Downregulation of miR-199a-5p promotes prostate adeno-carcinoma progression through loss of its inhibition of HIF-1α

Hypoxia-inducible factor-1 alpha (HIF-1α) plays key roles in cell survival under both hypoxia and normoxia conditions. Regulation of HIF-1α is complex and involves numerous molecules and pathways, including post-transcriptional regulation by microRNAs (miRNAs). Although upregulation of HIF-1α has been shown to promote prostate adenocarcinoma (PCa) progression, the mechanism by which miRNAs modulate HIF-1α in prostate cancer has not been clarified. Here, we show that miR-199a-5p is underexpressed in prostate adenocarcinoma. Artificial overexpression of miR-199a-5p decreased cell proliferation, motility, and tumor angiogenesis and increased apoptosis in PCa cell liness PC-3 and DU145 by directly targeting the 3’-untranslated region (UTR) of HIF-1α mRNA, which reduced HIF-1α levels as well as downstream genes transactivated by HIF-1α (such as VEGF, CXCR4, BNIP3 and BCL-xL). Abnormalities of miR-199a-HIF regulation may contribute significantly to PCa pathogenesis and progression.


INTRODUCTION
Prostate adenocarcinoma (PCa) is the most common non-skin cancer and a leading cause of cancer-related death in the United States and Europe [1]. PCa morbidity and mortality are also increasing dramatically in Asia, including China [2,3]. In the past few decades, improvements in screening, diagnostics and treatment have led to a consistent decrease in PCa mortality and an increase in overall survival rates [4]. Nonetheless, hormone-refractory PCa (HRPCa), which is resistant to therapeutic modalities after an initial response to androgen deprivation, remains a significant challenge in prostate cancer treatment [5]. Although numerous genes are implicated in PCa, the molecular mechanisms underlying its progression remain poorly understood.
Microarray screening has indicated that miR-199a-5p is downregulated in PCa, yet its role in PCa is still unknown [25]. Here, we demonstrate that miR-199a-5p suppresses HIF-1α via post-transcriptional regulation of its mRNA and that underexpression of miR-199a-5p in PCa contributes to high HIF-1α levels and PCa progression.

miR-199a-5p was significantly downregulated in prostate cancer
Expression of mature miR-199a-5p was investigated using stem loop RT-PCR. which showed high levels of miR-199a-5p expression were observed in benign prostate hyperplasia (BPH) and normal prostate (NP) tissue but very low levels in PCa tissue. miR-199a-5p was absent or barely detectable in DU145, PC3, LNCaP and CL1 cells ( Figure 1A, left panel). Based on real-time Q-PCR, the level of miR-199a-5p in PCa tissue was only 1/39.4 (±0.000021) of that in benign prostate tissue, and it was undetectable in the four PCa cell lines (<1/10,000 of the level in benign prostate tissue; Figure 1A, right panel).
To further evaluate miR-199a-5p expression in tissue samples, the expression level of miR-199a-5p was evaluated by in situ hybridization (ISH) in 67 benign tissue samples and 51 PCa tissue samples ( Figure 1B). The miR-199a-5p positivity rate in the benign prostate tissue samples (33/67, 49.3%) was significantly higher than that in the PCa tissue samples (15/51, 29.4%, P=0.038;). miR-199a-5p expression was inversely related to HIF-1α overexpression in prostate cancer HIF-1α expression in fresh PCa tissue samples and cell lines was analyzed by RT-PCR and Western blotting, revealing high levels of both HIF-1α mRNA ( Figure 1C) and protein ( Figure 1D) in PCa tissue and cell lines, in contrast to their absence in NP or BPH samples.
There was a significant inverse relationship between HIF-N and miR-199a-5p expression, as based on correlation analysis of the HIF-N protein level (determined by IHC) and miR-199a-5p level (determined by LNA-ISH) (r s =-0.393, P<0.01; Table 1). However, no significant correlation was found between HIF-C and miR-199a-5p (Table 1). The HIF-N protein level also correlated with the preoperative serum prostate-specific antigen (PSA) level, the Gleason score and the TNM stage. Conversely, miR-199a-5p expression alone was not significantly associated with PSA level, the Gleason score and the TNM stage ( Table 1).

Relationship of miR-199a-5p and HIF-N expression with tumor progression and PCa patient survival
In addition to the significantly different expression pattern of miR-199a-5p and HIF-1α in benign and cancerous prostate tissue samples (Tables 1 and 2; Figure 1), the positive rate of HIF-N in PCa with progression (27/58, 46.6%) was significantly higher than that in PCa without progression (15/59, 25.4%, P=0.021; Table 2).
Kaplan-Meier analysis showed an increased level of nuclear HIF-1α (HIF-N>3%) to be a significantly negative prognostic factor for both disease-specific survival (DSS) and progression-free survival (PFS) in PCa patients, as were the PSA level and TNM stage (P<0.05; Figure 2, Table 3). However, miR-199a-5p expression alone was not significantly associated with DSS or PFS ( Table 3).

miR-199a-5p targeted the 3'-UTR of HIF-1α
To demonstrate posttranscriptional regulation of HIF-1α mRNA by miR-199a-5p, luciferase reporter gene constructs were prepared. The potential seed sequence for miR-199a-5p and the flanking sequence of the HIF-1α 3'-UTR were cloned into luciferase reporter gene constructs. Constructs in which the seed sequences were mutated were also prepared.
With artificial overexpression of miR-199a-5p by infection with Ad-miR-199a-5p, dual reporter assays revealed significant downregulation of luciferase reporter gene activity by 62.1% (±0.012) for the pGL3-HIF-3'-UTR constructs. In contrast, luciferase reporter gene activity was largely restored when using the pGL3-Mut prostate cancer cell lines DU145, PC-3, LNCaP and CL1. (B) Further validation of miR-199a-5p loss in prostate cancer by LNA-ISH (nuclear counterstaining with methyl green) with U6 as a positive control. miR-199a-5p and U6 signals are blue-purple (original magnification ×400). (C) In contrast to miR-199a-5p, HIF-1α mRNA ( Figure 1C, upper panels, RT-PCR, the same actin control as in A) and protein ( Figure 1C, lower panels, Western blot, GAPDH as positive control) levels were significantly higher in prostate cancer tissue and cells than in benign prostate tissue. (D) Immunohistochemical analysis of the HIF-1α protein in tissue sections. Nuclear and cytoplasmic positivity is in brown. Nuclear counterstaining with hematoxylin (original magnification ×400). www.impactjournals.com/oncotarget HIF-3'-UTR, in which the seed sequence of the HIF-1α 3'-UTR was mutated ( Figure 3C).

Artificial miR-199a-5p overexpression by adenoviral vectors led to downregulation of the HIF-1α protein and genes downstream of HIF-1α
Concomitant with the artificial overexpression of mature miR-199a-5p by Ad-miR-199a infection, the HIF-1α protein level was significantly downregulated with miR-199a-5p though little change was observed for the level of HIF-1α mRNA ( Figure 4A). Moreover, the HIF-1α downstream genes BNIP3, BCL-xL, VEGF and CXCR4 were also significantly downregulated artificial overexpression ( Figure 4B).

DISCUSSION
In this study, we show that HIF-1α mRNA is a functionally relevant target of miR-199a-5p, the downregulation of which contributes to the oncogenic high level of HIF-1α in Pca [11,[16][17][18], as well as tumor progression and unfavorable patient survival.
A large body of experimental evidence supports that hypoxia-mediated or -independent increases of HIF-1α plays critical roles in tumorigenesis and progression of many cancers via HIF-1α-dependent activation of genes that promote cancer cell survival or proliferation (such as BCL-xL), spread (e.g., CXCR4), and angiogenesis (e.g., VEGF) [26][27][28][29]. High levels of HIF-1α protein and mRNA have been observed in PCa tissues and cell lines [30][31][32], and our data demonstrate that HIF-N (nuclear HIF-1α expression, the functional state) correlates significantly with TNM stage, Gleason score, tumor progression and poor survival (both DSS and PFS).
The miR-199a-5p expression status appears to vary in different tumors. For example, it has been reported to be upregulated in gastric cancer and pancreatic adenocarcinoma but downregulated in testicular cancer, ovarian cancer, multiple myeloma, colorectal cancer and hepatocellular carcinoma [19-24, 49, 52-54]. Different populations of cancer cells derived from epithelial ovarian cancers (EOC) may express different levels of miR-199a-5p [55,56]. Therefore, miR-199a-5p expression appears to be defined by the specific tissue and environment.
Data for miR-199a-5p expression and functions in prostate cancer are very limited [25,52,57]. Two microarray screening assays indicate that miR-199a-5p is downregulated in PCa and that miR-199a-5p cooperates with miR-181a and miR-30d to reduce PCa drug resistance by targeting GRP78. Our data demonstrated that miR-199a-5p is downregulated in PCa, but, in contrast to HIF-1α nuclear expression level, it alone was not significantly associated with patient survival, indicating miR-199a-5p was among many factors influencing HIF-1α level, which was a major determinant of tumor progression and patient survival in PCa. Several other genes have been shown to be targeted by miR-199a-5p [20-22, 52, 54, 58, 59]. Ets-1, V-ets erythroblastosis virus E26 oncogene homolog 1, is suppressed by miR-199a-5p, the loss of which may promote breast cancer cell invasion [59]. Inhibition of pro-cell proliferation gene CAC1 by miR-199a-5p reduces colorectal cancer cell growth and multidrug resistance [20]. DDR1, a collagen tyrosine kinase, is downregulated by miR-199a-5p, which decreases multiple myeloma cell invasiveness [49]. Nonetheless, it  -199a-5p. (B) The seed sequence for miR-199a-5p in the HIF-1α 3'-UTR is conserved across species. (C) Dual reporter gene assays were performed using pGL3 expression constructs with HIF-1α 3'-UTR regions containing the seed sequence inserted downstream of the luciferase coding sequence; the activity of the basic pGL3 construct was used as the baseline (pGL3-Promoter). With artificial expression of miR-199a-5p (by coinfection with AD-miR-199a-5p), reporter gene activity, represented by relative luciferase activity (firefly/Renilla), was significantly decreased when the miR-199a seed sequence of the HIF-1α 3'-UTR was present (pGL3-HIF-3'UTR); mutation of the seed sequence (pGL3-Mut HIF-3'UTR) significantly restored reporter gene activity. Expression of miR-199a-5p alone had no effect on reporter gene activity when no seed sequence was present.  has been reported that miR-199a-5p induces melanoma metastasis and angiogenesis by inhibiting ApoE, an inhibitor of metastasis and angiogenesis, through interaction at distinct ApoE receptors [22]. These results indicate complex and tissue-specific regulatory differences of miR-199a-5p.
Little is known regarding the molecular mechanisms responsible for miR-199a-5p downregulation in PCa. It has been reported that the transcription factors TWIST1 and EGR1 as well as the methylation status of miR-199a promoters may be involved [60][61][62]. Activation of the AKT pathway, which is associated with PCa progression, can also downregulate miR-199a-5p [49,63]. It is also intriguing that double-negative feedback loops exist between miR-199a-5p and its targets: miR-199a-5p targets and reduces expression of the Brm subunit of SWI/SNF, which negatively regulates Egr1 (a positive regulator of miR-199a-2, the precursor gene of miR-199a-5p) [61]. Clearly, further studies regarding the regulation of miR-199a-5p expression are needed.
In summary, we report that miR-199a-5p plays a tumor-suppressive role by directly targeting HIF-1α and thereby suppresses genes downstream of HIF-1α, such as VEGF, CXCR4, BNIP3 and BCL-xL. These data provide evidence for an additional dimension of the already rather complex regulation of HIF-1α in PCa, which may help in the development of miR-199a-5p-based strategies for treating this disease.  203 archived, formalin-fixed, paraffinembedded samples, including 134 prostate adenocarcinomas (121 needle biopsies, 13 transurethral resections of prostate samples) and 69 benign prostate tissues (all from needle biopsies), were used. In addition, 6 snap-frozen fresh tissue samples (2 cancerous, 2 benign prostate hyperplasia and 2 normal) obtained from prostatectomy specimens were included. All tissue samples were obtained from West China Hospital and were collected and used according to the ethical guidelines and procedures approved by the institutional supervisory committee. Exclusion criteria for the prostate adenocarcinoma group were the presence of another cancer or previous treatment (surgery, endocrine therapy or chemo-radiotherapy). The Gleason scores (GS) of the prostate adenocarcinomas were as follows: GS6 (8 cases, 6%); GS7 (38 cases, 28%); and GS8-10 (88 cases, 66%). The tumor-node-metastasis (TNM) stages were as follows: stage II, 10 cases (7%); stage III, 79 cases (59%); and stage IV, 45 cases (34%). This cohort of patients ranged in age from 53 to 86 years (mean, 69.2) and was treated by combined androgen blockade (surgical castration plus flutamide). Patients were followed by clinical and laboratory monitoring on a regular basis starting at definitive diagnosis. The diseasespecific survival (DSS)time was defined as the time from definitive diagnosis to disease-specific death, and the progression-free survival (PFS) time was defined as the time from definitive diagnosis to any of the following events after initial treatment: prostate-specific antigen failure, local progression, metastasis, or disease-specific death.

Real-time quantitative PCR
Real-time quantitative PCR (Q-PCR) was used together with stem-loop RT to quantitate mature miR-199a-5p. Q-PCR was performed with Light Cycler software 4.05 (Roche Diagnostics, Mannheim, Germany), as described. The β-actin gene was used as a control. The copy number of target genes (relative to β-actin) was determined using the 2 −ΔΔCt method: ΔΔCt=ΔCt exp − ΔCt con =(Ct exp-target −Ct exp-actin )−(Ct con-target −Ct con-actin ), where 'exp' represents the experimental group, 'con' the control group, and 'target' the gene of interest.
The locked nucleic acid in situ hybridization (LNA-ISH) signal intensity was recorded semiquantitatively, with 0 indicating no signal and 1, 2 and 3 indicating weak, moderate, and strong signals, respectively. The extent of LNA-ISH signal was defined as the percentage of cells showing signal and recorded as 0 (0%), 1 (1-30%), and 2 (>30%). An integrated score of LNA-ISH signal (obtained by the product of the intensity score and extent score) of 4 or more was designated as miR-199a-5p positive.

Immunohistochemistry
An anti-HIF-1α antibody used for immunohistochemistry (mouse monoclonal, 1:200, MAB5382; Chemicon, Billerica, MD). Immunostaining was performed as previously described [11]. The percentage of cells with positive nuclear (HIF-N) or cytoplasmic (HIF-C) immunostaining was examined by light microscopy following a protocol we previously described [64]. The cut-off values for HIF-N and HIF-C were 3% and score 2, respectively, as based on preliminary assays.

Recombinant adenoviral vectors for overexpression of miR-199a-5p
The pri-miR-199a (total sequence 422 bp) was amplified from HEK-293 cell genomic DNA with the indicated primers. The PCR product was cloned into pMD18-T (6011; TaKaRa, Dalian, China), verified by sequencing, and subcloned into the shuttle plasmid pAdTrack-CMV (designated pAdTrack-miR-199a). pAdTrack-miR-199a was linearized with PmeI and used to transform BJ5183-AD-1 cells harboring adenoviral pAdeasy-1 vectors (VXS0387; Stratagene, La Jolla, CA) for homologous recombination. Colonies were screened by plasmid miniprep and PacI restriction analysis to obtain clones with recombinant miR-199a (designated as AD-miR-199a). AD-miR-199a was amplified by repeated infection and verified by PCR. The pAdTrack-CMV empty vector was used as a control (designated as AD-Control). Titers and multiplicity of infection (MOI) values were determined according to the manufacturer's protocols.

Cell viability assay
Cells were collected and stained with trypan blue (200 mg/ml, T6146; Sigma, St. Louis, MO). The www.impactjournals.com/oncotarget number of viable cells was determined by microscopic examination.

Transwell invasion assay
AD-miR-199a-or AD-Control-infected PC-3 and DU145 cells (1.0×10 5 cells) were seeded in Matrigelcoated chambers with 8-µm pores (PIEP12R48; Millipore, Billerica, MA). Cells were suspended in serum-free medium and allowed to migrate toward a complete medium supplemented with 10% fetal bovine serum for 36 h. Non-invading cells were physically removed by scraping, and invading cells were stained with 0.1% crystal violet solution and counted under a light microscope.

Dual reporter gene assay
PC-3 cells were cultured in 24-well plates and transfected with 0.4 µg of the reporter constructs using Lipofectamine 2000 (11668019; Invitrogen, Carlsbad, CA). The pRL-CMV plasmid (E2261; Promega, Madison, WI) containing the Renilla luciferase gene (0.02 µg) was cotransfected as an internal control. Cells were infected with AD-miR-199a or AD-Control (MOI 100) at 4 h after transfection and collected 24 h later. Firefly and Renilla luciferase activities were assessed using a Luminometer TD-20/20 (Turner Design, Sunnyvale, CA).

Tubule-formation assay to measure in vitro angiogenesis of vascular endothelial cells
Culture plates (48-well) were coated with 100 µl of Matrigel (356234; BD, San Jose, CA) per well and incubated at 37°C for 1 h or until the gel solidified. A total of 5×10 4 human umbilical vein endothelial cells (HUVECs) were then seeded in each coated well and incubated with a tumor conditioned medium. The tumor conditioned medium was collected from cultures of PC-3 or DU145 cells that had been infected with AD-miR-199a or AD-Control for 48 h. After incubation at 37°C for 2, 4, 6, 8, 12, or 24 h, the HUVEC cells were examined for capillary-like network formation and imaged under a light microscope for each time point. Images were captured from 5 different fields in each well. Tubule formation was quantified by measuring the total tubule length and total number of branch points in triplicate wells using Image Pro Plus 6.0 (Media Cybernetics, Rochville, MD) software according to a published protocol [65].

Statistical analyses
The SPSS 10 program was used for general statistical and survival analyses. Fisher's exact test was used for between-group comparison, and Spearman rank order correlation was used for correlation analysis. The Kaplan-Meier method with the log-rank test was used for univariate survival analysis.

Author contributions
Xueqin Chen and Qiao Zhou designed the project and revised the manuscript. Xueqin Chen and Jinjing Zhong collected and analyzed data and prepared the manuscript. All authors contributed to data collection and writing of the manuscript.

CONFLICTS OF INTEREST
No conflicts of benefits declared.