A TrkB–STAT3–miR-204-5p regulatory circuitry controls proliferation and invasion of endometrial carcinoma cells

We examined TrkB protein and mRNA expression in the normal endometrium and endometrial cancer tissues using laser capture microdissection (LCM)/quantitative reverse transcription polymerase chain reaction (qRT-PCR) and immunohistochemistry. Our RT-PCR assays of normal and endometrial cancer cells captured by LCM (Figure 1A) revealed that mRNA transcript levels of TrkB appeared to be higher in the tumors than in the normal specimens, but overall the difference in TrkB mRNA expression between endometrial cancer tissues and the normal endometrium was not statistically significant (P > 0.05) (Figure 1B). Immunohistochemistry, on the other hand, demonstrated a markedly higher expression of TrkB in endometrial cancer tissues compared with the normal endometrium (P < 0.0001) (Figure 1C). The data suggest that TrkB is upregulated mainly at the posttranscriptional level in human endometrial cancer tissues.

We were interested in whether changes in TrkB expression impacted on the global profile of miRNA expression in endometrial cancer cells. Our microRNA array consisting of 1347 capture probes for mature human miRNAs showed marked changes in the expression of 98 miRNAs in HEC-1B^shTrkB cells whose TrkB expression was suppressed by short hairpin RNA (shRNA) against TrkB compared to HEC-1B cells (Figure 1D). TrkB overexpression also caused marked changes in 126 miRNAs in Ishikawa^TrkB cells ectopically expressing TrkB compared to Ishikawa cells (Figure 1D). Consistently, 76 miRNAs were found among the differentially expressed miRNAs in both HEC-1B^shTrkB cells (77.6%, 76/98) and Ishikawa^TrkB cells (60.3%, 76/126) (Table 1).

Separately, we surveyed the 3′-UTR of the TrkB promoter using three target-prediction algorithms, TargetScan, Pictar and MiRanda, to identify candidate miRNAs which may act as posttranscriptional modulators of TrkB expression. TargetScan, Pictar and MiRanda revealed 4, 3 and 37 candidate miRNAs, respectively (Figure 2A). Examination of the 76 differentially expressed miRNAs as identified by microarray analysis showed that miR-211-5p and miR-204-5p were the only two candidate miRNAs that were also identified by all three target-prediction algorithms to potentially bind to the 3′-UTR of the TrkB (Figure 2A). Furthermore, the miR-204-5p targeting site within the 3′UTR of TrkB (position 457–464) was highly conserved across five mammalian species (Additional file 1: Figure S1A). These intriguing findings suggest that miR-211-5p and miR-204-5p and TrkB are likely mutual modulators of their expression.

MiR-204-5p is reportedly dysregulated in endometrial carcinoma [18, 19]. To examine whether miR-204-5p modulated TrkB expression, we constructed vectors containing a wildtype or mutant TrkB 3′UTR directly fused downstream of the Firefly luciferase reporter gene (Figure 2B). The wildtype or mutant vector was co-transfected into 293 T cells with miR-204-5p mimic (miR-204 m) or its scrambled control (miR-204 m NC). Co-transfection assays showed that, compared with the scrambled miRNA control, miR-204 m significantly decreased the relative luciferase activities of 293 T cells transfected with the wildtype TrkB 3′UTR (P < 0.01) (Figure 2B). No reduction in luciferase activities was observed of 293 T cells transfected with the mutant TrkB 3′UTR.

Our finding that TrkB may be targeted by miR-204-5p led us to further delineate the actions of miR-204-5p on TrkB in endometrial cancer cells. Our RT-PCR assays revealed that Ishikawa^TrkB cells transfected with miR-204-5 m had markedly reduced TrkB mRNA transcript levels compared to Ishikawa^TrkB cells transfected with scrambled miRNA (P < 0.01) (Figure 2C). Immunoblotting assays also showed that Ishikawa^TrkB cells transfected with miR-204 m had markedly reduced TrkB levels (P < 0.05) (Figure 2D and E). Furthermore, HEC-1B^shTrkB cells transfected with a miR-204-5p inhibitor (miR-204i) had markedly higher TrkB mRNA transcript levels than HEC-1B^shTrkB cells transfected with scrambled miRNA (P < 0.01) (Figure 2F). Our immunoblotting assays additionally showed that HEC-1B^shTrkB cells transfected with miR-204i also had noticeably increased levels of TrkB (P < 0.01) (Figure 2G and H). These findings indicated that miR-204-5p negatively regulated the expression of TrkB.

Our microRNA microarray showed that miR-204-5p was downregulated in Ishikawa^TrkB cells. Further examination by qRT-PCR assays revealed that miR-204-5p expression was noticeably suppressed in Ishikawa^TrkB cells (vs. Ishikawa cells, P < 0.01), but markedly enhanced in HEC-1B^shTrkB cells (vs. HEC-1B cells, P < 0.01) (Figure 3A). These results further corroborated the microRNA array findings and indicated that TrkB suppresses miR-204-5p expression. Our qRT-PCR assays also showed that the mRNA transcript levels of TRPM3, which shares the same regulatory motif for transcription with miR-204 [17], were significantly decreased in Ishikawa^TrkB cells (P < 0.05), but markedly elevated in HEC-1B^shTrkB cells (P < 0.01) (Figure 3B), lending support to an inhibitory role of TrkB at the TRPM3/miR-204-5p locus.

The TrkB signaling pathway overlaps with the ERK/MAPK or JAK2/STAT3 signaling pathway in myeloma or endometrial cancer cells [20, 21], which prompted us to examine whether TrkB suppressed miR-204-5p expression via these signaling pathways. Our immunoblotting assays revealed that TrkB overexpression in Ishikawa^TrkB cells noticeably increased the phosphorylation of JAK2 and STAT3, which, however, was aborted by TrkB knockdown in HEC-1B^shTrkB cells (Figure 3C). No obvious activation was found by TrkB overexpression in ERK/MAPK pathway (Figure 3C), and effect of its inhibition (U0126 10 μM) on miR-204-5p expression (Additional file 2: Figure S2A-C). Moreover, inhibition of STAT3 by small interference RNA (siRNA) (Figure 3D) elevated the miR-204-5p expression in Ishikawa^TrkB and HEC-1B cells (P < 0.05) (Figure 3E), indicating that, indeed, TrkB activates the JAK2/STAT3 pathway in endometrial cancer cells. We further examined whether phospho-STAT3 could bind to three putative STAT-binding sites that are located near the TRPM3 promoter region upstream of miR-204-5p[22] (Figure 3F). Our ChIP assays showed that phospho-STAT3 could bind to all the three putative STAT-binding sites in Ishikawa^TrkB and HEC-1B cells, with the greatest binding to DW2 (Figure 3G). These data together suggest that TrkB overexpression upregulates the phosphorylation levels of STAT3, which binds to the STAT-binding sites near the TRPM3 promoter region upstream of miR-204-5p, leading to repression of miR-204 in endometrial cancer cells.

To investigate whether miR-204-5p modulated the growth of endometrial cancer cells, we transfected Ishikawa^TrkB cells with miR-204 m or its scrambled control and HEC-1B^shTrkB cells with miR-204i or its scrambled control. The MTT assays showed that compared with Ishikawa cells, Ishikawa^TrkB cells, which had markedly reduced miR-204-5p as measured by Taq Man PCR assays (Figure 4A, left panel), exhibited significantly increased growth (P < 0.05) (Figure 4A, right panel). Treatment with miR-204 m, however, significantly attenuated the growth of Ishikawa^TrkB cells (vs. scrambled control or non-transfected cells, P < 0.05) (Figure 4A, right panel). Conversely, compared to HEC-1B cells, HEC-1B^shTrkB cells, which had markedly increased miR-204-5p as measured by Taq Man PCR assays (Figure 4B, left panel), showed markedly reduced growth (P < 0.05) (Figure 4B, right panel). Treatment with miR-204i partially and yet significantly rescued the growth of HEC-1B^shTrkB cells (vs. scrambled control or non-transfected cells, P < 0.05) (Figure 4B, right panel). Furthermore, the clonogenic assays showed that miR-204 m caused a greater than 50% reduction in the number of colonies compared with that of Ishikawa^TrkB cells or Ishikawa^TrkB cells transfected with the scrambled control (P < 0.01 in both) (Figure 4C, Additional file 3: Figure S3A). By contrast, miR-204i noticeably increased the number of colonies compared with that of HEC-1B^shTrkB cells or HEC-1B^shTrkB cells transfected with the scrambled control (P < 0.01 in both) (Figure 4C, Additional file 3: Figure S3A). These results suggest that TrkB promotes while miR-204-5p suppresses the clonogenic growth of endometrial cancer cells.

We further assessed whether miR-204-5p modulated the migratory and invasive capacity of endometrial cancer cells. Our Transwell assays showed that Ishikawa^TrkB cells displayed an enhanced capacity of migration compared to Ishikawa cells, which, however, was markedly abated by miR-204 m (vs. non-transfected or scrambled control, P < 0.05) (Figure 4D, Additional file 3: Figure S3B). Conversely, HEC-1B^shTrkB cells exhibited reduced migration capacity compared to HEC-1B cells, which was enhanced by miR-204i (vs. non-transfected or scrambled control, P < 0.05) (Figure 4E, Additional file 3: Figure S3B). Furthermore, similar findings were observed in invasion assays of Ishikawa^TrkB cells and HEC-1B^shTrkB cells (Figure 4F and G, Additional file 3: Figure S3C). These results together demonstrate that TrkB increases while miR-204-5p suppresses the clonogenic growth, migration and invasion of endometrial cancer cells in vitro.

MiRNAs can target a series of mRNAs representing anywhere from several to hundreds of genes. To address whether the phenotypic effects of miR-204-5p expression are predominately due to the suppression of TrkB, rather than one of its other cellular targets, we additionally examined whether miR-204-5p and TrkB functioned in the same pathway in modulating clonogenic growth, migration and invasion of endometrial cancer cells. We abolished miR-204-5p activity by transfecting Ishikawa cells with miR-204i. The RT-PCR and Western blotting assays showed that in miR-204i-transfected cells, co-transfection with siTrkB significantly reduced TrkB expression as compared to co-transfection with siTrkB NC (Figure 5A). Furthermore, in the absence of miR-204-5p activity, the number of colonies in Ishikawa cells was increased (P < 0.01), while silencing of TrkB by specific siRNA resulted in a marked reduction in the number of colonies compared with the scrambled control (P < 0.01) (Figure 5B); and a similar reduction was observed in the number of migrated and invasive cells (P < 0.01) (Figure 5C and D). Furthermore, we chose Ishikawa cells, who do not express TrkB to see if miR-204-5p alone exerted any functional effect on EC cell lines. As expected, cells treated with miR-204 m had no obvious change in cellular proliferation, colony formation, migration and invasion compared with wildtype Ishikawa cells (Figure 5E-H). Together, these results suggest that TrkB is a functionally important downstream target of miR-204-5p that is involved in the clonogenic growth, migration and invasion of endometrial carcinoma cells.

We were interested in whether miR-204-5p suppressed the growth of xenografts bearing human endometrial carcinoma cells. We transfected Ishikawa^TrkB cells with miR-204-5p precursor or its scrambled control using lentiviruses (Figure 6A, Additional file 4: Figure S4A). Compared to the scrambled control, mouse xenografts bearing Ishikawa^TrkB cells overexpressing miR-204-5p showed a dramatic reduction in tumor size (P < 0.05) (Figure 6B and C) and tumor weight (P < 0.05) (Figure 6D). To verify its effects on protein expression, tumor tissue was embedded in paraffin and then stained with hematoxylin and eosin (H&E) for histological examination. As expected, TrkB and phospho-STAT3 levels were detectable in the control xenografts; however, the miR-204 expressing tumors had significantly lower levels of these proteins (Figure 6E, first four pairs of panels), thus verifying its role as a negative regulator of the TrkB/STAT3 pathway in vivo.

To determine whether miR-204-5p expression affects the in vivo proliferative ability of endometrial carcinoma cells, we examined the expression of two proliferation protein markers ki67 and PCNA by immunohistochemical staining (Figure 6E, last two pairs of panels). The ki67 proliferation index of the NC group [(92.80 ± 3.24)%] was markedly greater than that of the miR-204-5p group [(52.76 ± 9.62)%; P < 0.05]. Consistently, the PCNA proliferation index of the NC group [(91.30 ± 6.75)%] was significantly higher than that of the miR-204-5p group [(20.60 ± 11.46)%; P < 0.05], (Additional file 4: Figure S4B). These results indicate that miR-204-5p inhibits the tumorigenicity of endometrial carcinoma cells in vivo and further suggests a tumor-suppressive effect of miR-204-5p via the TrkB/STAT3 pathway.

To further determine the correlation between the clinicopathologic characteristics of endometrial cancer patients and miR-204-5p expression, we measured the expression levels of miR-204-5p in 25 normal endometrium samples and 71 endometrial cancer tissues by Taq Man PCR assays. We observed a significantly lower expression of miR-204-5p in endometrial cancer tissues compared with the normal endometrium (P < 0.0001) (Figure 7A). Moreover, Spearman correlation analysis showed a strong inverse correlation between the expression of TrkB and miR-204-5p (r = −0.2414, P < 0.05) (Figure 7B). Our RT-PCR assays further showed increasingly lower miR-204-5p levels as tumors progressed from FIGO stage I to III (stage I vs. III, P < 0.05) (Figure 7C). Though miR-204-5p levels were lower in type II vs. I tumors and tumors with myometrial invasion, no statistical association was observed between miR-204-5p expression and histological type, tumor grade, or myometrial invasion (P > 0.05) (Figure 7D-F). However, a statistically significant association was observed between miR-204-5p expression and lymph node metastasis with tumors showing positive lymph node metastasis having markedly lower levels of miR-204-5p (P < 0.05) (Figure 7G). Furthermore, analysis of the correlation of miR-204 expression and the overall survival (OS) of uterine corpus endometrioid carcinoma (UCEC) patients (n = 279) in an independent dataset from the Cancer Genome Atlas network further showed that though UCEC patients with high mR-204 expression exhibited a higher rate of OS, no significant difference in OS was noted between UCEC patients with high and low mR-204 expression (P > 0.05) (Figure 7H). However, high mR-204 expression was found to be associated with a higher likelihood of survival for UCEC patients (OS; 1.3164). These findings demonstrated that lower miR-201-5p expression is associated with advanced FIGO stages, lymph node metastasis and probably a lower chance for survival in endometrial cancer patients.

Primary tumor tissue samples were acquired from 110 treatment-naïve endometrial carcinoma patients who underwent hysterectomy with lymph node dissection at our institution between August 2009 and April 2012. The resected specimens were histologically examined by H&E and immunohistochemical staining. Among them, primary fresh tissues were collected from 71 of 110 corresponding patients immediately after surgical removal and snap-frozen in liquid nitrogen until further use. Patient demographic and baseline characteristics are shown in Table 2. In addition, 25 normal endometrium samples were obtained from patients who underwent hysterectomy due to other diseases than endometrial carcinoma. Tumor stage and grade were established according to the 2009 Federation International of Gynecology and Obstetrics (FIGO) surgical staging system [52]. 110 formalin-fixed, paraffin-embedded tissues were used for immunohistochemistry, and 71 fresh frozen samples from the same patients were used for LCM/qRT-PCR analysis. Acquisition of tissue specimens was approved by the Human Investigation Ethical Committee of the authors’ affiliated institution.

Ten to 20 serial frozen sections of 8 μm thickness were fixed in 70% ethanol for 2 min at −20°C and stained with HistoGene using a frozen section staining kit (Applied Biosystems, Foster City, CA). Then, the sections were rinsed in ice-cold RNA nuclease-free water at −20°C followed by incubation in xylene for 2 min at −20°C. After the sections were air-dried, the targeted cells were microdissected according to a UV cutting and laser capture procedure using the LCM system (Lecia, LMD 7000, Germany). Endometrial cancer cells and normal epithelial cells were captured onto CapSureMacro LCMcap (Applied Biosystems, Foster City, CA) to allow analysis of differential expression between cancer cells and normal endometrial cells.

Human endometrial cancer cell lines Ishikawa and HEC-1B and human embryonic kidney 293 T cells were obtained from the Chinese Academy of Sciences Committee Type Culture Collection (Shanghai, China). Ishikawa^TrkB cells stably expressing ectopic TrkB and HEC-1B^shTrkB cells stably expressing TrkB-specific short hairpin RNA (shRNA) were previously described [9]. Cells were maintained at 37°C in a humidified atmosphere containing 5% CO₂ in Dulbecco’s modified Eagle’s medium/F12 (Gibco, Auckland, NZ) supplemented with 10% fetal bovine serum (FBS) (Invitrogen, Carlsbad, CA).

For transfections, cells were seeded at 2 × 10⁵ cells/well in 6-well plates and after growth overnight were transfected with miR-204i, miR-204 m, miR-204 mimic negative control (miR-204 m NC), or miR-204 inhibitor negative control (miR-204i NC) (all from GenePharma, Shanghai, China) using Lipofectamine²⁰⁰⁰ (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. For siRNA knockdown of TrkB or STAT3, we used siRNA against TrkB (siTrkB) [53], or siRNA against STAT3 (siSTAT3) [54]. A scrambled siRNA (siTrkB-NC) was used as control. The sequences are listed in Table 3. Ishikawa cells were co-transfected with miR-204i and siTrkB, or miR-204i and siTrkB-NC using Lipofectamine²⁰⁰⁰.

Cells were seeded at 1.0 × 10⁵ cells per well of a 12-well plate, containing 1.2 mL regular medium. The media was then changed to phenol-red free medium with 0.5% stripped FBS for incubation at 37°C overnight. Immediately prior to treatment, the medium in the culture plates was aspirated, triply washed with PBS and replaced with fresh medium. U0126 (Cell Signaling Technology, Danvers, MA) dissolved in DMSO was subsequently added to each well. Concurrently, the same amount of DMSO was added to the control wells.

Total cellular RNA was isolated using a mirVana miRNA extraction Kit (Ambion, Austin, TX) according to the manufacturer’s instruction and labeled and hybridized using the Human MicroRNA Microarray Kit (Agilent Technologies, Santa Clara, CA) according to the manufacturer’s protocol. Hybridization signals were detected with an Agilent DNA microarray scanner G2505C, and scan images were analyzed using Agilent feature extraction software (v10.7.3). Data were analyzed using GeneSpring GX 12.0 software (Agilent Technologies). Values < 0.01 were set to 0.01 and each measurement was divided by the 75^th percentile of all measurements from the same samples. MiRNAs whose expression differed by at least 1.5-fold between Ishikawa and Ishikawa^TrkB cells (or between HEC-1B and HEC-1B^shTrkB) were selected.

Total cellular RNA was extracted using Tri-reagent (Molecular Research Center; Cincinnati, OH). For miRNA analysis, mature miRNA was reverse-transcribed from total RNA using a Taq Man MicroRNA Reverse Transcription kit. Real-time PCR was performed according to the manufacturer’s instructions. MiRNA expression was determined using the 2(^-△Ct) method and normalized against U6B. For quantification of TrkB and TRPM3, cDNA was generated using a Prime Script RT reagent Kit (TaKaRa; Dalian, China), and real-time PCR was performed on an ABI Prism 7000 Sequence Detection System with SYBR Premix Ex Taq (TaKaRa). The primer sequences are listed in Table 4. All experiments were performed in triplicate at least three times independently.

Chromatin immunoprecipitation (ChIP) assays were performed as previously described [22] using anti-phospho-STAT3 antibody (Tyr705) or rabbit isotype IgG (both from Cell Signaling Technology, Danvers, MA). STAT3-binding sites surrounding TRPM3 were predicted using in silico analysis (University of California, Santa Cruz Genome Browser and ENCODE database) as depicted previously [22] and by quantitative real-time PCR using SYBR green (Takara). Enrichment was calculated using the comparative Ct method and primers used are shown in Table 4 and were analyzed for specificity, linearity range, and efficiency to accurately evaluate occupancy (percentage of immunoprecipitation/input). VEGF was used as positive control and IgG as negative control.

For cell proliferation studies, cells were plated at 1 × 10³ cells/well in 96-well plates after 24-h serum starvation. Cell viability was determined at the appropriate time points using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium (MTT) assays with commercially available kits (Sigma; St. Louis, MO). Absorbance was measured at 490 nm using a Spectra Max 190 microplate reader (BIO-RAD; Hercules, CA). For clonogenic survival assays, cells were seeded at 2 × 10³ cells/plate in 6 well plates. After 2 weeks, cell colonies were stained with 0.1% crystal violet and counted.

Cells were plated serum-free at a density of 2 × 10⁵/well in invasion chambers (8 μm pore size; BD Biosciences, San Jose, CA) with or without Matrigel-coating. Medium containing 10% FBS was added into 24-well plates as a chemoattractant. After 6 h (migration assay) or 24 h incubation (invasion assay), cells were fixed with 4% paraformaldehyde for 1 h. Cells on the apical side of each insert were removed by mechanical scraping. Cells that migrated to the basal side of the membrane were stained with 0.1% crystal violet, visualized and counted under a Leica DMI 3000B microscope at 200 × magnification.

For Western blotting assays, conducted as previously depicted [9], antibodies against the following proteins were used: TrkB and phospho-TrkB (both from Abcam), phospho-P44/42 MAPK (thr204/tyr204), P44/42 MAPK, phospho-JAK2 (tyr1007/1008), JAK2, phospho-STAT3 (tyr705) and STAT3 (all from Epitomics). Protein bands were visualized by enhanced chemiluminescence (Pierce Biotechnology) and protein expression was normalized against β-actin.

The lentivirus vector expressing miR-204-5p was prepared using the Lenti-miR-204 miRNA Precursor Expression Construct according to the manufacturer’s protocol (Genechem, Shanghai) (Additional file 4: Figure S4A). Stable Ishikawa^TrkB cells containing the lentivirus vector carrying miR-204 or scramble hairpin control (Genechem) were established after selection with appropriate antibiotics.

For xenograft experiments, sixteen 5-week old female BALB/c nude mice (Chinese Academy of Sciences, Shanghai, China) were injected subcutaneously with 5 × 10⁶ Ishikawa^TrkB miRNA NC and Ishikawa^TrkB miR-204 cells, respectively, in the nape. Tumor size was monitored every 4 days by measuring the length and width with calipers, and tumor volumes were calculated with the formula: (L × W²) × 0.5 mm³, where L is the length and W is the width of each tumor. At the completion of the experiment, mice were sacrificed and the tumors were weighed, dissected, measured and photographed. The study protocol was approved by the local institution review board. The mice used in this experiment were maintained under specific pathogen-free conditions and handled in accordance with the NIH Animal Care and Use Committee regulations.

Paraffin embedded endometrial cancer and normal endometrium tissue sections (4 μm) were processed for immunohistochemistry as previously described [9]. Briefly, after deparaffinization and dehydration, specimens were boiled in 10 mM sodium citrate buffer to unmask antigens. Specimens were then blocked and incubated with primary antibody overnight at 4°C. Antibody binding was detected using Envision reagents (Boster bioengineering, Wuhan, China) according to the manufacturer’s instructions. For evaluation of TrkB expression, staining intensity was scored as 0 (negative), 1 (weak), 2 (medium), or 3(strong). The extent of staining was scored as 0 (0%), 1 (1%–25%), 2 (26%–50%), 3 (51%–75%), or 4 (76%–100%), according to the percentage of the positively stained areas in relation to the whole tumor area. The sum of the intensity score and the extent score was used as the final staining score (0–7) [55]. The results were assessed by two pathologists who were blinded to details regarding patient background.

Additionally, tissue sections (4 μm) of mouse xenografts were routinely prepared and immunohistochemistry was performed using standard avidin-biotin techniques. Anti-TrkB antibody (Abcam) and anti-phospho-STAT3 antibody (Epitomics) were used for the procedure. Brown staining of nuclei was regarded as positive. For in situ hybridization, sections of mouse xenografts were dewaxed and rehydrated, followed by digestion with proteinase K. A 5′ digoxin-labeled locked nucleic acid-modified miR-204-5p probe (Ambion Life Technologies,) was incubated on coverslips at 37°C overnight. Then, the sections were incubated with anti-digoxin antibody (Boster) at 37°C for 1 hour, followed by staining with nitro blue tetrazolium°-bromo-4-chloro-3-indolylphosphate. MiRNA-204-5p in the cytoplasm was stained purple.

We performed an in silico analysis of the association between miR-204 and OS of UCEC patients using published data from the Cancer Genome Atlas network [56] (Additional file 5). Clinical information was downloaded from the Additional file 5 “datafile.S1.1.KeyClinicalData.xls” and miRNA expression profiling by miR-seq was downloaded from the Additional file 5 “bcgsc.ca_UCEC.Illumina_miRNASeq.tar” [56]. The RPKM value of miRNAs in each sample was used to construct the expression profile for hsa-miR-204 and the median of miR-204 expression was used as the cutoff value for high and low expression of miR-204.

Data were presented as mean ± SD and analyzed by the SPSS 16.0 software (SPSS Inc., Chicago, IL). Student’s t-test was used for comparison between two groups, and one-way ANOVA followed by post-hoc Turkey’s test was used for comparison among multiple groups. Correlation was analyzed by Spearman test, and OS was assessed using standard log-rank test and by the Kaplan-Meier method. All P-values are two-sided, and P-values less than 0.05 indicated statistically significant difference. Each experiment was performed as least three times independently.