The present study was approved by the Institutional Review Board of Hunan Cancer Hospital and Affiliated Cancer Hospital of Xiangya Medical School (Changsha, China). All human materials were obtained with informed consent and approved by the Ethics Committee of the Hunan Cancer Hospital and Affiliated Cancer Hospital of Xiangya Medical School. Between January 2002 and December 2008, patients with ovarian metastasis of GC diagnosed pathologically at the Hunan Cancer Hospital were recruited. These human tissues were collected and examined with strict adherence to the protocol approved by this hospital. A total of 40 fresh GC tissues and corresponding ovarian metastatic tissue of gastric origin were collected from 40 female GC patients, with ages ranging between 43 and 80 years (median, 62.5 years), who had initially undergone resection at the Hunan Provincial Cancer Hospital (Changsha, China). The specimens included: Tumor tissue, adjacent normal tissue (at least 5 cm beyond the primary tumor margin), distant normal tissue and lymph nodes. The fresh tissue specimens were collected, washed with phosphate-buffered saline (PBS), cut into 1-mm³-sized pieces and maintained in Dulbecco's modified Eagle's medium with low glucose (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) containing 10% fetal bovine serum (FBS), penicillin (100 U/ml; Beijing Chemeebio Pharma-tech Co., Ltd., Beijing, China) and streptomycin (100 µg/ml; Beijing Chemeebio Pharma-tech Co., Ltd.). The tissues were subsequently incubated at 37°C in humid air with 5% CO₂. The medium was replaced every 3 days after the initial plating. When adherent fibroblast-like cells appeared after 10 days of culture, the attached cells were trypsinized and passaged (without dilution) into a new flask for further expansion. The cells of passage 4 were used for the experiments described. MSCs isolated from human bone marrow (hBM-MSCs) and the GC cell line BCG-823 from our institute were selected as controls and used for evaluation of the experimental results.

The growth curves of human gastric cancer-MSCs (hGC-MSCs), human gastric cancer ovarian metastatic tissue-MSCs (hGCOM-MSCs) and hBM-MSCs at passage 4 were seeded in 24-well plates (8,000 cells/well), followed by counting the number of cells per well on 14 successive days. The procedure was repeated three times and their growth curves were compared.

hGC-MSCs and hGCOM-MSCs (2.0×10⁶ cells) were trypsinized, washed twice with PBS and immunostained for 30 min on ice. A minimum of 5×10⁴ cells (in 100 µl PBS/0.5% BSA/2 mmol/l EDTA) were incubated with the following mouse anti-human monoclonal antibodies: CD73-FITC (561254), CD90-PE (555596), CD105-APC (561443), CD34-FITC (561819), CD45-APC-H7 (347463) and HLA-DR-FITC (555558; all from BD Biosciences, Franklin Lakes, NJ, USA). Labeled cells were analyzed using a flow cytometer (FACS Calibur; BD Biosciences).

All animal experiments were undertaken in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (National Institutes of Health, Bethesda, MD, USA), with the approval of the Hunan Provincial Cancer Hospital Review Board and Affiliated Cancer Hospital of Xiangya Medical School and also following Hunan province government Animal Care and Used Committee-approved protocols. A total of eight, 6-week-old female nonobese diabetic (NOD)/severe combined immunodeficiency mice (SCID; Laboratory Animal Center of Shanghai, Academy of Sciences, Shanghai, China) were housed in individually ventilated cages and maintained in a standard animal facility under controlled environmental conditions, at room temperature 26±2°C. BCG-823, hGC-MSCs or hGCOM-MSCs cells (5×10⁶) in 200 µl PBS were injected subcutaneously into the lower right flank of mice. The incidence of tumor formation was observed over 4 weeks.

hGC-MSCs, hGCOM-MSCs or hBM-MSCs were seeded at 5,000 cells/cm² in 35 mm diameter plates and cultured in DMEM with 10% FBS, with or without osteogenic supplements (0.1 nM dexamethasone, 10 mM β-glycerophosphate, 50 mg/l ascorbic acid and 4 µg/ml basic fibroblast growth factor; Sigma-Aldrich, St. Louis, MO, USA). The medium was changed three times each week and the cells were induced for 3 weeks. At the end of induction, the cells were subjected to alkaline phosphatase (ALP) staining followed by hematoxylin counterstaining (Zhingshan Golden Bridge, Beijing, China).

Total cellular RNA was isolated from hGC-MSCs and hGCOM-MSCs using TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.), according to the manufacturer's instructions. A total of 1.0 µg RNA was processed for cDNA synthesis with Superscript II reverse transcriptase using Oligo-dT orimers (Toyobo Co., Ltd., Osaka, Japan). PCR was performed using 1 µg of the cDNA sample with 0.3 U Taq polymerase (CinnaGen Co., Tehran, Iran) and 200 µM dNTPs, 10 pM of each primer, reaction buffer, and MgCl₂ (Takara Bio Inc., Otsu, Japan). A total of 1 µl of this mixture was used as a template for PCR to assess mRNA expression of ALP, dentin matrix acidic phosphoprotein 1 (DMP-1), osteocalcin (OSC) and osteopontin (OPN) using the ABI PRISM 7700 instrument (Applied Biosystems; Thermo Fisher Scientific, Inc., Foster City, CA, USA) with gene-specific primers [Table I, and described previously (23,24)] and the SYBR Green I (Thermo Fisher Scientific, Inc.) protocol. The PCR amplification was performed for 35 cycles, with the cycling conditions as follows: 94°C for 30 sec, 60°C (primer) for 30 sec, 72°C for 30 sec, with a final extension at 72°C for 10 min. PCR products were separated on a 1% agarose gel, stained with ethidium bromide and visualized under UV light. Relative expression ratios normalized to that of β-actin (Actb) were calculated.

All results are expressed as the mean ± standard deviation or median (range). The normality of data distribution was assessed by the Kolmogorov-Smirnov test. Data analysis was performed using SPSS version 16.0 (SPSS, Inc., Chicago, IL, USA). Receiver operating characteristic (ROC) curves were generated to determine the areas under the ROC curves and their 95% confidence intervals were calculated. A univariate test was used to examine the effect of each clinical variable on survival. Student's t-test was applied to determine statistical significance. P<0.05 was considered to indicate a statistically significant difference.

The first plastic-adherent cells were detected 1 week after tissue preparation. Primary cell cultures began to form colonies after 10 days of primary culture and reached a confluence of 90–95% at day 30 (Fig. 1A). Cells from human GC tissues and its ovarian metastatic tissues were seeded at low density on a 6-well plate and allowed to grow until 80% confluence. Cells were passaged at the same cell density.

The growth curves of hGCOM-MSCs, hGC-MSCs and hBM-MSCs are shown in Fig. 1B. Within 5 days after plating, the number of hGCOM-MSCs increased by almost two times that of hGC-MSCs and hBM-MSCs. In addition, hGCOM-MSCs demonstrated a higher cumulative population doubling level compared with hGC-MSCs and hBM-MSCs.

Specific cell-surface markers were selected to evaluate whether the hGC-MSCs and hGCOM-MSCs at passage 4 contained the MSC population. These experiments were performed with the objective of isolating a population of pluripotent-like stem cells derived from human GC and its ovarian metastatic deposits. The present study revealed that the two types of cells were CD73⁺, CD90⁺, CD105⁺, CD34⁻, CD45⁻ and HLA-DR⁻, which indicated characteristics of MSCs (6,25,26). The hGC-MSCs at passage 4 expressed high levels of CD73 (78.58%), CD90 (74.55%) and CD105 (77.15%). However, they were negative for the hematopoietic marker CD34 (0.74%), CD45 (0.18%) and HLA-DR (0.20%). The hGCOM-MSCs at passage 4 also expressed high levels of CD73 (83.24%), CD90 (76.32%) and CD105 (82.64%), but were negative for the hematopoietic markers CD34 (0.87%), CD45 (0.66%) and HLA-DR (0.53%), indicating the mesenchymal lineage of these cells (Fig. 2). Taken together, these results demonstrated that cells isolated from human GC tissues and their ovarian metastatic tissues possess the characteristics of MSC with regards to the expression of markers. RT-PCR results demonstrated that stem cell-associated genes, including Snail, Bmi-1, ABCG2 and Nanog, and mesenchymal lineage-associated genes, including α-SMA, CD73, CD44 and β-actin, were all expressed in hBM-MSCs, hGC-MSCs and hGCOM-MSCs (Fig. 3).

hGC-MSCs and hGCMO-MSCs failed to form tumors in NOD/SCID mice after >2 weeks (Fig. 4). However, GC cells (BGC-823) used as xenograft controls were able to form tumors 2 weeks after transplantation. Tumor formation was followed for 4 weeks.

In addition to their colony-forming ability and the expression of specific antigens on their cell surfaces, the capacity for tri-lineage differentiation is a key property of MSCs (27). The differentiation potential was evaluated by culturing the cells in osteogenic media. Long-term cultures (3 weeks) of cells grown in the presence of osteogenic media demonstrated the capacity to form Alizarin Red-positive condensed nodules with high levels of calcium covering the entire wells. The deposits were sparsely scattered throughout the adherent layer as single mineralized zones (Fig. 5A).

In addition, the expression levels of ALP, OPN, OSC and DMP-1 genes, considered to be osteo-specific genes, were analyzed at the mRNA level after 3 weeks of osteoinduction. The expression of these mRNAs (relative to Actb) in differentiated hGC-MSCs and hGCOM-MSCs was higher than that in undifferentiated hGC-MSCs and hGCOM-MSCs (Fig. 5B). These results indicate that hGC-MSCs and hGCOM-MSCs cultured in osteogenic media can differentiate into osteoblast-like cells.