The following antibodies were used: mouse anti-E-cadherin IgG2a (1:50,000; #610181); mouse anti-β-catenin IgG1 (1:10,000; #610153) (BD Biosciences, San Jose, CA, USA); mouse anti-N-cadherin IgG1 (1:1,000; sc-59987) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA); mouse anti-vimentin IgG1 (1:1,000; V5255); mouse anti-PSA-neural cell adhesion molecule (NCAM) IgM 5A5 (Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA, USA); anti-β-tubulin I IgG1 (1:1,000,000; T7816) (Sigma, St. Louis, MO, USA); and horseradish peroxidase (HRP)-labeled goat anti-mouse IgG (1:5,000; A0216) (Beyotime Institute of Biotechnology, Haimen, China).

The reagents used in this study were as follows: TGF-β1 (BD Biosciences), PNGase F (New England BioLabs, Inc., Ipswich, MA, USA), urea, DL-dithiothreitol, iodoacetamide, tetrachloroaurate (AuCl₄), Sepharose 4B, methanol, tunicamycin and 1-butanol (Sigma).

Mouse mammary epithelial cells (NMuMG), mouse mammary carcinoma cells (4T1), normal human breast cells (MCF10A) and human mammary carcinoma cells (MDA-MB-231; MB-231) were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The cells were cultured in Dulbecco's modified Eagle's medium (DMEM) or RPMI-1640 supplemented with 10% fetal bovine serum (FBS) (both from HyClone, Logan, UT, USA), 100 IU/ml penicillin and 100 µg/ml streptomycin (Gibco, Carlsbad, CA, USA) in a humidified 5% CO₂ atmosphere at 37°C. The NMuMG cells were cultured in DMEM containing 10 µg/ml insulin (Sigma). The MCF10A and MB-231 cells were cultured in DMEM containing 1% sodium pyruvate (Solarbio, Beijing, China). The 4T1 cells were cultured in RPMI-1640.

All procedures performed in experiments involving human participants were carried out in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

BC tissues were obtained from 20 patients with TNM stage I, II and III BC, and 4 normal breast tissue samples were obtained from surgical or autopsy specimens performed at the First Affiliated Hospital of Xi'an Jiaotong University (Xi'an, China). All the tissues were snap-frozen in liquid nitrogen and stored at −70°C until use. Informed consent was obtained from all patients in accordance with the Declaration of Helsinki. The present study was approved by the Research Ethics Committee of Jiangnan University (Wuxi, China).

The cells were cultured in 6-well plates, washed with phosphate-buffered saline (PBS), harvested, homogenized in T-PER lysis buffer (Thermo Fisher Scientific, Waltham, MA, USA) containing 10 U/ml aprotinin, and centrifuged at 14,000 × g at 4°C for 15 min. The super-natants were mixed with loading buffer (which contained 1% β-mercaptoethanol) and heated at 100°C for 10 min. Proteins were loaded on 7.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels and transferred onto a PVDF membrane. The membrane was blocked with 5% non-fat milk, incubated with primary antibody overnight at 4°C, probed with appropriate HRP-conjugated secondary antibody, visualized using the Pro-Light HRP kit (Tiangen Biotech Co., Ltd., Beijing, China), and photographed using a Molecular Imager Chemi DOC™XRS⁺ system (Bio-Rad, Richmond, CA, USA).

Total proteins from the tissues were isolated using an E.Z.N.A. DNA/RNA/Protein Isolation kit (Omega Bio-Tek, Doraville, CA, USA) according to the manufacturer's instructions. Glycans on proteins were released by PNGase F and desalted as previously described (18,19). The glycan sample was lyophilized and dissolved in methanol/H₂O (1:1, v/v) for further analysis.

Free sialic acids were labeled with DMB using a Sialic Acid Fluorescence Labeling kit (Takara Bio Inc., Otsu, Japan), according to the instructions provided by the manufacturer. Briefly, the cell lysate was hydrolyzed with 2 M acetic acid and labeled with DMB, as described in a previous study (20). A 10-µl aliquot of each sample was loaded on a Zorbax SB-C18 column (4.6×250 mm) (Agilent Technologies, Inc., Santa Clara, CA, USA) and eluted with methanol/acetonitrile/water (8:7:85, v/v/v) on a Waters e2695 HPLC system with a fluorescence detector (excitation, 373 nm; emission, 448 nm). Free sialic acids were quantified based on peak areas obtained from a defined standard.

Total RNA from the cells was isolated using an RNApure Tissue kit (CWBIO, Beijing, China) and RNA from human tissues was isolated using an E.Z.N.A. DNA/RNA/Protein Isolation kit (Omega Bio-Tek) according to the manufacturer's instructions. First-strand cDNA was synthesized using a ReverTra Ace-α^® kit (Toyobo, Osaka, Japan). Primers were designed using DNAman software as follows: mouse STX (167 bp) sense, 5′-CTTGGATGCGGAGAAGGAT and anti-sense, 5′-GGCACAAGTCTGGAAATGCT; mouse PST (126 bp) sense, 5′-GCGAACTGCCTATCCATCAC and antisense, 5′-TCACAGAATCTGGTGGCAAG; human STX (141 bp) sense, 5′-TCAGAACCAGAACCCAGTCA and antisense, 5′-CGACAGTCAGTTTCAAAGCC; human PST (106 bp) sense, 5′-ACTGA AAGTGCGAACTGCCT and antisense, 5′-GAGAAGACCTGTGCTGGGTC; mouse γ-tubulin (107 bp) sense, 5′-ATCTA CCTGTCGGAGCATGG and antisense, 5′-GCCTCCCGA TCTATGATGTC; and human β-actin (232 bp) sense, 5′-CTTCC TGGGCATGGAGTC and antisense, 5′-GCCGATCCACA CGGAGTA. Semi-quantitative PCR was performed as follows: 94°C, 4 min; 94°C, 45 sec, 60°C, 45 sec, 72°C, 15 sec for 30 cycles; 72°C, 5 min; quantitative PCR (qPCR) was performed using Ultra SYBR Mixture (Cat. no. CW0957; CWBIO) on a CFX96 Real-Time PCR detection system (Bio-Rad). The transcriptional levels of target genes were quantified using the 2^−ΔΔCt method, as previously described (21) and expressed as the means ± SD from triplicate experiments.

Phagokinetic gold sol assay was performed as previously described (22,23). Briefly, 1.8 ml of a 14.5 mM AuCl₄ solution and 6 ml of a 36.5 mM Na₂CO₃ solution were added to 11 ml distilled H₂O, heated to a boil, and then 1.8 ml of 0.1% formaldehyde was added. Colloidal gold solution (gold sol) was added to 24-well plates, and the plates were blocked with filter-sterilized 1% BSA, as previously described (22). Cells (2×10³) in complete culture medium were seeded onto the gold sol-coated wells and incubated for 18 h. Images were captured using an inverted microscope (model NA0.3OWD72; Sunny Optical Technology, Ningbo, China). The track areas of 50 cells were measured using the ToupView imaging system, as previously described (24) and expressed as square pixels.

Duplexes of 21 nucleotides of human and mouse STX siRNA target sequences (hSTXi or mSTXi) and negative control siRNA (NC), without homology to other known human and mouse genes, were designed and synthesized by Invitrogen (Carlsbad, CA, USA) as follows: hSTXi, 5′-GCCUGGAGAUAUUAUUCA UTT (sense); and mSTXi, 5′-CCUGAAGCCUGGAGACAU UTT (sense). siRNA (30 pmol) was transfected using Lipofectamine 2000 reagent (3 µl) (Invitrogen) and the 4T1 or MB-231 cells were examined 24 h following transfection. The suppression of the expression of STX was verified by semi-quantitative and quantitative RT-PCR.

The cells were cultured in 6-well plates and treated with or without TGF-β1 (5 ng/ml) for 48 h. Tunicamycin (4 µg/ml) was added together with TGF-β1 into 6-well plates, and the NMuMG or MCF10A cells were cultured for an additional 48 h, as previously described (25,26). The cells (5×10⁴) were plated in an upper Transwell insert (12/24-well Transwell; 8 µm polycarbonate membrane; Costar, Corning, NY, USA) in DMEM or RPMI-1640 medium containing 0.1 ml of 0.2% BSA (RuiTaibio, Beijing, China). A total of 0.6 ml of DMEM or RPMI-1640 medium supplemented with 10% FBS, serving as a chemoattractant, was deposited in the lower chamber. Following incubation for 16 h at 37°C in a 5% CO₂ atmosphere, the cells were washed with PBS and fixed with cold 4% buffered paraformaldehyde. The cells in the upper Transwell filter were removed with a cotton wool tip and stained with crystal violet. The Transwells were rinsed with deionized water and air-dried. The filters were photographed, and the cells in 5 random optical fields were counted to determine migration, as previously described (27,28).

Data were statistically analyzed using the Prism 5 software programs as previously described (29). Differences between the means were assessed by paired or unpaired Student's t-test, and P-values <0.05 were considered to indicate statistically significant differences.

As EMT is one of the main mechanisms involved in the development of BC metastasis (30), the expression of EMT-related markers in malignant 4T1 and MB-231 cells, in comparison with non-malignant NMuMG and MCF10A cells was determined by western blot analysis (Fig. 1A). The decreased expression of E-cadherin and β-catenin, as well as the increased expression of vimentin, were observed in the 4T1 and MB-231 cells. Tumor progression towards metastasis is often depicted as a multistage process in which malignant cells spread from their original site to colonize distant organs through acquired mobility and migration ability (31,32). In this experiment, increased motility (Fig. 1B) and migration (Fig. 1C) were observed in malignant 4T1 and MB-231 cells compared with normal NMuMG and MCF 10A cells.

In view of the findings that PSA is re-expressed in many types of tumor, we thus compared PSAs on N-glycans in these 4 cell lines. PSAs display unique modifications of N-glycans, particularly of NCAM (33). As shown in Fig. 2A, PSAs on N-glycans were hydrolyzed under mild acidic conditions, and free sialic acids (reflecting PSA levels) were detected using a fluorescence-labeling method followed by HPLC. In comparison with the NMuMG or MCF10A cells, the PSA level was slightly higher in the 4T1 cells and significantly higher in the MB-231 cells (Fig. 2B). These findings suggest that a high PSA expression is associated with the enhanced invasiveness and metastasis of BC cells.

The EMT process has been shown to be associated with cancer cell motility and metastasis. The non-malignant NMuMG and MCF10A cells were treated with TGF-β1 to induce EMT, as previously described (34). The treated cells displayed a spindle-like, elongated morphology (Fig. 3A), a reduced β-catenin expression, an increased vimentin expression, and exhibited the typical 'cadherin switch' from E-cadherin to N-cadherin (Fig. 3B). The TGF-β1-treated NMuMG cells displayed an enhanced motility and migration, whereas the TGF-β1-treated MCF10A cells exhibited only enhanced migration (Fig. 3C and D). The levels of free sialic acid released from N-glycan as a function of dose and time were examined in the NMuMG cells following treatment with TGF-β1. The NMuMG cells treated with 2 and 5 ng/ml TGF-β1 exhibited free sialic acid levels approximately 2.5- and 4-fold higher, respectively, than those of the untreated cells (Fig. 4A). In order to examine the effects of high levels of PSA on cell behavior, 5 ng/ml TGF-β1 was used in the following experiment. The PSA levels were higher in the TGF-β1-treated NMuMG cells at various time points (Fig. 4B). Similar results were obtained for the TGF-β1-treated MCF10A cells (Fig. 4C). Moreover, a high expression of polysialylated NCAM (PSA-NCAM) was observed in the NMuMG and MCF10A cells undergoing EMT (Fig. 4D).

In view of the correlation of high PSA levels with the facilitation of cell migration and cell motility during EMT, we examined the possibility that the downregulation of PSA reverses EMT in malignant cells. Semi-quantitative RT-PCR was used to evaluate the mRNA levels of PST and STX, two major polysialyltransferases responsible for PSA synthesis. The 4T1 and MB-231 cells exhibited high levels of STX, but undetectable levels of PST (data not shown). The silencing of PSA by the suppression of STX resulting in the decreased expression of PSA-NCAM (Fig. 5A and B), or by treatment with tunicamycin (which blocks N-glycan synthesis), significantly decreased the migration of both malignant cell lines (4T1 and MB-231) (Fig. 5C). The downregulation of PSA resulted in the decreased motility of the malignant BC cells (MB-231), but not of the 4T1 cells (Fig. 5C).

To further investigate the role of PSA during EMT, we used tunicamycin to inhibit N-glycan synthesis in normal cells undergoing EMT. The typical 'cadherin switch' from E-cadherin to N-cadherin and the enhancement of cell migration in the TGF-β1-treated MCF10A and NMuMG cells were reversed by the presence of tunicamycin (Fig. 6).

The patient characteristics and PSA levels in the normal tissues (n=4) and malignant tissues (n=20) are listed in Tables I and II. The PSA levels were lower (pmol level) in the normal tissues and higher in the malignant tumor tissues (Fig. 7A). The difference in the PSA levels between the normal tissues and advanced-stage BC tissues was significant (P<0.0001). The PSA level was consistently higher in the tumor tissues compared to the normal controls. In comparison with the controls, the PSA level was highest in the samples obtained from patients with TNM stage III BC (n=6, P<0.0001), followed by stage II (n=8, P<0.0001) and stage I (n=6, P=0.0085), suggesting that an increased PSA expression correlates with disease progression (Fig. 7B).

The aberrant expression of polysialyltransferases is often observed in malignant tumors and has thus been considered as a novel target for detection or treatment of metastatic cancer (35). As an enhanced level of PSA was observed in our clinical BC samples, we wished to examine the expression of polysialyltransferases between normal tissues and malignant tissues. We used semi-quantitative and quantitative RT-PCR to examine STX and PST expression at the mRNA level in BC tissues (Table I). STX was widely and highly expressed in all tissues examined (n=24) with no significant differences when compared with the normal controls (Fig. 7C and D). However, PST expression was significantly increased in the malignant tumor tissues compared to the normal tissues (Fig. 7C and E). PST was expressed distinctively in the tumor samples obtained from patients with stage I (n=6, 50%), II (n=8, 75%), and III (n=6, 83%) BC (Table II and Fig. 7E). The PST enzyme is clearly expressed in high-stage BC tissues, but not in normal tissues. The difference in the PST level between the normal tissues (n=4) and high-stage/lymph node-positive tissues (n=6, stage III) from patients with BC was significant at r²=0.8, indicating that PST expression is increased in BC tissues and correlates with cancer progression.