Attenuation of TGFBR2 expression and tumour progression in prostate cancer involve diverse hypoxia-regulated pathways

Paired PCa and adjacent normal tissue samples (collected postoperatively from September 2010 to February 2017) were obtained from 56 patients who were admitted to and underwent PCa resection in the Department of Urology, Tongji Hospital, Huazhong University of Science and Technology (Wuhan, China). Specimens were obtained under the informed consent of each patient and the approval of Ethics Committee of Tongji Hospital. All the diagnoses were validated by pathology reports. On removal of the surgical specimen, each sample was immediately frozen in liquid nitrogen and stored at − 80 °C until use. The clinicopathologic characteristics of patients are summarized in Additional file 1: Table S1.

Human prostate cancer cell lines (PC3, DU145 and LNCap) were obtained from the American Type Culture Collection (Manassas, VA), and a primary culture of human RWPE-1 was obtained from the Shanghai cell bank of the Chinese Academy of Sciences. Generally, PC3, DU145 and LNCap were cultured in RPMI-1640, while RWPE-1 was culture in K-SFM medium, supplemented with 10% fetal bovine serum (FBS) and placed in incubator with 5% CO2 at 37 °C. For hypoxic conditions, cells were cultured in a Forma Series II Water Jacket incubator with 1% O2, 5% CO2 and 94% N2 (Thermo Scientific).

Synthetic, chemically modified short single or double stranded RNA oligonucleotides and corresponding negative control were purchased from Ribo Biotech (Guangzhou, China). SiRNA sequences are available in Additional file 1: Table S2. For transfection, cells were seeded into 6-wells plate and incubated overnight, and then 100 nM of small RNA molecules were transfected into cells by using X-tremeGENE siRNA transfection reagent (Roche). Besides, plasmids were transfected into cells with X-tremeGENE HP DNA Transfection reagent (Roche).

Total RNA from tissues and cultured cells were isolated using TRIzol reagent (Invitrogen, USA). Total RNA measuring 1000 ng was reverse transcribed, and real-time quantitative polymerase chain reaction (RT-qPCR) was performed on the Mx3000P system (Stratagene, USA) according to the manufacturer’s instructions. A qRT-PCR analysis was performed with the Platinum SYBR Green qPCR SuperMix kit (Invitrogen, Carlsbad, CA). U6 snRNA and GAPDH were used as internal controls for miR-93 and TGFBR2, respectively. All primers sequences used are available in Additional file 1: Table S2.

Whole-cell lysates were prepared with RIPA lysis buffer of appropriate volume. Proteins were separated by 10% SDS-PAGE gel and transferred to PVDF membrane, following membrane blocking with 5% Albumin from bovine serum (BSA). Then the membranes were incubated in primary antibodies against targeted protein at 4 °C overnight, and after further incubated with HRP-conjugated secondary antibody, the bands were scanned. Antibodies used in experiments included: anti-TGFBR2 (Cell Signaling Technology, USA), anti-EZH2 (Abcam, USA), anti–HIF1A (Cell Signaling Technology, USA), anti-HIF2A (Cell Signaling Technology, USA), and anti-H3K27me3 (Abcam, USA). All primary antibodies except for internal control were used at 1:1000 dilution. GAPDH and α-tubulin (Boster, China) were used as the internal control at 1:5000 dilution.

To determine whether TGFBR2 expression can be elevated in prostate cancer cells by demethylation, we treated the cells with 5-aza-2′-deoxycytidine (5-Aza-DC). Cells were seeded in 6-well plates and treated with 5-Aza-DC (Sigma-Aldrich) at the final concentration of 10 μM, and culture for 72 h before the total RNA was extracted. The relative expression of TGFBR2 was measured by qRT-PCR with the untreated cells as control.

Genomic DNA from prostate cancer tissues and corresponding normal tissues and prostate cell lines was isolated using a simplified Proteinase K digestion method, and then bisulfited using Epitect Fast Bisulfite Conversion Kit(Qiagen) according to the manufacturer’s instructions. Then the methylation status of TGFBR2 promoter was determined by methylation specific PCR by amplified the genomic DNA after standard sodium bisulfate modification. Primer sequences which specifically recognized either the methylated or unmethylated TGFBR2 promoter were designed with Methprimer (http://​www.​urogene.​org/​cgi-bin/​methprimer/​methprimer.​cgi). The PCR conditions were: 95 °C for 5 min, followed by 40 cycles of 95 °C for 15 s, 57 °C for 30s, 72 °C for 30s; and a final extension at 72 °C for 10 min. PCR products were separated on 1.5% agarose gels which stained with Gel-Red, and visualized under UV illumination.

Proliferation of cells was detected by the MTS assay using the CellTiter 96®AQueous One Solution Cell Proliferation Assay kit (Promega). One thousand treated cells were seeded in each well in a 96-well plate, the absorbance at 490 nm was measured every 24 h from day 0 to 4.

For colony formation assay, 1000 treated cells were seeded into 6-wells plates and cultured for another 10–14 days. The colonies were stained with 0.5% crystal violet (Sigma, USA) and counted.

Cell apoptosis was detected via Annexin V Apoptosis Detection Kit APC (eBioscience, Affymetric). Briefly, the cells were washed in PBS and resuspended in 1 × binding buffer. Then the cells were stained by fluorochrome-conjugated Annexin V for 15 min at room temperature. After we washed the cells and resuspended them in binding buffer, we stained the cells with Propidium iodide (PI) staining solution and then analyzed them by flow cytometry.

About 5 × 10⁴ of PC3 and DU145 cells were plated in the upper chambers of 24-well Transwell plates (Corning) in FBS-free medium. Complete medium (10% FBS) was deposited in the lower chambers to serve as a chemo-attractant. After 12 h, cells remaining on the upper filter were removed, while cells that passed through the Transwell filter were stained by 0.5% crystal violet for 20 min. Images were taken of five random optical fields (200×) on each filter and cell number was quantified by utilizing the ImageJ software (National Institutes of Health, USA).

To evaluate cell invasion, Transwell membranes were coated with Matrigel (BD Biosciences) prior to seeding treated cells or control cells. The Matrigel served as a basement membrane barrier that cells would have to destroy in order to invade the lower chamber. After 24 h for PC3 and DU145, crystal violet staining and cell counting were performed as above.

After constructing the wild-type 3′-untranslated region (3’-UTR) of TGFBR2 using PCR, we sub-cloned it into the NotI and XhoI sites in the luciferase reporter plasmid psiCHECK2 (Promega). All used primers were included in Additional file 1: Table S3. DNA sequencing was used to validate the validity of plasmids. After co-transfection with siRNAs and plasmids, cells were cultured for another 48 h before harvested. Then the cells were lysed and assayed for Renilla and Firefly luciferase activity using Dual-Luciferase Reporter Assay System (Promega). The ratio of Renilla luciferase intensity to luciferase one was calculated for each assay.

Recent studies have provided evidence that hypoxia promote cancer progression in prostate cancer, while the exact effects of hypoxia on EMT are not fully understood. As TGF-β signalling plays critical roles in PCa, we sought to study the impact of hypoxia on the pivotal receptor, TGFBR2. The Cancer Genome Atlas (TCGA) data showed that TGFBR2 expression was significantly reduced in prostate cancer tissues (Additional file 1: Figure S1). Since high level of hypoxia and decreased TGFBR2 expression has been reported to occur in human or rat PCa [15, 16], we hypothesized that hypoxia might at least partly contribute to low expression of TGFBR2 in PCa. In an attempt to test our hypothesis, we cultured three of prostate cancer cell lines, PC3, DU145 and LNCap, under hypoxic conditions to evaluate the impacts of hypoxia. 24 h after the hypoxic treatment, cells were harvested and the protein and mRNA expression levels of TGFBR2 were examined by western blot and RT-PCR, respectively. As shown in Fig. 1a and Additional file 1: Figure S4, compared to cells in normoxia, all cell lines in hypoxia showed decreased levels of TGFBR2 protein, while the well-established hypoxia indicator HIF-1α protein expression was elevated. RT-PCR results also demonstrated that hypoxia treatment decreased the mRNA levels of TGFBR2, which was in accordance with the western blot results (Fig. 1b). These results indicated that TGFBR2 expression was attenuated by hypoxia pathway. However, since the transcription factors (HIF-1α and HIF-2α) always activated the transcription of downstream gene with hypoxia response element (HRE, consensus sequence G/ACGTG) on their promoter region [20], which seemed that TGFBR2 was not likely to be a direct downstream molecule of hypoxia. As a result, we turned to find other hypoxia-regulated pathways which involved in TGFBR2 down-regulation.

One of the common reasons of transcriptional silencing of tumour suppressor genes in cancer is CpG-island promoter hypermethylation. Previous studies have shown that methylation silencing of TGFBR2 resulted in lower expression of TGFBR2 in rat prostate cancer. However, we further validated the epigenetic changes of TGFBR2 in human PCa. Firstly, we treated the three prostate cancer cell lines with 10 μmol/L of DNA demethylation agent 5-Aza-2′-deoxycytidine (5-Aza-DC), and detected the expression level of TGFBR2 by RT-PCR. As seen in Fig. 2a, the TGFBR2 expression in all cell lines elevated significantly after the 5-Aza-DC treatment, which implied the potential regulatory roles of DNA methylation. Besides, we detected the methylation status of TGFBR2 promoter in our six paired prostate tissues by methylation specific PCR (MSP). Using MethPrimer algorithm, we detected CpG-island around the presumed transcriptional start site (TSS) and designed the methylation (M) and unmethylation (U) primers (Fig. 2b). As shown in Fig. 2c, only methylated DNA molecules were detected in six prostate cancer tissues, while only unmethylated DNA molecules were identified in corresponding noncancerous tissues. As a result, methylation silencing of TGFBR2 in prostate cancers contributed at least partly the decreased expression of TGFBR2.

It reported that rat prostate cancer cell lines had increased H3K4me2 or H3K9me3 at the Tgfbr2 promoter, while human prostate cancer cell lines were likely to have increased H3K27me3 at the TGFBR2 promoter [15]. As a histone methyltransferase, the enhancer of zeste homolog 2 (EZH2) is proved to exert its primary function to methylate H3K27me by transferring methyl group. EZH2 has been proved to exert important oncogenic functions in PCa, TCGA data also implies that high expression of EZH2 was associated with worse overall survival and disease-free survival (Additional file 1: Figure S2). Therefore, we hypothesized that EZH2 induced the epigenetic silencing of TGFBR2 by H3K27me3 or other epigenetic modifications. Firstly, we examined the expression relationship of EZH2 and TGFBR2 using the Pearson correlation coefficient analysis. Our panel of 56 PCa tissues showed significant inverse correlation between the expression of EZH2 and TGFBR2 (r = − 0.273, p < 0.01) (Fig. 2d). By processing the data from TCGA, we also found significant negative correlation between the expression of these two molecules in 568 cases of PCa (r = − 0.3479, p < 0.0001) (Additional file 1: Figure S3). To further validate our hypothesis, we used siRNA targeting EZH2 or EZH2 inhibitor 3-deazaneplanocin (DZNep) to knockdown EZH2 expression in DU145 cell lines. After the reduction of EZH2 and H3K27 trimethylation, the mRNA and protein expression levels of TGFBR2 elevated significantly (Fig. 2e, f, Additional file 1: Figure S5). Moreover, we conducted MSP with prostate cancer cells which were treated with DZNep or not. The results in Fig. 2g illustrated that DZNep could partially decreased the methylation levels of cancer cell lines and helped the cells restore the hypomethylation status, as both methylated and unmethylated DNA molecules could be detected in DZNep-treated cells. Besides, other epigenetic modulators, such as H3k9 methylation inhibitor BIX-01294 [24], had only slight effects on MSP results, which also proved the importance of EZH2-mediated H3K27me3 on regulating the TGFBR2 promoter methylation. These combined results demonstrated that EZH2 regulated TGFBR2 promoter hypermethylation could contribute to its epigenetic silencing in PCa.

In an effort to explore the mechanism of hypoxia-induced TGFBR2 down-regulation, we further tried to evaluate the impact of hypoxia on EZH2. Previous study showed the expression of EZH2 could be regulated by HIF-1α/HIF-1β dependent hypoxic pathway in prostate cancer [25]. Therefore, we detected the EZH2 mRNA and protein expression of prostate cancer cells under normoxic or hypoxic conditions. As we could see in Fig. 2h, i and Additional file 1: Figure S6, hypoxia induced both the mRNA and protein expression of EZH2. Taken together, hypoxia-induced EZH2 mediated the epigenetic silencing of TGFBR2 in PCa through promoter hypermethylation.

MicroRNAs (miRNAs) are a sort of small, endogenously expressed, well-conserved non-coding single stranded RNAs which can negatively regulate gene expression by binding directly to the 3′ untranslated region (3’-UTR) of corresponding target messenger RNAs (mRNAs) in a sequence-specific manner. Previous studies have identified a series of hypoxia-induced miRNAs in several cancers, such as miR-182 in PCa [22], and miR-25/93 in breast cancer [26]. Thus, we postulated that some deregulated hypoxia-responsive miRNAs might facilitate the PCa progression via reducing the expression of TGFBR2. By intersecting the potential miRNA-mRNA regulatory networks via an experimentally validated microRNA-target interactions database (miRTarBase) and well-known hypoxia-induced miRNAs in published articles, we extrapolated that miR-93 might be potential miRNAs which could regulate TGFBR2 expression and be induced by hypoxia. To validate whether miR-93 was a bona fide hypoxia-regulated miRNA in prostate cancer, we detected its expression in normoxic or hypoxic conditions. As shown in Fig. 3a, compared to cells in normoxia, expression of miR-93 was markedly increased in hypoxic conditions in all three PCa cell lines. To further confirm the results, we repeated the experiments in cells with siRNA-mediated HIF-1α knockdown, Additional file 1: Figure S7 showed that HIF-1α silencing was effective in prostate cancer cells, especially in DU145 and PC3. The experimental data showed that HIF-1α knockdown compromised the ability of hypoxia to induce miR-93 expression (Fig. 3b). These results combined together to confirm that miR-93 was a genuine target of hypoxia.

In an attempt to evaluate the potential of TGFBR2 as the direct target of miR-93, we firstly detected the expression pattern of miR-93 in PCa. We examined miR-93 expression levels in a panel of 45 human PCa tissues and found that the level of miR-93 was elevated in most of the malignant PCa tissues by 1.5 to 6 folds (Fig. 3c). Furthermore, compared to noncancerous prostate cells RWPE-1, the expression levels of miR-93 in three prostate cancer cell lines was significantly higher (Fig. 3d). To further investigate the clinical significance of elevated miR-93 expression, we assessed the association between miR-93 expression and several clinicopathologic characteristics of prostate cancer. Gleason score (GS) which reflects the differentiation and biological behavior of PCa is critical for PCa grading. In accordance with our prediction, higher GS accompanied with significantly higher expression of miR-93 (Fig. 3e). Besides, miR-93 expression levels are significantly elevated in patients of higher tumor stage (Fig. 3f). These findings demonstrated that higher expression of expression was correlated with the PCa progression and miR-93 had the potential to be useful biomarker for PCa.

To figure out if hypoxia-responsive miR-93 contributed to the attenuation of TGFBR2 expression via targeting the 3’-UTR of TGFBR2 mRNA, we performed correlation coefficient analysis between miR-93 and TGFBR2 using data of TCGA. We found significant negative correlation between the expression of miR-93 (MIMAT0000093) and TGFBR2 in 568 cases of PCa (r = − 0.5488, p < 0.0001) (Fig. 3g). Then we transfected the synthetical mimics and inhibitor of miR-93 into two PCa cell lines. As shown in Fig. 3h, i and Additional file 1: Figure S8, after transfection of miR-93 mimics, both mRNA and protein expression of TGFBR2 were significantly decreased, while TGFBR2 expression elevated after transfection of miR-93 inhibitor. According to seed sequence data of TGFBR2 via TargetScan (Fig. 3j), we constructed dual luciferase reporter of wild-type or mutant 3’-UTR of TGFBR2 mRNA and performed a dual luciferase reporter assay. As seen in Fig. 3k, miR-93 mimics had the ability to decrease the relative luciferase activity of wild-type TGFBR2, while the inhibitory effects were abolished when the seed sequence of TGFBR2 3’-UTR was mutated. These results proved that TGFBR2 was exactly the target of miR-93 in prostate cancer cells.

To explore the role of miR-93 in prostate cancer cells, we transfected PC3 and DU145 with miR-93 mimics or inhibitor to upregulate or decrease miR-93 expression, respectively. MTS assay showed that the proliferation rate of PC3 and DU145 cells was significantly repressed after silencing of miR-93, while the cell proliferation was dramatically accelerated following overexpression of miR-93 (Fig. 4a). Furthermore, the ability of colony formation was also affected by miR-93 mimics or antagonist in same manner of that of MTS (Fig. 4b, Additional file 1: Figure S9).

To determine whether miR-93could regulate the epithelial-mesenchymal transition of PCa cells, we performed in vitro gain-of-function or loss-of-function analyses by overexpressing or silencing miR-93 in PC3 and DU145 cells, and detected the effects of miR-93 on cell migration and invasion. Transwell migration and invasion assays were performed, which showed that ectopic expression of miR-93 significantly promoted the migration and invasion of PC3 and DU145 cells (Fig. 4c-f). In contrast, the migration and invasion ability of PCa cells was suppressed when endogenous miR-93 was silenced with miR-93 specific inhibitors (Fig. 4c-f). Besides, western blot and RT-qPCR showed that miR-93 mimics decreased the expression of epithelial marker E-cadherin, while increasing the expression of mesenchymal markers such as N-cadherin, Vimentin, Zeb1/2. In addition, miR-93 inhibitor exerted inversed regulatory roles on EMT markers (Fig. 4g, Additional file 1: Figure S10). These observations suggested that miR-93 could promote PCa cell migration and invasion in vitro.

As we had shown that miR-93 silencing could decrease the proliferation, migration, and invasion of PCa cells, we were curious that if TGFBR2 could mediate the oncogenic functions of miR-93 and if TGFBR2 silencing could restore these functions of PCa cells. Thus, we conducted a series of functional rescue assays to settle these puzzles. Cells which were co-transfected with miR-93 inhibitor and TGFBR2 siRNA showed significantly lower mRNA and protein expression of TGFBR2, compared to cells with single transfection of miR-93 inhibitor or additional negative control (Fig. 5a). These cells were then analyzed by flow cytometry apoptosis assay, which showed that additional TGFBR2 silencing decreased the cell ratio of early apoptosis in PC3 cells (Fig. 5b). Besides, TGFBR2 interference resulted in increased capability of colony formation in two PCa cell lines (Fig. 5c), which was paralleled with the effects of miR-93 overexpression in PCa cell lines. Subsequently, to test whether abrogation of TGFBR2 could rescue the pro-tumorigenic effects of miR-93, we conducted the Transwell migration essays. We found that TGFBR2 blocking promoted the restoration of migratory ability of PCa which were inhibited by miR-93 inhibitor (Fig. 5d). Taken together, our data presented here supported our hypothesis that abrogation of TGFBR2 restored the oncogenic functions of PCa cells which were compromised by miR-93 silencing, and cellular functions of miR-93 were mediated at least partly by directly targeting TGFBR2.