The LUAD and adjacent non-tumor tissue miRNA sequencing data and related clinical information for 521 LUAD patients were obtained from the TCGA Data Portal (as of March 2016). Patient exclusion criteria were as follows: i) first histologic diagnosis was not LUAD; ii) presence of another malignancy besides LUAD; iii) incomplete data for analysis; and iv) overall survival more than five years. A total of 463 LUAD patients were included in this study. miRNA expression profiles for normal lung tissue samples were obtained from adjacent non-tumor lung tissues (n=42). According to the staging system of the Union for International Cancer Control (UICC), 359 cases were well and moderately differentiated LUAD (stage I–II) and 104 were poorly differentiated LUAD (stage III–IV). In addition, among these 463 LUAD patients were 170 patients with lymphatic metastasis and 295 with non-lymphatic metastasis. The present study meets the publication guidelines of the TCGA.

In addition, 53 LUAD tissue specimens (tumor tissues and paired adjacent non-cancerous tissues) that we collected from Chinese Han population were obtained from the Nanjing Chest Hospital Medical School of Southeast University. Tissues were frozen with RNAlater (Ambion, Foster City, CA, USA) and stored at −80°C immediately after surgical resection until further analysis. Samples were collected with a detailed paper pathology report and a quality assessment report verifying collection of tumor and/or adjacent non-tumor lung tissues. Informed consent forms were obtained from all the patients. The present study was approved by the ethics committee of the Zhongda Hospital Southeast University.

To identify miRNAs abnormally expressed in LUAD compared with normal lung tissues, the ‘Level 3’ raw counts of miRNA expression from the TCGA database were analyzed using Illumina HiSeq Systems (Illumina, Inc., Hayward, CA, USA), 463 LUAD samples and 42 normal controls were not further normalized, because these data were already normalized by TCGA. Then, abnormally expressed miRNAs were compared in Level 3, including LUAD tumor tissues vs. adjacent non-tumor lung tissues, lymphatic vs. non-lymph node metastasis in LUAD patients and stage I–II vs. stage III–IV. Finally, intersection miRNAs were selected for further analysis. A flow chart for bioinformatics analysis is presented in Fig. 1.

According to the comparative analysis of LUAD miRNA sequencing data in TCGA, LUAD-specific intersection miRNAs were selected. The correlations between LUAD-specific intersection miRNAs and clinical features, including race, sex, age, TNM stage, lymphatic metastasis and patient outcome were further analyzed. Subsequently, to correlate specific intersection miRNAs with patient prognostic characteristics, the univariate Cox proportional hazards regression model was used to analyze the association between specific intersection miRNAs and LUAD patient survival. The Kaplan-Meier and log-rank methods (Mantel-Haenszel test) were used to test the equality of survival distributions in different groups subjected to comparison (19). Hazard ratios (HRs) for a 2-fold change in gene expression level from univariate Cox regression analysis were used to identify LUAD-specific intersection miRNAs associated with overall survival (20). miRNAs defined as having a protective signature showed HR<1 and those defined as high-risk had HR for death >1.

We randomly selected 5 specific key miRNAs in the above bioinformatics analysis and measured their actual expression levels in 53 diagnosed LUAD patients tumor tissues and adjacent non-tumor tissues by qRT-PCR. U6 was used as an internal normalized reference to confirm reliability and validity. Reverse transcription reactions using the ReverTra Ace^® qPCR RT kit (FSQ-101; Toyobo, Shanghai, China) was conducted in two steps according to the manufacturer's protocol. First, the mixture containing 1 µg of RNA samples was incubated in a 96-well plate for 5 min at 65°C and held at 4°C. Then, the 9 µl mixture, which comprised 2 µl 5X RT buffer, 0.5 µl RT Enzyme Mix, 0.5 µl miRNA-specific stem-loop RT primers (Guangzhou RiboBio, Co., Ltd., Guangzhou, China), and 6 µl ddH₂O, was incubated in a 96-well plate at 37°C for 15 min, 98°C for 5 min and subsequently held at 4°C.

Real-time PCR was performed to detect the expression level of the candidate miRNAs with the StepOne Plus™PCR System (Applied Biosystems, Foster City, CA, USA). QRT-PCR was then performed using Thunderbird™ SYBR^® qPCR Mix (QPS-201; Toyobo) according to the manufacturer's protocol. The PCR reaction components were 2 µl of cDNA, 10 µl of Thunderbird SYBR^® qPCR Mix, 0.6 µl (1 µl/pmol) PCR primers (ribobio), and 13.2 µl RNase-free water. The reaction was performed at 95°C for 1 min, followed by 45 cycles of 95°C for 15 sec, 60°C for 30 sec and 72°C for 30 sec. A dissociation curve was analyzed from 60 to 95°C. The Ct-value for each sample was calculated with the ∆∆Ct method (21), and fold change results were presented as 2^−∆∆Ct, where ∆∆Ct = (Ct_miRNAs - Ct_U6)_tumor - (Ct_miRNAs - Ct_U6)_{adjacent non-tumor tissues}.

DIANA-mirPath software (http://diana.cslab.ece.ntua.gr/pathways/) was applied to perform online gene enrichment analysis of putative targets of 25 LUAD-specific key miRNAs, comparing each set of miRNA targets to all known Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathways. Then LUAD-specific key miRNAs and signaling pathways were evaluated using the negative logarithm of P-value (−lg P-value), in which the value was proportional to relevance. Furthermore, interaction and reaction networks between genes were analyzed according to the pathway data available from the KEGG database (http://www.Genome.jp/kegg/pathway.html).

HBE cells and human LUAD cells (A549) were purchased from the Chinese Academy of Sciences Type Culture collection Cell Library Committee (Shanghai, China). HBE cells and A549 cells were separately cultured in Dulbeccos modified Eagles medium (DMEM) and 1640 cell-culture medium (HyClone Laboratories, Inc., Logan, UT, USA), supplemented with 10% fetal bovine serum (FBS; Gibco, Carlsbad, CA, USA), and 100 U/ml penicillin and 100 mg/ml streptomycin (HyClone Laboratories), in a humidified 5% CO₂ incubator at 37°C. The expression status of miR-30a-3p was further confirmed by qRT-PCR before they were used in experiments.

Transfections were performed using riboFect™ CP Transfection kit (Guangzhou RiboBio) according to the manufacturers instructions. A549 cells were trypsinized during logarithmic growth period and resuspended in normal growth medium at 1×10⁵ cells/ml. Cells (2×10⁵/well) were seeded in a 6-well plate with 1640 cell-culture medium the day before transfection, then transfected with 5 µl miR-30a-3p mimic (50 nM; Guangzhou RiboBio)/mimic negative control (Guangzhou RiboBio). The transfection complexes were dispensed into wells and incubated at 37°C for 24 h. After transfection, qRT-PCR was used to test the expression of miR-30a-3p transfected cells and confirm transfection efficiency.

The MTT [3-(4, 5-dimethyl)-2, 5-diphenyl-2H] assay was used to assess cell proliferation. Five thousand A549 cells were seeded into each well of a 96-well plate. Cells were transfected in triplicate repeated wells with 50 nM miR-30a-3p mimic and mimic negative control and then cultured in complete culture medium for 24 h. The culture supernatants were removed, and 200 µl MTT solution (0.5 mg/ml) was added into each well and incubated for 4 h at 37°C. Then supernatants were removed, and 150 µl dimethyl sulphoxide (DMSO) solution was added into each well and vibrated to dissolve formazan crystals. Optical density (OD) value was measured at 570 nm in a microtiter plate reader.

A549 cells (n=2×10⁵) were seeded into each well of a 6-well plate. Cells were transfected in sextuplicate repeated wells with 50 nM miR-30a-3p mimic and mimic negative control, and then were cultured in complete culture medium for 48 h. Cells were harvested by trypsinization in a tube and were washed twice in ice-cold phosphate-buffered saline (PBS). After staining with fluorescein isothiocyanate (FITC)-Annexin V and propidium iodide (PI) using the FITC Annexin V/PI apoptosis detection kit (Becton-Dickinson, Franklin Lakes, NJ, USA) according to the manufacturers protocol. The cells were analyzed for apoptosis using a flow cytometer (FACScan; Becton-Dickinson).

A549 cells were placed in 24-well plates (5×10⁴/well) and transfected with 50 nM miR-30a-3p mimic and mimic negative control, and then cultured in complete culture medium for 48 h. After incubation, cells were washed in PBS and incubated with the DNA dye Hoechst 33258 according to the manufacturer's protocol. The results were visualized under a fluorescent microscope with excitation at a wavelength of 350 nm and measured at 460 nm. Each test was carried out at least three times.

After homogenization of the transfected cells, protein concentration was measured using the BCA protein assay (Generay Biotech, Co., Ltd., Shanghai, China). Protein samples were separated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto nitrocellulose membranes. After membranes were blocked in 10% BSA for 2 h, immunoblotting was conducted by incubation with the primary antibodies overnight at 4°C. In the present study, GAPDH (D16H11) XP^® Rabbit mAb (37 kDa) (22) and AKT3 (E1Z3W) Rabbit mAb (60 kDa) (23) antibodies were purchased from Cell Signaling (Danvers, MA, USA). After incubation with HRP-conjugated goat anti-rabbit IgG (D110058; Sangon Biotech Shanghai, Co., Ltd., Shanghai, China) cells were incubated with secondary antibodies for 1 h at room temperature. Immunoreactive bands were visualized using a chemiluminescent substrate (Millipore, Darmstadt Germany) and analyzed with an automatic chemical luminescence/fluorescence image analysis system called the Gel Imaging System (Tanon Science & Technology, Co., Ltd, Shanghai, China).

The data are shown as mean ± SD and statistical comparisons were made using the Students t-test. The significance level was set as 0.05 as default to control the false discovery rate (FDR). Values of P<0.05 were considered statistically significant. Statistical analyses were performed using SPSS 19.0. Operating characteristic curve (ROC) was used to determine the specific key miRNAs as the sensitivity and specificity of detection of LUAD.

In the present study, a total of 1030 miRNAs were identified from TCGA ‘Level 3’ LUAD RNA-Sequencing data. Then 118 LUAD-associated abnormally expressed miRNAs were identified between 463 LUAD tumor tissue samples and 42 adjacent non-tumor tissues. Furthermore, we compared these 118 miRNAs between tumor stage and lymphatic metastasis. Seventy aberrantly expressed miRNAs were selected from comparisons of stage I–II (non-lymphatic metastasis) LUAD patient tissues and adjacent non-tumor lung tissues, 43 from comparisons of stage I–II (lymphatic metastasis) with adjacent non-tumor lung tissues, 89 from comparisons of stage III–IV (non-lymphatic metastasis) and adjacent non-tumor tissues and 88 from comparisons of stage III–IV (lymphatic metastasis) and adjacent non-tumor tissues. To further confirm data reliability, we selected 25 aberrantly expressed miRNAs (12 downregulated and 13 upregulated) from the intersection of the above four groups (Fig. 2 and Table I).

The 25 LUAD-specific intersection miRNAs were further analyzed according to their expression and clinical features (race, sex, age, TNM stage, lymphatic metastasis and patient outcome assessment at diagnosis in the TCGA database). Fifteen specific miRNAs were significantly aberrantly expressed in clinical feature comparisons (P<0.05; Table II).

One miRNA (miR-96-5p) was aberrantly expressed in race, 6 miRNAs (miR-133a-3p, miR-139a-3p, miR-30a-3p, miR-30c-2-3p, miR-127-5p and miR-196-5p) were aberrantly expressed in sex, 2 miRNAs (miR-338-5p and miR-378c) were aberrantly expressed in age, 4 miRNAs (miR-221-5p, miR-378a-3p, miR-451a and miR-486-5p) were aberrantly expressed in TNM stage, 3 miRNAs (miR-133a-3p, miR-139-5p and miR-221-5p) were aberrantly expressed in lymphatic metastasis and 6 miRNAs (miR-133a-3p, miR-139-5p, miR-221-5p, miR-30a-3p, miR-378c and miR-33a-5p) were aberrantly expressed in patient outcome.

A univariate Cox model was used to investigate the relationship between the the 25 LUAD-specific intersection miRNAs and overall survival. Three of these miRNAs were significantly associated with overall survival status in 463 LUAD patients (log-rank P<0.05): 2 (miR-378c and miR-221-5p) negatively (P<0.05) and 1 (miR-142-3p) positively (P<0.05) (Fig. 3).

Finally, we randomly selected 5 specific key miRNAs (miR-30a-3p, miR-30c-2-3p, miR-96-5p, miR-182-5p and miR-221-5p) to validate the reliability and validity of the results of the above miRNA analysis. We applied the paired t-test to assess the differences between the 53 newly diagnosed LUAD tumor tissues and the adjacent non-tumor lung tissues. The results showed that miR-30a-3p, miR-30c-2-3p and miR-221-5p were downregulated in LUAD tumor tissues compared with adjacent non-tumor lung tissues, while miR-96-5p and miR-182-5p were upregulated in LUAD tumor tissues (Fig. 4). The results from the qRT-PCR validation in 53 newly diagnosed LUAD patients were consistent with the above bioinformatics results (Table I).

ROC curve analysis demonstrated that the area under curve (AUC)=0.837, 0.819 and 0.835 for miR-30a-3p, miR-96-5p and miR-182-5p (P<0.01; Fig. 5A, C and D), which are the score both higher the cut-off (0.7) and could be considerable biomarker for early diagnosis of LUAD. ROC analysis measured an AUC=0.674 and 0.546 for miR-30c-2-3p and miR-221-5p (P<0.01; Fig. 5B and E), which both are scores very close to the cut-off (0.7) needed for a considerable biomarker for early diagnosis of LUAD.

To further confirm a possible link between miRNA and cell function, DIANA-mirPath software was used to perform differential expression analysis of miRNAs involved in signaling pathways. The bio-informatics analysis indicated that miR-30a-3p plays an important role in NSCLC (Fig. 6). It was also indicated that several genes were regulated by miR-30a-3p to activate the process of apoptosis in NSCLC (Fig. 7); for example, AKT3 played an important activating role.

miR-30a-3p was detected in analysis of miRNAs. QRT-PCR was used to detect the expression of miR-30a-3p expression in HBE cells and A549 cells. The results showed that the qRT-PCR findings were consistent with the TCGA findings (Fig. 8).

miR-30a-3p mimic was transfected into A549 cells to upregulate the expression of miR-30a-3p. qPCR was performed to determine the efficiency of transfection with the miR-30a-3p mimic. After transfection of the miR-30a-3p mimic, the expression of miR-30a-3p increased significantly (1852.04-fold), compared with negative control (Table III). Light microscopy revealed no significant difference in morphology of miR-30a-3p transfected A549 cells, nor was there any difference in cell density compared with the negative control group. Furthermore, the MTT assay was used to assess the effect of upregulated miR-30a-3p on the viability of A549 cells. The MTT assay showed that after upregulation of miR-30a-3p, the proliferation of A549 cells was significantly inhibited (P=0.000; Fig. 9A).

PI and Annexin V were used to dye A549 cells, and the apoptosis rate was assessed by flow cytometry. The total apoptosis rate of cells in the miR-30a-3p mimic group (17.100%) was significantly higher than among cells in the negative control (9.400%) (P=0.022; Fig. 9B). Furthermore, nuclear staining with Hoechst 33258 and electron microscopic examination demonstrated that in comparison with negative control cells, typical apoptotic morphological changes (such as cell shrinkage, condensation of cytoplasm and chromatin compaction), were more frequent in miR-30a-3p mimic group cells (Fig. 9C).

Western blot was applied to measure the effect of expression of the AKT3 gene on protein levels. The expression of the AKT3 gene significantly decreased (P=0.002; Fig. 10) protein levels in the miR-30a-3p mimic group compared with the negative control. This implies that miR-30a-3p might promote A549 cell apoptosis through negative regulation of the AKT3 gene.