Twenty GC tissues and paired normal gastric tissues were collected by endoscopic biopsy at Chonnam National University Hwasun Hospital (Jeonnam, Korea) for use in the preparation of mRNA and protein. For immunohistochemistry, tumor specimens were collected from 149 consecutive patients who underwent surgery between January 1998 and December 1999. None of the patients had received preoperative radiotherapy or chemotherapy. All underwent primary tumor resection with regional lymph node dissection. Formalin-fixed and paraffin-embedded tissue blocks were selected by viewing the original pathologic slides and choosing blocks that showed the junction between normal gastric epithelium and the tumor. The histologic grade was classified according to previously established criteria (21,22). Tumor staging used the American Joint Committee on Cancer (AJCC) staging system (23). Patient characteristics including gender, age at time of surgery, tumor size, stage, survival and follow-up information were obtained from hospital records and, when necessary, by contact with attending pathologists and physicians. The observation time was the interval between the time of surgery and last contact (death or last follow-up). The study group comprised 101 males and 48 females, with a mean ± standard deviation (SD) age of 58.0±11.1 years (range, 25–83 years). The mean ± SD size of the tumors was 4.5±2.9 cm (range, 0.2–20.0 cm). The mean follow-up period was 100.1 months (range, 0–176.4 months). All specimens were collected following the informed consent of patients. The study was approved by the Ethics Committee of Chonnam National University Hwasun Hospital.

Cell lines derived from AGS and SNU638 human GC cells were obtained from the American Type Culture Collection (Manassas, VA, USA). Cells were maintained in RPMI-1640 supplemented with 10% fetal bovine serum (both from Hyclone, Logan, UT, USA) and 1% penicillin-streptomycin (Sigma-Aldrich, St. Louis, MO, USA) in 5% CO₂ at 37°C. Gene knockdown was performed using the specific siRNA. Livin siRNA (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) and scramble siRNA (Qiagen, Valencia, CA, USA) were transfected with Lipofectamine™ RNAiMAX (Invitrogen, Carlsbad, CA, USA) for 48 h according to the manufacturer’s recommendations.

RNA isolation was performed using TRIzol reagent (Invitrogen) following the instructions provided by the manufacturer. Reverse transcription was carried out using 1 μg of RNA and MMLV transcription reagents (Invitrogen) according to the manufacturer’s recommendations. The amplification of specific DNA was performed with Taq polymerase and specific primers for Livin (5′-CACACAGGCCATCAGGACAAG-3′/5′-ACGGCACAAAGACGATGGAC-3′) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (5′-ACCACAGTC CATGCCATCAC-3′/5′-TCCACCACCCTGTTGCTGTA-3′, as an internal control).

Total cell extracts were prepared using Pierce^® RIPA buffer (Thermo Scientific, Rockford, IL, USA) with Halt™ protease and phosphatase inhibitor cocktail (Thermo Scientific). Total cell extracts were separated on polyacrylamide gels and then transferred to polyvinylidene fluoride membranes (Millipore, Billerica, MA, USA). The specific proteins were blotted with the primary antibody to Livin, Survivin and β-tubulin (Santa Cruz Biotechnology, Inc.); extracellular signal-regulated kinase (ERK), phospho-ERK, p38, phospho-p38, c-Jun NH2-terminal kinase (JNK), phospho-JNK, cleaved caspase-3, -7, -9, cleaved poly(ADP-ribose) polymerase (PARP), cyclin-dependent kinase 4 (CDK4), CDK6, cyclin D1, cyclin D3, cyclin B1, p21, p27, p57, p15 and p16 (Cell Signaling Technology, Danvers, MA, USA). The membranes were developed using the enhanced chemiluminescence detection system, horseradish peroxidase substrate (Millipore) and a model LS-4000 luminescent image analyzer (FujiFilm, Tokyo, Japan).

The invasive ability was calculated as the number of cells passing through the gelatin-coated Transwell filter chambers (Corning, NY, USA). Viable cells (2×10⁵) in 0.2% bovine serum albumin (BSA) were seeded in the upper chambers of Transwell units. Human plasma fibronectin (Calbiochem, La Jolla, CA, USA) as a chemoattractant was added to 0.2% BSA located in the lower chambers of the units. After 24 h of incubation in a 5% CO₂ humidified incubator, the cells on the upper surface of each filter were carefully removed with a cotton swab, and cells that had traversed the filter to invade the opposite surface of the filter were stained with Diff-Quik (Sysmex, Kobe, Japan). The number of invaded cells was determined in five random fields using light microscopy, and the mean value was calculated from data obtained from three separate chambers.

Cell migration was determined using the Culture-Inserts (2×0.22 cm²; Ibidi, Regensburg, Germany). To create a wound gap, cells were seeded on the Culture-Inserts, which were gently removed using sterile tweezers following a 24-h incubation. The progression of wound closure was photographed using an inverted microscope. The distance between gaps was normalized to 1 cm after capture of three random sites.

Cell viability was determined by the EZ-CyTox (tetrazolium salts, WST-1) cell viability assay kit (Daeil Lab Service Co., Seoul, Korea). After application of WST-1 reagent at 37°C at determined times, cell viability was measured using an Infinite M200 microplate reader with Magellan V6 data analysis software (both from Tecan, Grödig, Austria). All assays were performed three times in sets of three replicate wells.

For Annexin V staining, live cells were washed in phosphate-buffered saline (PBS) and then incubated with Annexin V fluorescein isothiocyanate (FITC; R&D Systems, Minneapolis, MN, USA). For cell cycle analysis, cells were incubated in 10 μg/ml ribonuclease A (Sigma-Aldrich) and 50 μg/ml propidium iodide (PI) at room temperature in the dark. BD Cell Quest^® v3.3 (Becton-Dickinson, San Jose, CA, USA) and WinMDI v2.9 (The Scripps Research Institute, San Diego, CA, USA) were used to analyze the population of Annexin V-positive cells and sub-G1 phase.

Paraffin tissue sections from patients were rehydrated through descending ethanol and were retrieved with citrate buffer (pH 6.0). Thereafter, endogenous peroxidase activity was quenched using Peroxidase-Blocking Solution (Dako, Carpinteria, CA, USA) and the tissues were incubated with polyclonal rabbit anti-human Livin in primary Diluent Solution (Invitrogen) overnight at 4°C. After washing, antibody binding was visualized using a Dako Real™ Envision horseradish peroxidase/3,3′-diaminobenzidine detection system (Dako). Stained tissues were photographed using a light microscope.

Assessment of immunostained specimens was performed independently by two observers without knowledge of the clinicopathological data. In the event of a discrepancy, a consensus was reached after further evaluation. The intensity of positive cancer cells was graded on a 4-point scale: 0, no staining of cancer cells; 1, weak staining; 2, moderate staining; and 3, strong staining. The percentage of staining of cancer cells was rated on a 4-grade scale: 0, none; 1, <10%; 2, 10–50%; 3, >50%. The intensity rating was multiplied by the percent staining rating to obtain an overall score. The mean overall score for 149 tumors analyzed was 4.0. This score was chosen as the cut-off point for discrimination of Livin expression (>4, positive expression; ≤4, negative expression).

For comparison of intergroups, data were derived from at least three independent experiments. The data are presented as means ± SD, and the Student’s t-test was used to determine statistical significance. The χ² test and Fisher’s exact test, where appropriate, were used to compare expression of Livin with various clinicopathological parameters. Actuarial survival rates of patients with positive or negative Livin expression were evaluated according to the Kaplan-Meier method and the differences were tested with a log-rank test. The statistical software program used was Statistical Package for the Social Sciences (SPSS/PC+ 15.0; SPSS, Chicago, IL, USA). A P-value <0.05 indicated a statistically significant difference.

We investigated the biologic roles of the Livin gene on oncogenic behavior using siRNA in AGS and SNU638 human GC cell lines. Livin gene expression consistently showed a specific reduction at the mRNA and protein levels in cells transfected with Livin siRNA (Fig. 1). To evaluate whether blocking Livin gene expression affects the oncogenic behavior of GC cells, cell migration, invasion and proliferation were assayed. The artificial wound gap in plates of the scramble siRNA-transfected AGS cells was significantly narrower than that of the Livin siRNA-transfected AGS cells at 12 h (P=0.049) (Fig. 2A). Transfection of Livin siRNA inhibited AGS and SNU638 cell invasion from 1145.8±562.5 and 799.2±243.3 invaded cells/field (scramble siRNA) to 286.0±175.8 and 119.2±58.7 invaded cells/field (P=0.017 and P=0.002, respectively) (Fig. 2B). Significant decreases in cell proliferation were observed after 72 h in the Livin siRNA-transfected AGS cells (P=0.022) and SNU638 cells (P=0.009), when compared to the scramble siRNA-transfected cells (Fig. 2C).

Apoptosis induced by transfection of siRNA was assayed using flow cytometry. The cell apoptotic rate induced by transfection of Livin siRNA was significantly increased, compared with that induced by transfection of the scramble siRNA (19.8 vs. 31.4%) in AGS cells, but Livin knockdown had a minimal influence on SNU638 cell apoptosis (15.7 vs. 21.7%) (Fig. 3A). Next, we investigated the activation of caspases, which are critical mediators of apoptosis. Expression of cleaved caspase-3, and -7 and PARP was upregulated in the AGS and SNU638 cells after transfection with Livin siRNA. The protein level of Survivin was reduced by transfection of Livin siRNA in AGS and SNU638 cells (Fig. 3B).

Flow cytometry was used to detect whether blocking of Livin gene expression alters cell cycle distribution. Transfection of Livin siRNA resulted in cell cycle arrest in the G0/G1 phase of AGS and SNU638 cells (Fig. 4A). Next, the effects of Livin on various CDK inhibitors (CDKIs), cyclins and CDKs, involved in cell cycle progression were assessed. The cyclin D1, CDK4 and CDK6 protein levels were significantly decreased following transfection of Livin siRNA in AGS and SNU638 cells, the p21 and p27 protein levels were significantly increased, and the p57, p15 and p16 protein levels were not altered in response to Livin knockdown (Fig. 4B).

The effect of Livin on stimulation of MAPK signaling pathways leading to apoptosis and cell cycle arrest in AGS and SNU638 cells was investigated. The phosphorylation levels of ERK1/2 and JNK were downregulated in Livin siRNA-transfected AGS and SNU638 cells. The phosphorylation level of p38 was not altered by transfection with Livin siRNA (Fig. 5).

To confirm the results of the GC cell line studies, the expression of Livin at the mRNA and protein levels was evaluated by RT-PCR and Western blotting in human GC tissues, paired normal gastric mucosa, and metastatic or non-metastatic lymph node tissues of the same patients acquired by endoscopic biopsy and from surgical specimens. In endoscopic biopsy specimens, Livin expression was upregulated in cancer tissues when compared to that in the paired normal mucosa at the mRNA and protein levels (P=0.006 and P=0.040, respectively) (Fig. 6). In surgical specimens, immunohistochemical staining of Livin protein was undetected or only weakly stained in the normal gastric mucosa. Immunohistochemical staining of the GC specimens localized Livin expression to the cancer cells, with no expression evident in the stromal compartment of the cancers (Fig. 7A). Immunohistochemical staining of Livin in metastatic lymph node tissues was significantly stronger than that in non-metastatic lymph node tissues (Fig. 7B). The score for immunohistochemical staining of Livin in metastatic lymph node tissues was significantly higher than that in non-metastatic lymph node tissues (P<0.001) (Fig. 7C).

To study the prognostic role of Livin in GC progression, we investigated the association between expression of the Livin protein immunohistochemically in formalin-fixed, paraffin-embedded tissue blocks obtained from 149 GC patients and the clinicopathological data, including survival. Expression of Livin protein was detected in 56 of the 149 (37.6%) GCs analyzed (Table I). The correlation between Livin expression and clinicopathological parameters is summarized in Table I. No significant correlation was found between Livin expression and various clinicopathological parameters including age, gender, tumor size, Lauren classification, histologic grade, depth of invasion, lymph node metastasis, distant metastasis, or stage. Analysis of the survival for all patients showed that Livin expression did not correlate with survival (P=0.144) (Fig. 8).