Nine breast cancer cell lines provided by Professor Tao (Cancer Epigenetics Laboratory, Department of Clinical Oncology, State Key Laboratory of Oncology in South China, Sir YK Pao Center for Cancer and Li Ka Shing Institute of Health Sciences, The Chinese University of Hong Kong and CUHK Shenzhen Research Institute, Hong Kong, China) were used in this study: BT549, MCF-7, MDA-MB-468, MDA-MB-231, SK-BR-3, T47D, YYC-B1, YCC-B3 and ZR-75-1. All the cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) with 10% fetal bovine serum (Invitrogen; Thermo Fisher Scientific, Inc.) and 100 U/ml penicillin/streptomycin at 37°C in moist air containing 5% CO₂.

Normal human adult tissue RNA samples were purchased commercially (Stratagene; Agilent Technologies, Inc., Santa Clara, CA, USA; EMD Millipore, Billerica, MA, USA; BioChain Institute, Inc., Newark, CA, USA). Primary breast cancer tumor tissues, matched adjacent non-malignant tissues and normal breast tissues were obtained from patients who had undergone primary surgery at the Department of Endocrinological and Breast Surgery, The First Affiliated Hospital of Chongqing Medical University (Chongqing, China), between January 2010 and March 2014. All samples were confirmed by pathologist physicians. The clinical and pathological data, including sex, age, tumor grade, tumor size, treatment and follow-up data, were available for the majority of the breast cancer samples. All patients signed informed written consent forms for participation in the study at initial clinical investigation. The present study was approved by the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University (approval no. 2010/2012-23). A multifunctional user-friendly online tool, Gene Expression Based Outcome for Breast Cancer Online (GOBO; http://co.bmc.lu.se/gobo), was used to analyze the expression level of MSX1 associated with molecular subtypes of breast cancer, ER status and histological grade as previously described (23).

Total RNA was isolated from tissues and cells using TRI Reagent (Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. Semi-quantitative RT-PCR was performed as described previously, using GAPDH as the control (24). The primer sequences used for PCR amplification are listed in Table I. RT-PCR was performed with 32 cycles for MSX1 and 23 cycles for GAPDH using Go-Taq Flexi DNA Polymerase (Promega Corporation, Madison, WI, USA).

The MDA-MB-468, MCF-7 and ZR-75-1 cell lines were used for pharmacological demethylation. Briefly 1×10⁶ cells were treated with 10 mmol/l Aza for 72 h and then with 100 nmol/l TSA (both from Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) for 24 h at 37°C. The cells were then harvested for DNA and RNA extraction, and further analysis.

Genomic DNA was extracted from the tumors, normal tissues and cell pellets using the QIAamp DNA mini kit (Qiagen GmbH, Hilden, Germany). DNA bisulfite treatment and MSP were performed as previously described (25). The bisulfite-treated DNA was amplified using MSP with a methylated-MSX1-specific or unmethylated-MSX1-specific primer set. The primers used for MSP and bisulfite sequencing are listed in Table I. The methylated and unmethylated MSP primer sets target the same CpG sites and have been tested previously to confirm that they do not amplify any genomic DNA without bisulfite treatment, and are therefore specific. MSP was performed for 40 cycles to amplify the unmethylated and methylated gene using AmpliTaq Gold DNA Polymerase (Applied Biosystems; Thermo Fisher Scientific, Inc.). The PCR products were analyzed on 2% agarose gel.

Total RNA was purified from a panel of fresh, paired primary breast tissues (20 tumors and the corresponding adjacent tissues) using TRI Reagent as previously described (26). RT-qPCR with Maxima SYBR-Green/ROX qPCR Master mix (MBI Fermentas; Thermo Fisher Scientific, Inc.) was used to detect gene expression (Table I) according to the manufacturer's protocol, using a HT7500 System (Applied Biosystems; Thermo Fisher Scientific, Inc.). Melting-curve analysis and agarose gel electrophoresis of the PCR products were performed to verify the specificity of the PCR and the identity of the PCR products. Each experiment was performed in triplicate and the relative expression levels of MSX1 in the breast tissues were normalized to those of β-actin. Data were normalized using the 2^−ΔΔCq method (27).

MDA-MB-231 cells infected with LV-MSX1 or LV-empty were trypsinized and resuspended. The cells were then cultured in triplicate in 96-well plates at a density of 1×10⁴ cells/well in a volume of 100 μl DMEM and allowed to grow overnight. Cell viability was quantified with a Cell Counting kit-8 assay (CCK-8; Dojindo Molecular Technologies, Inc., Kumamoto, Japan), in which only living cells are stained. After 24, 48 and 72 h, the medium was removed, and α-minimal essential medium (100 μl) containing 10 μl CCK-8 reagent was added to each well of precultured cells. Following incubation of the cells for 2 h at 37°C, the optical density was determined at a wavelength of 450 nm with an automatic microplate reader. The results are the means of three independent experiments conducted over several days.

MDA-MB-231 and MDA-MB-468 cells infected by LV-MSX1 or LV-empty were plated at a density of 300 cells/well in 6-well plates and cultured for 14 days in normal culture medium. The cells were then washed twice with PBS, fixed with 10% buffered formalin, dried and stained with 2% crystal violet. The surviving colonies (≥50 cells/colony) were manually counted in four different fields of vision and the mean values calculated. Each treatment was tested in triplicate.

Cultured MDA-MB-231 cells infected with LV-MSX1 or LV-empty were evenly seeded in 6-well plates and allowed to reach 100% confluence in DMEM. A straight wound was induced by dragging a 20-μl pipette tip through the confluent cell monolayer. The cells were incubated and allowed to migrate in the medium. At 0, 12, 24 and 36 h after wounding, the plates were washed twice with PBS to remove the dead cells, and images were captured in four random fields at a magnification of ×100 (Leica DMI4000B; Leica Microsystems, Ltd., Milton Keynes, UK). The rate of cell migration was quantified according to the percentage of repaired wound area, using Image Pro-Plus software (version 6.0; Media Cybernetics, Inc., Rockville, MD, USA). Each experiment was performed in triplicate.

For cell-cycle analysis, MDA-MB-231 cells and MDA-MB-468 cells infected with LV-MSX1 or LV-empty were seeded in 6-well plates (1×10⁶ cells/well) and incubated overnight at 37°C in 5% Co₂. The cells were digested by trypsin at 48 h after infection and then centrifuged at 1,000 × g for 5 min at room temperature. The cells were washed twice with PBS, fixed in 70% ethanol at 4°C for 2 h, and stained with propidium iodide (Sigma-Aldrich; Merck KGaA) for 30 min at 37°C. The samples were evaluated with the BD FACSCanto and the results were analyzed using BD FACSDIVA software (version 4.1; BD Biosciences, San Jose, CA, USA). All experiments were performed in triplicate, and one representative figure is shown for each cell type.

Cell migration and invasion assays were carried out using Transwell chambers (8 μm; Corning Incorporated, Corning, NY, USA). The cells were plated at a density of 2.5×10⁵ with RPMI-1640 in the upper well of each Transwell chamber. For invasion assay, 100 μl of Matrigel were added in the upper well first. The lower compartment was filled with DMEM supplemented with 10% fetal bovine serum. Following incubation for 24 h at 37°C with 5% CO₂ in a humidified atmosphere, the cells that had not migrated or invaded were removed from the upper surface of the filter with gentle swabbing. The cells on the lower surface of the filter were fixed in 4% paraformaldehyde, stained with 0.1% crystal violet staining solution for 15 min at room temperature, and counted by light microscopy at ×100 magnification in 10 images taken in five random fields. The averages were then calculated. The experiment was performed in triplicate.

The infected cells were harvested and lysed in M-PER Mammalian Protein Extraction Reagent (Pierce, Thermo Fisher Scientific, Inc.) containing a protease inhibitor cocktail (Sigma-Aldrich; Merck KGaA). The protein samples were incubated on ice for 10 min, and then centrifuged at 10,000 × g for 5 min at 4°C to remove the cell debris. The protein samples were quantified by BCA Protein Assay kit (Thermo Fisher Scientific, Inc.), 20 μg protein were separated using 10% SDS-PAGE gel and then electrophoretically transferred onto a polyvinylidene difluoride membrane (Bio-Rad Laboratories, Inc., Hercules, CA, USA). The membrane was blocked with 5% skimmed milk in Tris-buffered saline with Tween-20 for 1 h at room temperature, and then incubated overnight at 4°C with the following primary antibodies at the manufacturers' recommended dilutions: antibodies directed against β-catenin (#19807; Cell Signaling Technology, Inc., Danvers, MA, USA) active β-catenin (#4270; Cell Signaling Technology, Inc.), c-Myc (#1472-1; Epitomics; Abcam, Cambridge, UK), cleaved poly (ADP-ribose) polymerase (cleaved PARP; #9541), cleaved caspase-3 (#9661) (both from Cell Signaling Technology, Inc.) and cyclin D1 (CCND1; #1677-1; Epitomics, Abcam). The membranes were washed and then incubated with anti-mouse (#7076; Cell Signaling Technology, Inc.) or anti-rabbit (#7074; Cell Signaling Technology, Inc.) IgG, HRP-linked antibodies at a dilution of 1:3,000 at room temperature for 1 h. The protein bands were visualized using a Fusion FX5 system (Vilber Lourmat, Eberhardzell, Germany). The results were analyzed using ImageJ software (version 6.0; National Institutes of Health, Bethesda, MD, USA). GAPDH (sc-47724; Santa Cruz Biotechnology, Santa Cruz, CA, USA) was used as the endogenous protein for normalization.

Statistical analyses were performed using the Student's t-test, χ² test and Fisher's exact test to determine the significance of differences between groups. For all tests, p<0.05 was considered to indicate a statistically significant difference. Results are presented as the mean ± standard deviation.

The expression of MSX1 was assessed in a panel of normal human adult tissues, including breast tissues, and breast cancer cell lines, using semi-quantitative RT-PCR (Fig. 1). MSX1 was clearly expressed in all the normal human tissues, including normal breast tissue, at varying levels (Fig. 1A), but was frequently silenced or downregulated in breast cancer cell lines, with the exception of BT549 (Fig. 1C).

MSX1 expression was then examined at the mRNA level in primary breast tumors. qPCR demonstrated that MSX1 mRNA was downregulated in the breast cancer tissues compared with the normal breast tissues (p<0.05; Fig. 2A). The GOBO online tool was also used to assess MSX1 expression in 1,881 breast cancers. Gene-Set Analysis tumor demonstrated that the expression of MSX1 differs among different subtypes of breast cancer (28). The most aggressive subtype, the basal-like subtype, exhibited the lowest MSX1 expression, whereas the least aggressive subtype, the normal-like subtype, exhibited the highest expression of MSX1 (p<0.00001; Fig. 2B). The expression of MSX1 was also reduced with a negative estrogen receptor (ER) status and a higher histological grade (p<0.00001; Fig. 2C and D). These results indicate that MSX1 expression is frequently downregulated or silenced in breast cancer, and is associated with the malignant progression of breast cancer.

The involvement of promoter methylation in MSX1 silencing in breast cancer was evaluated. Bioinformatic analysis revealed a typical CpG island spanning the proximal promoter and the exon 1 region of the MSX1 gene (Fig. 1B). To determine whether promoter methylation leads to MSX1 silencing, MSX1 methylation in breast tumor cell lines was analyzed using MSP. The results demonstrated that MSX1 was methylated in seven breast cancer cell lines (MDA-MB-231, MDA-MB-468, MCF-7, T47D, ZR75-1, SK-BR-3 and YCC-B3), consistent with its silencing, but not in another two cell lines (BT549 and YCC-B1) in which MSX1 was unmethylated (Fig. 1C).

Pharmacological demethylation was conducted to test whether promoter methylation directly mediates the reduction of MSX1 levels in breast cancer cells. Three cell lines (MDA-MB-468, MCF-7 and ZR-75-1) lacking MSX1 expression were treated with Aza and the histone deacetylase inhibitor TSA. Following treatment, the expression of MSX1 in these cell lines was significantly increased compared with that prior to treatment, accompanied by decreased methylated alleles of MSX1 (Fig. 1D). These results indicate that promoter methylation is responsible for MSX1 silencing in breast cancer cells.

To ascertain whether MSX1 methylation occurs in primary breast tumors, MSX1 methylation by MSP was examined in 99 primary breast tumor tissues, 8 surgical margin tissues and 13 normal breast tissues. Aberrant MSX1 methylation was detected in 47 (47.5%) of the primary breast cancer tissues, but not in the surgical margin tissues or normal breast tissues (Fig. 3 and Table II). These results confirm the tumor-specific methylation of the MSX1 promoter. The association of MSX1 methylation with the clinicopathological features of breast cancer patients was also analyzed, where the clinicopathological features included age, tumor size, tumor grade, lymph-node metastasis, ER status, progesterone receptor status, human epidermal growth factor receptor 2 status and Ki67 status. However, no significant difference of MSX1 methylation status according to clinicopathological features of the patients was detected (Table III).

MDA-MB-231 and MDA-MB-468 cells, which are MSX1-silenced, were infected with LV-MSX1 and LV-empty plasmids. MDA-MB-231 and MDA-MB-468 cells stably overexpressing MSX1 were successfully generated, which was confirmed by microscopy and RT-PCR (Fig. 4A and B). CCK-8 cell proliferation and colony formation assays were performed to clarify the function of MSX1 in the proliferation of breast cancer cells. Cell viability was significantly reduced at 48 and 72 h after the infection of the MDA-MB-231 cells with MSX1 (p<0.05 and p<0.01, respectively; Fig. 4C). The colony formation assay demonstrated that MSX1-infected MDA-MB-231 and MDA-MB-468 cell colonies were reduced by 40 and 65% compared with those of the respective control cells (p<0.01; Fig. 4D). These results suggest that MSX1 inhibits the proliferation of breast cancer cells.

To investigate the mechanism underlying the growth-inhibitory effect of MSX1 on breast cancer cells, the cell cycle status of breast cancer cells was examined using flow cytometry. The number of cells arrested in the G1/S phase increased significantly in the MSX1-expressing MDA-MB-231 and MDA-MB-468 cells compared with the respective control cells, which was accompanied by a significant reduction in S-phase cells (p<0.01; Fig. 5A). Western blot analysis revealed upregulated expression levels of cleaved caspase-3 and cleaved PARP in the MSX1-expressing MDA-MB-231 cells compared with the control cells infected with empty vector. These results indicate that MSX1 causes G1/S cell cycle arrest and apoptosis in breast cancer cells.

Wound-healing and Transwell assays were performed to assess the effects of MSX1 expression on the migration and invasion of breast tumor cells. In the scratch wound-healing assay, the percentage wound closure of a confluent monolayer of MSX1-expressing MDA-MB-231 cells was significantly lower compared with that of control cells at 12 and 14 h post-wounding (p<0.05 and p<0.01, respectively; Fig. 6A). In the Transwell assays, the numbers of migrating and invading cells were significantly reduced in the MSX1-overexpressing cells compared with the control cells, suggesting that MSX1 inhibits breast cancer cell migration and invasion (p<0.01; Fig. 6B and C).

As the β-catenin signaling pathway is critical in the regulation of cell proliferation, whether MSX1 affects this pathway in breast cancer cells was investigated. The expression of active β-catenin was examined using western blotting. The expression levels of active β-catenin and its downstream targets c-Myc and CCND1 were reduced in the MSX1-expressing MDA-MB-231 cells compared with the control cells (Fig. 6D). These results indicate that MSX1 reduces the levels of active β-catenin and its downstream target genes, and thus may antagonize the WNT/β-catenin signaling pathway.