AR–PDEF pathway promotes tumour proliferation and upregulates MYC-mediated gene transcription by promoting MAD1 degradation in ER-negative breast cancer

BC cell lines MDA-MB-453 and SKBR-3 used in this study were purchased from Type Culture Collection of the Chinese Academy of Sciences, Shanghai, China. Results of gene profiling studies have shown that MDA-MB-453 cells are molecular apocrine (ER⁻/PR^−/AR⁺) BC cells and show high AR expression [14]. MDA-MB-453 cells were cultured in L15 medium (Gibco, USA) containing 10% foetal bovine serum (FBS; Gibco) and 1% penicillin/streptomycin (Life Technologies, USA) at 37 °C in an incubator lacking CO₂. SKBR-3 cells were cultured in RPMI 1640 medium supplemented with 10% FBS at 37 °C in a 5% CO₂ incubator. Next, the two cell lines were treated with 1 nM dihydrotestosterone (DHT; Sigma-Aldrich, USA) for 0 or 48 h or with different dose of DHT for 48 h.

Immunohistochemistry (IHC) analyses were performed as described previously [21]. Antibodies against AR (ab9474; dilution, 1:200), PDEF (ab197375; dilution, 1:200), MAD1 (ab175245; dilution, 1:200) and MYC (ab32072; dilution, 1:200) were purchased from Abcam. Anti-Ki67 antibody (sc-23,900; dilution, 1:200) was purchased from Santa Cruz Biotechnology. Normal breast tissue sections were processed simultaneously and were used as positive controls for AR and PDEF, and normal goat serum-substituted primary antibodies were used as negative controls. Two senior pathologists independently quantified IHC slides. IHC scores of PDEF were used to the multiplied result of percentage positivity and staining intensity in the stained tissue area, and total scores ranged from 0 to 6. Percentage positivity was scored as 0 (0–25%), 1 (26–50%) and 2 (> 50%), and staining intensity was scored as 0 (no staining), 1 (weak staining), 2 (moderate staining) and 3 (strong staining). A total score of ≥0 and ≤ 3 indicated negative PDEF expression, and a total score of ≥4 indicated positive PDEF expression [21]. AR expression was considered to be positive if nuclear staining was observed in > 10% tumour cells.

Western blotting was performed as described previously [21] by using the following primary antibodies: anti-AR antibody (ab9474; dilution, 1:3000), anti-PDEF antibody (ab53881; dilution, 1:1000; Abcam), anti-MAD1 antibody (ab175245; dilution, 1:3000), anti-MYC antibody (ab32072; dilution, 1:3000), anti-β-catenin antibody (ab32572; dilution, 1:3000; Abcam), anti-AKT antibody (sc-135,829; dilution, 1:3000; Santa Cruz Biotechnology), anti-phosphorylated AKT antibody (anti-p-AKT; sc-7985-R; dilution, 1:3000; Santa Cruz Biotechnology), anti-ERK antibody (sc-514,302; dilution, 1:3000; Santa Cruz Biotechnology), anti-phosphorylated ERK antibody (anti-p-ERK; sc-81,492; dilution, 1:3000; Santa Cruz Biotechnology) and anti-EGFR antibody (ab52894; dilution, 1:3000; Abcam).

Immunofluorescence staining was performed as described previously [21]. For this, BC tissue sections or cells were stained with the antibodies against AR (ab9474; dilution, 1:200) and PDEF (ab53881; dilution, 1:200). Quantification was performed using 4–6 independent fields.

Lentivirus infection was performed using Lenti-Pac™ HIV Expression Packaging Kit (GeneCopoeia, Guangzhou, China). Lentiviruses produced in 293 T cells were used to infect BC cells cultured in a medium containing 5 μg/mL polybrene. Lentiviral vectors expressing four independent shRNAs against PDEF or AR and those inducing PDEF or MAD1 overexpression were obtained from GeneCopoeia. After the infection, cells were selected using puromycin.

Cell lysates were generated using Complete Mini protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany). Total protein concentration in the cell lysates was measured using Pierce BCA protein assay kit (Thermo Scientific, Bonn, Germany) and was analysed using Eppendorf Master Photometer. Co-immunoprecipitation (Co-IP) assay was performed using the cell protein lysates and Pierce Co-IP kit (Thermo Scientific), according to the manufacturer’s protocol. For this, 10 μg anti-AR antibody (ab9474) or anti-PDEF antibody (ab53881) was incubated with a delivered resin and was covalently coupled. The antibody-coupled resin was incubated with the cell protein lysates overnight at 4 °C. Next, the resin was washed, and protein complexes bound to the antibody were eluted and examined by performing western blotting.

Chromatin immunoprecipitation (ChIP) assay was performed according to manufacturer’s (Millipore) instructions. Briefly, DNA in BC cell was cross-linked with histones by adding formaldehyde for 10 min at room temperature. BC cell were sonicated in SDS lysis buffer to produce cell lysates containing 300- to 500-bp chromatin fragments. Next, antibodies were incubated with Dynabead proteins A and G (Invitrogen) for 6 h, followed by overnight incubation with the sonicated cell lysates for chromatin collection. Amount of immunoprecipitated DNA was normalised to that in the input and was expressed relative to the amount of DNA present in a negative control intergenic region. Primers used for the ChIP assay are listed in supplemental document.

The upper chamber of a Transwell was coated with Matrigel (BD Bioscience, USA) for performing cell invasion assay. Briefly, BC cells (density, 1 × 10⁵ cells) were seeded and incubated in the upper chamber containing an FBS-deficient medium. The lower chamber was filled with a 10% FBS-containing medium. After incubation at 37 °C for 24 h, the cells in the upper chamber were removed with a cotton swab. The reverse face of the membrane contained cells that had invaded the membrane. The invaded cells were fixed with 4% paraformaldehyde and stained with Giemsa. Cell migration was assessed by performing wound-healing assay. For this, BC cells (density, 1 × 10⁶ cells) were cultured in a 3-cm dish and were wounded using a 100 μL plastic pipette tip. After 48 h, the size of the wound was measured and photographed.

Cell proliferation assay was performed using Cell Counting Kit-8 (Dojindo, Japan). Briefly, BC cells were plated in 96-well plates in triplicate at an approximate density of 3 × 10⁴ to 5 × 10⁴ cells per well and were cultured under a standard culture condition. The cells were then treated with the indicated reagent, and the number of cells per well was determined by measuring absorbance (450 nm) at indicated time points.

Cell cycle analysis was performed by staining BC cells with PI by using CycleTEST™ PLUS DNA reagent kit (BD Biosciences), according to the manufacturer’s instruction.

Approximately 500 BC cells were seeded in each well of a six-well plate and were incubated for 7 days. Colonies of these cells were fixed with methanol for 30 min and were stained with 0.1% crystal violet for 1 h.

Treated BC cells (density, 3 × 10⁶ cells) together with 100 μg Matrigel were inoculated into the mammary fat pads of 5-week-old female SCID mice. Tumour growth was recorded twice a week with a caliper-like instrument. Tumour volume was calculated using the formula tumour volume = (width² × length)/2. The mice were sacrificed after 6 weeks according to the guidelines for the welfare and use of animals in cancer research, and the final tumour volume and weight were determined. All in vivo experiments were reviewed and approved by the Animal Ethics Committee of TMUCIH and were performed according to the guidelines for the welfare and use of animals in cancer research and national law [22].

Tissues were fixed in 10% neutral-buffered formalin for 24 h, embedded in paraffin, cut into 4-μm-thick sections, deparaffinised with xylene and processed with a graded ethanol series. Next, the sections were stained with H&E and were observed using BX51 microscope (Olympus).

Data are presented as mean ± standard deviation (SD) of at least three independent experiments. Student’s t-test, χ² test and Fisher’s exact test were used to compare two groups by using SPSS 22.0 (IBM, Chicago, IL, USA). Kaplan–Meier test was used to estimate the OS and RFS. p < 0.05 was considered statistically significant.

AR and PDEF expression levels were first examined by performing IHC analysis of the 100 ER-negative BC tissues. PDEF showed a nuclear staining pattern, with little or no cytoplasmic or membranous staining. Of the 100 samples, 60 (60%) showed positive nuclear PDEF expression (PDEF⁺) and 69 (69%) showed positive nuclear AR expression (AR⁺). PDEF expression was associated with tumour grade (p = 0.032), pTNM stage (P = 0.011), lymphatic metastasis (P < 0.001) and AR expression (P < 0.001) (Table 1). PDEF⁺ tumours significantly more often showed AR positivity (Fig. 1a and b). AR and PDEF were more often co-expressed, with 55 (55%) cases having AR⁺PDEF⁺ tumours. Next, we examined the clinicopathological variables between AR⁺PDEF⁺ and others in the 100 ER-negative BC specimens and found that AR⁺PDEF⁺ was associated with pTNM stage (P = 0.047) and HER2 expression (P = 0.039) (Table 1). Results of western blotting showed that PDEF protein expression was higher in ER-negative BC tissues than the corresponding adjacent normal tissues (Fig. 1c). Kaplan–Meier survival analysis showed that high PDEF expression was associated with poor OS (P = 0.040) and RFS (P = 0.031); moreover, AR and PDEF co-expression was associated with poor OS (P = 0.043) and RFS (P = 0.027) (Fig. 1d).

ER-negative and AR-positive BC cell lines MDA-MB-453 [14, 23] and SKBR-3 [24] were treated with 1 nM DHT for 48 h to promote AR expression or were infected with an AR-shRNA-expressing lentiviral vector to inhibit AR expression. Upregulated AR expression promoted PDEF mRNA and protein overexpression, whereas downregulated AR expression significantly inhibited PDEF mRNA and protein expression (Fig. 2a). Next, we performed immunofluorescence staining to reassess AR and PDEF protein levels and obtained the same results (Fig. 2c). A time-course analysis of MDA-MB-453 and SKBR-3 cells showed that treatment with increasing DHT doses strongly increased PDEF mRNA expression (Fig. 2b). These results indicated a positive role of AR in the regulation of PDEF expression. We next examined whether AR protein physically interacted with PDEF protein. We found that AR co-immunoprecipitated with PDEF in both MDA-MB-453 (Fig. 2d) and SKBR-3 cells (Additional file 1: Figure S1a) because of DHT-induced interaction between these proteins. To further examine whether AR regulated PDEF expression, we examined four AR-binding regions in the PDEF locus mentioned in a previously established AR cistrome dataset [25]. The first one is located within the PDEF promoter, two enhancers (1 and 2) are located within the first intron of PDEF and the fourth one is located within the 3′-untranslated region of PDEF. Direct AR ChIP assay by using MDA-MB-453 cells showed DHT-induced recruitment of AR at the second enhancer of PDEF, thus confirming that PDEF was a direct target of AR (Fig. 2e).

We previously analysed PDEF protein expression and clinical outcome data in the 100 ER-negative BC tissues and found a significant correlation between high PDEF expression and poor OS. These findings strongly supported the potential role of PDEF in ER-negative BC tumorigenesis. Next, we evaluated AR and PDEF protein levels in the two ER-negative BC cell lines MDA-MB-453 and SKBR-3. Both AR and PDEF proteins were highly expressed in MDA-MB-453 cells but showed low expression in SKBR-3 cells (Fig. 3a). To determine whether PDEF expression affected cellular growth, SKBR-3 cells were infected with a PDEF-expressing lentiviral vector to promote PDEF expression and MDA-MB-453 cells were infected with a PDEF-shRNA-expressing lentiviral vector to inhibit PDEF expression (Fig. 3b and c). Results of the CCK-8 assay showed increased proliferation of PDEF-upregulated SKBR-3 cells and decreased proliferation of PDEF-downregulated MDA-MB-453 cells (Fig. 3d). Results of the cell cycle analysis showed that the percentage of S-phase cells was higher among PDEF-upregulated SKBR-3 cells than among control SKBR-3 cells but was lower among PDEF-downregulated MDA-MB-453 cells than among scramble shRNA-expressing MDA-MB-453 cells (Fig. 3e). To further examine the effect of PDEF in BC tumorigenesis, we examined the biological effect of PDEF on cancer cell invasion and migration by performing cell migration and invasion assays. Cellular PDEF overexpression significantly increased the invasion and migration of SKBR-3 cells, whereas PDEF knockdown decreased the invasion, migration and proliferation of MDA-MB-453 cells (Fig. 3f, g and h; Additional file 1: Figure S2a and b). Together, these results suggest that PDEF promotes ER-negative BC cell proliferation.

Studies have shown that AR facilitates the growth of ER-negative and AR-positive BC cells. In the present study, we found that PDEF was the downstream target gene of AR and that its expression was upregulated by AR. Moreover, we found that PDEF overexpression promoted the growth of ER-negative BC cells. To determine the molecular mechanism underlying the oncogenic effect of PDEF on ER-negative BC cell proliferation, we analysed AR downstream intracellular signalling components, including MEK/ERK, PI3K/AKT and MYC/MAD1. PDEF overexpression or downregulation in SKBR-3 or MDA-MB-453 cells, respectively, did not alter the expression of AKT, ERK, p-AKT and p-ERK, the key signal transducers acting downstream of AR (Fig. 4a and b). Moreover, we observed that PDEF overexpression or downregulation in SKBR-3 or MDA-MB-453 cells, respectively, did not affect AR expression, indicating the absence of a reciprocal regulatory loop between PDEF and AR. However, we found that MYC expression was significantly upregulated in PDEF-overexpressing SKBR-3 cells and was downregulated in PDEF-downregulated MDA-MB-453 cells, indicating a positive role of PDEF in regulating MYC expression. Moreover, we found that the expression of MAD1, a transcriptional repressor of MYC, was inhibited in PDEF-overexpressing SKBR-3 cells and was upregulated in PDEF-downregulated MDA-MB-453 cells, indicating a negative role of PDEF in regulating MAD1 expression.

PDEF co-immunoprecipitated with MYC in PDEF-upregulated SKBR-3 cells, which are suggested to show no interaction between PDEF and MYC (Fig. 4c). MAD1 functioned as an antagonist of MYC, and MAD1 overexpression in SKBR-3 and MDA-MB-453 cells downregulated MYC expression (Additional file 1: Figure S3a). Next, we examined whether PDEF upregulated MYC expression by promoting MAD1 degradation. We observed that PDEF co-immunoprecipitated with MAD1 in PDEF-downregulated MDA-MB-453 cells, which are suggested to show a negative interaction between PDEF and MAD1 (Fig. 4d). In addition, PDEF overexpression or downregulation in SKBR-3 or MDA-MB-453 cells, respectively, did not affect MYC mRNA levels (Fig. 4e), thus confirming that PDEF does not regulate MYC expression. However, PDEF upregulation in SKBR-3 cells decreased in MAD1 mRNA expression, whereas PDEF downregulation in MDA-MB-453 cells increased MAD1 mRNA expression, thus confirming that MAD1 is a direct target of PDEF. Results of a direct ChIP-qPCR assay showed PDEF recruitment at the MAD1 promoter (Fig. 4f). These data suggest that PDEF upregulates MYC-mediated gene transcription by promoting MAD1 degradation.

Because PDEF represses MAD1 expression, upregulation of MAD1 expression may suppress PDEF-mediated growth of ER-negative BC cells. To verify this hypothesis, we analysed PDEF-upregulated, simultaneous PDEF- and MAD1-upregulated and control SKBR-3 cells. We first confirmed MYC protein expression in SKBR-3 cells by performing western blotting. PDEF overexpression increased MYC expression; however, MAD1 expression suppressed PDEF overexpression-induced increase in MYC expression (Fig. 5a). Next, we performed the Transwell and colony-forming assays to evaluate the alteration in the invasive and stem-like properties of tumour cells. MAD1 expression suppressed PDEF overexpression-induced migration and proliferation of SKBR-3 cells (Fig. 5b and c). Results of the cell cycle analysis showed that MAD1 overexpression decreased PDEF upregulation-induced increase in the number of S-phase cells (Fig. 5d). To investigate the effect of MAD1 overexpression on PDEF, female nude mice were inoculated with stable PDEF-upregulated, stable simultaneous PDEF- and MAD1-upregulated and control SKBR-3 cell clones. We found that MAD1 over expression dramatically inhibited PDEF-mediated growth of (Fig. 5e and f) and reduced PDEF overexpression-induced Ki67 and MYC expression in (Fig. 5g, h and i) tumours isolated from the BC cell-inoculated nude mice. Moreover, compared with the mice inoculated with PDEF-overexpressing cells, the mice inoculated with MAD1-overexpressing cells did not show pulmonary metastasis. These findings indicate that upregulation of MAD1 expression can inhibit PDEF-induced proliferation of ER-negative BC cells.

We found that the AR–PDEF pathway promoted the growth of ER-negative BC cells and that PDEF was the downstream target gene of AR and was upregulated by AR. In addition, we found that PDEF promoted tumour proliferation and upregulated MYC-mediated gene transcription by promoting MAD1 degradation. Next, we investigated whether simultaneous inhibition of AR and PDEF expression further suppressed tumour proliferation compared with the inhibition of AR alone. For this, we analysed AR-downregulated (AR-deprived), simultaneous AR- and PDEF-downregulated (AR-/PDEF-deprived) and control MDA-MB-453 cell clones. MAD1 protein levels were higher in AR-/PDEF-deprived MDA-MB-453 cells than in AR-deprived and control MDA-MB-453 cells, whereas, MYC protein levels were lower in AR-/PDEF-deprived MDA-MB-453 cells than in AR-deprived and control MDA-MB-453 cells (Fig. 6a). Next, we examined functionally relevant changes in these cells. Results of the cell cycle analysis showed that the percentage of S-phase cells was lower among AR-/PDEF-deprived MDA-MB-453 cells than among AR-deprived and control MDA-MB-453 cells (Fig. 6b). Results of the colony-forming and CCK-8 assays showed that the proliferation of AR-/PDEF-deprived MDA-MB-453 cells was lower than that of AR-deprived and control MDA-MB-453 cells (Fig. 6c and f). Results of the wound-healing and Transwell assays showed that AR-/PDEF-deprived MDA-MB-453 cells showed significantly suppressed migration and invasion potential compared with AR-deprived and control MDA-MB-453 cells (Fig. 6d and e; Additional file 1: Figure S4a). To examine whether simultaneous inhibition of AR and PDEF expression was sufficient for inhibiting tumour proliferation and formation, female nude mice were inoculated with stable AR-downregulated, stable simultaneous AR- and PDEF-downregulated and control MDA-MB-453 cell clones. We found that the simultaneous inhibition of AR and PDEF expression dramatically inhibited tumour growth (Fig. 6g and h) and considerably reduced Ki67 and MYC expression (Fig. 6i, j and k). These results indicate that the simultaneous inhibition of AR and PDEF expression significantly suppresses the proliferation of ER-negative BC cells.

Thus, in the present study, we found that AR and PDEF are more often co-expressed and that AR directly upregulates PDEF expression, leading to its activation in ER-negative BC tissues and cells. Activated PDEF downregulates the expression of MYC transcriptional repressor MAD1, thus promoting its degradation and dissociation from MAX, the obligatory partner of MYC. In the absence of MAD1 competition, MYC forms heterodimers with MAX to sequentially induce gene transcription for promoting the proliferation of ER-negative BC cells (Fig. 6l).