mini kit (Qiagen GmbH, Hilden, Germany). One microgram each of the quantitated total RNA was converted into cDNAs using Transcriptor High Fidelity cDNA Synthesis kit (Roche Applied Science, Netherlands) following the manufacturer’s protocol. Quantitative real-time PCR was performed using 1X SYBR-Green Master Mix (Applied Biosystems), 10 pmol of each primer and 2 μl of the cDNA in an Mx3005P QPCR System (Agilent Technologies, Santa Clara, CA, USA) 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. GPX3 mRNA expression was also detected by conventional RT-PCR using Glod Taq Polymerase (Applied Biosystems) with an annealing temperature of 58°C. The GPX3 expression was normalized by the expression of β-actin. Oligonucleotide primers used for the PCR were: 5′-CAACCAATTTGGAAAACAGG-3′ and 5′-GTGGGAGGACAGGAGTTCTT-3′ for GPX3; and 5′-ATA GCACAGCCTGGATAGCAACGTAC-3′ and 5′-CACCTTCT ACAATGAGCTGCGTGTG-3′ for β-actin.

Cervical cancer cell lines (HeLa, SiHa, CaSki and ME-180) were purchased from the Korean Cell Line Bank (Seoul, Korea). Cells were maintained in minimum essential medium (Sigma-Aldrich, St. Louis, MO, USA) supplemented with fetal bovine serum to a final concentration of 10%, streptomycin (final concentration, 100 μg/ml) and penicillin (final concentration, 50 U/ml) at 37°C in a humidified incubator containing 5% CO₂. For the demethylation experiment, 5×10⁵ cells were grown for 24 h prior to 5-Aza-dC (Sigma-Aldrich) treatment. 5-Aza-dC in PBS was filtered and treated to the final concentration of 1 μM for 3 days, with fresh 5-Aza-dC changes at every 24 h. PBS-treated cells were grown alongside as untreated controls.

Genomic DNA was extracted from 4 cervical cancer cell lines (HeLa, SiHa, CaSki and ME-180) using QIAamp DNA mini kit (Qiagen GmbH, Hilden, Germany). MSP was based on a previous report (17). Genomic DNA (500 ng) was used for bisulfite conversion overnight following a protocol included in the EZ DNA Methylation kit (Zymo Research Corporation, Orange, CA, USA). Oligonucleotide primers used for amplifying the CpG islands for GPX3 promoter regions were: 5′-TATGTTATTGTCGTTTCGGGAC-3′ and 5′-GTCCGT CTAAAATATCCGACG-3′ for methylation-specific amplification; and 5′-TTTATGTTATTGTTTTGGGATG-3′ and 5′-ATCCATCTAAAATATCCAACACTCC-3′ for non-methylation-specific amplification. Amplification was performed in a cycle of 95°C for 10 min, 40 cycles of 94°C for 30 sec, 59°C for 30 sec and 72°C for 30 sec, followed by the final incubation at 72°C for 10 min. The amplified DNA products were resolved with electrophoresis on 1% agarose gels.

Gene expression data related to cervical cancer were downloaded from either Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo) or Array Express (www.ebi.ac.uk/arrayexpress). Gene expression profiling data containing cervical cancer cell lines, normal cervices and cervical cancer tissues (GSE 9750) and non-malignant cervical tissues and various FIGO stages of carcinomas (GSE 46857) were used to test the differences in GPX3 mRNA expression levels. A methylation profiling experiment performed on Illumina Human Methylation 27 BeadChip (GSE 30760) was used to probe methylation level of CpG islands of GPX3 in normal cervices and cervical tumor tissues. Microarray analyses were performed using BRB-Array Tools developed by Dr Richard Simon and BRB-Array Tools Development Team or using GeneSpring 12.6.

The present study included 50 cervical cancer patients who were diagnosed with cervical cancer in Yonsei University Hospital, Korea, between 2000 and 2004. Clinicopathological characteristics of the patients are shown in Table I. The present study was approved by the Institutional Review Board of Yonsei University Hospital. Formalin-fixed, paraffin-embedded surgical specimens were cut into 4 μM tissue sections and deparaffinized with xylene. After hydrating with graded ethanol, endogenous peroxidase activity was blocked with a mixture of methanol and H₂O₂ at a ratio of 40:1. Antigen retrieval was performed with antigen retrieval buffer (Dako, Carpinteria, CA, USA) and primary antibody incubation was performed at RT for 1 h (mouse monoclonal anti-human GPX3 antibody, working dilution; 1:100) (Abcam, Cambridge, MA, USA). Real™ EnVision™ HRP rabbit/mouse detection system (Dako) was used as secondary antibody. The sections were developed with 3,3′-diaminobenzidine (DAB) chromogen and counterstaining was carried out with hematoxylin. For analysis, patients were divided into two groups based on the expression level of GPX3; negative (no or <5% positive cells) and positive (>5% positive cells) group. The survival estimates were based on Kaplan-Meier survival curves along with 95% CI and the log-rank test was used to compare survival curves. A chi-square test was used to analyze the relationship between GPX3 protein expression and various clinicopathological parameters. Fisher’s exact test was used to determine the difference between matched normal and cancer tissues in GPX3 expression.

The mRNA expression of GPX3 was comparatively investigated in 6 normal cervix and 10 cervical cancer tissues. We found that GPX3 mRNA expression was significantly decreased in cervical cancer tissues compared to normal cervical tissues (Fig. 1A). We also examined the GPX3 expression profiles in other published genomic data. In a set of public microarray data containing normal cervices, cervical cancer tissues and cell lines (18), GPX3 showed significant downregulation in cervical cancer tissues and cell lines compared to normal cervical tissues (Fig. 1B). In other independent microarray data, GPX3 expression was again significantly repressed in cervical cancer compared to normal cervical tissues (Fig. 1C). However, there were no differences in expression levels of GPX3 among cancer tissues obtained from patients with different FIGO stages (Fig. 1D). Collectively, these results suggest that GPX3 is significantly downregulated in cervical cancer compared to its normal counterpart.

DNA methylation is one of the most prominent epigenetic mechanisms that control gene expressions and, therefore, disease progression (19–21). Previous results showed GPX3 downregulation in gastric cancer, Barrett’s adenocarcinomas and prostate cancer and their concurrent promoter hypermethylation (1,3,5–7). We searched the epigenome analysis database of normal and cancer tissues from the uterine cervix to examine the differential methylation degrees of GPX3. In a methylation profiling study comparing 152 normal uterine cervical tissues and 63 cervical cancer tissues (GSE 30760), GPX3 showed β values close to zero (no methylation in known CpG islands) in almost all normal tissues, whereas there was a significant trend of increase in β values in tumor samples, indicating an elevated degree of methylation in CpG islands of GPX3 (Fig. 2A). This is consistent with previous reports that GPX3 repression in cancer is due to hypermethylation or deletion. This finding shows that there is a higher degree of CpG methylation in GPX3 promoter region in cervical cancer tissues compared to normal cervical tissues.

To directly test whether GPX3 repression in cervical cancer is correlated with promoter methylation, we compared the GPX3 mRNA expression levels in cervical cancer cells with or without 5-Aza-dC treatment, a potent demethylating agent (22). In HeLa and SiHa, significant upregulation of GPX3 mRNAs, following a 5-Aza-dC treatment, were observed, whereas in CaSki and ME-180, relatively mild but still clear upregulation of GPX3 mRNAs were evident in the same conditions (Fig. 2B). To directly test the promoter methylation, bisulfite-converted genomic DNAs from the same cervical cancer cells were subjected to MSP. In HeLa and SiHa, only methylation-specific amplified DNA fragments were visible, indicating full methylation of CpG islands, whereas in CaSki and ME-180, both of the methylation-specific and unmethylation-specific amplified DNA fragments were visible, indicating partial methylation of CpG islands in these two cell lines. The degree and pattern of CpG methylation in different cervical cancer cell lines are consistent with those of upregulation of GPX3 following a 5-Aza-dC treatment. Therefore, GPX3 is downregulated in cervical cancer due to CpG hypermethylation, although methylation may not be the only mechanism suppressing its transcription.

GPX3 was mainly expressed in cytoplasm of the cells, and its expression was significantly increased in normal cervix (6/6, 100%) compared to cervical cancer tissues (22/50, 44.0%, p=0.023; Fig. 3A). Furthermore, lymph node metastases were observed more often in patients without GPX3 expression (73.7%) than in patients with GPX3 expression (26.3%, p=0.049; Fig. 3B). Patients without GPX3 expression also showed poorer prognosis than patients with GPX3 expression (p=0.033, median survival duration was 117 months for low GPX3 vs. 160.5 months for high GPX3; Fig. 3C).