One human bronchial epithelial cell line 16HBE and four NSCLC cell lines (A549, H358, SPC-A1 and H1299) were used in this study, all cell lines were obtained from the Cell Culture Center of the Shanghai Institute (Shanghai, China) and cultured in DMEM (Dulbecco's minimum essential medium) containing 10% fetal bovine serum (Gibco, Grand Island, NY, USA) at 37°C in a humidified incubator with 5% CO₂.

Fifty pairs of NSCLC tissues were obtained randomly from patients who underwent surgical resection at Changhai Hospital, Second Military Medical University between April 2014 and August 2015. None of the patients received chemotherapy or radiotherapy before surgery. All human materials were obtained with informed consent from patients and were approved by the ethics committees of the Second Military Medical University. The tissues were stored at −80°C.

We downloaded the microRNA data (accession number: GSE29248) from GEO databases in NCBI (http://www.ncbi.nlm.nih.gov/geo/). The microarray data were generated using Affymetrix U133A/B and Plus 2.0 platforms. After inter array quantile normalization, the expression levels of 50 miRNAs were visually generated as a heat map using GeneSpring GX, version 7.3 statistical software.

Oligonucleotides including miR-92a mimics, miR-92a inhibitor, PTEN siRNA and their negative control (NC) were purchased from Shanghai GenePharma Co. Ltd. A549 and H358 cells were plated in 6-well plates and transfected with 50 nM miRNAs or PTEN siRNA with Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. The coding sequences of PTEN were amplified by PCR and inserted into pcDNA3.0 vector to construct the PTEN over expression vector. The transfection of plasmids was conducted using Lipofectamine 2000 (Invitrogen). All the transfections were repeated more than three times independently.

Total RNA of the cultured cells and the tissues was extracted using TRIzol (Invitrogen) according to the manufacturer's instructions. Total RNA from each sample was reverse-transcribed to cDNA using the PrimeScript RT reagent kit (Takara, Tokyo, Japan). The procedure for qRT-PCR was previously described (20). To normalize the data for quantifcation of mRNA and miRNAs, the GAPDH and U6 (Applied Biosystems) were used as internal standards for miRNA and mRNA, respectively. The sequences of the PCR primers were as follows: PTEN forward, 5′-CGGCAGCATCAAATGTTTCAG-3′ and reverse 5′-AACTGGCAGGTAGAAGGCAACTC-3′; GAPDH forward, 5′-CTCCTCCTGTTCGACAGTCAGC-3′, and reverse 5′-CCCAATACGACCAAATCCGTT-3′. miR-92a forward 5′-CTGTCCTGTTATTGAGCACTGGTCTATGG-3′ and reverse 5′-AAGACATTAGTAACCCACCCCCATTCC-3′; U6 forward, 5′-CTCGCTTCGGCAGCACA-3′ and reverse, 5′-AACGCTTCACGAATTTGCGT-3′.

The effect of miR-92a on the viability of A549 and H358 cells was examined using the MTT assay. Briefly, A549 and H358 cells were seeded in a 96-well plate (8×10³ cells/well) in 100 µl growth medium after transfection. After 48 h incubation at 37°C, 20 µl of MTT solution (5 mg/ml) was added to each well, and the cells were continuously incubated for 4 h before 200 µl DMSO was added. The absorbance was read with a microplate reader (BioTek, Winooski, VT, USA) at 570 nm according to the manufacturer's instructions.

The effect of miR-92a on the apoptosis of A549 and H358 cells was examined using an Annexin V-FITC apoptosis detection kit (Life Technologies). After 24 h post-transfection, A549 and H358 cells were collected, centrifuged and resuspended in 100 µl binding buffer. Propidium iodide (PI) (3 µl) and Annexin V-FITC (6 µl) were added to each 100 µl sample and incubated for 15 min at room temperature in the dark. The samples were analyzed on a FACScalibur flow cytometer to determine rate of apoptosis.

The effect of miR-92a on cell invasion was evaluated using Transwell chamber assay (BD Biosciences, Bedford, MA, USA) according to the manufacturer's instructions. Briefly, 5×10⁴ A549 and H358 cells were seeded on the top chamber coated with Matrigel after transfection. The bottom chambers were filled with 20% FBS in DMEM. After incubated at 37°C for 48 h, the cells adherent to the upper surface of the filter were removed using a cotton swab. Then the migration cells were stained with crystal violet after fixed with 4% paraformaldehyde for 20 min, and the migrated cells were counted by averaging the total number of cells from triplicate determinations.

The effect of miR-92a on cell migration was evaluated with wound healing assays. Cells were plated in 6-well plates at 8×10⁵ cells/well, and after 48 h of transfection, the cell monolayer was scraped using a 10 µl micropipette tip, and then cultured with serum-free medium for 24 h. The gap distances of migrating cells were calculated from photomicrographs as previously described (21).

Total cellular proteins were lysed in RIPA buffer in the presence of proteinase inhibitor (Sigma, St. Louis, MO, USA). Concentrations of total cellular protein were determined using a BCA assay kit (Pierce, Rockford, IL, USA). Protein samples (25 µg) were separated by 8% SDS-PAGE and then transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore, Bedford, MA, USA). The membranes were incubated with primary antibodies at 4°C overnight using the following concentrations of PTEN, β-catenin, Vimentin, N-cadherin, E-cadherin, AKT, phospho-mTOR and mTOR (1:1000; Cell Signaling Technology, Danvers, MA, USA), phospho-AKT (Ser473) (1:500; Cell Signaling Technology) and anti-β-actin (1:500; Santa Cruz Biotechnology, Santa Cruz, CA, USA), followed by horseradish peroxidase-conjugated secondary antibody (anti-rabbit, 1:2000, Cell Signaling Technology). Anti-β-actin antibody was used as an internal control. The detected protein signals were visualized using the ECL method (22).

The wild-type PTEN 3′-UTR and mutated PTEN 3′-UTR were amplified and cloned downstream of the luciferase gene in a pGL3 reporter plasmid (Promega). The constructed vectors were named as wt-PTEN-PGL3 and mut-PTEN-PGL3, respectively. For the luciferase reporter assays, HEK293 cells were cultured in 24-well plates, and each well was co-transfected with wt (mut)-PTEN-PGL3 and miR-control, miR-195 mimics or inhibitor and cultured for 48 h. The luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA), according to the manufacturer's protocol.

A549 cells transfected with miR-92a inhibitor or the inhibitor NC and H358 cells transfected with miR-92a mimic or mimic NC were subcutaneously injected into four-week-old male athymic nude mice at 3×10⁶ cells in 0.2 ml PBS per mouse, 5 mice per group. After 5 weeks, mice were sacrificed and tumor weight was detected. All experiments were performed in the Animal Institute of Second Military Medical University and approved by the Medical Experimental Animal Care Commission of Second Military Medical University.

Statistical analysis was performed with the GraphPad Prism 5.0 software. Data are reported as mean ± SD. Statistical significance was calculated by Student's t-test among different groups. Pearson's or Spearman's analysis was used in correlation analysis. A P-value <0.05 was considered to indicate a statistically significant difference.

To explore the role of miRNAs in NSCLC, we first analyzed the differential expressed miRNAs via retrieving the microarray data in the GEO dataset (GSE29248). Cluster analysis based on the miRNA expression pattern indicated a significant difference between NSCLC cancer tissue and the adjacent normal tissue (Fig. 1A). It is well known that EMT play a key role in the metastasis of tumors (11,23), while miR-92a is reported to promote EMT progress in several other types of human cancer (24). For this reason, miR-92a was chosen as the candidate for further study.

To validate the expression of miR-92a obtained from miRNA microarray assay, qRT-PCR was applied to examine miR-92a expression in 50 pairs of NSCLC tissues, along with 34 metastasis and 16 non-metastatic NSCLC tissues. The result showed that miR-92a was significantly higher in NSCLC tissues than those of their matched adjacent normal tissues (Fig. 1B). miR-92a was also overexpressed in metastatic tissues compared with non-metastatic NSCLC tissues (Fig. 1C). To further investigate the correlation between miR-92a expression and prognosis of NSCLC patients, 50 patients were divided into 2 subgroups based on the mean of all samples: low miR-92a group (n=21): miR-92a expression ratio < median ratio; high miR-92a group (n=29): miR-92a expression ratio > median ratio. From Kaplan-Meier survival curve, we observed that patients with high miR-92a expression had significantly shorter overall survival than those with low miR-92a expression (P<0.001, log-rank test; Fig. 1D).

To validate whether upregulation of miR-92a was also present in NSCLC cell lines, we examined miR-92a expression in four NSCLC cell lines including A549, H358, SPC-A1, H1299, and a normal human bronchial epithelial cell line 16HBE acted as a control. It was found that the expression of miR-92a was also aberrantly upregulated in four NSCLC cell lines, as compared to 16HBE (Fig. 1E), and these results were similar with the detection in NSCLC tissues. These results indicated that miR-92a may be involved in the development of NSCLC.

Given the upregulation of miR-92a in NSCLC tissues and cell lines, we predicted that miR-92a may function as a tumor oncogene. To verify our hypothesis, miR-92a inhibitor was transfected into A549 cells and miR-92a mimic was transfected into H358 cells because A549 and H358 cells exhibited the highest and lowest expression levels of miR-92a in these NSCLC cell lines, respectively. qRT-PCR analysis showed that miR-92a expression was effectively reduced or enhanced by miR-92a inhibitor or mimic (Fig. 2A). The results of MTT revealed that miR-92a knockdown inhibited A549 cell proliferation, whereas miR-92a overexpression promoted H358 cell proliferation (Fig. 2B and C). Then, the percentage of apoptotic cells in miR-92a-transfected A549 and H358 cells was evaluated by flow cytometry. As shown in Fig. 2D–F, the ratio of apoptotic cells in A549 cells was increased, whereas decreased in H358 cells. To further study possible mechanisms through which miR-92a alters NSCLC cell apoptosis, we examined the expression of apoptosis-relevant proteins. As shown in Fig. 2G, miR-92a knockdown resulted in an obvious increased expression level of pro-apoptotic proteins including the cleaved-caspase-3, caspase-PARP and Bax in A549 cells.

To evaluate the effect of miR-92a on NSCLC tumor growth in vivo, A549 cells transfected with miR-92a inhibitor and H358 cells transfected with miR-92a mimic were subcutaneously injected into nude mice, and the growth of the resultant primary tumors were monitored. As expected, the tumor growth was significantly inhibited in miR-92a inhibitor transfected A549 cells (Fig. 2H), whereas significantly promoted in miR-92a mimic transfected H358 cells (Fig. 2I). All these results suggested that miR-92a executed an oncogenic effect in NSCLC by potentiating cell growth ability of NSCLC cells.

To explore whether miR-92a affects migration and invasion of NSCLC cells, wound healing and Transwell invasion assays were performed in A549 and H358 cells. Transfection of miR-92a inhibitor significantly attenuated the capacity of migration in A549 cells compared with the control cells without transfection (Fig. 3A), whereas the capacity of wound healing in H358 cells was significantly enhanced after treated with miR-92a mimic (Fig. 3B). Subsequently, the relative invaded cell number of A549 cells transfected with miR-92a inhibitor was significantly decreased compared with the control cells while the relative cell number was significantly increased in the miR-92a mimic transfected H358 cells (Fig. 3C), suggesting that miR-92a plays an important role in the regulation of NSCLC cellular motility, including the cell invasive and metastatic capacity.

Recent studies indicate that EMT plays an important role in the invasion and metastasis of NSCLC (25) and miRNAs have been recognized as key regulators of EMT (26), thus we assessed the ability of miR-92a to promote EMT in NSCLC cells. We found that the expression of E-cadherin known as the epithelial marker was significantly upregulated, whereas that of the mesenchymal marker N-cadherin, vimentin and β-catenin were significantly downregulated in A549 cells transfected with miR-92a inhibitor (Fig. 4A). In contrast, overexpression of miR-92a markedly decreased E-cadherin expression and increased N-cadherin, vimentin and β-catenin expression levels in miR-92a mimic transfected H358 cells (Fig. 4B). These results suggest that miR-92a controls NSCLC invasion and metastasis by regulating EMT.

To further characterize the molecular mechanisms involved in the oncogenic role of miR-92a in NSCLC cells, we searched for potential target genes of miR-92a using TargetScan and microRNA.org. PTEN was chosen as a target gene of miR-92a because its level is related with poor prognosis in lung cancer and facilitated EMT in colorectal cancer (CRC) (24,27–29). The predicted binding sites for miR-92a in the PTEN sequence are illustrated in Fig. 5A. Then, western blot analysis confirmed that overexpression of miR-92a markedly inhibited PTEN expression on protein level in H358 cells (Fig. 5B and D). In contrast, inhibition of miR-92a markedly promoted PTEN expression on protein level in A549 cells (Fig. 5C and D). Notably, overexpression of miR-92a in H358 cells or knockdown of miR-92a in A549 cells did not change PTEN mRNA level, which indicated miR-92a targeted PTEN mainly via translational inhibition (Fig. 5E). This finding agrees with the results reported by Zhang et al in colorectal cancer (24).

To determine whether PTEN is a direct target of miR-92a in NSCLC cells, luciferase reporter assay was performed. The result showed that overexpression of miR-92a significantly inhibited the luciferase activity of wild-type PTEN 3′UTR, while inhibition of miR-92a obviously promoted the luciferase activity of wild-type PTEN 3′UTR, but the activity of the mutant-type PTEN 3′UTR was not changed (Fig. 5F).

To further investigate the relationship between miR-92a and PTEN in NSCLC tissues, the level of PTEN mRNA in 50 NSCLC patients was measured by qRT-PCR (data not shown). As shown in Fig. 5G, Pearson's correlation analysis revealed that the expression of miR-92a was inversely correlated with the expression of PTEN in the 50 patients with NSCLC (r= −0.8492; P<0.001). These results indicated that miR-92a could directly bind anti-oncogenic PTEN and inhibited its expression in NSCLC.

It is reported that PTEN can negatively regulate the activity of Akt (30,31), and the PI3K/Akt pathway plays an important role in the invasion and metastasis of tumor cells (32–34). However, the effect of miR-92a on PTEN-mediated AKT signaling was not reported previously in NSCLC. Western blot analysis revealed that miR-92a knockdown in A549 cells robustly increased PTEN protein expression levels, inhibited the phosphorylated AKT (Ser 473) and phosphorylated-mTOR (Fig. 6A), whereas the inhibition of miR-92a in H358 cells exerted the opposite effect (Fig. 6B). However, miR-92a over expression or miR-92a knockdown did not change the expression levels of total AKT and mTOR protein (Fig. 6A and B). These results indicate that miR-92a promotes the activity of PI3K/AKT pathway in NSCLC cells.

To evaluate if PTEN is responsible for the oncogenic potential of miR-92a in NSCLC cells, gain- and loss-of-function studies of PTEN were performed. As expected, PTEN silencing by si-PTEN partially reversed the effects of miR-92a knockdown on cell migration and invasion in A549 cells (Fig. 7A and C). Conversely, over-expression of PTEN significantly alleviated the enhancement of cell migration and invasion induced by miR-92a mimic in H358 cells (Fig. 7B and D). Together, these data suggest that miR-92a exerts an oncogenic role by targeting PTEN/PI3K/AKT pathway in NSCLC.