TMAs (cat. no. 140317A; AlenaBio Biotechnology Ltd., Xi'an, China) with samples from healthy and breast cancer tissue, with stage and grade information, were purchased from US Biomax, Inc. (Rockville, MD, USA). For IHC analysis, TMA sections were deparaffinized in 100% xylene and rehydrated in graded ethanol solutions. The sections were then boiled under pressure in citrate buffer (pH 6.0) for 5 min for antigen retrieval. TMA sections were incubated at 37°C for 1 h with EpCAM (1:200 dilution; cat no. 21050-1-AP; Wuhan Sanying Biotechnology, Wuhan, China,) and COX-2 (1:100 dilution; cat. no. 12375-1-AP; Wuhan Sanying Biotechnology) antibodies in TBS containing 1% bovine serum albumin (Sangong Pharmaceutical Co., Ltd., Shanghai, China). After washing with PBS, the sections were incubated with anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibody (1:500 dilution; cat. no. SA00001-2; Wuhan Sanying Biotechnology). Signal development was performed by adding 250 µl 3,3′diaminobenzidine (Sangong Pharmaceutical Co., Ltd.) substrate solution to each slide and incubating for 3 min in the dark. Finally, slides were washed 3 times in water and drained. Images were captured using an Aperio ScanScope^® CS system (Nikon Instruments Inc., Vista, CA, USA). EpCAM or COX-2 positive staining on TMA sections was semi-quantitatively analyzed by two independent investigators using the following criteria: 0, background staining; 1, weakly positive; 2, moderately positive; 3, strongly positive staining.

The MCF-7 and MDA-MB-231 human breast adenocarcinoma cell lines were purchased from American Type Culture Collection (Manassas, VA, USA). MCF-7 cells were maintained in DMEM-F12 (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with 10% calf serum (Gibco; Thermo Fisher Scientific, Inc.), 1% penicillin/streptomycin, 1 mM sodium pyruvate, 1.5 g/l sodium bicarbonate, and 10 mM HEPES. MDA-MB-231 cells were cultured in Leibovitz's L-15 medium (Gibco; Thermo Fisher Scientific, Inc.), supplemented with 10% calf serum, 1% penicillin/streptomycin and 10 mM HEPES. All cells were incubated in a 5% CO₂ humidified atmosphere at 37°C.

Cells that were in the logarithmic growth phase were transfected with complementary DNAs encoding human EpCAM and control empty plasmid (cat. no. BC014785; Wuhan Sanying Biotechnology), or small interfering (si)RNA targeting EpCAM (si-EpCAM) and control scrambled siRNA (sequences 5′-UGCUCUGAGCGAGUGAGAATT-3′ and 5′-UUCUCACUCGCUCAGAGCATT-3′, respectively; GenePharma Co., Ltd., Shanghai, China), using Lipofectamine^® 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) as the delivery agent, according to the manufacturer's protocol. Subsequent experiments were performed 48 h post-transfection.

To prepare whole cell extracts, cells at 90% confluence were washed in PBS prior to incubation with lysis buffer containing 1% Triton X-100, 150 mM NaCl, 10 mM Tris (pH 7.4), 1 mM EDTA, 1 mM EGTA (pH 8.0), 0.2 mM Na₃VO₄, 0.2 mM phenylmethylsulfonyl fluoride and 0.5% Nonidet P-40 on ice for 10 min. Debris was removed from the lysates by centrifugation at 9,000 × g for 10 min at 4°C and the supernatants were collected. Protein concentration was determined with the Coomassie Protein Assay Reagent using bovine serum albumin as a standard. Equal amounts of protein (50 µg) were separated by 10% SDS-PAGE and transferred onto nitrocellulose membranes, which were blocked with TBS containing 0.5% Tween-20 and 5% fat-free dry milk for 2 h at 37°C. The membranes were then incubated for 3 h at 37°C with EpCAM (1:1,000 dilution; cat. no. 21020-1-AP; Wuhan Sanying Biotechnology), COX-2 (1:800 dilution; cat. no. 12375-1-AP; Wuhan Sanying Biotechnology) and GAPDH (1:2,000 dilution; cat. no. 10494-1-AP; Wuhan Sanying Biotechnology) primary antibodies. Following incubation with a HRP-conjugated anti-goat secondary antibody (1:1,000 dilution; cat. no. SA00001-2; Wuhan Sanying Biotechnology) for 40 min at 37°C, the protein bands were visualized using an enhanced chemiluminescence detection system (GE Healthcare Life Sciences, Chalfont, UK). Western blots presented are representative of at least 3 independent experiments. Protein expression was quantified using densitometry analysis with Labworks software version 4.6 (Labworks LLC, Lehi, UT, USA). Band intensity is expressed as the mean ± standard error of 3 experiments for each group. GAPDH was used as the loading control.

Immunohistochemical scores for EpCAM and COX-2 were tabulated, and the χ² test for trend analysis was performed to investigate the relationship between EpCAM and COX-2 expression and pathological diagnostic criteria for breast cancer. Spearman's correlation coefficient analysis was performed to test for positive or negative correlations between EpCAM and COX-2 expression across breast cancer subtypes and diagnostic parameters. Statistical significance was analyzed using GraphPad Prism 5 software (GraphPad Software, Inc., La Jolla, CA, USA). P<0.05 was considered to indicate a statistically significant difference.

To investigate the expression of EpCAM and COX-2 in breast cancer tissue, immunohistochemistry was performed on a series of 134 human breast cancer samples within TMAs. Representative EpCAM and COX-2 staining is presented in Fig. 1. A total of 92 (68.66%) samples highly expressed EpCAM and COX-2, whereas 19 (4.17%) samples exhibited low EpCAM and COX-2 expression; a total of 18 (13.43%) samples exhibited high EpCAM and low COX-2 expression, whereas 5 (3.73%) samples exhibited low EpCAM and high COX-2 expression. A positive correlation was revealed between EpCAM and COX-2 expression in breast cancer (r=0.63, P=0.009; Table I).

To examine the relationship between EpCAM and COX-2 expression and clinicopathological characteristics of the disease, the correlation between EpCAM and COX-2 expression and the following parameters was investigated: Age at the time of diagnosis, tumor differentiation, lymph node metastasis, as well as the expression of ER, PR, HER2, p53 and the proliferation marker Ki-67 (Table II). The results suggested that EpCAM and COX-2 expression were significantly correlated with the histological grade of the tumor (P<0.05). High expression of EpCAM and COX-2 was more frequently observed in higher grade (poorly differentiated) tumors compared with in lower grade tumors. Furthermore, a significant correlation was revealed between EpCAM and COX-2 expression and the expression of ER, PR and Ki-67 (P<0.05); however, no correlation was apparent with lymph node metastasis and p53 expression. Notably, the expression of EpCAM was positively correlated with the expression of HER2 (P<0.05), whereas no correlation was revealed between COX-2 and HER2 expression (Table II).

The aforementioned results demonstrated that EpCAM and COX-2 expression were positively correlated in breast cancer tissue samples. The expression levels of EpCAM and COX-2 were also detected in two breast cancer cell lines. Western blot analysis demonstrated that EpCAM and COX-2 protein expression levels varied between these two cell lines. The expression of EpCAM and COX-2 appeared higher in MDA-MB-231 cells compared with MCF-7 cells (Fig. 2).

To investigate whether EpCAM was involved in the regulation of COX-2 in breast cancer, MCF-7 and MDA-MB-231 breast cancer cells were transfected with EpCAM overexpression plasmid or control. The results demonstrated that overexpression of EpCAM promoted the expression of COX-2 (Fig. 3). Furthermore, MCF-7 and MDA-MB-231 cells were transfected with siEpCAM or control in order to silence the expression of EpCAM. Western blot analysis revealed that EpCAM silencing decreased the expression of COX-2 in both cell lines (Fig. 4). These results suggested that EpCAM may be involved in the regulation of COX-2 expression.