Gene expression data were downloaded from the GSE35988 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE35988) (14) and GSE21034 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE21034) (15). Detailed information on patients can be obtained from the Oncomine database (https://www.oncomine.org/resource/main.html). Gene set enrichment analysis (GSEA) was used to identify pathway gene sets that were correlated with the miR-195 expression profile (http://www.broadinstitute.org/gsea/index.jsp). The gene sets were derived from the Molecular Signatures Database. The normalized ES was used to compare the analysis results across the various gene sets.

Human prostate cancer cells, DU-145 and PC-3, were purchased from Shanghai Cell Bank, Chinese Academy of Sciences. DU-145 cells were cultured in Dulbecco's modified Eagle's medium (DMEM; Hyclone, Beijing, China) plus 10% fetal bovine serum (FBS; Hyclone, Gaithersburg, MD, USA), 100 mg/ml streptomycin and 100 U/ml penicillin (Hyclone, Beijing, China). PC-3 cells were cultured in DMEM/F12 (Hyclone, Beijing, China) medium supplemented with 10% FBS and antibiotics. Cells were incubated at 37°C in a humidified atmosphere of 5% CO₂ in air.

Cells were plated in 6-well plates in their normal growth medium without antibiotics for 24 h before transfections. Transient transfections of miRNA mimics/siRNA (both from GenePharma) were carried out using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions, when cells reached 50–70% confluency. miR-195 mimics, the negative control (miR-NC), HMGA1 siRNA or the universal scrambled negative control (NC siRNA) were purchased from GenePharma (Shanghai, China).

The transfected cells were plated at 3,000/well in 96-well plates. Every 24 h after plating, respectively, 20 µl of 5 mg/ml MTT in PBS was added per well; cells were lysed after 4 h by addition of 200 µl dimethyl sulfoxide (DMSO; Sigma). Absorbance was measured at 570 nm.

After transfection, the cells were seeded in 6-well plates at a density of 3,000 cells/well and cultured for 9–14 days until visible colonies appeared. Then the cells were stained with crystal violet. The number of colonies was counted only if they contained more than 50 cells.

Cell cycle distribution was detected by flow cytometry (FCM). After 48 h of transient transfections, the cells were harvested by trypsinisation and fixed in 70% ice-cold ethanol the overnight. Then cells were washed with cold PBS and resuspended in propidium iodide (PI) nuclear staining for cell cycle analysis.

Whole cell extracts were prepared in RIPA buffer containing protease inhibitor PMSF (both from Beyotime, Shanghai, China). Protein concentration was measured using the BCA assay (Beyotime) according to the manufacturer's instructions. Total protein was electrophoresed by SDS-PAGE. Then the proteins were transferred to a polyvinylidene fluoride membrane (Millipore, Billerica, MA, USA) and blocked for 1 h with 5% skim milk at room temperature. Incubation with primary antibodies was conducted overnight at 4°C. The blots were incubated with horseradish peroxidase labelled secondary antibodies and the signal was detected using ECL (Beyotime). The following antibodies were used for western blotting: rabbit anti-GAPDH (1:500; Xianzhi Biotechnology, Hangzhou, China), rabbit anti-HMGA1 (1:500; Cell Signaling Technology, USA), HRP-labelled goat anti-rabbit secondary antibody (1:3,000; Zhongshan Golden Bridge Biotechnology, Beijing, China).

HMGA1 3′UTR-luciferase reporter vectors were cloned into psiCHECK-2™ vectors. In brief, wide and mutant 3′UTR regions were chemically synthesised and cloned into the XhoI and NotI sites of the psiCHECK-2™ vector (Promega, Madison, WI, USA).

For the luciferase reporter assay, the cells were cultured in 24-well plates and transiently co-transfected with 50 nM miRNA (miR-195 mimics or scrambled miR-195 negative control) and 250 ng reporter vectors (wild-type reporter vectors or mutant-type reporter vectors), using Lipofectamine™ 2000. Luciferase activities were measured using a Dual-Luciferase assay kit (Promega) according to the manufacturer's instructions at 48 h post-transfection

The lentiviral packaging kit was purchased from Open Biosystems (Huntsville, AL, USA). The lentivirus carrying hsa-miR-195 or hsa-miR-negative control (NC) was packaged following the manufacturer's manual. Stable cell lines were established by infecting the lentivirus into PC-3 cells and selection by puromycin.

Nude mice (female BALB/c-nu, 4-weeks old) were purchased from the Shanghai Experimental Animal Center (Chinese Academy of Sciences, Shanghai, China), and maintained under special pathogen-free (SPF) conditions. Six mice were randomly divided into two groups. PC-3 cells stably expressing miR-195 were injected subcutaneously into the flank of nude mice (4×10⁶ cells in 100 µl). PC-3 cells stably expressing miR-NC were used as the negative control. Tumor size was measured using a vernier caliper every week, and tumor volume was calculated according to the formula: Volume = 0.5 × length × width². Immunohistochemical staining assay was performed as previously described (2).

In the GSE21304 data, we analyzed the miR-195 expression pattern in different pathological grade PCa tissues and found that miR-195 expression was downregulated in high grade PCa tissues compared to low grade tissues (Fig. 1A). Next, the relationship between miR-195 expression and patient prognosis was investigated. The result indicated that a high level of miR-195 was associated with longer disease-free survival (Fig. 1B).

To assess the role of miR-195, we overexpressed miR-195 in prostate cancer cell lines (DU145 and PC3) followed by RNA sequence and bioinformational analysis. GSEA analysis showed that there was negatively enriched expression of gene sets involved in cell cycle progression in the miR-195-transfected cells (Fig. 2A). Consistent with the GSEA analysis, miR-195 overexpression inhibited DU145 and PC3 cell proliferation in the MTT assay (Fig. 2B). miR-195 overexpression also decreased clonogenicity of the DU145 and PC3 cells compared to that noted in the miR-NC (Fig. 2C). FCM analysis showed that overexpression of miR-195 led to a significant increase in the percentage of G0/G1 phase DU145 and PC3 cells (Fig. 2D). This indicated that miR-195 overexpression suppressed the growth of prostate cancer cells.

To find the target of miR-195, bioinformatic analysis was performed using TargetScan Human 6.2 and miRanda. HMGA1 was found to possess potential miR-195 target sites within its 3′UTR (Fig. 3A). Guided by this finding, we performed western blot analysis in the DU145 and PC3 cells that were transfected with miR-195 or miR-NC. The results showed that overexpression of miR-195 inhibited HMGA1 expression (Fig. 3B). In order to investigate HMGA1 as a direct target of miR-195, transient transfection of the HMGA1 3′UTR plasmid along with miR-195/miR-NC was performed in the DU145 and PC3 cells. A significant decrease was observed in luciferase activities when compared with the mutant plasmid (Fig. 3C). These results showed that HMGA1 is a direct target of miR-195.

In the GSE35988 dataset, we initially analyzed the HMGA1 expression pattern in benign prostate hyperplasia (BPH) tissues and different PCa tissues and found that HMGA1 expression was elevated in the CRPC tissues compared to the level noted in the BPH tissues. Furthermore, HMGA1 expression was significantly increased in the CRPC tissues when compared with that noted in the androgen-dependent prostate cancer (ADPC) tissues (Fig. 4A). GSEA was used to evaluate the pathways that were differentially expressed between patients with high levels of HMGA1 expression and those with low levels of HMGA1 expression. The data revealed that HMGA1 regulates genes primarily associated with PCa cell cycle progression (Fig. 4B). Next, we investigated the correlation between HMGA1 expression and survival using Kaplan-Meier survival curve analysis with a log-rank comparison. PCa samples expressing higher than median levels of HMGA1 were associated with decreased biochemical relapse-free survival and overall survival relative to those with HMGA1 levels lower than the median (P<0.05) in the Oncomine data (Fig. 4C and D).

Then, HMGA1 was knocked down in the DU145 and PC3 cells by introducing siRNA. The knockdown efficiency was verified through western blot analysis and reduced expression of HMGA1 was observed (Fig. 5B). MTT assay showed that low expression of HMGA1 inhibited the proliferation of the PCa cells (Fig. 5A). FCM analysis also indicated that low expression of HMGA1 induced G0/G1 phase arrest (Fig. 5C).

In the previous experiment, miR-195 was found to be downregulated in human PCa tissues and to play an important role in proliferation and cell cycle distribution. We further examined the effect of miR-195 in a nude mouse PCa xenograft model. Overexpression of miR-195 inhibited PCa xenograft growth in vivo (Fig. 6A). Immunohistochemical analysis revealed that HMGA1 was markedly decreased following miR-195 treatment (Fig. 6B).