Silencing of Glutathione Peroxidase 3 through DNA Hypermethylation Is Associated with Lymph Node Metastasis in Gastric Carcinomas

Abstract

Gastric cancer remains the second leading cause of cancer-related death in the world. H. pylori infection, a major risk factor for gastric cancer, generates high levels of reactive oxygen species (ROS). Glutathione peroxidase 3 (GPX3), a plasma GPX member and a major scavenger of ROS, catalyzes the reduction of hydrogen peroxide and lipid peroxides by reduced glutathione. To study the expression and gene regulation of GPX3, we examined GPX3 gene expression in 9 gastric cancer cell lines, 108 primary gastric cancer samples and 45 normal gastric mucosa adjacent to cancers using quantitative real-time RT-PCR. Downregulation or silencing of GPX3 was detected in 8 of 9 cancer cell lines, 83% (90/108) gastric cancers samples, as compared to non-tumor adjacent normal gastric samples (P

Full text

Silencing of Glutathione Peroxidase 3 through DNA
Hypermethylation Is Associated with Lymph Node
Metastasis in Gastric Carcinomas
Dun-Fa Peng
1
, Tian-Ling Hu
1
, Barbara G. Schneider
2
, Zheng Chen
1,5
, Ze-Kuan Xu
5,6
, Wael El-Rifai
1,3,4
*
1Department of Surgery, Vanderbilt University Medical Center, Nashville, Tennessee, United States of America, 2Department of Medicine, Vanderbilt University Medical
Center, Nashville, Tennessee, United States of America, 3Department of Cancer Biology, Vanderbilt University Medical Center, Nashville, Tennessee, United States of
America, 4Department of Veterans Affairs Tennessee Valley Healthcare System, Nashville, Tennessee, United States of America, 5Department of General Surgery, the First
Affiliated Hospital of Nanjing Medical University, Nanjing, China, 6Institute of Tumor Biology, Jiangsu Province Academy of Clinical Medicine, Nanjing, China
Abstract
Gastric cancer remains the second leading cause of cancer-related death in the world. H. pylori infection, a major risk factor
for gastric cancer, generates high levels of reactive oxygen species (ROS). Glutathione peroxidase 3 (GPX3), a plasma GPX
member and a major scavenger of ROS, catalyzes the reduction of hydrogen peroxide and lipid peroxides by reduced
glutathione. To study the expression and gene regulation of GPX3, we examined GPX3 gene expression in 9 gastric cancer
cell lines, 108 primary gastric cancer samples and 45 normal gastric mucosa adjacent to cancers using quantitative real-time
RT-PCR. Downregulation or silencing of GPX3 was detected in 8 of 9 cancer cell lines, 83% (90/108) gastric cancers samples,
as compared to non-tumor adjacent normal gastric samples (P,0.0001). Examination of GPX3 promoter demonstrated DNA
hypermethylation ($10% methylation level determined by Bisulfite Pyrosequencing) in 6 of 9 cancer cell lines and 60% of
gastric cancer samples (P= 0.007). We also detected a significant loss of DNA copy number of GPX3 in gastric cancers
(P,0.001). Treatment of SNU1 and MKN28 cells with 5-Aza-29Deoxycytidine restored the GPX3 gene expression with a
significant demethylation of GPX3 promoter. The downregulation of GPX3 expression and GPX3 promoter hypermethyla-
tion were significantly associated with gastric cancer lymph node metastasis (P= 0.018 and P= 0.029, respectively). We also
observed downregulation, DNA copy number losses, and promoter hypermethylation of GPX3 in approximately one-third of
tumor-adjacent normal gastric tissue samples, suggesting the presence of a field defect in areas near tumor samples.
Reconstitution of GPX3 in AGS cells reduced the capacity of cell migration, as measured by scratch wound healing assay.
Taken together, the dysfunction of GPX3 in gastric cancer is mediated by genetic and epigenetic alterations, suggesting
impairment of mechanisms that regulate ROS and its possible involvement in gastric tumorigenesis and metastasis.
Citation: Peng D-F, Hu T-L, Schneider BG, Chen Z, Xu Z-K, et al. (2012) Silencing of Glutathione Peroxidase 3 through DNA Hypermethylation Is Associated with
Lymph Node Metastasis in Gastric Carcinomas. PLoS ONE 7(10): e46214. doi:10.1371/journal.pone.0046214
Editor: Regine Schneider-Stock, Institute of Pathology, Germany
Received April 16, 2012; Accepted August 29, 2012; Published October 10, 2012
This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for
any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Funding: This work was funded by National Institutes of Health R01CA106176 and R01CA93999. The funders had no role in study design, data collection and
analysis, decision to publish, or preparation of the manuscript.
Competing Interests: Co-author Wael El-Rifai is a PLOS ONE Editorial Board member. This does not alter the authors’ adherence to all the PLOS ONE policies on
sharing data and materials.
* E-mail: wael.el-rifai@Vanderbilt.edu
Introduction
Gastric cancer (GC) is the fourth most common cancer in the
world [1,2] with about 900,000 new cases diagnosed in the world
each year [2]. Gastric cancer remains the second leading cause of
cancer-related deaths worldwide [2]. Although there has been a
decline in the overall incidence of distal gastric cancer during the
past decades, the incidence is rising for adenocarcinomas of the
proximal part of the stomach in the Western world [2]. It is
generally agreed that gastric cancer is a multifactor disease in
which Helicobacter pylori (H. pylori) infection plays a crucial role,
especially for distal gastric cancer [3–5]. Accumulating data
indicate that H. pylori infection generates high levels of reactive
oxygen species (ROS) from multiple sources [6]. Activated
neutrophils, for example, are the main source of ROS production
in the H. pylori-infected stomach; however, H. pylori itself can also
produce ROS [7]. In addition, extensive recent studies have
revealed that H. pylori-induced ROS production in gastric
epithelial cells might affect gastric epithelial cell signal transduc-
tion, resulting in gastric carcinogenesis [8–10]. Excessive ROS
production in the stomach promotes DNA damage in gastric
epithelial cells, suggesting its involvement in gastric carcinogenesis
[7,11,12].
Normal cells have intact antioxidative properties that protect
cells from ROS-induced DNA damage and cell injury [13–15].
Among these systems, the glutathione peroxidase family (GPXs) is
a major antioxidative enzyme family that catalyzes the reduction
of hydrogen peroxide, organic hydroperoxide, and lipid peroxides
by reduced glutathione [15–19]. Glutathione peroxidase 3
(GPX3), also named plasma glutathione peroxidase, is the only
known selenocysteine containing an extracellular antioxidant
isoform [20]. The human GPX3 gene is approximately 10 kb in
length, spanning 5 exons on chromosome 5q32. [20,21]. It has
been reported that GPX3 catalyzes the reduction of hydrogen
peroxide and lipid peroxides and is a major scavenger of ROS
produced during normal metabolism or after oxidative insult
PLOS ONE | www.plosone.org 1 October 2012 | Volume 7 | Issue 10 | e46214

[22,23]. GPX3 is selectively expressed in normal human tissues,
including the gastrointestinal tract. However, downregulation of
GPX3 has been recently reported in multiple human cancers such
as prostate, esophageal, and bladder cancer [24–26], suggesting its
importance in human tumorigenesis. In the present study, we
examined GPX3 gene expression, gene promoter methylation
status, and copy number in a panel of primary gastric cancers and
correlated it with clinicopathological parameters.
Materials and Methods
Ethics Statement
De-identified human tissue samples were obtained from the
archives of pathology at Vanderbilt University (Nashville, TN,
USA) and from the National Cancer Institute Cooperative Human
Tissue Network (CHTN). The use of specimens was approved by
the Institutional Review Board at Vanderbilt University Medical
Center. All patients provided written consent, and samples were
collected after surgical resection. All tissue samples that were
included in this study were collected from tissues that remained
after the completion of diagnosis and are otherwise discarded.
Tissue samples
All tissue samples were obtained from the pathology archives at
Vanderbilt University (Nashville, TN, USA) and from the
National Cancer Institute Cooperative Human Tissue Network
(CHTN). The use of specimens from the archival tissue repository
was approved by the Institutional Review Board. All tissue samples
included in this study were coded and collected from tissues that
remained after the completion of diagnosis and that were
otherwise discarded. Among 108 patients, 70 male and 38 female,
ages ranging from 38–87 years of age with a median age of 64
years. All tumors were histologically verified. The gastric
adenocarcinomas ranged from well-differentiated (WD) to poorly
differentiated (PD), stages I to IV, with a mix of intestinal- and
diffuse-type tumors. The ‘‘normal’’ samples in this study were the
gastric mucosal epithelial tissues adjacent to the cancers without
neoplastic changes.
Cell lines
Nine gastric cancer cell lines (AGS, MKN28, MKN45,
MKN75, KATO3, SNU1, SNU5, SNU16, and RF1) were
purchased from American Type Culture Collection (Manassas,
VA, http://www.atcc.org) and Riken (Ibaraki, Japan; http://
www.brc.riken.go.jp/lab/cell/english). Cells were maintained in
either DMEM medium or RPMI 1640 medium with a supplement
of 10% fetal bovine serum and antibiotics. All cell lines were
cultured at 37uC with 5% CO
2
.
Quantitative Real-Time Reverse Transcription PCR (qRT-
PCR) Analysis of GPX3
Total RNA was isolated using the RNeasy Mini-kit (Qiagen,
Valencia, CA, USA). Single-stranded cDNA was subsequently
synthesized using the iScript cDNA Synthesis Kit (Bio-Rad,
Hercules, CA, USA). Expression of GPX3 was evaluated for a set
of 153 frozen primary human samples including 108 samples of
gastric carcinoma and 45 samples of normal gastric mucosa
adjacent to cancers. For 24 tumors, matching normal mucosa was
available from the same patients. The GPX3 primers (forward 59-
GCCGGGGACAAGAGAAGT-39and reverse 59-GAGGACG-
TATTTGCCAGCAT-39) were designed using the online soft-
ware, Primer 3 (http://frodo.wi.mit.edu/). The qRT-PCR was
performed using an iCycler (Bio-Rad) with the threshold cycle
number determined by use of iCycler software, version 3.0.
Reactions were performed in triplicate and the threshold numbers
were averaged. Results for the GPX3 gene were normalized to
HPRT1 gene (forward 59-TTGGAAAGGGTGTT
TATTCCTCA-39and reverse 59-TCCAGCAGGTCAGC AAA-
GAA-39), which had minimal variation in all normal and tumor
samples tested, and is therefore considered to be a reliable and
stable reference gene for RT-PCR. Expression was calculated by
use of the formula 2
(Rt–Et)
/2
(Rn–En)
as previously described [27,28].
For all of the primary gastric carcinoma samples, the gene was
considered downregulated, if the relative mRNA expression was
#0.5 [29].
DNA Bisulfite Treatment and Pyrosequencing Analysis
DNA was purified using a DNeasy Tissue Kit (Qiagen). The
bisulfite modification of the DNA from cell lines and tissues was
performed using an EZ DNA Methylation-Gold Kit (Zymo
Research, Orange, CA), according to the manufacturer’s protocol.
The GPX3 promoter CpG island was identified by using a CpG
island online search tool (http://www.uscnorris.com), as previous-
ly described [28]. The Pyrosequencing primers were designed
using PSQ Assay Design Software (Biotage, Uppsala, Sweden).
The forward primer sequence was AGGTGGGGAGTT-
GAGGGTAA, the reverse biotin-labeled primer sequence was
Biotin-TCCCAACCACCTTTCAAAC, and the sequencing
primer was GGGAGTTGAGGGTAAGT. A 40 ng aliquot of
modified DNA was subjected to polymerase chain reaction (PCR)
amplification of the specific promoter region using the above
primers and the Platinum PCR SuperMix High Fidelity Enzyme
Mix (Invitrogen, Carlsbad, CA). The PCR products were checked
by gel electrophoresis to confirm the size of the product and rule
out the formation of primer dimers. The specific PCR products
were then subjected to quantitative Pyrosequencing analysis using
a Biotage PyroMark MD System (Biotage), following the protocol
provided by the manufacturer. The results were analyzed by Pyro
Q-CpG 1.0.9 software (Biotage). Based on the methylation levels
in the normal samples, we used 10% methylation as a cutoff for
the identification of DNA hypermethylation of the GPX3
promoter. Statistical analysis was performed to detect significant
changes in the frequencies of DNA methylation of the CpG sites
between tumor and normal samples.
5-Aza-29Deoxycytidine and Trichostatin-A Treatment
For validation of the role of promoter DNA hypermethylation
in transcriptional regulation of GPX3 in vitro, gastric cancer cell
lines SNU1 and MKN28 were used. SNU1 and MKN28 cells
were maintained in Dulbecco’s modified Eagle’s medium
(DMEM) supplemented with 10% fetal bovine serum (FBS) and
antibiotics (Invitrogen). Cells were seeded at low density for
24 hours and then treated with 5 mM 5-Aza-29deoxycytidine (5-
Aza, Sigma-Aldrich, St. Louis, MO) for 72 hours or 300 nM
Trichostatin-A (TSA, Wako, Osaka, Japan) for 24 hours. Total
RNA and DNA were isolated and purified by RNeasy and
DNeasy Tissue kits (Qiagen), as described above. DNA methyl-
ation levels of the CpG nucleotides of the GPX3 promoter were
determined by Pyrosequencing. The GPX3 mRNA expression
levels were determined by qRT-PCR, as described above.
Immunofluorescence staining of GPX3 protein in SNU1
cells
To check GPX3 protein expression after 5-Aza treatment, we
performed immunofluorescence staining against GPX3 in SNU1.
SNU1 cells treated with 5 mM 5-Aza (Sigma-Aldrich) for 72 hours
and/or 300 nM Trichostatin-A (TSA, Wako) for 24 hours, fixed
GPX3 Silencing and Metastasis
PLOS ONE | www.plosone.org 2 October 2012 | Volume 7 | Issue 10 | e46214

with fresh 4% paraformaldehyde for 45 min at room temperature,
followed by permeabilization with 0.1% Triton X-100 in 0.1%
sodium citrate for 2 min on ice. Cells were then incubated with
10% normal goat serum (Invitrogen) for 20 min at room
temperature. Cells were incubated with primary antibody against
GPX3 (Rabbit, 1:500; Abnova, Taiwan) overnight at 4uC followed
by secondary goat anti-rabbit antibody conjugated with Alexa
Fluor 488 (1:1000, Invitrogen) at room temperature for 45 min.
The slides were mounted using Vectashield with DAPI (Vector
Laboratories, Burlingame, California, USA) and viewed under a
fluorescence microscope.
Measurement of DNA Copy Number by Quantitative PCR
(qPCR)
For evaluation of relative DNA copy numbers, we performed
the qPCR amplifications using an iCycler (Bio-Rad). PCR
reactions were prepared in a total volume of 20 ml containing
template DNA (40 ng), with the threshold cycle number deter-
mined by use of iCycler software version 3.0. Primers were
designed using the online software Primer 3 (http://frodo.wi.mit.
edu/). The forward and reverse primers for GPX3 genomic DNA
were 59-CCCCTTCAGTAGGGCCTAAG-39and 59-
TTCTTCAGGACCAGGACCAC-39, respectively. The primers
were obtained from Integrated DNA Technologies (Coralville,
Iowa). Reactions were performed in triplicate, and the threshold
numbers (CT) were averaged. The results were normalized to the
average CT of both b-Actin and GAPDH, which generated
similar results and had minimal variation in all normal and tumor
samples tested. DNA copy number was calculated in the same way
as the quantification of qRT-PCR for mRNA expression [29] and
normalized to the average level of 10 blood DNA samples of
normal individuals in which the GPX3 copy number are not
affected ( 2 copies, equal to copy number ratio 1.0). Loss of DNA
copy number was considered at a relative cutoff ratio #0.5.
Construction of GPX3 expression adenoviral system
The full length of GPX3 coding sequence with Flag-tag was
amplified from normal cDNA by PCR using Platinum PCR
SuperMix High Fidelity (Invitrogen) and was cloned into the
pACCMV.pLpA plasmid. The pACCMV.pLpA-GPX3 plasmid
was then co-transfected with pJM17 plasmid into 293 AD cells to
generate and propagate the recombinant GPX3-expressing
adenoviral particles as previously described [30]. The viruses
were plaque purified, and the titer of the virus was determined
using the Adeno-X qPCR Titration Kit (Clontech, California,
USA) following the manufacturer’s instructions.
Colony formation assay
AGS cells were infected with 25 MOI control or GPX3-
expressing adenoviral particles. 48 hours after infection, cells were
split and seeded (500 cells/well) in 6-well plates. Cells were
cultured at 37uC for another 2 weeks. Cells then were stained with
0.05% crystal violet. The images of the plates were analyzed using
Image J software. Each experiment was set in triplicate and
statistical analysis was done using Prism software. The relative
number of colonies in GPX3-expressing cells was adjusted to
control cells.
Colony formation in soft agar
To check if GPX3 was also involved in anchorage-independent
growth, soft agar colony formation assay was performed. In brief,
AGS and MKN28 cells were infected with 25 MOI control or
GPX3-expressing adenoviral particles. 48 hours after infection,
cells were split and seeded (1.0610
5
cells/well) in a 0.35% noble
agar (Sigma) mixed with culture medium (on the top of 0.5%
noble agar with medium) in 6-well plates. Cells were cultured at
37uC for another 2 weeks. The images of the plates were captured
under a microscope and analyzed using Image J software. Each
experiment was set in triplicate and statistical analysis was done
using Prism software. The relative number of colonies in GPX3-
expressing cells was adjusted to control cells.
Scratch wound healing assay
To check if GPX3 has a role in cell migration, we performed a
scratch wound healing assay in AGS cells [31]. AGS cells were
infected with 25 MOI GPX3-expressing or control adenoviruses.
When cells were 100% confluent, a scratch was made across the
plates using a pipette’s tip. Cells were cultured in complete
medium and images were taken every 24 hours to monitor the
wound healing process.
Statistical Analysis
GraphPad Prism software version 4.0 (GraphPad Prism
Software, La Jolla, CA) was used for all of the statistical analyses.
The Student ttest was used to compare the DNA methylation level
and mRNA ratios between normal and gastric cancers in matched
samples (paired t-test) and unmatched samples (unpaired t-test). In
addition, we analyzed the association between DNA methylation
and clinicopathological factors. The correlation between the DNA
methylation level and mRNA expression was determined by
Spearman’s rank correlation. The comparison of GPX3 gene
expression or DNA methylation level with clinicopathological
factors was made by Chi-square/Fisher’s exact tests or unpaired t-
test. All Pvalues were based on two-tailed tests and differences
were considered statistically significant when the Pvalue was
#.05.
Results
GPX3 was frequently downregulated or silenced in
gastric carcinomas
The qRT-PCR analysis demonstrated that GPX3 mRNA
expression was frequently downregulated in gastric cancer cell
lines (8/9 cell lines examined, Table 1) and in primary gastric
cancer tissue samples (90/108, 83%) as compared to 45 normal
samples (P,.0001; Figure 1A). Further analysis of 24 paired tumor
and normal samples confirmed the significant downregulation of
GPX3 mRNA expression in tumors as compared with their
corresponding normal samples (Figure 1B). Moreover, about one-
third (36/108, 33%) of primary gastric cancers and 6/9 of the
cancer cell lines (Table 1) showed complete silencing of GPX3
mRNA expression, as indicated by the absence of a detectable
signal. Notably, about 33% (15/45) of the ‘‘normal’’ gastric tissues
adjacent to cancers without neoplastic changes also displayed
downregulation (relative expression, #0.5) of GPX3 mRNA, as
normalized to the value of the average of all normal samples.
Promoter DNA hypermethylation of the GPX3 gene
correlates with downregulation of mRNA expression
The DNA methylation changes, which were determined by
Pyrosequencing technology, were not associated with the patients’
ages. A representative DNA methylation scheme of each CpG sites
in the GPX3 promoter of 4 matched normal and tumors is shown
in Figure 2A. Quantitative analysis of GPX3 promoter DNA
methylation, indicated increased promoter DNA methylation
levels of all tested CpG nucleotides in tumor samples compared
GPX3 Silencing and Metastasis
PLOS ONE | www.plosone.org 3 October 2012 | Volume 7 | Issue 10 | e46214

with normal samples (Figure 2B). The average methylation level
for the GPX3 gene promoter in gastric cancers was significantly
higher than that in normal samples (P= 0.007, Figure 2C).
Hypermethylation of the GPX3 promoter ($10%) was detected in
60% (36/60) of the gastric cancers and 6/9 cancer cell lines
(Table 1), consistent with our results showing GPX3 downregu-
lation. We observed that 39% (9/23) of the normal gastric tissues
adjacent to cancers showed an increased DNA methylation level of
the GPX3 promoter from 10% to 15%. Analysis of DNA
methylation in 22 tumor samples and their matching tumor-
adjacent normal gastric mucosae demonstrated a similar signifi-
cant increase in the level of DNA methylation in tumors compared
with their controls (P= 0.0086, Figure 2D). We next analyzed the
promoter DNA methylation against mRNA expression levels in all
samples. As shown in Figure 3A, samples with hypermethylation
(.10%) had significantly lower levels of GPX3 expression
(P= 0.035) compared with samples with absent or low promoter
methylation levels (#10%). Using the Spearman rank correlation,
we found a significant inverse correlation between promoter
methylation and mRNA expression of GPX3 (coefficient
r=20.32, P= 0.024; Figure 3B). These results suggest that the
hypermethylation of the GPX3 promoter region is one of the
factors involved in the suppression of its mRNA expression in
gastric cancers.
5-Aza-29Deoxycytidine (5-Aza) and Trichostatin-A (TSA)
treatment restored GPX3 expression in silenced gastric
cancer cell lines
As shown in Figure 4, the 5-Aza treatment of the SNU1 (A) and
MKN28 (B) cell lines with its fully methylated GPX3 promoter
(Table 1) restored GPX3 mRNA expression, and this restoration
of gene expression was associated with promoter demethylation.
TSA treatment alone had no effect in restoring the GPX3
expression or in altering the methylation levels. However,
administration of TSA following 5-Aza had a significant additive
effect in restoring gene expression. Furthermore, TSA treatment
following 5-Aza led to additional gene expression with a further
decrease in the methylation level of GPX3. To check if 5-Aza
treatment also restored GPX3 protein level, we performed an
immunofluorescence assay using antibody against GPX3 in SNU1
cells. As shown in Figure 4C, there was a significant increase in the
GPX3 green immunofluorescence signal after 5-Aza and 5-Aza-
TSA treatments as compared to DMSO control, suggesting that 5-
Aza and 5-Aza-TSA can restore the GPX3 protein expression in
these cells.
Figure 1. GPX3 is downregulated in gastric cancers. 108 gastric cancer samples and 45 tumor-adjacent histologically normal gastric mucosa
samples were analyzed by real-time RT-PCR for GPX3 expression, which was significantly downregulated in gastric cancers as compared to adjacent
normal samples (A). A similar result was confirmed in 24 matched normal and tumor samples from the same patients (B).
doi:10.1371/journal.pone.0046214.g001
Table 1. DNA copy number, methylation level and gene
expression of GPX3 in gastric cancer cell lines.
Sample
DNA Copy
number ratio % Methylation
mRNA expression
ratio
Normal 1.0 5 1
AGS 0.5 3 0.1
MKN28 0.6 91 0
MKN45 0.3 92 0
MKN75 0.6 91 0
KATO3 0.2 88 0
SNU1 0.4 94 0
SNU5 0.8 4 0.9
SNU16 0.3 6 0.1
RF1 0.6 24 0
Copy number of the GPX3 gene was determined using qPCR and normalized to
10 normal blood DNA samples as described in the Methods section. The real
copy number of each sample equals to the above copy number ratio times 2.
DNA methylation levels were determined using Pyrosequencing and are shown
as an average level of the 8 CpG sites in the GPX3 promoter region. GPX3 mRNA
expression fold was determined using real-time RT-PCR and normalized to the
average value of normal gastric samples as described in the Methods section.
doi:10.1371/journal.pone.0046214.t001
GPX3 Silencing and Metastasis
PLOS ONE | www.plosone.org 4 October 2012 | Volume 7 | Issue 10 | e46214

Loss of DNA copy numbers cooperates with DNA
hypermethylation for silencing GPX3 mRNA expression in
gastric cancer
Although GPX3 promoter hypermethylation correlated statis-
tically with low gene expression levels, we detected silencing of
GPX3 mRNA expression in 83% (90/108) of gastric cancers,
whereas promoter hypermethylation was seen in only 60% (36/60)
of gastric cancers. These findings prompted us to find out whether
loss of copy numbers could be a contributing factor in silencing
GPX3 expression. Evaluation of relative DNA copy numbers of
the 9 gastric cancer cell lines clearly demonstrated that 5 of the 9
cell lines have lost GPX3 copy number (copy number ratio#0.5)
as compared to the normal cells, which have a copy number of 2
(Table 1). In particular, in AGS and SNU16, loss of DNA copy
number seems to account for GPX3 downregulation. While in the
other 6 cell lines (MKN28, MKN45, MKN75, KATO3, SNU1,
and RF1), both loss of copy number and DNA hypermethylation
are associated with the silencing. The only cell line (SNU5) lacking
both DNA hypermethylation and copy number loss also expresses
GPX3 similar to the level in normal samples. The DNA copy
number of twenty-two gastric cancer samples and 22 tumor-
adjacent ‘‘normal’’ stomach samples were compared with the
average copy number of 10 blood samples from normal
individuals. The GPX3 copy numbers in the 10 blood samples
were used as an accurate reference for calculating the ratio and
normalization in gastric tissues. As shown in Figure 5, there was a
significantly lower level of GPX3 DNA copy numbers in the
gastric cancers and the tumor-adjacent ‘‘normal’’ stomach tissue
samples, as compared to that in normal blood samples (P= 0.001).
The comparison of GPX3 copy number in 22 matched tumor and
Figure 2. GPX3 promoter was hypermethylated in gastric cancers. DNA methylation level of 8 CpG sites in the GPX3 promoter was
quantitated by Pyrosequencing. (A) A schematic profile of GPX3 methylation of the 8 CpG sites was examined in 4 matched normal (N) and tumor (T)
samples. (B) The DNA methylation level of each of the 8 CpG sites in 23 normal samples and 60 tumor samples were measured. * P,0.01. (C) The
average DNA methylation level of the 8 CpG sites in 23 normal samples and 60 tumor samples is shown. (D) The average DNA methylation level of the
8 CpG sites in 22 matched normal and tumor samples from the same patients is indicated.
doi:10.1371/journal.pone.0046214.g002
GPX3 Silencing and Metastasis
PLOS ONE | www.plosone.org 5 October 2012 | Volume 7 | Issue 10 | e46214

Figure 3. GPX3 promoter methylation level inversely correlates with GPX3 gene expression. (A) GPX3 expression level in gastric cancers
with GPX3 hypermethylation ($10%) was significantly lower than that in gastric cancers without hypermethylation (,10%). (B) Spearman analysis of
all normal and tumor samples with GPX3 gene expression and DNA methylation demonstrated a significant inverse correlation(r = 20.32) between
DNA hypermethylation and gene expression.
doi:10.1371/journal.pone.0046214.g003
Figure 4. 5-Aza treatment restored GPX3 gene expression in gastric cancer cell lines with silenced GPX3. SNU1 and MKN28 cancer cells
were treated with 5-Aza and TSA as described in the Methods section. In A (SNU1 cells) and B (MKN28 cells), relative GPX3 expression ratios
normalized to HPRT are shown in the left panel. DNA methylation levels of corresponding samples are shown on the right panel. C shows
immunofluorescence staining of GPX3 protein in SNU1 cells. 5-Aza, 5-Aza-29deoxycytidine. TSA, Trichostatin-A.
doi:10.1371/journal.pone.0046214.g004
GPX3 Silencing and Metastasis
PLOS ONE | www.plosone.org 6 October 2012 | Volume 7 | Issue 10 | e46214

normal samples demonstrated lower levels of DNA copy numbers
in tumors (P= 0.058).
Loss of GPX3 correlated with tumor lymph node
metastasis
Based on the available clinicopathological information, we
found both GPX3 hypermethylation and downregulation of
GPX3 expression significantly correlated with lymph node
metastasis (P= 0.029 and P= 0.018 for DNA methylation and
gene expression, respectively; Figure 6). No significant correlation
was found between either DNA methylation or gene expression,
and other parameters examined, such as patient age, sex, tumor
grade, and invasion depth. GPX3 DNA copy number did not
correlate with lymph node metastasis and other parameters,
possibly due to the small number of samples.
Reconstitution of GPX3 in gastric cancer cells did not
suppress tumor cell growth but inhibited tumor cell
migration
Because GPX3 has been suggested to be a potential tumor
suppressor in prostate cancer [26], we hypothesized that GPX3
may have the similar tumor suppressor function in gastric cancer.
We reconstituted GPX3 gene expression in AGS and MKN28 cell
lines and performed colony formation and soft agar colony
formation assays. Interestingly, we did not observe any significant
difference in the number of colonies between GPX3-expressing
AGS cells and the control cells (Figure 7A). Similar results were
obtained by soft agar assay in AGS cells and MKN28 cells
(Figures 7B&7C). In contrast, scratch wound healing assay data
indicated that GPX3-expressing AGS cells exhibited significantly
slower repair of the scratched wound as compared to control cells,
suggesting that GPX3 expression could impair cell migration
(Figure 8).
Discussion
The glutathione peroxidase family (GPXs) is a major anti-
oxidative enzyme family that catalyzes the reduction of hydrogen
peroxide, organic hydroperoxide, and lipid peroxides [15–19]. In
this process, GPXs convert oxygen superoxide and hydrogen
peroxide, a major ROS which have been reported to induce
oxidative DNA damage [30], to harmless intermediate products,
and therefore play critical roles in protecting cells from DNA
damage. GPX3 is a secreted form of the GPX family that is readily
detectable in plasma and mucosal surfaces, and detoxifies ROS
before it can enter into cells [17]. H. pylori infection is the main risk
factor for gastric tumorigenesis which leads to pro-inflammatory
response and generation of high levels of ROS in stomach with a
significant increase in oxidative DNA damage [7]. ROS-induced
DNA damage leading to disruption of genomic integrity has been
shown to be an important cause of human cancers [14]. The loss
of GPX-mediated activities may be associated with early stages of
inflammation-mediated carcinogenesis [16]. Therefore, our results
showing frequent loss of GPX3 expression in gastric cancer may
underscore the failure in the cellular antioxidant system which is
the first line of defense against detrimental ROS activity.
Figure 5. GPX3 copy number loss in gastric cancers. GPX3 copy
number was determined by qPCR and normalized to both beta-actin
and GAPDH of the same samples. The results were then compared to 10
blood DNA samples from normal individuals in which GPX3 copy
number was assumed to be 2.0 (copy number ratio equals to 1.0 in
Table 1). Loss of GPX3 copy number was detected in both gastric cancer
samples and adjacent normal samples.
doi:10.1371/journal.pone.0046214.g005
Figure 6. Dysfunction of GPX3 correlated with lymph node metastasis. (A) Loss of GPX3 gene expression correlated with lymph node
metastasis. (B) Increase in GPX3 methylation correlated with lymph node metastasis. (C) GPX3 copy number change did not correlate with lymph
node metastasis.
doi:10.1371/journal.pone.0046214.g006
GPX3 Silencing and Metastasis
PLOS ONE | www.plosone.org 7 October 2012 | Volume 7 | Issue 10 | e46214

Downregulation of GPX3 has been reported in human cancers
such as lung, ovarian, bladder, esophageal, and prostate cancer
[24–26,28,32], as well as in gastric cancers [33,34]. Mono-allelic
hypermethylation and inactivation of GPX3 in benign precursor
lesions, metaplasia, and dysplasia of the esophagus has been
reported; while inactivation of both alleles was detected in invasive
carcinoma [25]. Complete inactivation of the GPX3 gene by
genetic loss of one allele and methylation-mediated silencing of the
remaining allele was also reported in prostate cancer [26]. In the
current study, we have demonstrated promoter hypermethylation
and DNA copy number loss as possible mechanisms mediating
silencing of GPX3 in gastric cancers, although other mechanisms
such as gene mutation and miRNA regulation may also be
involved and need to be further studied. The occurrence of both
genetic and epigenetic alterations in GPX3 suggests that silencing
of this gene could be a critical event in the multi-step gastric
tumorigenesis cascade. Of note, we observed mRNA downregu-
lation, promoter hypermethylation and DNA copy number loss of
GPX3 in a number of tumor-adjacent ‘‘normal’’ gastric mucosae
samples. These tumor-adjacent ‘‘normal’’ tissues, although histo-
logically normal, they usually have some degree of inflammation
and could possibly have changes at the molecular level. Therefore,
our data suggest that silencing of GPX3 is possibly an early event
in gastric tumorigenesis. Given the known function of GPX3, its
Figure 7. Reconstitution of GPX3 did not suppress tumor cells growth. (A) Colony formation assay in AGS cells. (B) Soft agar colony
formation assay in AGS cells. (C) Soft agar colony formation assay in MKN28 cells. Right panels in A, B and C demonstrate the quantitative data using
Image J software.
doi:10.1371/journal.pone.0046214.g007
GPX3 Silencing and Metastasis
PLOS ONE | www.plosone.org 8 October 2012 | Volume 7 | Issue 10 | e46214

downregulation is expected to result in impairment in the
antioxidant capacity of cells with possible accumulation of
oxidative DNA damage at the early stage of gastric tumorigenesis.
Further studies on GPX3 methylation, inflammation and H. pylori
infection are needed to fully elucidate the relationship among
them.
A recent study has demonstrated that GPX3 can suppress
prostate cancer growth and metastasis [26]. Unfortunately, we did
not confirm the similar tumor suppression role of GPX3 in two
gastric cancer cell lines (AGS and MKN28), suggesting that GPX3
may have differential functional roles in different organs.
Nonetheless, our scratch wound healing assay data indicated that
GPX3-expressing cells have impairment in their migration
capacity. This is in accordance with our findings showing a
significant correlation of GPX3 downregulation and hypermethy-
lation with lymph node metastasis. Further studies are necessary to
confirm this finding using alternative assays and to elucidate the
underlying mechanisms of how GPX3 potentially regulates cancer
metastasis.
In conclusion, our results suggest that GPX3 gene inactivation
by promoter methylation and/or copy number loss is a frequent
finding in gastric cancer that correlates with increased incidence of
lymph node metastasis. This loss may account for a reduced
antioxidant capacity and accumulation of oxidative DNA damage
observed in gastric tumorigenesis. Further studies to explore the
functions of GPX3 in gastric tumorigenesis are required to address
its potential role in the development and progression of gastric
cancer, in particular, its potential roles in tumor cell migration and
metastasis.
Author Contributions
Conceived and designed the experiments: DFP WER ZKX BS. Performed
the experiments: TLH ZC DP. Analyzed the data: DFP WER.
Contributed reagents/materials/analysis tools: DFP WER. Wrote the
paper: DFP WER BS.
References
1. Parkin DM, Pisani P, Ferlay J (1999) Estimates of the worldwide incidence of 25
major cancers in 1990. International journal of cancer Journal international du
cancer 80: 827–841.
2. Crew KD, Neugut AI (2006) Epidemiology of gastric cancer. World journal of
gastroenterology : WJG 12: 354–362.
3. Pandey R, Misra V, Misra SP, Dwivedi M, Kumar A, et al. (2010) Helicobacter
pylori and gastric cancer. Asian Pacific journal of cancer prevention : APJCP 11:
583–588.
4. Piazuelo MB, Epplein M, Correa P (2010) Gastric cancer: an infectious disease.
Infectious disease clinics of North America 24: 853–869, vii.
5. Pritchard DM, Crabtree JE (2006) Helicobacter pylori and gastric cancer.
Current opinion in gastroenterology 22: 620–625.
6. Augusto AC, Miguel F, Mendonca S, Pedrazzoli J Jr, Gurgueira SA (2007)
Oxidative stress expression status associated to Helicobacter pylori virulence in
gastric diseases. Clinical biochemistry 40: 615–622.
7. Handa O, Naito Y, Yoshikawa T (2010) Helicobacter pylori: a ROS-inducing
bacterial species in the stomach. Inflammation research : official journal of the
European Histamine Research Society [et al] 59: 997–1003.
8. Handa O, Naito Y, Yoshikawa T (2007) CagA protein of Helicobacter pylori: a
hijacker of gastric epithelial cell signaling. Biochemical pharmacology 73: 1697–
1702.
9. Smoot DT, Elliott TB, Verspaget HW, Jones D, Allen CR, et al. (2000)
Influence of Helicobacter pylori on reactive oxygen-induced gastric epithelial
cell injury. Carcinogenesis 21: 2091–2095.
10. Lamb A, Yang XD, Tsang YH, Li JD, Higashi H, et al. (2009) Helicobacter
pylori CagA activates NF-kappaB by targeting TAK1 for TRAF6-mediated Lys
63 ubiquitination. EMBO reports 10: 1242–1249.
11. Bagchi D, McGinn TR, Ye X, Bagchi M, Krohn RL, et al. (2002) Helicobacter
pylori-induced oxidative stress and DNA damage in a primary culture of human
gastric mucosal cells. Digestive diseases and sciences 47: 1405–1412.
12. Papa A, Danese S, Sgambato A, Ardito R, Zannoni G, et al. (2002) Role of
Helicobacter pylori CagA+infection in determining oxidative DNA damage in
gastric mucosa. Scandinavian journal of gastroenterology 37: 409–413.
13. Halliwell B (1999) Antioxidant defence mechanisms: from the beginning to the
end (of the beginning). Free radical research 31: 261–272.
14. Strange RC, Spiteri MA, Ramachandran S, Fryer AA (2001) Glutathione-S-
transferase family of enzymes. Mutation research 482: 21–26.
15. Hayes JD, McLellan LI (1999) Glutathione and glutath ione-dependent enzymes
represent a co-ordinately regulated defence against oxidative stress. Free radical
research 31: 273–300.
16. Brigelius-Flohe R, Kipp A (2009) Glutathione peroxidases in different stages of
carcinogenesis. Biochimica et biophysica acta 1790: 1555–1568.
17. Brigelius-Flohe R (2006) Glutathione peroxidases and redox-regulated tran-
scription factors. Biological chemistry 387: 1329–1335.
18. Wingler K, Brigelius-Flohe R (1999) Gastrointestinal glutathione peroxidase.
BioFactors 10: 245–249.
19. Herbette S, Roeckel-Drevet P, Drevet JR (2007) Seleno-independent glutathione
peroxidases. More than simple antioxidant scavengers. The FEBS journal 274:
2163–2180.
20. Yoshimura S, Suemizu H, Taniguchi Y, Watanabe K, Nomoto Y, et al. (1991)
Molecular cloning of human plasma glutathione peroxidase gene and its
expression in the kidney. Nucleic acids symposium series: 163–164.
21. Yoshimura S, Suemizu H, Taniguchi Y, Arimori K, Kawabe N, et al. (1994)
The human plasma glutathione peroxidase-encoding gene: organization,
sequence and localization to chromosome 5q32. Gene 145: 293–297.
22. Takebe G, Yarimizu J, Saito Y, Hayashi T, Nakamura H, et al. (2002) A
comparative study on the hydroperoxide and thiol specificity of the glutathione
peroxidase family and selenoprotein P. The Journal of biological chemistry 277:
41254–41258.
23. Comhair SA, Erzurum SC (2005) The regulation and role of extracellular
glutathione peroxidase. Antioxidants & redox signaling 7: 72–79.
24. Chen B, Rao X, House MG, Nephew KP, Cullen KJ, et al. (2011) GPx3
promoter hypermethylation is a frequent event in human cancer and is
associated with tumorigenesis and chemotherapy response. Cancer letters 309:
37–45.
25. Lee OJ, Schneider-Stock R, McChesney PA, Kuester D, Roessner A, et al.
(2005) Hypermethylation and loss of expression of glutathione peroxidase-3 in
Barrett’s tumorigenesis. Neoplasia 7: 854–861.
26. Yu YP, Yu G, Tseng G, Cieply K, Nelson J, et al. (2007) Glutathi one peroxidase
3, deleted or methylated in prostate cancer, suppresses prostate cancer growth
and metastasis. Cancer research 67: 8043–8050.
27. El-Rifai W, Moskaluk CA, Abdrabbo MK, Harper J, Yoshida C, et al. (2002)
Gastric cancers overexpress S100A calcium-binding proteins. Cancer research
62: 6823–6826.
28. Peng DF, Razvi M, Chen H, Washington K, Roessner A, et al. (2009) DNA
hypermethylation regulates the expression of members of the Mu-class
Figure 8. Reconstitution of GPX3 expression suppressed tumor
cell migration. AGS cells were infected with Ad-GPX3 and Ad-CTRL.
When cells were confluent (48 hours after viral infection), a scratch
wound healing assay was performed. The scratch gap was periodically
monitored and recorded.
doi:10.1371/journal.pone.0046214.g008
GPX3 Silencing and Metastasis
PLOS ONE | www.plosone.org 9 October 2012 | Volume 7 | Issue 10 | e46214

glutathione S-transferases and glutathione peroxidases in Barrett’s adenocarci-
noma. Gut 58: 5–15.
29. Soutto M, Peng D, Razvi M, Ruemmele P, Hartmann A, et al. (2010) Epigenetic
and genetic silencing of CHFR in esophageal adenocarcinomas. Cancer 116:
4033–4042.
30. Peng D, Belkhiri A, Hu T, Chaturvedi R, Asim M, et al. (2011) Glutathione
peroxidase 7 protects against oxidative DNA damage in oesophageal cells. Gut.
31. Liang CC, Park AY, Guan JL (2007) In vitro scratch assay: a convenient and
inexpensive method for analysis of cell migration in vitro. Nature protocols 2:
329–333.
32. He Y, Wang Y, Li P, Zhu S, Wang J, et al. (2011) Identification of GPX3
epigenetically silenced by CpG methylation in human esophageal squamous cell
carcinoma. Digestive diseases and sciences 56: 681–688.
33. Jee CD, Kim MA, Jung EJ, Kim J, Kim WH (2009) Identification of genes
epigenetically silenced by CpG methylation in human gastric carcinoma.
European journal of cancer 45: 1282–1293.
34. Zhang X, Yang JJ, Kim YS, Kim KY, Ahn WS, et al. (2010) An 8-gene
signature, including methylated and down-regulated glutathione peroxidase 3, of
gastric cancer. International journal of oncology 36: 405–414.
GPX3 Silencing and Metastasis
PLOS ONE | www.plosone.org 10 October 2012 | Volume 7 | Issue 10 | e46214