Identification of novel molecular markers for detection of gastric cancer cells in the peripheral blood circulation using genome-wide microarray analysis

Abstract

Although metastasis or relapse is a leading cause of death for patients with gastric cancer, the hematogenous spread of cancer cells remains undetected at the time of initial therapy. The development of novel diagnostic molecular marker(s) to detect circulating gastric cancer cells is an issue of great clinical importance. We obtained peripheral blood samples from 10 patients with gastric cancer who underwent laparotomy and 4 healthy volunteers. Microarray analysis consisting of 30,000 genes or ESTs was carried out using eight gastric cancer tissues and normal gastric mucosae. We selected 53 genes up-regulated in gastric cancer compared to normal gastric mucosae from our microarray data set, and, among these, identified five candidate marker genes (TSPAN8, EPCAM, MMP12, MMP7 and REG3A) which were not expressed in peripheral blood mononuclear cells (PBMCs) from 4 healthy volunteers. We further carried out semi-quantitative nested reverse transcription-polymerase chain reaction (RT-PCR) for HRH1, EGFR, CK20 and CEA in addition to the five newly identified genes using PBMCs of patients with gastric cancer, and found that expression of one or more genes out of the nine was detected in 80% of the patients with gastric cancer. Moreover, the numbers of genes expressed in PBMCs were ≤2 and ≥2 in all vascular invasion-negative cases and in 5 of 6 positive cases, respectively, showing significant differences between the two groups (P=0.041). Nested RT-PCR analysis for the set of nine marker genes using PBMCs may provide the potential for detection of circulating gastric cancer cells prior to metastasis formation in other organs.

Introduction

Gastric cancer causes approximately 800,000 deaths worldwide per year and is still one of the leading causes of cancer-related death in the world (1). Most gastric cancers at an early stage can be cured by surgical resection; however, patients with advanced gastric cancers have worse prognosis than those with early stage disease (2). Although metastasis or relapse is the main cause of death for patients with gastric cancer (3), the hematogenous spread of malignant cells remains undetected at the time of initial therapy. During the development of cancer, tumor cells may detach from the primary tumor and disseminate into the lymph system and/or blood circulation, and grow in the bone marrow, liver, kidney and other organs, which is called micrometastasis (4). Micrometastasis is barely detected by routine biochemical and histopathological assays or graphical methods, such as X-ray, CT and MRI (3). Detection of circulating tumor cells at the mRNA level [reverse transcription-polymerase chain reaction (RT-PCR)] in blood samples of patients with cancer could serve as a unique and easy diagnostic tool to predict cancer recurrence and to monitor treatment effectiveness (5–7). However, molecular marker(s) that detect circulating gastric cancer cells for routine clinical use have not yet been identified. Hence, the development of novel diagnostic molecular marker(s) to detect circulating gastric cancer cells is an issue of great clinical importance.

Carcinoembryonic antigen (CEA) is a well-known tumor marker and has been used to detect small amounts of adenocarcinoma cells in the blood, peritoneal wash or other body fluids (8–12). However, the expression of CEA mRNA is not specific to cancer cells and often produces false-positive results (13). Profiling of gene expression patterns on genome-wide microarrays enables investigators to perform comprehensive characterization of molecular activities in cancer cells (14–17). Systematic analysis of expression levels for thousands of genes is also a useful approach for identifying molecular markers to detect small amounts of circulating cancer cells (18). In this study, we identified genes whose expression had been altered during gastric carcinogenesis using genome-wide information obtained from 8 cases on a microarray consisting of 30,000 transcribed elements. Based on the results of the microarray assay, we identified five candidate genes for the specific detection of circulating gastric cancer cells at the mRNA level. We suggest that such information may lead ultimately to improve the prognosis of patients with gastric cancers.

Materials and methods

Blood and tissue samples

Blood samples were obtained from 10 patients with gastric cancer who underwent laparotomy and 4 healthy volunteers after obtaining informed consent. Heparinized blood samples (5 ml) from the 10 patients with gastric cancer were obtained from a peripheral artery through a catheter used for monitoring arterial blood pressure during surgical operation. Peripheral venous blood was obtained from 4 healthy volunteers for control after discarding the initial 10 ml of blood to protect the mixture from epithelial cells. Clinicopathological characteristics of the 10 patients are shown in Table I. Clinical stage of each patient was judged according to the Union for International Cancer Control (UICC) TNM classification. Among the 10 patients with gastric cancer, 8 primary gastric cancer tissues and corresponding non-cancerous gastric mucosae from surgically resected tissues were obtained at Sapporo Medical University and Douto Hospital after each patient had provided informed consent. The samples that had been confirmed histologically as gastric adenocarcinoma were used for microarray study. These samples were immediately frozen and stored at −80°C. All cancer tissues were obtained from the margin of the tumor mass, while non-cancerous tissues were obtained from corresponding normal mucosae of the same stomach. This study was approved by the Ethics Committee of Sapporo Medical University, School of Medicine, Hokkaido, Japan.

Table I.

Characteristics of patients included in the nested RT-PCR analysis of PBMCs.

Parameters	No. of patients
Gender (male:female)	5:5
Age range (average), in years	41–82 (61.9)
Depth of tumor invasion (T1:T2:T3:T4)	1:6:2:1
Lymph node metastasis (N0:N1:N2:N3)	4:3:2:1
Distant metastasis (M0:M1)	10:0
Liver metastasis (H0:H1)	10:0
Peritoneal metastasis (P0:P1)	9:1
Peritoneal lavage cytology (CY0:CY1)	9:1
Stage (I:II:III:IV)	4:1:2:3
Lymphatic invasion (ly0:ly1–3)	2:8
Vessel invasion (v0:v1–3)	4:6

RNA extraction of blood samples

We prepared peripheral blood mononuclear cells (PBMCs) using Ficoll (Amersham Biosciences, Buckinghamshire, UK) and extracted total RNA using TRIzol (Invitrogen, Inc., Carlsbad, CA, USA) according to the manufacturer’s instructions. Before the synthesis of cDNA, deoxyribonuclease I (DNase I) (Nippon Gene, Japan) was added to each sample of total RNA according to the manufacturer’s instructions.

Analysis of microarray

Total RNA was extracted from each gastric tissue using TRIzol according to the manufacturer’s instructions. To guarantee the quality of RNAs, total RNA extracted from the residual tissue of each case was electrophoresed on a denaturing agarose gel, and the quality was confirmed by the presence of rRNA bands. After treatment with DNase I, T7-based RNA amplification was carried out as described previously with a few modifications (19). Using 5 μg of total RNA from each tissue sample as starting material, one round of amplification was performed; the amount of each amplified RNA (aRNA) was measured by a spectrophotometer. A mixture of normal gastric mucosae from 8 patients was prepared as a universal control and was amplified in the same manner; 2.5 μg of aRNAs from each cancerous tissue and from the control was reversely transcribed in the presence of Cy5-dCTP and Cy3-dCTP, respectively (15). AceGene 30K-1 Chip Version (Hitachi Software Engineering Co., Japan) was used for microarray analysis. The procedures for hybridization, washing, photometric quantification of signal intensities of each spot and normalization of data were according to the manufacturer’s instructions. To normalize the amount of mRNA between tumors and controls, the fluorescence intensities of Cy5 (gastric cancer) and Cy3 (control) for each target spot were adjusted so that the mean Cy5/Cy3 ratio of 30,000 genes equaled 1. Genes were categorized into three groups according to the cancer/normal ratio of their mean signal intensity: up-regulated (expression ratio >5.0), down-regulated (expression ratio <0.2) and unchanged expression (expression ratio between 0.2 and 5.0).

Semi-quantitative RT-PCR

To validate the result of the microarray analysis, we examined the expression levels of the genes up-regulated in gastric cancer by means of semi-quantitative RT-PCR analysis. Total RNAs (3 μg) extracted from each cancerous tissue and normal gastric mucosa were reversely transcribed for single-stranded cDNAs using oligo(dT)12–18 primer with Superscript II reverse transcriptase (Life Technologies, Inc.). Each single-stranded cDNA was diluted for subsequent PCR amplification. A housekeeping gene, GAPDH, served as the internal control. The PCR reaction was conducted at 95°C for 5 min, and then for 30 cycles at 95°C for 30 sec, 60°C for 30 sec and 72°C for 1 min followed by 72°C for 10 min, in the Gene Amp PCR System 9700 (Perkin-Elmer Applied Biosystems, Foster City, CA, USA).

Nested RT-PCR using blood samples

We performed nested RT-PCR using total RNAs extracted from PBMCs to accurately examine mRNA levels of the candidate marker genes. Initially, RT-PCR was carried out as described above. In nested RT-PCR, 1 ml of the initial PCR product, 4 ml of 10X PCR buffer, 200 mmol/l dNTP mixture, 0.2 mmol/l primers and 1 unit Taq DNA polymerase (Takara) were added to a 40-ml aliquot of the reaction mixture. The PCR reaction was conducted at 95°C for 5 min, and then 30 cycles at 95°C for 30 sec, 60°C for 30 sec and 72°C for 1 min followed by 72°C for 10 min, in the Gene Amp PCR System 9700. The RT-PCR products were detected using 2% agarose gel electrophoresis. The primer sequences are summarized in Table II.

Table II.

Primer sequences for semi-quantitative nested RT-PCR.

Gene	Forward primer	Reverse primer
TSPAN8	5′-TCAACTTCTTGTTCTGGCTATGT-3′	5′-TATAGCTTTGGCATGGTCTCTGC-3′
EPCAM	5′-TGATCCTGACTGCGATGAGAGC-3′	5′-CAGCTTTCAATCACAAATCAGT-3′
MMP12	5′-AACCAGCTCTCTGTGACCCCA-3′	5′-TCCAAGGATGTTAGGAAGCAAC-3′
MMP7	5′-TCTCTGGACGGCAGCTATGCG-3′	5′-AATAAGACACAGTCACACCATAA-3′
REG3A	5′-GTATCTTGGATGCTGCTTTCCTG-3′	5′-GTATGACAAAATGAAGAGACTGA-3′
HRH1	5′-TACAAGGCCGTACGACAACACT-3′	5′-TCTGCTGTTCTTCTATGGTGCCT-3′
EGFR	5′-ATGTCCCCACGGTACTTACTCCC-3′	5′-TCTTAACAATGCTGTAGGGGCTC-3′
CK20	5′-TGGATTTCAGTCGCAGA-3′	5′-ATGTAGGGTTAGGTCATCAAAG-3′
CEA	5′-TTCTCCTGGTCTCTCAGCTGGG-3′	5′-AATGCTTTAAGGAAGAAGCAA-3′

Results

Identification of up- or down-regulated genes in the gastric cancers

We extracted RNAs from eight primary gastric cancer tissues and corresponding normal gastric mucosae as control, and carried out gene expression analysis using a microarray consisting of 30,000 genes or ESTs. We then selected genes from our data set according to the criterion that the cancer/ normal ratio of the mean signal intensity of a given gene was >5.0 or <0.2, and 53 genes were identified as up-regulated and 123 genes as down-regulated in the gastric cancer tissues compared to the normal gastric mucosa (Tables III and IV). The up-regulated genes represented a variety of functions, including genes associated with signal-transduction pathways (SFRP4 and TSPAN8), genes encoding transcription factors (TRIM33), genes involved in various metabolic pathways (ADH4, USP33, RNF128, MAN2A1, UBD and GCNT3), apoptosis (SPP1 and RIPK2), chemokines (CCL20), DNA replication and recombination (SNRPA1), cell adhesion and cytoskeleton (LAMB3, EPCAM, MMP7 and COL1A1), cell-cell signaling (CEACAM6 and CXCL9), cell cycle (CDC2, BUB1 and CCNB2), cell proliferation (REG1B and REG3A), or other functions (SPINK4, TMC5, LGALS2, KYNU, DDX58, LY96, UMPS and RNF157).

Table III.

Genes up-regulated in advanced gastric cancer.

No.	Accession no.	Gene symbol	Description
1	NM_006507	REG1B	Regenerating islet-derived 1 β (pancreatic stone protein, pancreatic thread protein)
2	NM_004577	PSPH	Phosphoserine phosphatase
3	NM_138938	REG3A	Regenerating islet-derived 3 α
4	NM_014471	SPINK4	Serine peptidase inhibitor, Kazal type 4
5	NM_001105249	TMC5	Transmembrane channel-like 5
6	NM_002426	MMP12	Matrix metallopeptidase 12 (macrophage elastase)
7	NM_002423	MMP7	Matrix metallopeptidase 7 (matrilysin, uterine)
8	NM_002483	CEACAM6	Carcinoembryonic antigen-related cell adhesion molecule 6 (non-specific cross reacting antigen)
9	NM_001786	CDC2	Cell division cycle 2, G1 to S and G2 to M
10	NM_000582	SPP1	Secreted phosphoprotein 1 (osteopontin, bone sialoprotein I, early T-lymphocyte activation 1)
11	NM_004751	GCNT3	Glucosaminyl (N-acetyl) transferase 3, mucin type
12	NM_000574	CD55	CD55 molecule, decay accelerating factor for complement (Cromer blood group)
13	NM_002443	MSMB	Microseminoprotein, β
14	NM_004336	BUB1	UB1 budding uninhibited by benzimidazoles 1 homolog (yeast)
15	NM_015017	USP33	Ubiquitin-specific peptidase 33
16	NM_004591	CCL20	Chemokine (C-C motif) ligand 20
17	NM_017633	FAM46A	Family with sequence similarity 46, member A
18	NM_000088	COL1A1	Collagen, type I, α 1
19	XR_017717	ADAMTSL3	ADAMTS-like 3
20	NM_138938	REG3A	Regenerating islet-derived 3 α
21	NM_017934	PHIP	Pleckstrin homology domain interacting protein
22	XR_016124		Similar to p21-activated kinase 2
23	NM_006398	UBD	Ubiquitin D
24	NM_002358	MAD2L1	MAD2 mitotic arrest deficient-like 1 (yeast)
25	NM_002483	CEACAM6	Carcinoembryonic antigen-related cell adhesion molecule 6 (non-specific cross-reacting antigen)
26	NM_173164	IPO9	Importin 9
27	NM_003014	SFRP4	Secreted frizzled-related protein 4
28	NM_004616	TSPAN8	Tetraspanin 8
29	NM_002354	EPCAM	Epithelial cell adhesion molecule
30	NM_006498	LGALS2	Lectin, galactoside-binding, soluble, 2
31	NM_002372	MAN2A1	Mannosidase, α, class 2A, member 1
32	NM_003937	KYNU	Kynureninase (L-kynurenine hydrolase)
33	NM_003821	RIPK2	Receptor-interacting serine-threonine kinase 2
34	NM_00108039	ITGA7	Integrin, α 7
35	NM_000670	ADH4	Alcohol dehydrogenase 4 (class II), π polypeptide
36	NM_014314	DDX58	DEAD (Asp-Glu-Ala-Asp) box polypeptide 58
37	NM_006418	OLFM4	Olfactomedin 4
38	NM_198187.3	ASTN2	Astrotactin 2
39	NM_015364	LY96	Lymphocyte antigen 96
40	NM_000574	CD55	CD55 molecule, decay accelerating factor for complement (Cromer blood group)
41	NM_018964	SLC37A1	Solute carrier family 37 (glycerol-3-phosphate transporter), member 1
42	NM_018455	CENPN	Centromere protein N
43	NM_001710	CFB	Complement factor B
44	NM_033020	TRIM33	Tripartite motif-containing 33
45	NM_003090	SNRPA1	Small nuclear ribonucleoprotein polypeptide A'
46	NM_000373	UMPS	Uridine monophosphate synthetase (orotate phosphoribosyl transferase and orotidine-5′-decarboxylase)
47	NM_144584	C1orf59	Chromosome 1 open reading frame 59
48	NM_052916.2	RNF157	Ring finger protein 157
49	NM_006332	IFI30	Interferon, γ-inducible protein 30
50	NM_002416	CXCL9	Chemokine (C-X-C motif) ligand 9
51	NM_001017402	LAMB3	Laminin, β 3
52	NM_004701	CCNB2	Cyclin B2
53	NM_194463	RNF128	Ring finger protein 128

Table IV.

Genes down-regulated in advanced gastric cancer.

No.	Accession no.	Gene symbol	Description
1	NM_004190	LIPF	Lipase, gastric
2	NM_020143	PNO1	Partner of NOB1 homolog (S. cerevisiae)
3	NM_000257	MYH7	Myosin, heavy chain 7, cardiac muscle, β
4	NM_015173	TBC1D1	TBC1 (tre-2/USP6, BUB2, cdc16) domain family, member 1
5	NM_005408	CCL13	Chemokine (C-C motif) ligand 13
6	NM_174929	ZMIZ2	Zinc finger, MIZ-type containing 2
7	NM_004747	DLG5	Discs, large homolog 5 (Drosophila)
8	NM_024872.2	DOK3	Docking protein 3
9	NM_201653	CHIA	Chitinase, acidic
10	NM_003893	LDB1	LIM domain binding 1
11	NM_012455.2	PSD4	Pleckstrin and Sec7 domain containing 4
12	NM_005213	CSTA	Cystatin A (stefin A)
13	NM_005416	SPRR3	Small proline-rich protein 3
14	NM_014989	RIMS1	Regulating synaptic membrane exocytosis 1
15	NM_001018005	TPM1	Tropomyosin 1 (α)
16	NM_213589	RAPH1	Ras association (RalGDS/AF-6) and pleckstrin homology domains 1
17	NM_004898	CLOCK	Clock homolog (mouse)
18	NM_013292		Fast skeletal myosin light chain 2
19	NM_020321	ACCN3	Amiloride-sensitive cation channel 3
20	NM_002754	MAPK13	Mitogen-activated protein kinase 13
21	NM_013443	ST6GALNAC6	ST6 (α-N-acetyl-neuraminyl-2,3-β-galactosyl-1,3)-N-acetylgalactosaminide α-2,6-sialyltransferase 6
22	NM_001042453		Serine/threonine protein kinase MST4
23	NM_032646	TTYH2	Tweety homolog 2 (Drosophila)
24	NM_015089		p53-associated parkin-like cytoplasmic protein
25	NM_003609	HIRIP3	HIRA interacting protein 3
26	NR_002219	BIRC5	Baculoviral IAP repeat-containing 5 (survivin)
27	NM_000068	CACNA1A	Calcium channel, voltage-dependent, P/Q type, α 1A subunit
28	NM_203377	MB	Myoglobin
29	NM_003768	PEA15	Phosphoprotein enriched in astrocytes 15
30	NM_053013	ENO3	Enolase 3 (β, muscle)
31	XR_018802	PI4K2A	Phosphatidylinositol 4-kinase type 2 α
32	NM_003725	HSD17B6	Hydroxysteroid (17-β) dehydrogenase 6 homolog (mouse)
33	NM_006063	KBTBD10	Kelch repeat and BTB (POZ) domain containing 10
34	NM_012288	TRAM2	Translocation associated membrane protein 2
35	NM_000730	CCKAR	Cholecystokinin A receptor
36	NM_000290	PGAM2	Phosphoglycerate mutase 2 (muscle)
37	NM_199354	PRB1	Proline-rich protein BstNI subfamily 1
38	XR_019039	ACTB	Actin, β
39	NM_006478	GAS2L1	Growth arrest-specific 2 like 1
40	NM_024674	LIN28	Lin-28 homolog (C. elegans)
41	NM_001070	TUBG1	Tubulin, γ 1
42	NM_015654	NAT9	N-acetyltransferase 9
43	NM_003643	GCM1	Glial cells missing homolog 1 (Drosophila)
44	NM_006901.2	MYO9A	Myosin IXA
45	NM_017785	CCDC99	Coiled-coil domain containing 99
46	NM_025135	FHOD3	Formin homology 2 domain containing 3
47	NM_022566	MESDC1	Mesoderm development candidate 1
48	NM_198255	TERT	Telomerase reverse transcriptase
49	NM_018231		Amino acid transporter
50	NM_002458	MUC5B	Mucin 5B, oligomeric mucus/gel-forming
51	NM_001001522	TAGLN	Transgelin
52	NM_002631	PGD	Phosphogluconate dehydrogenase
53	NM_006984	CLDN10	Claudin 10
54	NM_004359	CDC34	Cell division cycle 34 homolog (S. cerevisiae)
55	NM_001824	CKM	Creatine kinase, muscle
56	NM_002274	KRT13	Keratin 13
57	XR_019039	ACTB	Actin, β
58	NM_000477	ALB	Albumin
59	NM_001519.2	BRF1	BRF1 homolog, subunit of RNA polymerase III transcription initiation factor IIIB (S. cerevisiae)
60	NM_006790	MYOT	Myotilin
61	NM_021948	BCAN	Brevican
62	NM_001142404.1	CD164	CD164 molecule, sialomucin
63	BC050364.1	C7orf13	Chromosome 7 open reading frame 13
64	NM_005177	ATP6V0A1	ATPase, H+ transporting, lysosomal V0 subunit a1
65	NM_020393	PGLYRP4	Peptidoglycan recognition protein 4
66	XM_937007	FRMPD3	FERM and PDZ domain containing 3
67	NM_024754	PTCD2	Pentatricopeptide repeat domain 2
68	NM_001098511	KIF2A	Kinesin heavy chain member 2A
69	NM_025058	TRIM46	Tripartite motif-containing 46
70	AK126458.1	MYO15B	Myosin XVB pseudogene
71	NM_018659	CYTL1	Cytokine-like 1
72	NM_002965	S100A9	S100 calcium binding protein A9
73	NM_032566	SPINK7	Serine peptidase inhibitor, Kazal type 7 (putative)
74	NM_001669	ARSD	Arylsulfatase D
75	NM_206820	MYBPC1	Myosin binding protein C, slow type
76	NM_003200	TCF3	Transcription factor 3 (E2A immunoglobulin enhancer binding factors E12/E47)
77	NM_031413	CECR2	Cat eye syndrome chromosome region, candidate 2
78	NM_017539	DNAH3	Dynein, axonemal, heavy chain 3
79	NM_017426	NUP54	Nucleoporin 54 kDa
80	NM_002020	FLT4	Fms-related tyrosine kinase 4
81	NM_007320	RANBP3	RAN binding protein 3
82	NM_005286	NPBWR2	Neuropeptides B/W receptor 2
83	NM_006428	MRPL28	Mitochondrial ribosomal protein L28
84	NM_014280.2	DNAJC8	DnaJ (Hsp40) homolog, subfamily C, member 8
85	NM_020679	MIF4GD	MIF4G domain containing
86	NM_001823	CKB	Creatine kinase, brain
87	NM_000477	ALB	Albumin
88	NM_001927	DES	Desmin
89	NM_005416	SPRR3	Small proline-rich protein 3
90	NM_022468	MMP25	Matrix metallopeptidase 25
91	NM_016599	MYOZ2	Myozenin 2
92	NM_000243	MEFV	Mediterranean fever
93	NM_002272	KRT4	Keratin 4
94	NM_003279	TNNC2	Troponin C type 2 (fast)
95	NM_006685	SMR3B	Submaxillary gland androgen regulated protein 3 homolog B (mouse)
96	NM_014760	TATDN2	TatD DNase domain containing 2
97	NM_006928	SILV	Silver homolog (mouse)
98	NM_016522		Neurotrimin
99	NM_000760	CSF3R	Colony stimulating factor 3 receptor (granulocyte)
100	NM_003167	SULT2A1	Sulfotransferase family, cytosolic, 2A, dehydroepiandrosterone (DHEA)-preferring, member 1
101	NM_183360	DTNB	Dystrobrevin, β
102	NM_001711	BGN	Biglycan
103	NM_023077	C1orf163	Chromosome 1 open reading frame 163
104	NM_015926.4	TEX264	Testis expressed 264
105	NM_006757	TNNT3	Troponin T type 3 (skeletal, fast)
106	NM_002675	PML	Promyelocytic leukemia
107	XR_018113	GAPDH	Glyceraldehyde-3-phosphate dehydrogenase
108	NM_021245	MYOZ1	Myozenin 1
109	NM_000383	AIRE	Autoimmune regulator
110	NM_006846	SPINK5	Serine peptidase inhibitor, Kazal type 5
111	XM_939725	AP1S2	Adaptor-related protein complex 1, sigma 2 subunit
112	NM_024505	NOX5	NADPH oxidase, EF-hand calcium binding domain 5
113	NM_020145	SH3GLB2	SH3-domain GRB2-like endophilin B2
114	NM_016192	TMEFF2	Transmembrane protein with EGF-like and two follistatin-like domains 2
115	NM_006472	TXNIP	Thioredoxin interacting protein
116	NM_031866	FZD8	Frizzled homolog 8 (Drosophila)
117	NM_003808	TNFSF13	Tumor necrosis factor (ligand) superfamily, member 13
118	NM_015503	SH2B1	SH2B adaptor protein 1
119	NM_014047	C19orf53	Chromosome 19 open reading frame 53
120	NM_022754	SFXN1	Sideroflexin 1
121	NM_003061	SLIT1	Slit homolog 1 (Drosophila)
122	NM_003047	SLC9A1	Solute carrier family 9 (sodium/hydrogen exchanger), member 1 (antiporter, Na+/H+, amiloride sensitive)
123	NM_021991	JUP	Junction plakoglobin

On the other hand, the down-regulated genes included those associated with various metabolic pathways (CKM, ARSD and BGN), small molecule transport (ACCN3, ATP6V0A1 and SFXN1), signal transduction (FLT4, NPBWR2, CSF3R and FZD8), cell cycle regulation (TBC1D1, DLG5, GAS2L1, CDC34 and SH2B1), cell adhesion (RAPH1, CLDN10, BCAN, CD164 and JUP), transcription factors (LDB1, CLOCK, GCM1, BRF1, TCF3, PML and AIRE), cell-cell signaling (PGD, S100A9, CCL13 and RIMS1) or other functions.

Identification of candidate genes as molecular markers for the detection of circulating gastric cancer cells in human peripheral blood

Of the 53 genes that were up-regulated in the gastric cancer compared to the normal gastric tissues, we identified five candidate marker genes [tetraspanin 8 (TSPAN8), epithelial cell adhesion molecule (EPCAM), matrix metallopeptidase 12 (MMP12), matrix metallopeptidase 7 (MMP7) and regenerating islet-derived 3 α (REG3A)] for the detection of circulating gastric cancer cells in peripheral blood in accordance with the following criteria: i) no or weak expression in human normal tissues in the published database (20), ii) no expression in PBMCs from 4 healthy volunteers by nested RT-PCR. In addition to the above five newly identified genes, we analyzed histamine receptor H1 (HRH1) since a previous study reported that this gene was overexpressed in gastric cancer cells, and the expression of this gene satisfied the above criteria (21). Moreover, three candidate marker genes [keratin 20 (CK20), epidermal growth factor receptor (EGFR) and carcinoembryonic antigen (CEA)], which have been reported to be promising markers for the detection of cancer cells, were further analyzed (11,13,22,23).

Association of the expression of the nine marker genes for the detection of circulating gastric cancer cells with clinicopathological parameters by nested RT-PCR

We carried out semi-quantitative nested RT-PCR analysis of the nine candidate marker genes for the detection of circulating cancer cells using PBMCs of patients with gastric cancer. Of the nine candidate genes, the expression of MMP12 and CEA mRNAs was positive in 40% of the patients with gastric cancer. However, the expression of the other seven genes was positive in ≤30% of the patients, respectively (Table V). We then investigated a combined effect of the expression of the nine genes on the detection of circulating cancer cells. Expression of one or more genes out of the nine was detected in 80% of the patients with gastric cancer by nested RT-PCR (Table VI).

Table V.

Positive ratio of the nine marker genes for detection of circulating gastric cancer cells.

Marker genes	TSPAN8	EPCAM	HRH1	CK20	MMP12	MMP7	EGFR	REG3A	CEA
Positive cases (%)	20	30	30	20	40	10	10	20	40

Table VI.

Number of positive genes in 10 cases by nested RT-PCR.

Cases	GC-1	GC-2	GC-3	GC-4	GC-5	GC-6	GC-7	GC-8	GC-9	GC-10
No. of positive genes	1	0	2	3	5	6	1	2	0	2

We further investigated the association of the expression of the nine candidate marker genes with clinicopathological parameters of the 10 cases. We focused on four parameters: vascular invasion (v factor), lymphatic invasion (ly factor), lymph node metastasis (N factor) and pathological stage I–IV, and investigated the association of these parameters with the total number of positive genes in the PBMCs of each patient (Fig. 1). Of the four parameters, the numbers of genes expressed in the PBMCs were ≤2 in all of the vascular invasion-negative cases (v 0), while the numbers of genes were ≥2 in 5 of 6 positive cases (v 1–3), exhibiting a significant difference between the two groups (P=0.041; Fig. 1A). However, no significant association was observed for the other three parameters (Fig. 1B–D), suggesting that the combined expression analysis of the nine marker genes using PBMCs detected micrometastasis through vascular invasion in the primary gastric cancer tissues.

Figure 1.

Relationships between clinicopathological parameters and the expression of the set of nine marker genes in PBMCs of gastric cancer patients. Associations of the total number of marker genes which were expressed in the PBMCs with (A) vascular invasion (v factor), (B) lymphatic invasion (ly factor), (C) lymph node metastasis (N factor) and (D) pathological stage are shown. The score indicates the number of marker genes expressed in PBMCs of each patient. N.S., not significant (P>0.05). Student’s t-test was used for A and B, and Cochran-Armitage test was used for C and D.

Discussion

Microarrays, at present, are widely used to analyze expression of thousands of genes simultaneously in cancer tissues. In the present study, we identified five genes (TSPAN8, EPCAM, MMP12, MMP7 and REG3A) as potential markers for the detection of circulating cancer cells in the peripheral blood of patients with gastric cancer through genome-wide gene expression profiling in combination with nested RT-PCR. Some of these genes have previously been reported to be up-regulated in gastric cancer cells; however, they have not previously been designated for the detection of circulating gastric cancer cells by nested RT-PCR. Furthermore, the combined expression analysis of the five genes and four previously reported marker genes, HRH1, EGFR, CK20 and CEA, revealed that one or more mRNAs among the nine genes could be detected in 80% of the patients with gastric cancer by nested RT-PCR, suggesting that a set of nine marker genes is more sensitive than a single marker gene for detection of circulating gastric cancer cells. In this study, we did not investigate the association of distant metastasis with expression of the nine marker genes since no patients had distant metastasis among the 10 studied patients. Although we could not exclude false-positive cases due to non-malignant epithelial cells which may have contaminated the blood samples during collection and which may have expressed the targeted transcripts (18), pathological v factor showed a significant association with the total number of marker genes expressed in the PBMCs of the patients. Hence, the set of nine marker genes may be promising for the detection of minimal amounts of circulating gastric cancer cells prior to the metastatic growth of gastric cancer cells in organ(s).

Among the five marker genes which were newly identified in the microarray analysis, we identified epithelial cell adhesion molecule (EPCAM) which is a member of a family of type I membrane proteins and pan-epithelial differentiation antigen expressed in many types of carcinomas (24–28). Magnetic beads or structures coated with EPCAM monoclonal antibodies have been recently used for circulating cancer cell separation (29–31). Although we did not compare the accuracy of the detection of gastric cancer cells by these methods to that of nested RT-PCR since we did not conduct the former assays, 30% of patients with gastric cancer exhibited EPCAM-positivity in PBMCs by nested RT-PCR. Further clinical study investigating the relationship between the clinical outcome of patients and EPCAM expression in PBMCs by nested RT-PCR may clarify whether this method could be clinically applied for the detection of circulating gastric cancer cells. Two matrix metalloproteinases, MMP7 and MMP12, were among the five marker genes which were newly identified in this study. MMPs are a family of zinc-dependent proteolytic enzymes capable of cleaving extracellular matrix proteins, and the expression of MMPs in cancer tissue has been reported to be associated with the risk of metastasis (32–38). These two MMPs may play important roles in tumor invasion and the formation of metastasis in gastric cancer.

In conclusion, five novel marker genes were designated for the detection of circulating gastric cancer cells. The nested RT-PCR analysis for the set of nine marker genes, TSPAN8, EPCAM, MMP12, MMP7, REG3A, HRH1, EGFR, CK20 and CEA, using PBMCs of patients with gastric cancer may provide the potential for the detection of circulating gastric cancer cells prior to the formation of metastasis in other organs. Our data suggest that early detection and personalized therapy for gastric cancers, by prescribing the appropriate treatment to patients with a high risk of metastasis, may be achievable by utilizing specific sets of marker genes according to the approach shown here.