The human GC cell lines, SGC-7901, AGS, BGC-823 and MKN-45, and human umbilical vein endothelial cells (HUVECs) were purchased from the Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences (Shanghai, China). All cells were cultured at 37°C in 5% CO₂ and at saturation humidity in RPMI-1640 medium containing 10% fetal bovine serum.

Gastric tumor and adjacent non-tumorous tissues (≥3 cm from tumor) were obtained from 25 patients with GC who underwent curative surgery (D2 radical resection) at the Second Affiliated Hospital of Soochow University (Suzhou, China) between January 2014 and December 2015. Patients that had undergone radiotherapy/chemotherapy prior to surgery were excluded from the study. These tissues were used for reverse transcription polymerase chain reaction (RT-PCR). None of the patients had received radiotherapy or chemotherapy prior to surgery. This study was approved by the Ethics Committee of the Second Affiliated Hospital of Soochow University, and all patients were fully informed of the experimental procedures.

Total RNA was extracted and isolated from tissue samples using TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA), according to the manufacturer's protocol. Reverse transcription of RNA was performed using a reverse transcription kit (Promega, Madison, WI, USA).

In brief, 1 µg of total RNA from each sample was reverse transcribed following the manufacturer's protocol. SYBR Green reagent (Applied Biosystems; Thermo Fisher Scientific, Inc.) was used for RT-qPCR to analyze mRNA expression. The PCR primers were designed according to the human PBX3 and GAPDH cDNA sequences in GenBank, as follows: PBX3 forward, 5′-GAGCTGGCCAAGAAATGCAG-3′ and reverse, 5′-GGGCGAATTGGTCTGGTTG-3′; and GAPDH forward, 5′-GGACCTGACCTGCCGTCTAG-3′ and reverse, 5′-GTAGCCCAGGATGCCCTTGA-3′. GAPDH acted as the constitutive control. Relative expression ratios of PBX3 in each paired tumor to non-tumor tissue sample were calculated using the 2-ΔΔCq method (20).

The pCMV6-PBX3 plasmid was purchased from Origene (Origene Technologies, Inc., Rockville, MD, USA). The pCMV6-PBX3 or pCMV6-vector was then transfected into SGC-7901 cells using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.), in accordance with the manufacturer's protocol.

For depletion of endogenous PBX3 expression, short hairpin RNA (shRNA) targeting 592–613 bp was inserted into a lentivirus vector as previously described (6). Lentiviral vectors were then transfected into MKN-45 GC cells. Stably transfected cells were selected by treatment with 5 µg/ml blasticidin and were used for identification and additional research.

GC cells were harvested and lysed using radioimmunoprecipitation assay buffer (Solarbio, Beijing, China) containing 1% phenylmethylsulfonyl fluoride protease inhibitors. A bicinchoninic acid assay kit (Pierce Biotechnology, Inc., Rockford, IL, USA) was used to measure the total protein concentration. Equivalent amounts of protein (50 µg) were separated by 10% SDS-PAGE, and the resolved proteins were transferred to polyvinylidene difluoride membranes. The membranes were blocked with 5% skim milk for 2 h and then incubated with primary antibodies overnight at 4°C. Primary antibodies were as follows: PBX3 (polyclonal rabbit; cat. no. 12571-1-AP; dilution, 1:500; Proteintech Group, Inc., Rosemont, IL, USA); E-cadherin (monoclonal rabbit; cat. no. 3195; dilution, 1:500; Cell Signaling Technology, Inc., Danvers, MA, USA); N-cadherin (monoclonal rabbit; cat. no. 13116; dilution, 1:500; Cell Signaling Technology, Inc.); vimentin (monoclonal rabbit; cat. no. 5741; dilution, 1:1,000; Cell Signaling Technology, Inc.); p-AKT (Ser473) (monoclonal rabbit; cat. no. 4060; dilution, 1:500; Cell Signaling Technology, Inc.); total AKT (monoclonal rabbit; cat. no. 4691; dilution, 1:500; Cell Signaling Technology, Inc.) and GAPDH (monoclonal mouse; cat. no. ab8245; dilution, 1:10,000; Abcam, Cambridge, UK). Membranes were then incubated with secondary antibody for 2 h at room temperature and were visualized using an enhanced chemiluminescence detection system (GE Healthcare Life Sciences, Chalfont, UK) in accordance with the manufacturer's protocol.

For cell migration and invasion assays, a total number of 1×105 GC cells (SGC-7901-PBX3, SGC-7901-NC, MKN-45-NC and MKN-45-PBX3/sh) were suspended in serum-free RPMI-1640 medium (HUVEC cells were suspended in supernatant from GC cells) and plated in Transwell chambers (8 µm for 24-well plate; Costar; Corning Incorporated, Corning, NY, USA) with or without Matrigel (BD Biosciences, San Jose, CA, USA), according to the manufacturer's protocol. For each assay, medium containing 10% FBS was added to the lower chamber as a chemoattractant. Subsequent to 24 h culture, the cells were fixed by 10% formalin and stained by 0.5% crystal violet. Finally, images of cells in the lower chamber were captured and the number of cells was counted by inverted microscopy (×200 magnification).

HUVEC cells were cultured in tumor supernatant and plated in a 96-well plate with Matrigel (BD Biosciences) at a concentration 3×104 cells/well. Tumor supernatant was collected from PBX3-overexpressing groups and PBX3-silenced groups subsequent to 24 h culture in RPMI-1640 medium. Following 12 h incubation at 37°C in a 5% CO₂ atmosphere, tubes were photographed by microscopy and evaluated by Image Pro Plus software (Media Cybernetics, Inc., Rockville, MD, USA).

To examine matrix metallopeptidase (MMP)-9 activity, gelatin zymography was performed using 10% SDS-PAGE gels containing 1 mg/ml gelatin (Sigma-Aldrich; Merck Millipore, Darmstadt, Germany). In brief, GC cells were cultured in serum-free RPMI-1640 medium for 24 h and the supernatant was then collected and centrifuged at 201 × g for 5 min. The gels were washed twice in renaturation buffer (2.5% Triton X-100) for 30 min each time to remove SDS subsequent to electrophoresis and then incubated at 37°C for 24 h in a reaction buffer consisting of 50 mM Tris-HCl (pH 7.5), 5 mM CaCl2 and 150 mM NaCl. The gels were then stained with 0.5% Coomassie brilliant blue R-250 (Sigma-Aldrich; Merck Millipore) for 2 h and destained in buffer (30% methanol; 10% acetic acid). Clear transparent bands in the background of blue staining represented gelatinase activity.

Four-week-old male BALB/c nude mice (n=10; Institute of Zoology, Chinese Academy of Sciences, Beijing, China) were used to evaluate the role of PBX3 in peritoneal spreading in vivo. Nude mice were housed at a specific pathogen-free environment in the Animal Laboratory Unit, School of Medicine, Soochow University. Mice received humane care and the study protocols were performed according to a protocol approved by the Institutional Animal Care and Use Committee (IACUC) at Soochow University. A total number of 2×106 SGC-7901-PBX3 or SGC-7901-NC cells were injected into the abdominal cavity of the 5 mice of the PBX3 and NC groups, respectively. Mice were euthanized on the 30th day subsequent to injection, and the abdominal masses were imaged and photographed. All experiments were performed in accordance with the official recommendations of the Chinese Animal Community at Soochow University School of Medicine (Suzhou, China).

Student's t-test was used to examine the statistical differences between the two groups. Data are shown as mean ± standard deviation. Statistical analyses were performed using SPSS 19.0 software (IBM, Armonk, NY, USA). P<0.05 was considered to indicate a statistically significant difference.

The levels of PBX3 in GC and adjacent non-tumor tissues were examined by RT-qPCR assay. The results showed that the PBX3 mRNA level was markedly increased in GC tissues compared with the corresponding non-tumor tissues (Fig. 1A; P<0.01). This finding suggests that PBX3 may act as an oncogenic gene in GC. Notably, a previous study showed that PBX3 expression was associated with poor tumor differentiation, invasion depth and clinical stage in GC (16).

Western blot analysis was then used to examine the expression of PBX3 at the protein level in GC cell lines. It was found that the expression levels were different in different GC cell lines (Fig. 1B). The results indicated that MKN-45 GC cells inherently express high levels of PBX3, whereas SGC-7901 GC cells express low levels of PBX3. Overexpression of PBX3 in SGC-7901 and attenuation of PBX3 expression in MKN-45 GC cells were confirmed at the protein level by western blot analysis (Fig. 1C). Stably transfected cells were used for further experiments.

Cancer metastasis is the leading cause of cancer-associated mortality and it has been challenging to study due to a series of rare, stochastic events involved in cancer metastasis (21). Therefore, the present study examined the effects of PBX3 on GC mobility by Transwell assay. It was observed that the cell migration rate and invasion rate were each increased in PBX3 overexpressing groups compared with their respective control groups (Fig. 2A-C; P<0.01). The opposing results were obtained subsequent to silencing of PBX3 in MKN-45 cells (Fig. 2A-C; P<0.01). This finding suggests that the mobility of GC cells changed with the expression of PBX3.

It is well known that tumor angiogenesis is a key role in cancer progression. To examine the effects of PBX3 on tubular formation of HUVEC cells, tumor supernatant was collected from GC cells to suspend HUVEC cells. In total, 3×104 HUVEC cells were plated onto a 96-well plate that was coated with Matrigel. The 96-well plate was then incubated for 12 h at 37°C in 5% CO₂. Following 12 h incubation, increased tubule forming ability was observed in PBX3-overexpressing groups compared with the control groups (Fig. 3A), such as number of tubules (Fig. 3B), and number of intersections (Fig. 3C). Converse results were achieved in PBX3-knockdown groups compared with the control groups (Fig. 3A-C).

Therefore, the effect of PBX3-overexpressing GC cells on the migration of HUVEC cells. It was observed that supernatant from PBX3-overexpressing GC cells could stimulate the migration of HUVEC cells. The mechanism of the promotion of HUVEC cell migration by PBX3-overexpressing GC cells remains unclear.

EMT is known to play important roles in the progression of tumors, particularly in cancer metastasis. It was found that overexpression of PBX3 elevated the migration and invasion rates of GC cells. EMT has an unelucidated effect on these processes. To clarify this, the EMT-associated proteins in GC cells were examined by western blot analysis. It was observed that the N-cadherin and vimentin protein levels were increased, while E-cadherin was decreased in PBX3-overexpressing GC cells compared with their respective control cells (Fig. 4A and B). Converse findings were observed after PBX3 was knocked down in MKN-45 GC cells (Fig. 4A and B).

Furthermore, AKT signaling is strongly associated with EMT, invasion and metastasis in a variety of human tumors. Therefore, AKT activity was evaluated in PBX3-overexpressing GC cells. Western blot analysis showed that the level of p-AKT (ser473) was increased in PBX3-overexpressing GC cells, whereas opposing results were observed in PBX3-silencing GC cells (Fig. 4A and C). Additionally, it was observed that overexpression of PBX3 elevated the activity of MMP-9 in GC cells by gelatin zymography assay (Fig. 4D and E). However knockdown of PBX3 made no significant change in the activity of MMP-9 in GC cells. These findings suggest that PBX3-induced EMT may be via the AKT signaling pathway in GC cells.

To examine the effects of PBX3 on peritoneal spreading and metastasis in vivo, SGC-7901-PBX3 and SGC-7901-NC GC cells were injected into the abdomens of nude mice. Extensive peritoneal spreading was observed in the PBX3-overexpressing group compared with the control group (Fig. 5A). There were significantly more visible peritoneal nodules in the PBX3-overexpressing group than in the control group (4.600±0.748 vs. 2.200±0.374; P<0.05; Fig. 5B). Thus overexpression of PBX3 promotes GC invasion and metastasis in vivo as well as in vitro.