MDA-MB-231, T47D, and MCF-10A cell lines were purchased from the Chinese Academy of Sciences Cell Bank (Shanghai, China). The cells were cultured in Dulbecco's modified Eagle medium (Invitrogen, Waltham, USA) supplemented with 10% fetal bovine serum (Gibco, New York, USA). All of the cell lines were cultured in an atmosphere containing 5% CO₂ at 37°C.

We constructed a Pokemon overexpression vector according to the previous method [27], with minor modifications. Briefly, full-length human Pokemon was obtained by standard polymerase chain reaction (PCR) amplification from MCF7 cell cDNA and was cloned into the green fluorescence protein (GFP)-linked pcDNA3.1 plasmid. Pokemon small interfering RNA (siRNA) and negative siRNA were bought from GenePharma (Shanghai, China).

Cells were lysed using Trizol reagent (Life Technology, Rockville, USA). The cell lysates were mixed with chloroform and divided into aqueous phase and organic phase. Total RNA was extracted from the supernatant. Reverse-transcription was completed using PrimeScript™ RT Reagent Kit (Takara, Dalian, China), which catalyzes RNA to cDNA. Polymerase chain reaction for cDNA was induced by SYBR Green PCR Master Mix (Takara). Final detection for transcriptional expression was performed using a 7500 real-time PCR System (Applied Biosystems, Foster City, USA). The mRNA expression levels were normalized to the house-keeping gene GAPDH and were calculated using the 2^−ΔΔCt method.

Gene microarray was performed as previously described [28]. Briefly, after isolation of total RNA and reverse transcription, cDNA was then subjected to gene expression profiling using the Affymetrix Human Gene 1.0T arrays (Genechem, Shanghai, China). Data were obtained by Genspring 7.0, Expression Console version 1.1.1, and DNA-chip Analyzer 2008 (dChip). The gene expression data were analyzed using the SBC Analysis System (Genechem). The retrieved data showing a fold-change ≥ 1.5 were filtered out. These genes were then functionally classified based on Gene Ontology Database, Affymetrix Database, and DAVID 6.7 Functional Annotation Database.

Cells were transfected with Pokemon siRNA and treated with TGFβ1 or not for 48 hours. EdU (5-ethynyl-2′-deoxyuridine, 50 µM; Cell-Light™ EdU Apollo^®488 In Vitro Imaging Kit; RiboBio, Guangzhou, China) was added to further culture for 2 hours. According to the manufacturer's protocol, the EdU-containing medium was removed and 4% paraformaldehyde was used to fix cells at room temperature for 30 minutes. After removing the paraformaldehyde, lysine (2 mg/mL in deionized water) was added under shaking for 5 minutes, followed by 2 phosphate-buffered saline (PBS) washes. Then, Apollo 480 fluorescent azide reaction buffer was added and allowed to react for 30 minutes in darkness, before washing with 0.5% Triton X-100. After staining with Hoechst dye for 30 minutes, cells were washed with PBS and were finally stored in 100 µL PBS. Images were obtained using a fluorescence microscope. The percentage of EdU positive cells was calculated as follows: (EdU Incorporated Cells/Hoechst Stained Cells) × 100%.

Cells were treated with Pokemon siRNA for 24 hours, transferred into a 96-well plate with a density of 4 × 10³ cells per well, and then treated with TGFβ1 or not for 48 hours. Then 20 µL of MTS reaction buffer was added per well, and cells were further cultured for 2 to 4 hours at 37°C. The optical density values represented the cell viability and were obtained using a microplate reader at a wavelength of 490 nm.

After transfection, cells were cultured in normal culture medium for 14 days. Cells were fixed with 4% paraformaldehyde for 30 minutes at room temperature and then washed twice with PBS. Wright-Giemsa Stain (Baso, Zhuhai, China) was used to stain cell colonies according to the protocol. Briefly, 10 drops of Giemsa solution A were added and allowed to stain for 3 minutes. This was then rapidly mixed with 20 drops of Giemsa solution B, followed by shaking of the cell plates for 5 minutes. The stain solution was then washed with flowing water and finally the plates were air dried. Colony number was counted by direct visual counting.

Total proteins were obtained from the supernatant lysis buffer of breast cancer cells. A BCA protein assay kit (Beyotime, Shanghai, China) was used to measure the protein concentrations according to the manufacturer's instruction. Proteins with different molecular weights were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, Bedford, USA). Membranes were incubated in 5% non-fat milk for 1 hour at room temperature, then at 4°C with primary antibodies overnight, followed by 1 hour at room temperature with the corresponding secondary antibodies. Protein bands were shown by ECL Reagent (Beyotime) and were photographed using the Fluorchem E System (Cell Biosciences, Santa Clara, USA). The antibodies used for GFP-Pokemon, Smad4, and β-actin were obtained from CST (Boston, USA). The secondary antibodies were obtained from Proteintech Group (Chicago, USA).

Cells were collected and fixed with 75% ethanol at −20°C and were blocked with 2% bovine serum albumin in PBS for 1 hour at room temperature. Then, cells were incubated with primary rabbit anti-TGFβ (1:100, CST) or rabbit anti-Smad4 (1:100, CST) antibodies overnight at 4°C. After 3 washes with PBS, cells were incubated with secondary antibodies (anti-rabbit IgG conjugated with Alexa Fluor 594, 1:1,000, CST) for 1 hour. Cell nuclei were shown by DAPI (or 4′,6-diamidino-2-phenylindole) staining. Images were obtained using a laser scanning confocal microscope (LSM510; Carl Zeiss, Jena, Germany).

In order to obtain total protein, transfected cells were harvested and lysed using lysis buffer containing protease inhibitors. Co-immunoprecipitation assay was performed using a co-immunoprecipitation kit (Pierce, Rockford, USA) according to the manufacturer's instructions. Briefly, 3 mg of total protein for each treatment was used in this study, pre-cleared with sepharose resins. The supernatants were divided equally into 2 tubes and incubated into columns containing 1 µg immobilized anti-Pokemon or anti-SP1 antibody (CST). The immunocomplexes were covalently associated with sepharose resins. The resins were eluted 5 times and the co-immunoprecipitates were separated on an SDS-PAGE gel before transferring to a PVDF membrane for analysis with anti-Pokemon or anti-SP1 antibodies.

MCF-7 or MDB-MA-231 cells were seeded into a 6-well plate (2 × 10⁵ cells per well) and were incubated overnight. The cells were then co-transfected with either 0.3 µg/well of Smad4 firefly luciferase reporter plasmid constructs (pLuc366 or pLuc207) or the control pGL3-Basic vector (Promega, San Luis Obispo, USA), then respectively co-transfected with or without Pokemon- and SP1-expressing plasmids. The renilla luciferase plasmid was also co-transfected to correct for variations in transfection efficiency (50 ng/well). After incubating for 48 hours, the cells in each well were separately collected and the luciferase activity was measured using a microplate luminometer. Final data were calculated as the ratio of firefly luciferase activity versus renilla luciferase activity units (n = 3).

Smad4 promoter probes (Probe 1 and Probe 2) were synthesized by Sangon Biotech (Shanghai, China). Probe 1: 5′-AGCGGGAGGCGGGGGCAGCC-3′ for the wild-type promoter (Probe); 5′-AGCAAAAGGCAAAAACAGCC-3′ for the mutant promoter (mProbe). Probe 2: 5′-AAAGCTGCGGGGGAAAAG-3′ for the wild-type promoter; 5′-AAAGCTGCAAAAAAAAAG-3′ for the mutant promoter. Probes were labeled with biotin. Nuclear extracts of MDB-MA-231 cells were prepared using a nuclear extraction kit (Pierce). In brief, an electrophoretic mobility shift assay (EMSA) was performed as follows: nuclear extracts (20 µg) and labeled probes (200 fM) were incubated for nuclear extract binding reaction; probes without nuclear extracts were used as the negative control; a combination of nuclear extracts, probes (Probe or mProbe), and unlabeled probe (com-Probe, 200×) were used for the competition reaction; a combination of nuclear extracts, probes, and IgG, anti-Pokemon or anti-SP1 antibodies were used for the supershift reaction. DNA-protein complexes were loaded onto a 6% native polyacrylamide gel. After electrophoresis, separated DNA was transferred to a nylon membrane before being cross-linked and photographed.

Analyses in this study were performed using the GraphPad Prism statistical software (GraphPad Software, La Jolla, USA). Experimental data were represented as the mean ± standard error of the mean. Statistical analysis was performed via one-way analysis of variance followed by Dunnett's test for multiple comparisons. The p-values less than 0.05 were considered statistically significant.

In order to verify whether Pokemon acts as a pro-oncoprotein or a tumor suppressor protein, and whether Pokemon promotes or inhibits cell proliferation arrest in breast cancer cells, cell proliferation ability-related experiments were applied (Figure 1). For alternative Pokemon downregulation or TGFβ treatment in MDB-MA-231 cells, DNA replication ability verified by EdU incorporation assay was significantly suppressed (Figure 1A and B). The cell proliferation rate, measured by MTS assay, was reduced (Figure 1C). Cell viability observed by optical photograph was reduced (Figure 1D and E) and cell colony numbers were less than control cells (Figure 1F and G). Co-treatment of Pokemon siRNA and TGFβ proteins presented contrasting results. This data highlights the key inhibition role of Pokemon in TGFβ1-induced cell proliferation arrest in breast cancer cells.

To elucidate the mechanism by which Pokemon is involved TGFβ-dependent cell proliferation arrest in breast cancer cells, we used gene microarray to screen for signaling related to Pokemon expression in MDB-MA-231 cells. Several signaling pathways, such as those of the metabolism of lipids and lipoproteins, hemostasis, and the adaptive immune system, were revealed to be responsive to Pokemon alteration (Figure 2A). We also found that “pathway in cancer”-associated genes, such as TGFβ, Smad4, TGFα, MYC, and BIRC3I, were repressed by ectopic Pokemon expression in MDB-MA-231 cells (Figure 2B). Ectopic Pokemon expression also inhibited the expression of Smad4, as detected by immunofluorescence (IF) assay and western blotting (Figure 2C-F).

We further determined the effect of Pokemon on the expression of proliferation-associated genes in the TGFβ pathway by using PCR gene expression array in breast cancer cells with Pokemon siRNA. This analysis identified multiple genes involved in TGFβ signaling-associated cellular proliferation regulated by Pokemon in breast cancer cells. The expression levels of 13 key genes associated with cellular proliferation during TGFβ signaling were analyzed. Pokemon siRNA was found to result in the downregulation of CYCLIND1, CDKN1A, GADD45B, and CDC6, and the upregulation of TGFβ1 in both MDB-MA-231 and T47D cells. TGFβ2, TGFβ3, CDKN1B, IGFBP1, ATF3, EMP1, PTK2, and ACVR1 showed distinct responses to the downregulation of Pokemon in MDB-MA-231 and T47D cells (Figure 3A and B). The levels of the genes above, with the exception of IGFBP1, were all downregulated after Pokemon siRNA transfection in MCF-10A cells (Figure 3C). Western blotting demonstrated that the protein levels of CYCLIND1 and CDKN1A also decreased after Pokemon knockdown in three breast cancer cell lines (Figure 3D). These results suggest that Pokemon regulates multiple proliferation associated genes in the TGFβ pathway and could thus be an essential modulator in TGFβ pathway-induced proliferation inhibition in breast cancer cells.

To further identify the mechanism of Pokemon-induced Smad4 repression, we used a dual luciferase reporter system to assess the role of Pokemon in the transcription-regulation of Smad4. Two specific luciferase plasmids containing the Smad4 proximal promoter sequence, pLuc366 and pLuc207, were used to test Smad4 promoter activity. As a result, Pokemon was found to reduce Smad4 luciferase activity in MDB-MA-231 cells (Figure 4A). Furthermore, an EMSA was used to confirm whether the Pokemon protein bound to the Smad4 promoter sequences. Two specific Smad4 promoter probes (Probe 1 and Probe 2) with a GC box sequence were used to bind with the nucleus extraction obtained from MDB-MA-231 cells. Pokemon antibody was found to be unable to trigger the supershift of the protein DNA complex (Figure 4B), indicating that Pokemon has no direct protein-DNA interactions with the Smad4 promoter sequences. SP1, which was previously proved to regulate Smad4 promoter activity during invasion, was found to bind with Probe 1. Moreover, the ectopic expression of Pokemon was revealed to decrease the binding affinity of SP1 with Probe 1. The co-immunoprecipitation assay also showed the direct interaction between SP1 and Pokemon in MDB-MA-231 cells (Figure 4C), and a dual luciferase reporter assay verified that the SP1 enhanced-promoter activity of Smad4 was abrogated by Pokemon in MDB-MA-231 cells (Figure 4D). These results indicated that Pokemon functions as a negative modulator of Smad4 expression by interacting with SP1, leading to the separation of SP1 from the Smad4 promoter sequence, leading to decreased transcription activity of Smad4 in breast cancer cells.

After confirming that Pokemon decreases the expression of Smad4 and blocks breast cancer cell growth in vitro, the correlation between the expression of these 2 proteins was identified in clinical samples. To further validate the correlation between Pokemon and Smad4 expression in a broad range of clinical samples, we analyzed public gene expression microarray datasets of breast cancer tissues from 2 The Cancer Genome Atlas (TCGA) datasets: Breast Invasive Carcinoma (TCGA, Cell 2015; http://www.cbioportal.org/study?id=brca_tcga_pub2015#summary) and Breast Invasive Carcinoma (TCGA, Nature 2012; http://www.cbioportal.org/study?id=brca_tcga_pub#summary). Consistent with our in vitro data, the Pokemon mRNA levels showed a significantly negative correlation with Smad4 mRNA levels in different subtypes of breast cancer cases across the two independent datasets. Pokemon showed the highest mRNA levels in the subtype of Luminal A and the lowest levels in the basal-like subtype (Figure 5A and B, Supplementary Tables 1 and 2).