A549, H1299 (American Type Culture Collection, Manassas, VA, USA) and bulk-selected H1299 were cultured in RPMI-1640 medium (Gibco, Grand Island, NY, USA), supplemented with 10% FBS (Biowest, Barcelona, Catalonia, Spain), 1% penicillin and streptomycin and 1% sodium pyruvate (Gibco) at 37°C in a humidified atmosphere of 5% CO₂.

Twelve NSCLC tumor samples and their corresponding normal tissues were collected from Shanghai Pulmonary Hospital, snap-frozen in liquid nitrogen and stored at −70°C until analysis, between August 2012 and January 2014. None of the patients received pre-surgical chemo/radiation therapy. All NSCLC patients gave written informed consent for the use of clinical specimens for medical research. The study was reviewed and approved by the Committee for Ethical Review of Research Involving Human Subjects at Shanghai Pulmonary Hospital.

The pEGP-miR-512 construct was cloned as described (19). In brief, the genomic fragment of hsa-miR-512 precursor was amplified. The PCR primers were: 5′-TCGAGGATCCACCTTTCACTATTGAGCA ACA-3′ and 5′-TCGAGCTAGCTGAAGTCCAGTTTAT GGCGCA-3′. The PCR product was digested with EcoRI and NheI restriction enzymes and cloned into the pEGP-miR vector (Cell Biolabs, San Diego, CA, USA). The construct was designated as pEGP-miR-512.

All transfection were carried out using Lipofectamine™ 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's procedure. For H1299 cells stably expressing miR-512, H1299 cells transfected with pEGP-miR-512 were selected with 5 μg/ml puromycin (Sigma, St. Louis, MO, USA) for two weeks. The bulk-selected H1299 cells stably expressing miR-512 were maintained in RPMI-1640 (Gibco) with 2.5 μg/ml puromycin (Gibco). Bulk cultures were used to avoid clone specific effects.

Total RNA was extracted with Trizol (Invitrogen), and miRNA was extracted with miRcute miRNA Isolation kit (Tiangen, Shanghai, China). The RNA was converted to cDNA using Primer Script RT Reagent kit (Takara, Tokyo, Japan), which was incubated for 15 min at 37°C followed by 5 min at 85°C. The specific reverse transcription (RT) primer of miR-512-5p was 5′-GTCGTATCCAGTGCAGGGTCCGAGGTATTCGC ACTGGATACGACGAAAGTG-3′. Real-time quantitative PCR was run on a Light-Cycler Roche 480 (Roche Applied Science, Shanghai, China) with reactions consisted of a hot start (10 min at 95°C), followed with 40 cycles of 15 sec at 95°C and 60 sec at 60°C. The expression level of miR-512-5p and p21 was normalized to U6 and GAPDH, respectively. Each sample was run in triplicate to ensure accuracy. The PCR primers for miR-512-5p were 5′-CGGCGGCACTCAGCCTTGAGGG-3′ and 5′-GTGCAGGGTCCGAGGT-3′. The PCR primers for p21 were 5′-CTACTGAGGAGCCAGCGTCTA-3′ and 5′-CTGCCCATCATCATGACCT-3′. The PCR primers for GAPDH were 5′-CTTAGATTTGGTCGTATTGG-3′ and 5′-GAAGATGGTGATGGGATT-3′.

miR-512-5p mimic and miR-512-5p inhibitor were synthesized by Biotend (Shanghai, China), and p21 siRNA were synthesized by GenePharma (Shanghai, China). The sequence of miR-512-5p inhibitor was 5′-GAAAGUGCCCUCAAGGCUGAGUG-3′. Two pairs of p21 siRNA were synthesized. The sequences of p21 siRNA-1 (sip21-1) were 5′-AAUGGCGGGCUGCAU CCAGGA-3′ and 5′-UCCUGGAUGCAGCCCGCCAUU-3′, and p21 siRNA-2 (sip21-2) were 5′-CUUCGACUUUGUCA CCGAG-3′ and 5′-CUCGGUGACAAAGUCGAAG-3′.

Apoptosis rate was assessed by the FITC-Annexin V/PI Apoptosis Detection kit (BD Pharmingen, San Diego, CA, USA) according to the manufacturer's instruction. At 48 h after transfection, cells were collected and assayed with Annexin V binding buffer containing FITC-Annexin V and PI for 15 min at 37°C in the dark. Cells were analyzed by a flow cytometer (BD FACSCalibur, BD Bioscience, San Jose, CA, US) and CellQuest software.

Cells were seeded on a 12-well plate well, and the culture medium was changed to RPMI-1640 medium without FBS 4 h after transfection. The concentrations of glucose and L-lactate were measured after another 36 h of incubation with a glucose test kit (Sigma) and L-lactate assay kit (Sigma), respectively.

Cell Counting Kit-8 (Dojindo Laboratories, Kumamoto, Japan) was used to perform cell proliferation assays according to the protocol of the manufacturers on days 1, 2, 3 and 4 post-plating. Briefly, 5000 cells were plated in 96-well culture plate per well and were incubated in 10 μl CCK-8 per well for 2 h, then examined. Absorbance was measured at 450 and 630 nm (used as reference) with a Microplate Reader ELx808 (Bio-Tek Instruments, Winooski, VT, USA). Each experiment was performed in sextuplicate.

For cell migration assay, 5×10⁴ cells in 0.1 ml media containing 1% FBS were seeded into the upper chamber, with 0.6 ml of medium (10% FBS) under the upper chamber. The cells were then cultured for 72 h, and the non-invading cells were wiped off with a cotton swab. Cells under the chamber were fixed with methanol, stained with DAPI (Sigma), and washed with phosphate-buffered saline (PBS). All of the cells on the bottom of the chamber were counted under the microscope.

H1299 cells were co-transfected with 50 pmol miR-512-5p, 1 μg firefly luciferase reporter containing the wild- or mutant-type putative target site in 3′ UTR of p21 gene and 0.15 μg Renilla luciferase control vector pRL-TK (Promega, Madison, WI, USA). Assays were assessed 48 h later using a SpectraMax M5 plate reader (Molecular Devices, Sunnyvale, CA, USA). Co-transfections of luciferase report construct and pRL-TK with miR-512-5p or NC were performed with the same procedure.

Immunoblot was performed as follows. Whole cells were lysed in RIPA buffer (CW, Shanghai, China) supplemented with protease inhibitors (100 μg/ml PMSF, Sigma-Aldrich; Complete Protease Inhibitor Cocktail, Roche) for 15 min. Protein extracted from lysates were submitted to SDS-PAGE (10%) and transferred to polyvinylidenedifluoride plus membrane (Merck Millipore, Darmstadt, Germany). Membranes were incubated in 5% non-fat milk prepared with Tris-buffered saline containing 0.1% Tween-20 (TBST) at room temperature (RT) for 1.5 h, and the membranes were immunoblotted with rabbit anti-p21 polyclonal antibody (1:1000, Proteintech, Chicago, IL, USA) or with mouse anti-tubulin monoclonal antibody (1:5000, Sigma) overnight at 4°C, washed two times using TBST at RT, and then incubated in secondary antibodies against rabbit or mouse IgG (1:5000, KPL, Gaithersburg, MD, USA) for 1.5 h at RT. The membranes were washed three times and antigen-antibody complexes were visualized by enhanced chemiluminescence (GE Healthcare, Little Chalfont, Buckinghamshire, UK) following the manufacturer's instructions.

Data are presented as mean ± SD of at least three independent experiments for each cellular experimental group. Samples were analyzed by two-tailed unpaired Student's t test, and a P<0.05 was considered statistically significant. We used one, two, and three asterisks in the figures to represent P<0.05, P<0.01, and P<0.001, respectively. All histograms were drawn using GraphPad Prism 5.

miR-512-5p is reported to be significantly upregulated in cervical intraepithelial neoplasia specimens compared to normal cervical epithelium samples (13). To assess the expression levels of miR-512-5p in NSCLC, we collected 12 pairs of flash-frozen tumor vs. normal lung tissues from 12 NSCLC patients. Of the 12 NSCLC patients, 9 patients were subtyped as adenocarcinoma, 3 were subtyped as squamous carcinoma (data not shown). Expression of miR-512-5p was examined by RT-qPCR. Relative miR-512-5p expression levels in NSCLC tumors were presented as-fold change (2^{ΔCt tumor−ΔCt normal}) of tumor versus paired normal tissues, expression level of normal tissues was assigned as 1.0. Compared to paired normal tissues, miR-512-5p was repressed in tumor samples in 83% (10/12) of NSCLC cases; while two paired samples showed increased expression of miR-512-5p in tumor tissue (Fig. 1). This result suggested miR-512-5p might act as a tumor suppressor in NSCLC.

To explore whether miR-512-5p played a role in the progress of NSCLC cells, we set out to analyze the expression level of miR-512-5p in A549 and H1299 with and without RA treatment at 10 μM or 100 μM at different time points. Cells treated with dimethyl sulfoxide (DMSO) were used as control. As shown in Fig. 2A, the expression of miR-512-5p increased about 3-fold compared with that of control when exposed to 10 μM RA for 24, 48 and 72 h in A549 cells. As the RA concentration was expanded to 100 μM, the expression of miR-512-5p increased by 5-, 17- and 3-fold at 24, 48 and 72 h, respectively. In H1299 cells, the expression of miR-512-5p increased less than 4-fold compared with that of control when treated with 10 μM RA. However, when exposed to 100 μM RA, the expression level of miR-512-5p increased to about 10-fold at these three time points (Fig. 2B).

A549 and H1299 cells were cultured for 48 h after transfection with 50 nM miR-NC or miR-512-5p mimics, then stained with FITC-Annexin V and PI, and analyzed by flow cytometry. We found that cells overexpressing miR-512-5p had a larger population of apoptotic cells than control cells. In A549 cells, overexpression of miR-512-5p caused 40% increase of apoptotic cells compared with that of control cells. In H1299 cells, the increase of apoptotic cells was about 46% (Fig. 3).

The Warburg effect of energy metabolism is critical to the survival and proliferation of cancer cells (20). We asked whether miR-512-5p affected cancer cell energy metabolism. To this end, we transfected A549 and H1299 with miR-512-5p and 4 h later the culture medium was changed to RPMI-1640 without FBS. After another 36 h of culture, the glucose consumption and lactate production were detected. As shown in Fig. 4, the rates of glucose consumption and lactate production were strongly decreased by miR-512-5p overexpression in A549 and H1299 cells.

We assessed cell growth rate at day 1, 2, 3 and 4 by CCK-8 assay to study the role of miR-512-5p on cell proliferation in A549, H1299 and H226 cells. The cells were transfected with either miR-512-5p or miR-NC. Compared with miR-NC, miR-512-5p mimics exhibited almost no effect on cell proliferation on A549, H1299 and H226 cells (Fig. 5).

Metastasis is a process that involves multiple steps, including cancer cell adhesion, invasion and migration (21). To explore the role of miR-512-5p in these processes, we transfected A549 cells with miR-NC or miR-512-5p and seeded cells into the transwell chambers. After incubation for 72 h, medium in chamber was removed and all of the cells on the bottom of the chamber were counted under the microscope. We found overexpression of miR-512-5p attenuated A549 cells migration by 39% compared to that of control (Fig. 6).

In the screening of genes that could be affected by miR-512-5p overexpression, we found that p21 was down-regulated by miR-512-5p. A549 and H1299 cells were transiently transfected with miR-NC or miR-512-5p mimics for 48 h, and the expression of p21 mRNA and protein were detected by real-time PCR and western blotting. As shown in Fig. 7A, overexpression of miR-512-5p decreased p21 mRNA by 71% and 91% in A549 and H1299 cells, respectively. Compared with cells transfected with miR-NC, p21 protein was decreased significantly in cells transfected with miR-512-5p (Fig. 7B). The experiments showed miR-512-5p inhibited the expression of p21 in NSCLC cells.

To explore the mechanism of miR-512-5p in downregulating p21, we first performed a computational screen for genes with complementary sites to miR-512-5p in its 3′-UTR using open-access software including TargetScan (www.targetscan.org), PicTar (http://pictar.bio.nyu.edu), Sanger microRNA target (http://microrna.sanger) to check if p21 was a target of miR-512-5p. However, no binding site in p21 was identified. We then carried out a blast to examine whether there was complementary site between p21 3′-UTR and miR-512-5p. We found a potential binding site of miR-512-5p within p21 3′-UTR (Fig. 7C). To test whether p21 is a direct target of miR-512-5p, mutation in the potential binding site was generated in the p21 3′-UTR sequence of in the complementary site (Fig. 7C). p21 3′-UTR fragment containing wild-type or mutant type binding site was cloned into pGL3-prometor vector. Luciferase reporter construct containing wild-type 3′-UTR was cotransfected with pRL-TK, as well as miR-NC or miR-512-5p mimics. Furthermore, luciferase reporter construct containing wild-type 3′-UTR or mutant type 3′-UTR was cotransfected with miR-512-5p mimics and pRL-TK. As shown in Fig. 7D, the luciferase activity of wild-type 3′-UTR cotransfected with miR-512-5p decreased by 36% compared with the wild-type 3′-UTR cotransfected with miR-NC. Whereas, the luciferase activity of miR-512-5p cotransfected with wild-type p21 3′-UTR was decreased by 17% compared with that cotransfected with mutant-type p21 3′-UTR (Fig. 7E). Collectively, these results indicated that miR-512-5p could bind to p21 3′-UTR to repress p21 expression.

We identified that p21 is a target gene of miR-512-5p. To evaluate how p21 is involved in the functioning of miR-512-5p, we used p21 small interfering RNA (siRNA) to knock down the expression of p21 and investigated NSCLC cell apoptosis. siRNAs against p21 (sip21) were synthesized according to previous studies (22,23), and the silencing efficiency was examined. A siRNA lacking homology to the genome was used as negative control. Both sip21-1 and sip21-2 had good silencing efficiency in A549 and H1299 cells, and sip21-2 showed better silencing efficiency (Fig. 8A and B). Thus, we used sip21-2 (sip21 for short) for further analyses. Cell apoptosis assay was then performed. As shown in Fig. 8C and D, knockdown of p21 significantly promoted cell apoptosis in A549 and H1299 cells. The results were in congruent with miR-512-5p overexpression.

To confirm that miR-512-5p may affect cell apoptosis through targeting p21, we also detected the impact of p21 siRNA on cell apoptosis in the presence of miR-512-5p inhibitor in H1299 cells stably expressing miR-512-5p. To clone miR-512-5p into pEGP-miR vector, we amplified a 300 bp genomic fragment containing miR-512-5p according to the instruction. This fragment contains the precursor of miR-512-3p, which can also transcribe miR-512-3p. We selected H1299 cells stably expressing miR-512 (designated as H512) or EGP control. To avoid clone-specific effects, we used bulk-selected cells for functional analysis.

We then detected the impact of p21 siRNA on cell apoptosis in the presence of miR-512-5p inhibitor in H512. The efficiency of miR-512-5p inhibitor in knocking down miR-512-5p expression was examined by RT-PCR. The result showed miR-512-5p inhibitor inhibited 82% of miR-512-5p expression compared with inhibitor NC (data not shown). miR-512-5p inhibitor was transfected into H512 to knockdown the expression of miR-512-5p, and apoptosis assays were performed. As shown in Fig. 9, miR-512-5p inhibitor attenuated apoptosis in H512 cells (Fig. 9A and B). In the presence of p21 siRNA, the effect of miR-512-5p inhibitor on apoptosis could be rescued (Fig. 9C and D).

To further verify that miR-512-5p affected NSCLC cell glycolysis through targeting p21, we transfected A549 and H1299 with sip21 for 4 h and then the culture medium was changed to RPMI-1640 without FBS. After incubation for 36 h, we detected glucose consumption and lactate production. As expected, the rates of glucose consumption and lactate production were strongly decreased by sip21 in A549 cells (Fig. 10A and B). In H1299 cells, glucose consumption showed a trend of decreasing (Fig. 10C), while the lactate production had a significant decrease (Fig. 10D). The results were consistent with miR-512-5p overexpression.