We screened the expression spectrum of circRNA in Gene Expression Omnibus (GEO) datasets (http://www.ncbi.nlm.nih.gov/gds/) until October 2020. We screened out the GSE101684(GPL21825, Agilent-074301 Arraystar Human CircRNA microarray V2, Agilent Technologies Inc, California, USA) dataset, which included 4 normal lung tissues and 4 LUAD tissues. Meanwhile, We downloaded miRNAseq and mRNAseq data of LUAD from the TCGA database by means of the Data Transfer Tool (obtained from GDC Apps)(https://tcga-data.nci.nih.gov/) with the closing date of 30 October 2020. The miRNA profiles contained 483 LUAD tissue samples and 45 nearby healthy lung tissue samples, and the profiles of mRNAs contained 497 LUAD tissues and 54 adjacent normal tissues. This study did not require ethical approval or informed consent since we used public data from GEO and TCGA.
Differential expression analysis
The procession of download files including data calibration, normalization, and log2 conversion was conducted by means of R packages. The screening of circRNAs with differential expressions (DEcircRNAs) was performed utilizing the Limma package, and the criteria included adjusted P-value < 0.05 accompanied with |log 2 (fold change [FC])| > 1.5. In addition, the screening of miRNAs and mRNAs with differential expressions (DEmiRNAs and DEmRNAs, respectively) was performed utilizing the edgeR package, and the threshold values included adjusted P-value < 0.05 accompanied with |log 2 (fold change [FC])| > 1. Hierarchical clustering analysis of DEcircRNAs was performed using R package “heatmap” and volcano plots of DEmRNAs and DEmiRNAs were drawn using R package “gplots”.
Construction of the competing endogenous RNA regulatory network.
The prediction of miRNA response elements (MREs) and corresponding target miRNAs was conducted using the Cancer Specific CircRNA database (http://gb.whu.edu.cn/CSCD/). Only miRNAs that presented in both target miRNAs and DEmiRNAs on the basis of the TCGA database were filtered for network construction. Next, three databases (miRDB、miRTarBase and TargetScan) were applied to the prediction of the target mRNAs of DEmiRNAs. The precondition of being selected as mRNA candidates was the recognition of all three databases. The intersection between the candidate mRNAs and DEmRNAs was subsequently performed for screening the target DEmRNAs of DEmiRNAs, and the overlapping mRNAs were retained for constructing the network of ceRNAs. The pairs of miRNA-mRNA and circRNA-miRNA were combined to construct the regulation network of circRNA-miRNA-mRNA. Ultimately, Cytoscape v3.7.2 (http://www.cytoscape.org) was used to visualize the network.
PPI network establishment and module analysis
For the assessment of the interaction among DEmRNA varieties, a PPI network was established by means of the Search Tool for the Retrieval of Interacting Genes (STRING) online tool (STRINGdb: https://string-db.org/), which can provide comprehensive interactions between proteins and genes. The PPI network was constructed with a combined score of > 0.9 as the screening criterion. The cytoHubba app was then used to extract hubgenes from the PPI network, which was visualized by Cytoscape.
To evaluate the effect of the expression level of hubgenes on the overall survival of LUAD patients, we entered the 6 miRNAs and 10 hubgenes into the Kaplan-Meier Plotter website to draw the Kaplan Meier curve. Log-rank test was then carried out and p< 0.05 was considered significant.
Cells and culture
A549, H1299 and BEAS-2B (human immortalized normal epithelial cells) cells were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Cells were cultured in RPMI 1640 medium and DMEM with 10% foetal bovine serum (Gibco, Carlsbad, CA, USA) and L-glutamine (Invitrogen, Carlsbad, CA, USA) at 37 °C in a 5% CO2 atmosphere.
RNA extraction, cDNA synthesis and quantitative real-time PCR analysis
The detailed processes were performed as we previously described(11). Primers used in the study were listed in Table1. The primer of miR-let-7c was designed by Accurate biology, and mRQ3’ primer was supplied with the kit. The CT values of the gene mRNA levels were normalized to those of β-actin and U6. The △△Ct method was applied to calculate the relative quantities of these mRNAs. Each experiment was performed in triplicate.
The small interfering RNA (siRNA) sequences corresponding to the target sequences and inhibitor of miR-let-7c were directly synthesized (GenePharma). The siRNA constructs were as follows: siRNA-circ_0039908-1: 5’-GUUUACUGUGAGAUAUCAATT-3’ (sense)and 5’-UUGAUAUCUCACAGUAAACTT-3’ (antisense), siRNA-circ_0039908-2: 5’- AUUGUUUACUGUGAGAUAUTT-3’ (sense)and 5’- AUAUCUCACAGUAAACAAUTT-3’ (antisense). Transfection of siRNA and inhibitor into cells were performed with Lipofectamine 2000 according to the instructions of the manufacturer.
Cell proliferation analysis
Cell proliferation was evaluated using CCK-8 assays. Cells were digested and plated at a concentration of 3× 103 cells per well in 96-well plates under standard cell culture conditions. A Cell Counting Kit-8 (Boster, Wuhan, China) was used to detect cell proliferation after culture for 24, 48, and 72h.
Colony Formation Assay
A colony formation assay was used to confirm malignant transformation. Three thousand cells were seeded in 3ml of RPMI 1640 medium supplemented with 10% foetal bovine serum and incubated at 37 °C with 5% CO2. The number of colonies formed after 14 days was counted using ImageJ.
Migration and invasion assay
For the migration assay, cells were plated into the upper chamber of a Transwell insert in medium containing 1% FBS. The difference between the migration assay and invasion assay is that in invasion assay, the upper chamber is precoated with a Matrigel matrix (BD Science, Sparks, MD, USA). In the upper chamber, 3×104 cells were seeded for the migration assay, while 5×104 cells were seeded for the invasion assay. Then, 800 μl of complete medium was added to the lower chamber. After 24h of incubation, the cells that invaded from the upper surface to the lower chamber were fixed with methanol and stained with crystal violet. Finally, five random fields from each Transwell sample were selected to examine under a light microscope.
We used Student’s t-test (two-tailed) and two-way ANOVA to analyse the results when required. P<0.05 indicates that a difference between groups was significant. All data were analysed using GraphPad Prism 8.