The function and mechanism of COX-2 in angiogenesis of gastric cancer cells

Human gastric cancer cell line SGC7901 was cultivated in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal calf serum, penicillin (100 U/ml) and streptomycin (100 μg/ml), in a CO₂ incubator (Forma Scientific) [5]. Human umbilical vein endothelial cells (HUVEC-12; ATCC, Manassas, VA) were grown in Kaighn's modification of Ham's F12 medium (ATCC) with 2 mM Lglutamine, 1.5 g/l sodium bicarbonate, 0.1 mg/ml heparin, 0.03 mg/ml endothelial cell growth supplement and 10% FBS.

The siRNA oligos for COX-2 were designed according to previous report. Target sequences were aligned to the human genome database in a BLAST search to ensure that the choosing sequences were not highly homologous with other genes. For oligo-1, S: 5'-tttgcatcgatgtcaccatagaacatctatggtgacatcgatgcttttt-3', AS: 5'-ctagaaaaagcatcgatgtcacc atagatgttctatggtgacatcgatg-3' For annealing to form DNA duplexes, 100 μM of each S and AS oligos was used. The duplexes were diluted and then ligated with mU6pro vector which previously digested by the Bbs I/Xba I restriction enzyme and gel purified at room temperature for 30 min. The products were transformed into DH5α competent cells. Ampicillin-resistant colonies were chosen, identified by restriction digestion and further confirmed by DNA sequencing.

SGC7901 cells were planted in six-well plates and cultured in drug-free medium. At 90-95% confluence, cells were washed twice with PBS, grew in 2 ml of DMEM without antibiotics. Using Lipofectamine™ 2000 reagent (Invitrogen, Inc. Carlsbad CA), 2 μg of mU6pro-COX-2siRNA plasmids were transfected into cells according to the manufacturer's instructions. The cells transfected with mU6pro vector alone were served as negative control. Forty-eight hours later, cells were placed in growth medium containing G418 (GIBCO) for clone selection. The expression levels of COX-2 in G418-resistant clones were evaluated by western blot analysis.

All of the PCR products were separated on ethidium bromide stained agarose, and visualized with UV as described previously [6].

The western blot was done as described previously. In brief, total cellular proteins were prepared and then quantified by Bradford method [7]. A measure of 80 ug of lysates were electrophoresed in 12% SDS-PAGE and blotted on a nitrocellulose membrane (Immoblin-P, Millipore, Bedford, MA, USA). Membranes were blocked with 5% fat-free milk powder at room temperature and incubated overnight with antibody at 4°C. After three washes for 15 min in PBS-T, the membrane was incubated with the HRP-conjugated goat anti-mouse IgG antibody (Wuhan, Hubei, China) for 1 h at room temperature. The enhanced chemiluminescence (Amersham Life Science, Piscataway, NJ, USA) was added and monitored for the development of color.

Cells were seeded on a 96-well plate at 3 × 10⁴ cells/well. Each sample had four replicates. The medium was replaced at 2-day intervals. Viable cells were counted by the 3-[4,5-dimethylthiazol-2-yl]- 2,5-diphenyltetrazolium bromide (MTT) assay after 2, 4, 6, and 8 days.

Female athymic nu/nu mice, 5-6 weeks of age, were obtained from FMMU Experimental Animal Co. (Shaanxi, China) and housed in a pathogen-free facility for all of the experiments. The logarithmically growing cells were trypsinized and resuspended in D'Hanks solution, and 5 × 10⁶ cells in 0.2 ml were injected subcutaneously into the left flank of mice [8]. Experimental and control groups had at least 6 mice each. Tumors were measured twice weekly with microcalipers, and the tumor volume was calculated according to the formula: volume = length × (width²)/2.

Tumor microvessel densities (MVD) were quantified by anti-CD31 immunohistochemistry. Briefly, tumor sections from nude mice were cut using a LEICA cryostat and the paraffin sections were mounted on positively charged Superfrost slides and dried overnight. The immunostaining was done according to standardized protocols.

Tube formation assay was performed as described previously (Chia et al, 2010). Briefly, Confluent HUVEC cells were harvested and diluted in DMEM with 10% FBS, which were then seeded on Matrigel-coated 24-well plates. Cell culture medium was then replaced by conditioned medium. After 16 h, Matrigel was fixed, stained with H & E and examined under inverted microscope. The mean tube length in five random fields per well was quantified by computer software.

Briefly, confluent monolayer of HUVEC was cultured with non-growth factor containing media for 12 h before harvesting. Harvested cells were suspended in serum-free DMEM199 and HUVEC cells were seeded onto tissue culture inserts in triplicate. The inserts were removed after 8 h culture and washed with PBS. Non-migrated cells on the upper surface of the inserts were removed by wiping with cotton swabs. The inserts were fixed in neutral buffered formalin solution, stained with hematoxylin and eosin (H & E) and mounted on microscope slides. HUVEC migration was quantitated by counting the number of cells in three random fields (!200) per insert.

The gene expression was compared between SGC7901-siRNA and SGC7901-vector cells for three times [9]. RNA was extracted from 80-90% confluent cells using Trizol and purified with RNeasy spin columns (Qiagen, Valencia, CA) according to the manufacturers' instructions. Quality of the RNA was ensured before labeling by analyzing 20 to 50 ng of each sample using the RNA 6000 NanoAssay and a Bioanalyzer 2100 (Agilent, Palo Alto, CA). Samples with a peak ratio of 1.8 to 2.0 were considered suitable for labeling. Cy3- or Cy5-labeled cDNA was generated and the Cy3/Cy5 single-stranded cDNA/cot1 DNA pellet was resuspended in hybridization buffer, then the hybridization mix was applied to GEArray Q Series Human Angiogenesis Gene Array. The ratios of gene expression were considered to be significant if they were 2 or 0.5 in at least two independent experiments. Genes were assigned to functional families based on information from LocusLink and PubMed.

Data were presented as mean ± standard deviation (S.D.) unless otherwise specified. Comparisons between groups were made using the Student-Newman-Keuls test or the Kruskal-Wallis test. All data were analyzed using the SPSS software package (SPSS Inc, Chicago, USA). A value of P < 0.05 was considered significant.

As Figure1 showed, SGC7901 cells were transfected and then one resistant clone (SGC7901-siRNA) with significantly decreased COX-2 expression and one vector transfected control clone (SGC7901-vector) were selected. The results of MTT assay showed that down-regulation of COX-2 might significantly decrease the proliferation of SGC7901 cells (Figure2A). As shown in Figure2B, down-regulation of COX-2 might inhibit the malignant growth of SGC7901 cells in vivo.

As shown in Figure3, the number of endothelial cells within the tumors formed by COX-2-downregulating cells was less than that of tumors formed by control cells. In order to investigate the angiogenic property of COX-2 in endothelial cells, the in vitro tube formation of HUVEC was assessed. As shown in Figure4, 5, down-regulation of COX-2 might suppress cell tube formation and migration in HUVEC.

Using cDNA microarray, genes were identified differentially expressed between different transfected SGC7901 cells. Compared with control cells, a total of 23 genes were found to be differentially expressed in COX-2-downregulating cells, including FGF4, PDGF-BB, PDGFRB, PF4, TGFB2, TGFBR1, VEGF, FLT1, FLK 1, angiopoietin-1, angiopoietin-2, Tie2, IFNA1, PRL, PTN, SCYA2, SPARC, TNFSF15, PECAM1, MMP2, SERPINF1, THBS2 and OPN. To confirm the microarray findings, RT-PCR and western blot were undertaken in gastric cancer cells. Down-regulation of COX-2 might inhibit VEGF, Flt-1, Flk-1/KDR, angiopoietin-1, tie-2, MMP2 and OPN (Figure6).