Download Identification of Genes with Differential Expression in Acquired Drug

272 Vol. 10, 272–284, January 1, 2004
Clinical Cancer Research
Identification of Genes with Differential Expression in Acquired
Drug-Resistant Gastric Cancer Cells Using High-Density
Oligonucleotide Microarrays
Hio Chung Kang,1 Il-Jin Kim,1 Jae-Hyun Park,1
Yong Shin,1 Ja-Lok Ku,1 Mi Sun Jung,1
Byong Chul Yoo,2 Hark Kyun Kim,2 and
Jae-Gahb Park1,2
1
Laboratory of Cell Biology, Cancer Research Institute and Cancer
Research Center, Seoul National University, Seoul, Korea, and
2
Research Institute and Hospital, National Cancer Center, Goyang,
Gyeonggi, Korea
ABSTRACT
Purpose: A major obstacle in chemotherapy is treatment failure due to anticancer drug resistance. The emergence of acquired resistance results from host factors and
genetic or epigenetic changes in the cancer cells. The purpose of this study was to identify differentially expressed
genes associated with acquisition of resistance in human
gastric cancer cells.
Experimental Design: We performed global gene expression analysis in the acquired drug-resistant gastric cancer cell lines to the commonly used drugs 5-fluorouracil,
doxorubicin, and cisplatin using Affymetrix HG-U133A microarray. The gene expression patterns of 10 chemoresistant
gastric cancer cell lines were compared with those of four
parent cell lines using fold-change and Wilcoxon’s test for
data analysis.
Results: We identified over 250 genes differentially expressed in 5-fluorouracil-, cisplatin-, or doxorubicin-resistant gastric cancer cell lines. Our expression analysis also
identified eight multidrug resistance candidate genes that
were associated with resistance to two or more of the tested
chemotherapeutic agents. Among these, midkine (MDK), a
heparin-binding growth factor, was overexpressed in all
drug-resistant cell lines, strongly suggesting that MDK might
contribute to multidrug resistance in gastric cancer cells.
Conclusions: Our investigation provides comprehensive
gene information associated with acquired resistance to an-
Received 7/11/03; revised 10/20/03; accepted 10/21/03.
Grant support: Research Grant 2002 from the National Cancer Center,
Korea, and the BK21 (Brain Korea 21) Project for Medicine, Dentistry,
and Pharmacy.
The costs of publication of this article were defrayed in part by the
payment of page charges. This article must therefore be hereby marked
advertisement in accordance with 18 U.S.C. Section 1734 solely to
indicate this fact.
Notes: Drs. H. C. Kang, I-J. Kim, and J-H. Park contributed equally to
this work.
Requests for reprints: Jae-Gahb Park, National Cancer Center, 809
Madu-dong, Ilsan-gu, Goyang, Gyeonggi, 411-764, Korea. Phone: 8231-920-1501; Fax: 82-31-920-1511; E-mail: [email protected].
ticancer drugs in gastric cancer cells and a basis for additional functional studies.
INTRODUCTION
Gastric cancer is one of the most common cancers worldwide. Although the occurrence rate of gastric cancer has decreased, Asian countries such as Korea, China, and Japan, and
some European and South American countries still have a high
incidence of the disease (1, 2). Many chemotherapeutic agents
have been used to treat gastric cancer patients, but the emergence of drug resistance has prevented successful treatment in
many cases. The two major forms of drug resistance are intrinsic
resistance, in which previously untreated tumor cells are inherently insensitive to the chemotherapeutic agent, and acquired
resistance, in which treated tumor cells become insensitive after
drug exposure (3). To date, many research groups have studied
the various mechanisms of drug resistance, hoping to overcome
this major obstacle in chemotherapy. Researchers have determined that acquired drug resistance is multifactorial, in that it
involves host factors and genetic and epigenetic changes, as well
as numerous molecular events (4). The resistance itself may be
due to decreased drug accumulation, alteration of intracellular
drug distribution, reduced drug-target interaction, increased
detoxification response, cell-cycle deregulation, increased
damaged-DNA repair, and reduced apoptotic response (5).
However, although researchers believe that multiple factors
participate in chemoresistance, most studies have focused on a
limited number of candidate genes. For example, it has been
well known that overexpression of the multidrug resistance gene
(MDR1) is associated with cancer cells that have drug resistance. However, little is known about the genes differentially
expressed in a variety of drug-resistant cancer cells, especially
in gastric cancer (6). It is hoped that the recently developed
techniques for genome-wide expression analysis will provide
additional information, novel candidate genes associated with
cancer drug resistance, and perhaps new therapeutic targets.
Microarray technologies have been widely used for comprehensive gene expression analysis as well as mutation and
single nucleotide polymorphism detection (7–14). In particular,
large-scale microarray analysis of gene expression enables researchers to analyze simultaneous changes in thousands of
genes and identify significant patterns. Because the genomewide expression analysis in a variety of drug-resistant gastric
cancer cells has not yet been performed, we have used the
recently developed Affymetrix HG-U133A high-density oligonucleotide microarray for analysis of the global gene expression.
The drugs 5-fluorouracil (5-FU), doxorubicin, and cisplatin
are widely used in the treatment of various malignancies: 5-FU
is a well-known antimetabolite that acts as a thymidylate synthase inhibitor (3); doxorubicin targets topoisomerase II by
interfering with the catalytic cycle (15); and cisplatin interca-
Clinical Cancer Research 273
lates into DNA, leading to DNA damage in the cancer cells (16).
In this study, we have examined genes that are differentially
expressed in 5-FU-, doxorubicin-, or cisplatin-induced chemoresistant gastric cancer cell lines, as compared with their drugsensitive parent cell lines. We identified genes showing altered
expression in resistant cell lines, as well as several potential
multidrug resistance candidate genes that were associated with
resistance to two or more of the chemotherapeutic agents.
MATERIALS AND METHODS
Cell Lines and Cell Culture. Four 5-FU-resistant gastric
cancer cell lines (SNU-620R-5-FU/1000; SNU-638R-5-FU/
50000; SNU-668R-5-FU/4000; SNU-719R-5-FU/600), 3 doxorubicin-resistant cell lines (SNU-620R-DOX/300; SNU-668RDOX/50; SNU-719R-DOX/100), and 3 cisplatin-resistant cell
lines (SNU-620R-CIS/2000; SNU-638R-CIS/400; SNU-668RCIS/400) were created from four different gastric carcinoma cell
lines (SNU-620; SNU-638; SNU-668; SNU-719) established by
Park et al. (17). All 14 cell lines were cultured in RPMI 1640
supplemented with 10% fetal bovine serum (HyClone, Logan,
UT), 20 mM HEPES, and 100 units/ml penicillin-streptomycin
(Invitrogen, Carlsbad, CA) and the indicated concentrations of
drugs in Table 1 (5-FU; doxorubicin; cisplatin) in a humidified
incubator at 37°C in an atmosphere of 5% CO2 until 80 –90%
confluence was achieved.
Drugs and Selection of Drug-Resistant Cells. 5-FU
was purchased from Choongwae Pharma Corp. (Gyeonggi, Korea), doxorubicin from Dong-A Pharmaceutical Co. Ltd. (Seoul,
Korea), and cisplatin from Ildong Pharmaceutical Co. Ltd.
(Seoul, Korea). For production of the resistant cell lines, the four
parent gastric cancer cells were initially exposed to the various
drugs at concentrations indicated by the respective IC50 values
determined by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay (described below). The drug concentration was increased 2– 4-fold after at least 8 weeks of
continuous drug exposure. Fresh drugs were added by gradually
increasing to the final concentrations shown in Table 1. Stable,
drug-resistant cell lines were selected and cultured in the presence of the final drug concentrations. The establishment period
of the drug-resistant cell lines varied from 11 months to 2 years.
Before microarray experiments, cells were maintained for 1
week without drugs to eliminate acute effects.
MTT Assay. Sensitivities of the drug-resistant and parent cell lines to 5-FU, doxorubicin, and cisplatin were deter-
Table 1
mined by MTT assay as described previously (18). Briefly,
single-cell suspensions were obtained by pipette disaggregation
of the floating cells or by trypsinization of monolayer cultures.
The number of cells plated into 96 wells was determined after
preliminary cell growth studies so that untreated cells were in
exponential growth phase at the time of initial harvest and at the
end of the 4-day incubation. Equal number of cells was inoculated into each well in RPMI 1640 supplemented with 10% fetal
bovine serum. For each drug, 5–10 concentrations were used,
covering a 3–5-log concentration range that was chosen to span
the 50% inhibitory concentration determined by preliminary
assays. After 4 days of culture, MTT (Sigma Chemical Co., St.
Louis, MO) was added to each well and was incubated at 37°C
for an additional 4 h. The medium was aspirated from plates
leaving about 30 ␮l of medium in each well. Care was taken not
to disturb the formazan crystals at the bottom of the wells. One
hundred fifty ␮l of DMSO was added to each well, and the
plates were placed on a shaker for 15 min to solubilize the
formazan crystals. The plates were then read immediately at 540
nm on a scanning multiwell spectrophotometer (ELISA reader;
Biotek Instruments Inc., Burlington, VT). All of the data points
represent the mean value of a minimum of six wells. Table 1
shows the IC50 representing the drug concentration resulting in
50% growth inhibition and the relative resistance calculated
from the ratio of the IC50 of the drug-resistant cell lines versus
that of the parent cell lines.
RNA Preparation and Affymetrix GeneChip Hybridization. Total RNA was extracted using the Trizol reagent
(Life Technologies, Inc. Carlsbad, CA) according to the manufacturer’s instructions. Genes expressed in 10 drug-resistant cell
lines and their parent cell lines were analyzed on a high-density
oligonucleotide microarray (HG-U133A; Affymetrix, Santa
Clara, CA) containing 22,282 transcripts. Target preparation and
microarray processing procedures were performed as described
in the Affymetrix GeneChip Expression Analysis Manual (Affymetrix, Santa Clara, CA). Briefly, the extracted total RNA
was purified with an RNeasy kit (Qiagen, Valencia, CA).
Twenty ␮g of total RNA was used to synthesize double-strand
cDNA with SuperScript II reverse transcriptase (Life Technologies, Inc. Rockville, MD) and a T7-(dT)24 primer (Metabion,
Germany). Then, biotinylated cRNA was synthesized from the
double-stranded cDNA using the RNA Transcript Labeling kit
(Enzo Life Sciences, Farmingdale, NY) and was purified and
fragmented. The fragmented cRNA was hybridized to the oli-
Drug-resistant gastric cancer cell lines, drug concentrations, IC50, and relative resistance used in the study
Drugs
Resistant cell lines
Drug concentrations
(␮g/ml)
5-Fluorouracil (5-FU)
SNU-620R-5-FU/1000
SNU-638R-5-FU/50000
SNU-668R-5-FU/4000
SNU-719R-5-FU/600
SNU-620R-DOX/300
SNU-668R-DOX/50
SNU-719R-DOX/100
SNU-620R-CIS/2000
SNU-638R-CIS/400
SNU-668R-CIS/400
1
50
4
0.6
0.3
0.05
0.1
2
0.4
0.4
Doxorubicin (DOX)
Cisplatin (CIS)
IC50 (␮g/ml)
Parent
Resistant
Relative
resistance
0.03 ⫾ 0.01
0.04 ⫾ 0.05
2.38 ⫾ 0.12
0.40 ⫾ 0.06
0.45 ⫾ 0.19
0.20 ⫾ 0.01
0.04 ⫾ 0.01
4.30 ⫾ 0.30
0.81 ⫾ 0.21
6.52 ⫾ 0.84
5.64 ⫾ 2.57
⬎500
164 ⫾ 36.5
2.59 ⫾ 0.14
3.21 ⫾ 0.95
0.68 ⫾ 0.25
0.26 ⫾ 0.06
25.779 ⫾ 0.939
4.339 ⫾ 1.466
22.384 ⫾ 7.655
188
⬎12,500
69
6.5
7.1
3.4
6.5
6
5.4
3.4
274 Gene Expression Profiling in Drug-Resistant Gastric Cancer Cells
Table 2
Genes differentially expressed in 5-fluorouracil (5-FU)-resistant gastric cancer cells
620R-5-FU/1000 638R-5-FU/50000 668R-5-FU/4000 719R-5-FU/600
Affymetrix
identification
Symbol
Description
Function
Genes overexpressed
202589_at
TYMS
214321_at
NOV
212877_at
201467_s_at
202718_at
202284_s_at
212614_at
218437_s_at
201141_at
203816_at
203409_at
208791_at
209674_at
212957_s_at
205366_s_at
209035_at
208920_at
209627_s_at
200687_s_at
209120_at
205548_s_at
211998_at
Thymidylate synthetase dTMP biosynthesis
Nephroblastoma
Cell growth
overexpressed gene
KNS2
Kinesin 2 (60–70 kD) Cell organization and
biogenesis
NQO1
NAD(P)H
Detoxification response
dehydrogenase,
quinone 1
IGFBP2 Insulin-like growth
Cell growth
factor binding protein
2 (36 kD)
CDKN1A Cyclin-dependent kinase Cell cycle
inhibitor 1A (p21,
Cip1)
Homo sapiens mRNA; Unknown
cDNA
DKFZp586N012
Unknown
LZTFL1 Leucine zipper
transcription factorlike 1
GPNMB Glycoprotein
Regulation of cell
(transmembrane) nmb proliferation
DGUOK Deoxyguanosine kinase Guanosine metabolism
DDB2
Damage-specific DNA DNA repair
binding protein 2
(48 kD)
CLU
Clusterin (complement Lipid metabolism,
apoptosis
lysis inhibitor, SP40,40, sulfated
glycoprotein 2,
testosterone-repressed
prostate message 2,
apolipoprotein J)
CRY1
Cryptochrome 1
G-protein coupled
(photolyase-like)
photoreceptor
Homo sapiens mRNA; Unknown
cDNA
DKFZp434F172
HOXB6
Homeo box B6
Transcription factor
MDK
Midkine (neurite
Cytokine, growth factor
growth-promoting
factor 2)
SRI
Sorcin
Cell motility, cell-cell
signaling
OSBPL3 Oxysterol binding
Lipid metabolism,
protein-like 3
transport
SF3B3
Splicing factor 3b,
mRNA splicing
subunit 3, 130 kD
Homo sapiens cDNA: Unknown
FLJ22189 fis, clone
HRC01043
BTG3
BTG family, member 3 Cell cycle
H3F3B
H3 histone, family 3B DNA binding
(H3.3B)
Genes down-regulated
204351_at
S100P
217867_x_at
205428_s_at
218677_at
204990_s_at
S100 calcium binding
protein P
BACE2
␤-site APP-cleaving
enzyme 2
CALB2
Calbindin 2, (29 kD,
calretinin)
LOC57402 S100-type calcium
binding protein A14
ITGB4
Integrin, ␤ 4
Criteria
F
F
a
E
E
F
F.C.b
p
F.C.
56.1 0.00002 ⫺2.6
10.3 0.000046 ⫺1.6
2.0 0.000046
p
F.C.
p
F.C.
p
0.99998
0.934434
11.2 0.00002
2.3 0.000389
1.9 0.000438
16.0 0.001651
1.9
0.004481
14.7 0.000078
1.0 0.532344
F
E
⫺1.4 0.99998
4.1
0.00225
3.4 0.000027
F
E
1.5 0.00002
9.6
0.00002
4.3 0.000865 ⫺1.1 0.5
F
E
2.6 0.000023
10.0
0.000865
2.3 0.000023
1.3 0.033304
F
E
3.8 0.330589
5.1
0.00002
5.2 0.000114
1.9 0.000273
2.4 0.002753
1.6
0.000101
8.0 0.016731
1.9 0.013078
F
10.6 0.00002
F
E
5.5 0.000865
2.1
0.000101
4.3 0.00004
1.9 0.252851
F
F
E
E
2.5 0.000023
2.0 0.000273
1.0
3.3
0.5
0.00002
5.9 0.000046
3.2 0.000692
2.2 0.000346
1.2 0.284967
F
E
4.1 0.00002
1.7
0.002032
2.1 0.000114
1.7 0.002032
F
E
2.0 0.000027
1.4
0.002032
3.9 0.000966
2.2 0.263341
F
3.6 0.094279
1.8
0.002753
2.2 0.493524
1.9 0.002753
F
F
2.2 0.48058
2.4 0.00002
3.6
3.1
0.00002
0.00002
1.7 0.000692
1.8 0.033304
2.0 0.094279
1.8 0.004073
F
2.0 0.057676
1.8
0.005409
2.7 0.004073
1.9 0.033304
1.9 0.00003
2.9
0.00002
1.5 0.000618
1.9 0.000865
1.6 0.035785
1.6
0.000068
2.8 0.000966
1.8 0.018128
F
E
F
F
E
1.7 0.000078
2.3
0.00002
1.8 0.008511
1.9 0.001201
F
F
E
E
1.8 0.00002
1.5 0.000023
1.5
1.6
0.001486
0.000241
2.4 0.00002
2.5 0.000023
1.5 0.033304
1.6 0.003355
Calcium binding
F
E
1.5 0.999448
25.5
0.99998
63.1 0.99998
2.6 0.99997
Protein catabolism
F
E
1.7 0.99997
1.8
0.999693
22.0 0.997247
1.5 0.996645
Calcium binding
F
4.0 0.996645
12.6
0.999922 ⫺1.2 0.088938
1.7 0.999833
Unknown
F
E
2.1 0.99998
5.7
0.99998
3.2 0.999508
1.5 0.999948
Cell adhesion
F
E
2.2 0.99998
6.1
0.99998
1.8 0.999853
2.2 0.99998
Clinical Cancer Research 275
Table 2
Continued
620R-5-FU/1000
Affymetrix
identification Symbol
Description
202887_s_at RTP801 HIF-1 responsive
RTP801
204109_s_at NFYA
Nuclear transcription
factor Y, ␣
202847_at
PCK2
Phosphoenolpyruvate
carboxykinase 2
(mitochondrial)
208813_at
GOT1
Glutamic-oxaloacetic
transaminase 1,
soluble
222025_s_at OPLAH 5-Oxoprolinase (ATPhydrolyzing)
201250_s_at SLC2A1 Solute carrier family 2
(facilitated glucose
transporter),
member 1
202068_s_at LDLR
Low density lipoprotein
receptor (familial
hypercholesterolemia)
202267_at
LAMC2 Laminin, gamma 2
[nicein (100 kD),
kalinin (105 kD),
BM600 (100 kD),
Herlitz junctional
epidermolysis
bullosa)]
208613_s_at FLNB
Filamin B, ␤ (actin
binding protein 278)
204394_at
POV1
Prostate cancer
overexpressed gene 1
205455_at
MST1R Macrophage stimulating
1 receptor (c-metrelated tyrosine
kinase)
Function
a
F.C.
E
2.3
1.5
Criteria
b
638R-5-FU/50000
668R-5-FU/4000
719R-5-FU/600
p
F.C.
p
F.C.
p
F.C.
p
0.99998
2.4
0.99998
4.1
0.999973
3.3
0.99998
0.777251
1.7
0.994067
2.4
0.98579
5.5
0.955851
Apoptosis
F
Transcription factor
F
Glucose metabolism
F
E
3.5
0.99998
2.0
0.99997
1.9
0.999727
2.7
0.99998
Amino acid
metabolism
F
E
1.6
0.999886
2.5
0.999932
1.6
0.989865
4.1
0.99998
ATP hydrolysis
F
E
2.9
0.99998
2.6
0.999948
3.0
0.99998
1.2
0.997247
Glucose transport
F
E
1.7
0.998514
2.0
0.999759
1.3
0.964215
4.2
0.99997
Lipid metabolism,
endocytosis
F
E
1.5
0.999654
2.8
0.999911
1.9
0.99998
2.3
0.99994
Cell adhesion
F
E
2.5
0.999508
2.0
0.999948
1.2
0.5
2.8
0.99998
Cytoskeletal anchoring
F
E
2.4
0.998664
4.7
0.999965
⫺1.0
0.242585
2.0
0.999932
Oncogenesis
F
E
1.8
0.999654
1.8
0.999382
3.1
0.977068
1.5
0.999382
Protein modification
F
1.6
0.925732
1.80
0.999226
2.6
0.999308
1.8
0.993497
a
Criteria shows how the genes were selected by data analysis. F indicates that the gene was selected by fold-change. E indicates that the gene
was selected by Wilcoxon’s test. F E indicates that the gene was selected by both fold-change and Wilcoxon’s test.
b
F.C., fold change.
gonucleotide microarray, which was washed and stained with
streptavidin-phycoerythrin. Scanning was performed with an
Agilent Microarray Scanner (Agilent Technologies, Palo
Alto, CA).
Data Analysis. GeneChip analysis was performed based
on the Affymetrix GeneChip Manual (Affymetrix Inc., Santa
Clara, CA) with Microarray Analysis Suite (MAS) 5.0, Data
Mining Tool (DMT) 2.0, and Microarray Database software. All
of the genes represented on the GeneChip were globally normalized and scaled to a signal intensity of 500. Fold changes
were calculated by comparing transcripts between parent and
acquired drug-resistant cell lines. The Microarray Analysis Suite
software used the Wilcoxon’s test to generate detected (present
or absent) and changed (increased or decreased) calls, and used
the calls to statistically determine whether a transcript was
expressed or not, and whether it was relatively increased, decreased, or unchanged. After being filtered through a “present”
call (p ⬍ 0.05), a transcript was considered differentially expressed when it satisfied one of the following two conditions: 1)
by fold change, transcripts increased or decreased ⬎1.5-fold; 2)
by one-sided Wilcoxon’s rank test, transcripts’ average fold
change exceeded 1.5, with an “increased” (p ⬍ 0.003) or
“decreased” (p ⬎ 0.997) call. In the case of 5-FU, all transcripts
meeting the above conditions in at least three of four cell lines
were considered differentially expressed. Hierarchical clustering
and dendrogram figures were generated using Cluster and TreeView software (http://rana.stanford.edu). A three-dimensional
graph of the multidimensional scaling was created using SPSS
(SPSS Inc., Chicago, IL) and SigmaPlot (SPSS Inc.). To evaluate the statistical significance of eight genes differentially
expressed in two or three of the drug subsets, paired t test was
performed using SPSS (SPSS Inc.).
Quantitative and Semiquantitative RT-PCR. We selected two genes for real-time quantitative reverse transcription
(RT)-PCR and seven genes for semiquantitiative RT-PCR for
validation of the microarray data. Five ␮g of total RNA was
used for creation of single-stranded cDNA using the SuperScript
Preamplification System for First Strand cDNA Synthesis (Life
Technologies, Inc., Rockville, MD). The cDNA was diluted and
quantitatively equalized for PCR amplification. For real-time
quantitative RT-PCR of MDK (midkine) and BIRC5, TaqMan
PCR method using a 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA) was performed according to
the manufacturer’s instructions. We used the primers and probes
276 Gene Expression Profiling in Drug-Resistant Gastric Cancer Cells
provided as Assays-on-Demand Gene Expression Products (Applied Biosystems). The expression of seven genes (MDK, FOSB,
NQO1, DDB2, ITGB4, ABCC1, and IGFBP2) was verified by
semiquantitative RT-PCR. The primer sets for PCR amplification were as follows: MDK forward 5⬘-ATGCAGCACCGAGGCTTCCT-3⬘, reverse 5⬘-ATCCAGGCTTGGCGTCTAGT-3⬘;
FOSB forward 5⬘-GAGAGGGGAAGAGACAAAGT-3⬘, reverse 5⬘-CTTCATCCTCACACAGGACT-3⬘; NQO1 forward
5⬘-TGGAGAATATTTGGGATGAG-3⬘, reverse 5⬘-AATCCAGGCTAAGGAATCTC-3⬘; DDB2 forward 5⬘-GGAGATATCATGCTCTGGAA-3⬘, reverse 5⬘-GGCTACTAGCAGACACATCC-3⬘; ITGB4 forward 5⬘-TTCCAAATCACAGAGGAGAC-3⬘, reverse 5⬘-CTTGAGGTTGTCCAGATCAT-3⬘; ABCC1 forward 5⬘-CTGACAAGCTAGACCATGAATGT-3⬘, reverse 5⬘-TCACACCAAGCCGGCGTCTTT-3⬘; and IGFBP2
forward 5⬘- TTCCAGTTCTGACACACGTA-3⬘, reverse 5⬘GACACAGGGGTTCAAAAATA –3⬘.
PCR was carried out with 1 ␮l of cDNA as follows: initial
denaturation at 94°C for 5 min followed by 25–30 cycles of
94°C for 30 s, 55°C for 30 s, 72°C for 1 min, followed by a final
elongation at 72°C for 7 min.
Western Blot Analysis. To investigate the correspondence between mRNA and protein of MDK, we performed Western blot analysis in SNU-620, SNU-620R-5-FU/1000, SNU620R-CIS/2000, SNU-638, and SNU-638R-CIS/400. MDK
protein was detected with an MK antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Western blot analysis was performed as described previously (19).
RESULTS
Gene Selection from Microarray Data Analysis. Because the high-density oligonucleotide microarray contains a
large number of probes, two different statistical analysis methods were used in parallel to select genes that were differentially
expressed in drug-resistant gastric cancer cells. First, we investigated genes that showed altered expression patterns in drugspecific cell line subsets composed of the four 5-FU-, three
doxorubicin-, and three cisplatin-resistant cell lines. In the
5-FU-resistant gastric cancer cell lines, a total of 38 genes were
selected as having significant fold-change and Wilcoxon’s test
results. Twenty-two genes were up-regulated and 16 were
down-regulated in at least three of the four 5-FU-resistant cell
lines. In the doxorubicin-resistant cell lines, we found over 200
differentially expressed genes, many of which were overexpressed. In the cisplatin-resistant cell lines, we identified 27
differentially expressed genes, 19 of which were up-regulated
and 8 of which were down-regulated. After the individual gene
selection in each drug subset, we next screened for genes that
were differentially expressed in more than one of the drug
subsets, i.e. multidrug resistance genes. We identified eight
genes that were differentially expressed in more than one drug
subset; only one of the eight was differentially expressed in all
three subsets.
Validation of Microarray Results. To verify the expression of the genes identified in microarray experiments, either
real-time quantitative RT-PCR or semiquantitative RT-PCR was
performed using the same RNA as that used in the microarray
analysis. We tested two genes (MDK and BIRC5) for real-time
quantitative RT-PCR and seven genes (MDK, FOSB, NQO1,
DDB2, ITGB4, ABCC1, and IGFBP2) for semiquantitative RTPCR, and found that the results were in good agreement with
those from the microarray data, in that the observed differences
were not significant.
Differentially Expressed Genes in 5-FU-Resistant Gastric Cancer Cells. In 5-FU-resistant gastric cancer cell lines,
we identified 38 differentially expressed genes (Table 2), most
of which are involved in cell proliferation, metabolic pathways,
cell growth, and cell organization. In addition, a subset of
differentially expressed genes was associated with signaling,
responses to external stimuli, and cell adhesion. In particular, we
observed up-regulation of cell growth regulators (NOV and
IGFBP2) and genes involved in nucleobase, nucleotide, and
nucleic acid metabolism (HOXB6, SF3B3, TYMS, DGUOK,
and DDB2). Also, the overexpression of TYMS (thymidylate
synthetase) was consistent with the action of 5-FU as a TYMS
inhibitor. The signal transducers sorcin (SRI), MDK, and NQO1
were up-regulated, and the calcium-binding protein SRI was
up-regulated in all of the 5-FU-resistant cell lines. A novel
hypoxia-inducible factor 1 (HIF-1)-responsive gene, RTP801,
recently identified as involved in both pro- and antiapoptotic
activities (20), was down-regulated, as were several calcium
ion-binding molecules (S100P, CALB2, and LDLR).
Differentially Expressed Genes in Doxorubicin-Resistant
Gastric Cancer Cells. Table 3 shows the top 54 up-regulated
genes among ⬎200 genes and 20 down-regulated in doxorubicin-resistant gastric cancer cells. Many of the up-regulated
genes are associated with the cell cycle, including genes involved with regulation of the cell cycle (FOSB, CDC2, CDC20,
CDKN3, and MKI67), control of the mitotic cell cycle (BUB1,
BUB1B, RRM1, and RRM2), DNA replication (TOP2A and
MCM4), and antiapoptosis (BIRC5-survivin). Among the ATPbinding cassette (ABC) transporters, ABCC1 was up-regulated
and ATP2B1 was down-regulated. In our microarray data, the
expression of the MDR1 gene, which encodes a P-glycoprotein,
was not detected. The chloride transporter CLIC4, responders to
external stimulus (TBL1X and MBP), and transcription factor
NFAT5 were all down-regulated in doxorubicin-resistant gastric
cancer cells.
Differentially Expressed Genes in Cisplatin-Resistant
Gastric Cancer Cells. We identified 27 genes that were differentially expressed in cisplatin-resistant gastric cancer cells,
19 of which were up-regulated and 8 of which were downregulated (Table 4). The differentially expressed genes in
cisplatin-resistant cells showed the lowest fold change levels
among three different drug subsets. The up-regulated genes
included cell proliferation regulators (IGFBP6, FTH1, and
GRN), genes associated with cell cycle (CDKN1A), stress responders (HSPA1B), transporters (ATP7A), cell adhesion molecules (PCDHGB7 and JUP), and metabolic factor (SF3B3). In
contrast, the DNA repair and cell cycle checkpoint gene NBS1,
and the ubiquitin-conjugate enzyme and apoptosis suppressor
HIP2 were down-regulated.
Genes Differentially Expressed in Two or Three of the
Drug Subsets. We identified eight genes that were differentially expressed in more than one of the three drug subsets. For
example, MDK (p ⫽ 0.0007) was overexpressed by ⬎1.5-fold
in 9 of 10 drug-resistant gastric cancer cell lines. The remaining
Clinical Cancer Research 277
Table 3
Genes differentially expressed in doxorubicin (DOX)-resistant gastric cancer cells
SNU-620R-DOX/300 SNU-668R-DOX/50 SNU-719R-DOX/100
Affymetrix
identification
Symbol
Genes overexpressed
219918_s_at
FLJ10517
202768_at
202805_s_at
218663_at
213007_at
218355_at
203967_at
202870_s_at
204886_at
219588_s_at
202095_s_at
218741_at
204023_at
219000_s_at
218726_at
212621_at
203276_at
209773_s_at
204441_s_at
209053_s_at
204962_s_at
203214_x_at
209642_at
201467_s_at
213599_at
219510_at
203755_at
202503_s_at
Description
Hypothetical protein
FLJ10517
FOSB
FBJ murine
osteosarcoma viral
oncogene homolog B
ABCC1
ATP-binding cassette,
sub-family C (CFTR/
MRP), member 1
HCAP-G
Chromosome
condensation
protein G
FLJ10719
Hypothetical protein
FLJ10719
KIF4A
Kinesin family
member 4A
CDC6
CDC6 cell division
cycle 6 homolog
(Saccharomyces
cerevisiae)
CDC20
CDC20 cell division
cycle 20 homolog
(S. cerevisiae)
STK18
Serine/threonine
kinase 18
FLJ20311
Hypothetical protein
FLJ20311
BIRC5
Baculoviral IAP repeatcontaining 5
(survivin)
MGC861
Hypothetical protein
MGC861
RFC4
Replication factor C
(activator 1) 4,
37 kDa
MGC5528
Hypothetical protein
MGC5528
DKFZp762E1312 Hypothetical protein
DKFZp762E1312
KIAA0286
KIAA0286 protein
LMNB1
Lamin B1
RRM2
Ribonucleotide reductase
M2 polypeptide
POLA2
Polymerase (DNAdirected), ␣ (70 kD)
WHSC1
Wolf-Hirschhorn
syndrome candidate 1
CENPA
Centromere protein A,
17 kDa
CDC2
Cell division cycle 2, G1
to S and G2 to M
BUB1
BUB1 budding
uninhibited by
benzimidazoles 1
homolog
NQO1
NAD(P)H
dehydrogenase,
quinone 1
OIP5
Opa-interacting protein 5
POLQ
Polymerase (DNA
directed), theta
BUB1B
BUB1 budding
uninhibited by
benzimidazoles 1
homolog beta
KIAA0101
KIAA0101 gene product
Function
Criteria
a
F.C.b
p
F.C.
p
F.C.
p
Unknown
F
E
7.6
0.000114
3.1
0.000046
49.9
0.00002
Regulation of
transcription
F
E
4.3
0.000027
16.9
0.000865
22.6
0.00002
Transport, drug
resistance
F
E
34.5
0.00002
5.7
0.00006
1.7
0.00002
Unknown
F
24.8
0.00006
10.4
0.005933
3.4
0.00002
15.5
0.00002
19.7
0.000052
1.4
0.001651
F
21.0
0.000068
3.4
0.015426
7.9
0.00004
Cell cycle regulator F
5.3
0.000147
17.5
0.000023
8.3
0.024755
5.7
0.000101
20.4
0.000101
3.5
0.000046
4.0
0.000492
19.6
0.011045
3.6
0.002032
E
Unknown
Microtubule motor
Regulation of cell
cycle
F
Regulation of cell
cycle
Unknown
F
F
E
6.8
0.000027
16.4
0.000167
3.0
0.000035
Anti-apoptosis
F
E
4.8
0.00013
8.2
0.00004
12.1
0.00002
Unknown
F
E
16.2
0.000214
6.1
0.001201
2.2
0.001486
DNA replication
F
E
2.4
0.000023
19.3
0.00004
1.9
0.000346
Unknown
F
7.0
0.038415
7.4
0.061522
7.9
0.000101
Unknown
F
4.9
0.000114
7.3
0.004925
10.2
0.001832
Unknown
Structural molecule
DNA synthesis
F
F
F
4.2
13.5
4.6
0.000273
0.001077
0.00002
13.9
5.1
12.9
0.000389
0.069813
0.00002
3.1
2.0
3.1
0.000147
0.001486
0.00002
Unknown
F
11.0
0.00712
2.1
0.5
7.4
0.030967
Oncogenesis
F
7.3
0.00004
10.4
0.000147
2.5
0.00003
Chromosome
organigenesis
Cell cycle
F
12.9
0.002753
3.7
0.041201
3.5
0.000023
F
E
5.5
0.000035
12.0
0.00002
2.3
0.000023
Mitotic checkpoint
F
E
3.7
0.000346
11.5
0.00003
4.4
0.000023
E
1.4
0.002032
7.6
0.00002
9.7
0.00002
Detoxification
response
E
E
E
E
Unknown
DNA repair
F
F
E
3.3
10.7
0.00003
0.213188
9.0
4.3
0.000273
0.583567
5.0
2.1
0.000114
0.010135
Mitotic checkpoint
F
E
7.6
0.00002
4.6
0.000189
3.4
0.000027
Unknown
F
E
6.1
0.00002
6.4
0.00002
2.9
0.00002
278 Gene Expression Profiling in Drug-Resistant Gastric Cancer Cells
Table 3
Continued
SNU-620R-DOX/300
Affymetrix
identification
Symbol
202589_at
TYMS
202580_x_at FOXM1
209714_s_at CDKN3
Description
Thymidylate synthetase
Forkhead box M1
Cyclin-dependent kinase
inhibitor 3
202240_at
PLK
Polo-like kinase
(Drosophila)
221520_s_at FLJ10468 Hypothetical protein
FLJ10468
210052_s_at C20orf1
Chromosome 20 open
reading frame 1
205345_at
BARD1
BRCA1 associated
RING domain 1
204026_s_at ZWINT
ZW10 interactor
203270_at
DTYMK
Deoxythymidylate
kinase (thymidylate
kinase)
218782_s_at PRO2000 PRO2000 protein
208079_s_at STK6
Serine/threonine
kinase 6
212021_s_at MKI67
Antigen identified by
monoclonal antibody
Ki-67
218009_s_at PRC1
Protein regulator of
cytokinesis 1
201292_at
TOP2A
Topoisomerase (DNA) II
␣ 170 kDa
201477_s_at RRM1
Ribonucleotide reductase
M1 polypeptide
222037_at
MCM4
MCM4 minichromosome
maintenance deficient
4 (S. cerevisiae)
202338_at
TK1
Thymidine kinase 1,
soluble
218039_at
ANKT
Nucleolar protein ANKT
203362_s_at MAD2L1 MAD2 mitotic arrest
deficient-like 1 (yeast)
204146_at
PIR51
RAD51-interacting
protein
204538_x_at NPIP
Nuclear pore complex
interacting protein
218115_at
FLJ10604 Hypothetical protein
FLJ10604
206102_at
KIAA0186 KIAA0186 gene product
219148_at
TOPK
T-LAK cell-originated
protein kinase
207165_at
HMMR
Hyaluronan-mediated
motility receptor
(RHAMM)
208808_s_at HMGB2
High-mobility group
box 2
Genes down-regulated
221705_s_at FLJ21168 Hypothetical protein
FLJ21168
203803_at
PCL1
Prenylcysteine lyase
208003_s_at NFAT5
Nuclear factor of
activated T-cells 5,
tonicity-responsive
207755_at
Homo sapiens cDNA
FLJ13892 fis, clone
THYRO1001656,
215936_s_at KIAA1033 KIAA1033 protein
207700_s_at NCOA3
Nuclear receptor
coactivator 3
201867_s_at TBL1X
Transducin ␤-like 1Xlinked
218659_at
FLJ10898 KIAA1685 protein
Function
Criteria
a
SNU-668R-DOX/50
SNU-719R-DOX/100
p
F.C.
p
F.C.
p
7.8
7.6
3.5
0.000052
0.00002
0.00002
6.2
5.2
2.3
0.00002
0.000023
0.000114
1.4
2.5
9.4
0.00004
0.00002
0.00002
F
7.1
0.000307
4.7
0.044149
2.9
0.000068
Unknown
F
3.2
0.000618
8.4
0.003355
3.1
0.000027
Unknown
F
E
6.8
0.00002
2.8
0.00002
5.0
0.00002
Tumor suppressor
F
E
7.2
0.00225
4.5
0.001336
2.7
0.00013
Kinetochore function
DNA metabolism
F
E
E
4.4
3.3
0.000027
0.000618
7.0
9.6
0.000046
0.000078
2.9
1.3
0.000046
0.00249
Cell cycle
Mitotic protein kinase
F
F
E
6.1
6.9
0.00002
0.00002
4.8
3.5
0.00002
0.000147
3.1
2.9
0.005409
0.00002
Cell cycle regulator
F
E
5.7
0.000052
4.3
0.000346
3.2
0.000027
Cytokinesis
F
E
6.8
0.00002
2.4
0.00002
3.7
0.00002
DNA topoisomerase
F
E
6.5
0.00002
2.8
0.00002
3.4
0.00002
DNA synthesis
F
E
3.8
0.000068
7.1
0.000273
1.6
0.000189
DNA replication
F
E
6.4
0.00002
3.9
0.000046
2.1
0.000492
Thymidine kinase
F
E
4.9
0.00006
4.9
0.000035
2.1
0.002032
Unknown
Mitotic checkpoint
F
F
E
E
4.7
6.3
0.00002
0.000027
3.7
2.8
0.000307
0.000273
3.1
2.3
0.00002
0.000114
DNA repair
F
E
6.2
0.00002
2.5
0.000389
2.6
0.000346
Unknown
F
E
6.7
0.00002
2.5
0.00002
2.0
0.00002
Unknown
F
E
3.6
0.000147
3.8
0.000346
2.7
0.000035
Unknown
Protein kinase
F
F
E
E
4.1
2.9
0.00002
0.00003
3.6
3.7
0.00002
0.00002
2.2
3.2
0.00002
0.00013
Cell motility
F
2.9
0.000492
1.9
0.004481
4.8
0.000023
Transcription factor
F
E
3.9
0.00002
2.8
0.000052
2.9
0.00002
Unknown
F
E
11.2
0.999611
2.4
0.999308
3.2
0.998514
Protein catabolism
Transcription factor
F
F
E
E
9.1
5.7
0.999922
0.999911
2.4
5.2
0.99998
0.99998
3.6
3.6
0.999899
0.99998
Unknown
F
4.3
0.991489
5.4
0.999886
2.4
0.949447
Unknown
Transcription coactivator
Signal transduction
F
F
E
3.1
4.2
0.99987
0.999833
3.8
2.7
0.99998
0.999853
3.1
2.2
0.999654
0.945978
F
E
2.7
0.99998
3.4
0.99998
2.9
0.99987
Unknown
F
E
3.1
0.99998
2.6
0.999948
3.2
0.99998
dTMP biosynthesis
Transcription factor
Cell cycle arrest
F
F
Cell cycle
E
E
E
F.C.
b
Clinical Cancer Research 279
Table 3
Continued
SNU-620R-DOX/300
Affymetrix
identification
209750_at
Symbol
NR1D2
Description
Nuclear receptor
subfamily 1, group D,
member 2
222158_s_at LOC51029 CGI-146 protein
212638_s_at WWP1
WW domain-containing
protein 1
208615_s_at PTP4A2
Protein tyrosine
phosphatase type
IVA, member 2
218930_s_at FLJ11273 Hypothetical protein
FLJ11273
202551_s_at CRIM1
Cysteine-rich motor
neuron 1
220892_s_at PSA
Phosphoserine
aminotransferase
202033_s_at RB1CC1
RB1-inducible coiledcoil 1
209281_s_at ATP2B1
ATPase, Ca⫹⫹
transporting, plasma
membrane 1
201559_s_at CLIC4
Chloride intracellular
channel 4
203973_s_at CEBPD
CCAAT/enhancer
binding protein
(C/EBP), ␦
214168_s_at TJP1
Tight junction protein 1
(zona occludens 1)
Function
Criteria
a
b
SNU-668R-DOX/50
SNU-719R-DOX/100
F.C.
p
F.C.
p
F.C.
p
Transcription factor
F
E
3.2
0.99998
3.0
0.999948
2.6
0.999973
Unknown
Protein interaction
F
F
E
3.1
2.6
0.99998
0.99998
3.3
3.8
0.999977
0.99998
2.2
1.8
0.999932
0.995519
Dephosphorylation
F
E
3.3
0.99998
2.7
0.999977
1.7
0.999034
Unknown
F
E
3.3
0.99997
2.2
0.99997
2.0
0.99998
Insulin-like growth
factor receptor
Enzyme
F
E
2.7
0.998923
2.1
0.999977
2.1
0.99997
E
1.3
0.99751
3.8
0.99997
1.5
0.999973
Nuclear architecture
F
1.8
0.999899
2.8
0.99997
1.6
0.952736
Transporter, ATP
binding
F
2.2
0.999693
1.8
0.958799
1.7
0.945978
Chloride channel
F
1.5
0.999308
1.9
0.99775
1.9
0.997968
Transcription factor
F
2.2
0.999954
1.5
0.998349
1.6
0.252851
Intercellular junction
assembly
F
2.1
0.945978
1.7
0.916174
1.5
0.990708
E
a
Criteria shows how the genes were selected by data analysis. F indicates that the gene was selected by fold-change. E indicates that the gene
was selected by Wilcoxon’s test. F E indicates that the gene was selected by both fold-change and Wilcoxon’s test.
b
F.C., fold change.
cell line (SNU-719R-DOX/100) showed a 1.4-fold MDK overexpression, which was slightly below the 1.5-fold cutoff value
used to determine significance. CDKN1A (p ⫽ 0.088), SF3B3
(p ⫽ 0.029), and LZTFL1 (p ⫽ 0.00074) were overexpressed in
both 5-FU- and cisplatin-resistant gastric cancer cells. NQO1
(p ⫽ 0.044), TYMS (p ⫽ 0.079), and IGFBP2 (p ⫽ 0.012) were
overexpressed in both 5-FU-resistant and doxorubicin-resistant
cell lines, whereas CALB2 (p ⫽ 0.051) was down-regulated in
these cells.
DISCUSSION
Recently, several studies on the drug-sensitivity and drugresistance in either untreated human cancer cell lines or drugexposed cells have been performed using microarray technologies (7, 9, 11, 21–22). They have revealed the correlations
between gene expression and drug activity, as well as identification of genes differentially expressed in drug-sensitive and
drug-resistant cancer cells. Also, several microarray studies on
the identification of genes with altered expression in human
gastric cancers have been performed (8, 10, 12). From these
studies, numerous genes have been identified as being associated with gastric cancer development and progression, some of
which will be used as novel chemotherapeutic targets for the
treatment or prevention of gastric cancer. However, although the
emergence of drug resistance is recognized as a major burden in
all cancer treatment, few microarray studies have sought to
identify candidate genes associated with drug resistance in gastric cancer. Here, we established 12 drug-resistant gastric cancer
cell lines from 4 different gastric cancer cell lines (SNU-620,
SNU-638, SNU-668, and SNU-719) by repeated exposure to the
chemotherapeutic drugs 5-FU, doxorubicin, and cisplatin. The
resistant cell lines have acquired drug-resistance over the longterm period from 11 months to 2 years, by increasing drug
dosage, and confirmed their stable resistance by repeated drug
sensitivity assays (MTT assay). It has previously been shown
that gastric cancer cells are more sensitive to doxorubicin and
cisplatin than to 5-FU (23). In support of this, we found that the
relative resistances of 5-FU-resistant cell lines measured by
MTT assay were higher than those of doxorubicin- and cisplatin-resistant cell lines. Of the 12 drug-resistant cell lines, two
(the doxorubicin-resistant cell line from SNU-638 and the
cisplatin-resistant cell line from SNU-719) were excluded from
our microarray analysis because MTT assay determined that
their relative resistances were ⬍2-fold, which was regarded as a
low degree of resistance. Thus, we used a total of 14 cell lines
(10 drug-resistant and 4 parent cell lines) for genome-wide
expression analysis to identify drug resistance candidate genes
in drug-resistant gastric cancer cell lines.
It was suggested that each cancer cell represents a different
pattern of drug-resistance gene expression even within cells
clonally derived from the same cancer, and may be expected to
exhibit a considerable amount of heterogeneity with respect to
280 Gene Expression Profiling in Drug-Resistant Gastric Cancer Cells
Table 4
Genes differentially expressed in cisplatin (CIS)-resistant gastric cancer cells
SNU-620R-CIS/2000 SNU-638R-CIS/400 SNU-668R-CIS/400
Affymetrix
identification
Symbol
Genes overexpressed
218437_s_at
LZTFL1
202284_s_at
CDKN1A
209035_at
MDK
213258_at
TFPI
36711_at
MAFF
221766_s_at
C6orf37
203571_s_at
215836_s_at
APM2
PCDHGB7
213603_s_at
203851_at
HSPC022
IGFBP6
216041_x_at
GRN
212992_at
LOC113146
213861_s_at
200688_at
221934_s_at
201015_s_at
218078_s_at
205198_s_at
218611_at
202581_at
200748_s_at
Leucine zipper
transcription factorlike 1
Cyclin-dependent kinase
inhibitor 1A (p21,
Cip1)
Midkine (neurite
growth-promoting
factor 2)
Tissue factor pathway
inhibitor
Homolog F (avian)
Chromosome 6 open
reading frame 37
Adipose specific 2
Protocadherin ␥
subfamily B, 7
HSPC022 protein
Insulin-like growth
factor binding protein
6
Granulin
Hypothetical protein
BC011859
DKFZP586D0919 DKFZP586D0919
protein
SF3B3
Splicing factor 3b,
subunit 3, 130 kDa
Homo sapiens cDNA
FLJ35885 fis, clone
TESTI2009018
JUP
Junction plakoglobin
ZDHHC3
Zinc finger, DHHC
domain
containing 3
ATP7A
ATPase, Cu⫹⫹
transporting, ␣
polypeptide
IER5
Immediate early
response 5
HSPA1B
Heat shock 70 kDa
protein 1B
FTH1
Ferritin, heavy
polypeptide 1
Genes downregulated
218841_at
MGC5540
203788_s_at
SEMA3C
201581_at
DJ971N18.2
202906_s_at
NBS1
202346_at
HIP2
204632_at
RPS6KA4
222155_s_at
FLJ11856
213622_at
COL9A2
a
Description
Hypothetical protein
MGC5540
Domain, secreted,
(semaphorin) 3C
Hypothetical protein
DJ971N18.2
Nijmegen breakage
syndrome 1 (nibrin)
Huntingtin interacting
protein 2
Ribosomal protein S6
kinase, 90 kDa,
polypeptide 4
Putative G-protein
coupled receptor
GPCR41
Collagen, type IX, ␣ 2
Function
Criteria
a
F.C.b
p
F.C.
p
F.C.
p
Unknown
F
2.1
0.019624
2.0
0.000438
5.5
0.145682
Cell cycle
F
3.6
0.37888
2.0
0.000023
1.5
0.003699
Cytokine, growth factor F
E
1.9
0.00004
3.1
0.00002
2.0
0.000189
F
E
2.2
0.00013
3.3
0.00002
1.6
0.00225
Transcription activating
factor
Unknown
F
E
1.3
0.002825
1.3
0.000236
4.3
⬍0.000001
E
1.7
0.000167
3.3
0.00002
1.7
0.00002
Unknown
Cell adhesion
E
1.7
1.9
0.000273
0.5
1.3
3.2
0.000618
0.00003
3.6
1.5
0.000027
0.038415
E
E
1.6
2.4
0.00003
0.000307
2.5
1.3
0.000023
0.000027
2.3
2.3
0.000023
0.000023
E
2.1
0.00002
2.2
0.00002
1.3
0.000052
E
2.5
0.00013
1.4
0.000214
1.4
0.00002
Plasma protein
Unknown
Signal transduction
F
F
Positive regulation of
cell proliferation
Unknown
Unknown
F
1.9
0.021224
1.6
0.000966
1.8
0.467656
mRNA splicing
F
1.5
0.5
1.8
0.161038
2.0
0.131343
Unknown
F
1.5
0.000552
2.2
0.000035
1.5
0.138386
Cell adhesion
Unknown
F
F
1.5
1.6
0.000046
0.003041
1.7
1.9
0.00004
0.000346
1.8
1.6
0.000273
0.069813
Copper ion transporter,
ATP binding
F
1.6
0.000167
1.7
0.003041
1.5
0.088938
E
Unknown
E
1.4
0.003041
1.6
0.000241
1.7
0.000241
Heat shock response
E
1.3
0.000068
1.9
0.00002
1.3
0.00249
Iron binding, cell
proliferation
E
1.6
0.000147
1.3
0.00002
1.7
0.000046
Unknown
F
2.0
0.994591
1.8
0.934434
4.1
0.973302
Drug resistance
F
2.1
0.703911
2.2
0.971234
1.9
0.998799
Unknown
F
1.9
0.98579
1.5
0.994067
2.0
0.999932
DNA repair
F
1.9
0.990708
1.6
0.98579
1.7
0.998923
Ubiquitin protein
enzyme
Protein kinase
F
2.0
0.999833
1.5
0.999948
1.5
0.99288
F
1.6
0.996301
1.8
0.998514
1.5
0.454766
Unknown
F
1.7
0.999135
1.7
0.999886
1.5
0.99994
Skeletal development
F
1.6
0.5
1.7
0.925732
1.6
0.545234
E
Criteria shows how the genes were selected by data analysis. F indicates that the gene was selected by fold-change. E indicates that the gene
was selected by Wilcoxon’s test. F E indicates that the gene was selected by both fold-change and Wilcoxon’s test.
b
F.C., fold change.
Clinical Cancer Research 281
Fig. 1 Multidimensional scaling and cluster analysis of acquired drugresistant gastric cancer cells. A, three-dimensional graph of multidimensional scaling generated by SPSS and SigmaPlot. Using the 22,282
transcripts contained in the Affymetrix HG-U133A oligonucleotide
microarray, we arranged the whole expression pattern of each drugresistant gastric cancer cell line in the three-dimensional graph. The
overall expression patterns follow the parent cell specificity rather than
drug specificity. 5-FU, 5-fluorouracil; DOX, doxorubicin; CIS, cisplatin. B, cluster analysis of the drug-resistant gastric cancer cell lines
(vertical axis) and genes selected for fold-change (⬎1.5-fold) and Wilcoxon’s test in three of four 5-fluorouracil (5-FU)-resistant, or all
doxorubicin (DOX)- and cisplatin (CIS)-resistant cell lines (horizontal
axis).
drug resistance (4). Because our four gastric cancer parent cell
lines were genetically different (17), the overall gene expression
pattern of each resistant cell line tended to be more similar to the
same-parent-originated resistant cell lines than to the samedrug-treated resistant cell lines, as shown in multidimensional
scaling (Fig. 1A).
Our microarray data provided information on genes that
were differentially associated with resistance to specific chemotherapeutic drugs, as well as those differentially associated with
two or more of the drugs. In this study, cells were maintained
without drugs before microarray experiments to avoid making
note of transcriptional changes caused by the insult of drugs
themselves. Thus, this precaution may not be compatible to
detect dynamic changes in response to the drugs themselves,
like acquired transcriptional activation. Although 5-FU, doxorubicin, and cisplatin exert their chemotherapeutic effects in
different ways, we found eight genes that were differentially
expressed in association with resistance to more than one of the
tested drugs, and one that was overexpressed in association with
resistance to all three tested drugs.
The doxorubicin-resistant cell lines provided the most differentially expressed genes. This suggests that the doxorubicinresistant cells, although derived from different gastric cancer
patients, contain more consistent molecular changes than the
other drug-resistant cells. When we performed clustering analysis between all drug-resistant cell lines and the differentially
expressed genes, we observed that the doxorubicin-resistant cell
lines were clustered together (Fig. 1B). A previous study demonstrated that at different time points, etoposide-resistant melanoma cells showed stable gene deregulation, whereas cisplatinand fotemustine-resistant cells showed substantial variation
(11). Together with our results, this suggests that anticancer
drugs targeting topoisomerase II, such as doxorubicin and etoposide, exhibit coherent resistance in different cell types and
during different periods.
Cell cycle deregulation is an important molecular event in
the acquisition of drug resistance (5). Most of the genes we
identified as overexpressed in doxorubicin-resistant gastric cancer cells were involved in the cell cycle, including the mitotic
cell cycle-associated genes BUB1, BUB1B, CDKN3, RRM1, and
RRM2. It was previously shown that cell cycle genes had an
interesting expression pattern in microarray experiment after
exposure to doxorubicin in MCF-7 cells and a subset of these
genes was also constitutively overexpressed in MCF-7 doxorubicin-resistant cells (7). Thus, these results suggest that cell
cycle genes might significantly contribute to the doxorubicin
resistance. In our microarray result, the up-regulations of BIRC5
were observed in doxorubicin-resistant cells. BIRC5, also
termed survivin, is a protein responsible for inhibiting apoptosis
and preventing cell death (24). Negative correlation between
BIRC5 (survivin) and 5-FU derivatives in untreated human
cancer cell lines was suggested (9), implying that the overexpression of BIRC5 might be associated with drug-resistance.
RRM2, one of two ribonucleotide reductase subunits, is a ratelimiting enzyme in DNA synthesis and repair (25). Previously,
it was reported that the overexpression of RRM2 mRNA and
protein was found in gemcitabine-resistant cells, implicating the
gene in drug resistance (25). Finally, the ATP-binding cassette
(ABC) transporter ABCC1 (MRP1) was up-regulated in all
doxorubicin-resistant gastric cancer cells, as compared with
ABCC1 expressions in the parent cells. Drug resistance through
drug efflux pumps has been well described, and the ABC
transporters are also known as energy-dependent drug efflux
pumps (4). In both our doxorubicin-resistant cells and their
parent gastric cancer cells, ABCB1 (MDR1) expressions were
282 Gene Expression Profiling in Drug-Resistant Gastric Cancer Cells
not detectable. It was previously demonstrated that the levels of
MDR1 RNA was relatively low in gastric cancer cell lines,
whereas intermediate or high levels were present in colorectal
carcinoma cell lines (6, 18).
A total of 38 genes were differentially expressed in 5-FUresistant gastric cancer cell lines. The relationship between the
overexpression of TYMS and 5-FU resistance is well-characterized (3), and was noted in this investigation except in one case
(SNU-638R–5-FU/50000). Although SNU-638R–5-FU/50000
represented the highest degree of resistance to 5-FU (⬎12,500fold), TYMS expression was slightly decreased in these cells per
our microarray data. This may be explained by the proposition
that high doses of 5-FU reduce the activity of TYMS (26).
Because the SNU-638R–5-FU/50000 cells were cultured at high
5-FU concentrations (380 ␮M) over almost 2 years, it is possible
that TYMS transcription levels had been down-regulated, and
that the acquired 5-FU resistance was not dependent on the
TYMS mechanism. A more possible explanation for the TYMS
down-regulation in SNU-638R-5-FU/50000 cells is that TYMS
may be mutated and has a decreased affinity for 5-fluoro-dUMP.
In our microarray data, DYPD expression was not significant in
5-FU resistant cells, although a previous report suggested that
DYPD expression was negatively correlated with 5-FU sensitivity in cancer cells (21). Another up-regulated gene is that for
SRI, a soluble resistance-related calcium-binding protein known
to be overexpressed in various drug-resistant cancer cells (27,
28), although this is the first identification of its association with
5-FU resistance. All four of the 5-FU-resistant cell lines showed
up-regulated SRI expression, and, although increased expression
of SRI is believed to result from gene amplification (28), it is not
yet known how SRI confers multidrug resistance. CLU (clusterin), also overexpressed in all of the 5-FU-resistant cell lines,
is a ubiquitous glycoprotein that is highly overexpressed in
some normal and malignant tissues undergoing apoptosis (29).
The overexpression of CLU has been shown to inhibit apoptotic
cell death in cisplatin-treated renal cell carcinoma, suggesting
that CLU confers a chemoresistance phenotype through its antiapoptotic activity (29, 30). Moreover, it has been demonstrated
that suppression of CLU expression by an antisense oligonucleotide had a chemosensitizing effect (29, 30). From these
results, it can be suggested that CLU is a potent candidate gene
for gastric cancer cell drug resistance.
In cisplatin-resistant gastric cancer cells, the differentially
expressed genes were mainly associated with stress response,
transport, cell cycle, and metabolism. ATP7A, a copper-transporting P-type adenosine triphosphatase, was overexpressed in
cisplatin-resistant gastric cancer cells. The closely related
ATP7B is well known for conferring resistance to cisplatin (31),
but little is known about the association of ATP7A with cisplatin
resistance. ATP7A is highly homologous with ATP7B and has a
similar export function for copper. A recent report (32) showed
ATP7A overexpression in a cisplatin-resistant ovarian cancer
cell line. These results suggest a potential role for ATP7A in
cisplatin resistance, as well as ATP7B, although the mechanism
has not yet been elucidated (32). The Fos gene family members,
FOS and FOSB, showed up-regulation in two of three cisplatinresistant cells (not listed in Table 4), consistent with previous
reports suggesting a relationship between Fos gene expression
and drug resistance (33).
In this study, we have identified eight genes whose differential expression was associated with resistance to more than
one of the three chemotherapeutic drugs tested. Of these, the
MDK gene was consistently overexpressed in all 5-FU-, doxorubicin-, and cisplatin-resistant gastric cancer cells compared
with the expressions in their parent cells, strongly suggesting
that it may contribute to multidrug resistance. The correspondence between mRNA and protein of MDK was observed in
SNU-620R-5-FU/1000 and SNU-638R-CIS/400 by Western
blot analysis. MDK, a heparin-binding growth factor, is associated with promoting neuronal survival, inducing neurite outgrowth (34 –36), and has been implicated in carcinogenesis and
angiogenesis. Found at low levels in normal adult tissue, MDK
is frequently overexpressed in esophageal, gastric, colon, pancreatic, hepatocellular, lung, breast, urinary bladder carcinoma,
neuroblastoma and Wilms’ tumors (35, 36). Transfection of an
MDK antisense oligodeoxynucleotide into CMT-93 mouse rectal carcinoma cells markedly suppressed tumor growth, suggesting that MDK could be a potent target for cancer therapy (36,
37). MDK has cytoprotective activity and has rescued cisplatininduced apoptotic cell death in both murine kidney and cultured
G401 cells, implying that MDK confers a cellular growth advantage by functioning as an antiapoptotic factor (34). Recently,
the overexpression of HB-EGFGF (heparin-binding epidermal
growth factor-like growth factor) was observed in the cisplatinresistant and 5-FU-resistant gastric tumors by microarray analysis, suggesting that HB-EGFGF is a candidate chemoresistantrelated gene of gastric cancer (38). Interestingly, these results
support that heparin-binding growth factors such as MDK or
HB-EGFGF might be one of the important genes contributing to
chemoresistance in gastric cancers. Accordingly, we hypothesize that the overexpressed MDK can contribute to the growth
progression of gastric cancer cells with an acquired multidrug
resistance and may be a potent target to restore chemosensitivity. To clarify the relationship between MDK overexpression
and multidrug resistance, it will be necessary to further investigate whether the inhibition of MDK overexpression in drugresistant cancer cells leads to recovered sensitivity.
In summary, we have identified genes that are differentially
expressed in 5-FU-, doxorubicin-, and cisplatin-resistant gastric
cancer cells. Some of the identified genes were previously
known to be associated with drug-resistance. To date, most
studies on drug resistance in gastric cancer have focused on
candidate gene work targeting a limited number of drug resistance genes identified in other cancers. In contrast, our use of
microarray technology for a genome-wide screen of 10 drugresistant gastric cancer cell lines has provided useful information on differentially expressed genes and possible new candidate multidrug resistance genes in gastric cancer cells. Although
further in vivo validations for the identified genes are required
because the materials used in the study were induced drugresistant cells in vitro, this important information may lead to
the discovery of new drug resistance targets, and perhaps to the
development of better cancer chemotherapy strategies.
REFERENCES
1. Bae, J-M., Won, Y-J., Jung, K-W., Suh, K-A., Ahn, D-H., and Park,
J. G. Annual report of the Central Cancer Registry in Korea-1999: based
Clinical Cancer Research 283
on registered data from 128 hospitals. Cancer Res. Treat., 33: 367–372,
2001.
2. Kelley, J. R., and Duggan, J. M. Gastric cancer epidemiology and
risk factors. J. Clin. Epidemiol., 56: 1–9, 2003.
3. Banerjee, D., Mayer-Kuckuk, P., Capiaux, G., Budak-Alpdogan, T.,
Gorlick, R., and Bertino, J. R. Novel aspects of resistance to drugs
targeted to dihydrofolate reductase and thymidylate synthase. Biochimica Biophysica Acta, 1587: 164 –173, 2002.
4. Gottesman, M. M. Mechanisms of cancer drug resistance. Annu.
Rev. Med., 53: 615– 627, 2002.
5. Larsen, A. K., Escargueil, A. E., and Skladanowski, A. Resistance
mechanisms associated with altered intracellular distribution of anticancer agents. Pharmacol. Ther., 85: 217–229, 2000.
6. Kang, M. S., Kim, H. S., Han, J. A., Park, S. C., Kim, W. B., and
Park, J. G. Characteristics of human gastric carcinoma cell lines with
induced multidrug resistance. Anticancer Res., 17: 3531–3536, 1997.
7. Kudoh, K., Ramanna, M., Ravatn, R., Elkahloun, A. G., Bittner,
M. L., Meltzer, P. S., Trent, J. M., Dalton, W. S., and Chin, K-V.
Monitoring the expression profiles of doxorubicin-induced and doxorubicin-resistant cancer cells by cDNA microarray. Cancer Res., 60:
4161– 4166, 2000.
8. Hippo, Y., Taniguchi, H., Tsutsumi, S., Machida, N., Chong, J-M.,
Fukayama, M., Kodama, T., and Aburatani, H. Global gene expression
analysis of gastric cancer by oligonucleotide microarrrays. Cancer Res.,
62: 233–240, 2002.
9. Dan, S., Tsunoda, T., Kitahara, O., Yanagawa, R., Zembutsu, H.,
Katagiri, T., Yamazaki, K., Nakamura, Y., and Yamori, T. An integrated
database of chemosensitivity to 55 anticancer drugs and gene expression
profiles of 39 human cancer cell lines. Cancer Res., 62: 1139 –1147,
2002.
10. Sakakura, C., Hagiwara, A., Nakanishi, M., Shimomura, K., Takagi,
T., Yasuoka, R., Fujita, Y., Abe, T., Ichikawa, Y., Takahashi, S.,
Ishikawa, T., Nishizuka, I., Morita, T., Shimada, H., Okazaki, Y.,
Hayashizaki, Y., and Yamagishi, H. Differential gene expression profiles of gastric cancer cells established from primary tumour and malignant ascites. Br. J. Cancer, 87: 1153–1161, 2002.
11. Wittig, R., Nessling, M., Will, R. D., Mollenhauer, J., Salowsky, R.,
Münstermann, E., Schick, M., Helmbach, H., Gschwendt, B., Korn, B.,
Kioschis, P., Lichter, P., Schadendorf, D., and Poustka, A. Candidate
genes for cross-resistance against DNA-damaging drugs. Cancer Res.,
62: 6698 – 6705, 2002.
12. Ji, J., Chen, X., Leung, S. Y., Chi, J-T. A., Chu, K. M., Yuen, S. T.,
Li, R., Chan, A. S. Y., Li, J., Dunphy, N., and So, S. Comprehensive
analysis of the gene expression profiles in human gastric cancer cell
lines. Oncogene, 21: 6549 – 6556, 2002.
13. Kim, I-J., Kang, H. C., Park, J.-H., Ku, J-L., Lee, J-S., Kwon, H-J.,
Yoon, K-A., Heo, S. C., Yang, H-Y., Cho, B. Y., Kim, S. Y., Oh, S. K.,
Youn, Y-K., Park, D-J., Lee, M-S., Lee, K-W., and Park, J-G. RET
oligonucleotide microarray for the detection of RET mutations in multiple endocrine neoplasia type 2 syndromes. Clin. Cancer Res., 8:
457– 463, 2002.
14. Kim, I-J., Kang, H. C., Park, J-H., Shin, Y., Ku, J-L., Lim, S-B.,
Park, S. Y., Jung, S-Y., Kim, H. K., and Park, J-G. Development and
applications of a ␤-catenin oligonucleotide microarray: ␤-catenin mutations are dominantly found in the proximal colon cancers with microsatellite instability. Clin. Cancer Res., 9: 2920 –2925, 2003.
15. Robert, J., and Larsen, A. K. Drug resistance to topoisomerase II
inhibitors. Biochimie, 80: 247–254, 1998.
16. Kartalou, M., and Essigmann, J. M. Mechanisms of resistance to
cisplatin. Mutat. Res., 478: 23– 43, 2001.
17. Park, J-G., Yang, H-K., Kim, W. H., Kang, M-S., Lee, J-H., Oh,
J. H., Park, H-S., Yeo, K-S., Kang, S. H., Song, S-Y., Kang, Y. K.,
Bang, Y-J., Kim, Y. I., and Kim, J-P. Establishment and characterization
of human gastric carcinoma cell lines. Int. J. Cancer, 70: 443– 449,
1997.
18. Park, J-G., Kramer, B. S., Steinberg, S. M., Carmichael, J., Collins,
J. M., Minna, J. D., Gazdar, A. F. Chemosensitivity testing of human
colorectal carcinoma cell lines using a tetrazolium-based colorimetric
assay. Cancer Res., 47: 5875–5879, 1987.
19. Kadomatsu, K., Hagihara, M., Akhter, S., Fan, Q-W., Muramatsu,
H., Muramatsu, T. Midkine induces the transformation of NIH3T3 cells.
Br. J. Cancer, 75: 354 –359, 1997.
20. Shoshani, T., Faerman, A., Mett, I., Zelin, E., Tenne, T., Gorodin,
S., Moshel, Y., Elbaz, S., Budanov, A., Chajut, A., Kalinski, H., Kamer,
I., Rozen, A., Mor, O., Keshet, E., Leshkowitz, D., Einat, P., Skaliter,
R., and Feinstein, E. Identification of a novel hypoxia-inducible factor
1-responsive gene, RTP801, involved in apoptosis. Mol. Cell. Biol., 22:
2283–2293, 2002.
21. Scherf, U., Ross, D. T., Waltham, M., Smith, L. H., Lee, J. K.,
Tanabe, L., Kohn, K. W., Reinhold, W. C., Myers, T. G., Andrews,
D. T., Scudiero, D. A., Eisen, M. B., Sausville, E. A., Pommier, Y.,
Botstein, D., Brown, P. O., and Weinstein, J. N. A gene expression
database for the molecular pharmacology of cancer. Nat. Genet., 24:
236 –244, 2000.
22. Staunton, J. E., Slonim, D. K., Coller, H. A., Tamayo, P., Angelo,
M. J., Park, J., Scherf, U., Lee, J. K., Reinhold, W. O., Weinstein, J. N.,
Mesirov, J. P., Lander, E. S., and Golub, T. R. Chemosensitivity
prediction by transcriptional profiling. Proc. Natl. Acad. Sci. USA, 98:
10787–10792, 2001.
23. Park, J-G., Kramer, B. S., Lai, S. L., Goldstein, L. J., and Gazdar,
A. F. Chemosensitivity patterns and expression of human multidrug
resistance-associated MDR1 gene by human gastric and colorectal carcinoma cell lines. J. Natl. Cancer Inst. (Bethesda), 82: 193–198, 1990.
24. Blanc-Brude, O. P., Yu, J., Simosa, H., Conte, M. S., Sessa, W. C.,
and Altieri, D. C. Inhibitor of apoptosis protein surviving regulates
vascular injury. Nat. Med., 8: 987–994, 2002.
25. Zhou, B., and Yen, Y. Characterization of the human ribonucleotide
reductase M2 subunit gene; genomic structure and promoter analyses.
Cytogenet. Cell Genet., 95: 52–59, 2001.
26. Elstein, K. H., Mole, M. L., Setzer, W., Zucker, R. M., Kavlock,
R. J., Rogers, J. M., and Lau, C. Nucleoside-mediated mitigation of
5-fluorouracil-induced toxicity in synchronized murine erythroleukemic
cells. Toxicol. Appl. Pharmacol., 146: 29 –39, 1997.
27. Beyer-Sehlmeyer, G., Hiddemann, W., Wörmann, B., and Bertram,
J. Suppressive subtractive hybridization reveals differential expression
of sergylcin, sorcin, bone marrow proteoglycan, and prostate-tumorinducing gene I (PTI-1) in drug-resistant and sensitive tumour cell lines
of haematopoetic origin. Eur. J. Cancer, 35: 1735–1742, 1999.
28. Parekh, H. K., Deng, H. B., Choudhary, K., Houser, S. R., and
Simpkins, H. Overexpression of sorcin, a calcium-binding protein, induces a low level paclitaxel resistance in human ovarian and breast
cancer cells. Biochem. Pharmacol., 63: 1149 –1158, 2002.
29. Miyake, H., Hara, I., Kamidono, S., and Gleave, M. E. Synergistic
chemosensitization and inhibition of tumor growth and metastasis by the
antisense oligodeoxynucleotide targeting clusterin gene in a human
bladder cancer model. Clin. Cancer Res., 7: 4245– 4252, 2001.
30. Lee, C., Jin, R., Kwak, C., Jeong, H., Park, M., Lee, N., and Lee, S.
Suppression of clusterin expression enhanced cisplatin-induced cytotoxicity on renal cell carcinoma cells. Urology, 60: 516 –520, 2002.
31. Nakayama, K., Kanzaki, A., Ogawa, K., Miyazaki, K., Neamati, N.,
and Takebayashi, Y. Copper-transporting p-type adenosine triphosphatase (ATP7B) as a cisplatin based chemoresistance marker in ovarian
carcinoma: comparative analysis with expression of MDR1, MRP1,
MRP2, LRP, and BCRP. Int. J. Cancer, 101: 488 – 495, 2002.
32. Katano, K., Kondo, A., Safaei, R., Holzer, A., Samimi, G.,
Mishima, M., Kuo, Y-M., Rochdi, M., and Howell, S. B. Acquisition of
resistance to cisplatin is accompanied by changes in the cellular pharmacology of copper. Cancer Res., 62: 6559 – 6565, 2002.
33. Hollander, M. C., and Fornace, A. J. J. Induction of fos RNA by
DNA-damaging agents. Cancer Res., 49: 1687–1692, 1989.
34. Qi, M., Ikematsu, S., Ichihara-Tanaka, K., Sakuma, S., Muramatsu,
T., and Kadomatsu, K. Midkine rescues Wilms’ tumor cells from
cisplatin-induced apoptosis: regulation of Bcl-2 expression by midkine.
J. Biochem., 127: 269 –277, 2000.
284 Gene Expression Profiling in Drug-Resistant Gastric Cancer Cells
35. Takei, Y., Kadomatsu, K., Matsuo, S., Itoh, H., Nakazawa, K.,
Kubota, S., and Muramatsu, T. Antisense oligodeoxynucleotide targeted
to midkine, a heparin-binding growth factor, suppresses tumorigenicity
of mouse rectal carcinoma cells. Cancer Res., 61: 8486 – 8491, 2001.
36. Sakitani, H., Tsutsumi, M., Kadomatsu, K., Ikematsu, S., Takahama, M., Iki, K., Tsujiuchi, T., Muramatsu, T., Sakuma, S., Sakaki, T.,
and Konishi, Y. Overexpression of midkine in lung tumors induced by
N-nitrosobis(2-hydroxypropyl)amine in rats and its increase with progression. Carcinogenesis (Lond.), 20: 465– 469, 1999.
37. Takei, Y., Kadomatsu, K., Itoh, H., Sato, W., Nakazawa, K., Kubota, S., and Muramatsu, T. 5⬘-,3⬘-Inverted thymidine-modified antisense
oligodeoxynucleotide targeting midkine. Its design and application for
cancer therapy. J. Biol. Chem., 277: 23800 –23806, 2002.
38. Suganuma, K., Kubota, T., Saikawa, Y., Abe, S., Otani, Y., Furukawa, T., Kumai, K., Hasegawa, H., Watanabe, M., Kitajima, M.,
Nakayama, H., and Okabe, H. Possible chemoresistance-related genes
for gastric cancer detected by cDNA microarray. Cancer Sci., 94:
355–359, 2003.