RETRACTED: Circular RNA circ_0001946 acts as a competing endogenous RNA to inhibit glioblastoma progression by modulating miR‐671‐5p and CDR1

2.1. Gene expression profiles

The high‐throughput sequencing data for three paired glioma tissue and normal brain tissue samples were downloaded from the Gene Expression Omnibus database, a publicly available database. The series accession number was GSE86202, and the platform was GPL16791. In the analysis, we used three kinds of blob selector including HTSeq, FindCirc, and CIRC_FINDER. Among them, HTSeq was used to identify mRNAs, and FindCirc and CIRC_FINDER were used for circRNAs. By means of next‐generation sequencing, the presence and quantity of RNA in a biological sample at a given moment in time were ascertained with RNA‐seq, averting many limitations of other transcriptomic approaches. Data quality control was assessed with FastQC. The GENCODEv.19 Gene Transfer format file was used as a transcript reference (GENCODE annotation). We used RNA‐seq to match the sequencing data to the genome and then extracted the fragments for undirected comparison, followed by recombination and comparison with the genome.

2.2. Differential expression analysis

The statistical software R and Bioconductor package DESeq. 2 were applied to screen differentially expressed mRNAs and circRNAs. MRNAs with differential expression were selected with the threshold set to |log₂FC| > 2 (fold change [FC]) and an adjusted p < 0.01. The p value was adjusted by the Benjamini‐Hochberg procedure. Differentially expressed circRNAs were screened under the parameters |log2FC| > 2 and p < 0.05. Though an intersection analysis (Venn diagram), 22 common circRNAs with differential expression, including circ_0001946, were identified in the glioma tumor tissue.

CircNet (http://circnet.mbc.nctu.edu.tw/) was used to detect miRNA sponges and to filter a gene or miRNA of interest to establish the associated gene–miRNA–circRNA regulatory network. CircInteractome (https://circinteractome.nia.nih.gov/) was used to identify potential miRNA binding sites on circRNAs.

2.3. Glioma cell line culture and transfection

Glioma cell lines (U87 and U251 cells) and human microglia (HM) were purchased from the American Type Culture Collection (Washington, DC) and maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 50 U/ml penicillin G, and 250 µg/ml streptomycin at 37°C in a humidified environment with 5% CO₂.

miR‐671‐5p mimics/inhibitor and the corresponding negative control were purchased from GenePharma (Shanghai, China). Short interfering RNAs (siRNAs) for knocking out circ_0001946 (circ_CDR1) were designed on the Thermo Fisher Scientific website and synthesized by Sangon Biotech (Shanghai, China). pcDNA3.1 (Cyagen, SuZhou, China) was used to overexpress circ_0001946 and CDR1. Transfections were performed in serum‐free medium 24 hr after plating with Lipofectamine 2000 (Invitrogen). After 6 hr, the transfected cells were placed in complete medium and maintained at 37°C in 5% CO₂. The siRNA sequences used are listed below:

si‐circ_0001946‐1: 5′‐GCAATATCCAGGGTTTCCGAT‐3′;

si‐circ_0001946‐2: 5′‐TGCAATATCCAGGGTTTCCGA‐3′; and

si‐circ_0001946‐3: 5′‐TGTCTGCAATATCCAGGGTTT‐3′.

2.4. Quantitative real‐time PCR

Total RNA was obtained from cultured cells using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. To detect the circ_0001946 and miR‐671‐5p levels, quantitative real‐time PCR (qRT‐PCR) was performed using SYBR Premix DimerEraser (Takara) with a 7900HT system. GAPDH/U6 were used as internal controls. FCs were calculated by the relative quantification method (2^−ΔΔCt ). The primers sequences of the evaluated genes are listed in Table 1.

2.5. Dual‐luciferase reporter assay

The targeting site was predicted by TargetScan (http://www.targetscan.org/vert_71/). Dual‐luciferase reporter plasmids were constructed using pmirGLO dual‐luciferase vectors (GenePharma). The sequences of miR‐671‐5p, circ_0001946, and CDR1 were cloned into the vectors separately. Lipofectamine 2000 (Invitrogen) was used to cotransfect GBM cells with circ_0001946‐wt/circ_0001946‐mut or CDR1‐wt/CDR1‐mut together with the miR‐671‐5p mimics/negative control. After a 48 hr induction, luciferase activity was assessed using a dual‐luciferase reporter kit (Promega, Madison, WI). The normalization of the relative firefly luciferase activity was carried out using the Renilla luciferase activity as a reference. The sequences were synthesized by Sangon Biotech.

2.6. Western blot analysis

Cells were washed three times with prechilled phosphate‐buffered saline (PBS). The cells were lysed in radio‐immunoprecipitation assay (RIPA) lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1% Triton X‐100, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate [SDS], 1 mM phenylmethanesulfonyl fluoride (PMSF), and a protease inhibitor mixture). The total protein lysates were separated by SDS‐polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes. The membranes were blocked with skim milk for 1 hr at room temperature, and then the blot was incubated with primary antibodies recognizing rabbit CDR1 (PA5‐68142, 1:1,000) and GAPDH (ab9485, 1:2,500) at 4°C overnight. After washing with tris‐buffered saline tween (TBST) buffer three times, the blot was then incubated with a secondary goat anti‐rabbit IgG antibody (ab6721, 1:10,000) conjugated with horseradish peroxidase. Finally, the protein was detected using a super signal protein detection kit (Pierce). Both the anti‐GAPDH and goat anti‐rabbit IgG antibodies were purchased from Abcam (Cambridge, MA). The anti‐CDR1 antibody was obtained from Thermo Fisher Scientific (Cambridge, MA).

2.7. Transwell assay

Cell suspensions (5 × 105 cells/ml) were prepared with serum‐free medium. For the transwell migration assays, 100 μl cell suspension (5 × 104 cells) and 200 μl serum‐free medium was added in the top chamber of a 24‐well cell invasion assay plate (Costar) containing a polycarbonate transwell filter without Matrigel. A total of 500 μl chemokine was placed in the lower chamber. For the invasion assays, the polycarbonate filter was coated with Matrigel, and the other experimental procedures were analogous to those of the migration assays. The cells were cultured at 37°C for 8 hr for the migration assays or 36 hr for the invasion assays. The cells in the top chamber were then removed with cotton swabs. Then cells that migrated and invaded the lower membrane surface were washed twice with PBS and fixed in 4% paraformaldehyde for 30 min. The cells were washed again using PBS, and 0.1% crystal violet was used to stain the cells for 20 min. After the cells were washed at least three times with double‐distilled H₂O, the membrane was dried and subsequently observed using a light microscope. The assays were performed in triplicate, and the data are presented as the mean ± standard error.

2.8. Colony formation assays

Cells were harvested, seeded in 6‐well plates and incubated at 37°C, and the medium was replaced every 3 days. After incubating for 7–10 days, the cells were washed twice with PBS, fixed with 95% ethanol for 10 min, stained with 0.1% crystal violet for 10 min, and subsequently washed with PBS three times. The number of visible colonies was counted, and the representative colonies were captured for each well. Each experiment was performed in triplicate.

2.9. Flow cytometry analysis

The U87 and U251 cell lines were first digested and stained with an Annexin V‐FITC/propidium iodide (PI) Apoptosis Detection kit (Biouniquer Technology, Nanjing, China) according to the instructions of the manufacturer. The cell concentration then reached a density of 1 × 105. Then, 100 μl suspension was incubated with annexin V‐FITC and PI at room temperature for 10 min. A flow cytometer (BD Pharmingen, San Diego, CA) was used to detect cell apoptosis.

2.10. Xenograft mouse models

All mouse experiments were approved by Shengjing Hospital of China Medical University. All animals used were 6‐week‐old female BALB/c nude mice (Animal Research Centre, Perth, Australia). A subcutaneous xenograft model was established by injecting 5 × 106 U87 cells transfected with empty vector (control), si‐circ_0001946, circ_0001946, circ_0001946+miR‐671‐5p, or circ_0001946+CDR1 subcutaneously into the hip area on both sides of the nude mice. After 5 weeks, the mice were killed, and the subcutaneous xenograft tumors were isolated. The tumors were extracted and stored at −80°C. Tumor volume was calculated by using the equation: volume (mm³) = A × B ²/2, where A is the largest diameter, and B is the diameter perpendicular to A.

2.11. Immunohistochemical staining

For immunohistochemical analysis, tissue sections were deparaffinized, rehydrated and subjected to high‐temperature antigen retrieval in 10 mM citrate buffer (pH 6.0). An anti‐Ki67 antibody (ab15580, 1:2,000; Abcam) was applied for 2 hr in blocking buffer (2.5% bovine serum albumin (BSA), 5% goat serum, and 0.3% Triton X‐100 in PBS), followed by species‐appropriate secondary Alexa Fluor 488 dye‐conjugated antibodies (Amersham). Fluorescent antibody‐labeled slides were mounted in DAKO fluorescent mounting medium containing 1 μg/ml Hoechst counterstain. Horseradish peroxidase (HRP)‐conjugated secondary antibodies were visualized by diaminobezidin (DAB) staining (Vector Laboratories). Apoptotic cells were detected with an ApopTag Detection kit (Chemicon International). Images were obtained with an Axiovert S100 TV inverted fluorescence microscope (Zeiss) and Open Lab 3.5.1 software or with an Axiovert 100 inverted microscope (Zeiss) equipped with a Hamamatsu Orca digital camera.

2.12. Statistical analysis

Statistical analysis was carried out using GraphPad Prism 6.0 (GraphPad Software, Inc., San Diego, CA), and the results are presented as the mean ± standard deviation. To determine statistical significance, Student's t test and one‐way analysis of variance were performed for comparisons of two or more groups. When the p value was less than 0.05, the difference was considered statistically significant.

3.1. CDR1 and circ_0001946 levels were both downregulated in GBM

RNA‐seq was performed to detect mRNAs and circRNAs. A total of 23,000 mRNAs were identified by HTSeq, 493 and 254 circRNAs were identified by FindCirc and CIRC_FINDER, respectively. Then, DESeq. 2 was used to identify differentially expressed mRNAs (log₂FC > 2, adjusted P < 0.01) and circRNAs (log2FC > 2, p < 0.05). In total, 2,437 mRNAs were differentially expressed in GBM, 85 differentially expressed circRNAs were detected in the FindCirc group, and 41 differentially expressed circRNAs were detected in the CIRC_FINDER group. Figure 1a,b shows the top 20 differentially expressed mRNAs and circRNAs (CIRC_FINDER group), respectively.

Using a Venn diagram, 22 common circRNAs were detected between the FindCirc group (85 circRNAs) and CIRC_FINDER group (41 circRNAs). Then, mRNAs matching the 22 circRNAs were identified through circBase (http://www.circbase.org/). By intersecting these mRNAs with the 2,437 differentially expressed mRNAs identified above, CDR1 was identified. To characterize miRNA–circRNA–CDR1 interactions, CircNet was used to search and predict CDR1 sponge miRNAs. Figure 1c shows 14 miRNAs were predicted to sponge CDR1, including hsa‐miR‐671‐5p, hsa‐miR‐149‐3p, hsa‐miR‐377‐5p, and hsa‐miR‐7106‐5p. Among these results, several miRNAs were predicted to target hsa_circ_0001946 as well as sponge CDR1, including miR‐671‐5p.

3.2. Circ_0001946 and CDR1 were expressed at low levels, and miR‐671‐5 was expressed at high levels in GBM cell lines

To validate the results of the RNA‐seq and differential analysis, we performed qRT‐PCR to determine the expression levels of circ_0001946, miR‐671‐5p, and CDR1 in the GBM cell lines and HM. The results showed that compared to the HM group, the GBM cell line groups had low expression of circ_0001946 and CDR1 and high expression of miR‐671‐5p (Figure 2a–c; p < 0.001), consistent with the bioinformatics analysis.

3.3. miR‐671‐5p interacted with circ_0001946 and CDR1

miR‐671‐5p binding sites were found in circ_0001946 in the CircInteractome analysis, and the result of the TargetScan ( http://www.targetscan.org/vert_71/ ) analysis suggested that CDR1 might be a target of miR‐671‐5p as well. A dual‐luciferase reporter gene assay was used to verify these potential relationships, and the relative luciferase activity was significantly decreased in the cells cotransfected with circ_0001946‐wt and the miR‐671‐5p mimics or CDR1‐wt and the miR‐671‐5p mimics compared with the cells cotransfected with circ_0001946‐wt and NC or CDR1‐wt and NC, respectively, (Figure 2d,e), which confirmed the predicted target relationships.

3.4. CDR1 expression was positively regulated by circ_0001946 and negatively regulated by miR‐671‐5p

We designed pcDNA3.1‐circ_0001946 and three siRNAs (si‐circ_0001946‐1, si‐circ_0001946‐2, and si‐circ_0001946‐3). QRT‐PCR was used to study their effects on the expression of circ_0001946, and the results showed that the circ_0001946‐expressing plasmid could increase the circ_0001946 RNA level significantly, whereas si‐circ_0001946‐3 had the best silencing effect on the circ_0001946 RNA level in the GBM cells and was used for the next experiment (Figure 3a; p < 0.01). In addition, the expression of miR‐671‐5p was upregulated in the miR‐671‐5p mimics group but downregulated in the miR‐671‐5p inhibitor group, indicating that the miR‐671‐5p mimics and inhibitor were highly functional in the GBM cells (Figure 3b; p < 0.01). In addition, the western blotting and qRT‐PCR results showed that CDR1 expression was significantly upregulated with the overexpression of circ_0001946, whereas the silencing of circ_0001946 markedly reduced CDR1 expression (Figure 3c,d; p < 0.01), suggesting that circ_0001946 positively regulated the expression of CDR1. In addition, the overexpression of miR‐671‐5p significantly suppressed CDR1 expression, which was upregulated when miR‐671‐5p was inhibited (Figure 3e,f; p < 0.01), indicating that the expression of CDR1 was negatively regulated by miR‐671‐5p.

3.5. Circ_0001946 antagonized miR‐671‐5p to regulate GBM

To verify whether circ_0001946 and miR‐671‐5p can affect cell progression in GBM, we used a colony formation assay to determine their influence on cell colony formation, and the result showed circ_0001946 could suppress cell colony formation; however, miR‐671‐5p promoted colony formation. In addition, we found that circ_0001946 counteracted the effect of miR‐671‐5p (Figure 4a,b). A flow cytometry assay was used to measure cell apoptosis (Figure 4c,d), and transwell migration and invasion assays were performed to determine the migration and invasion of glioma cells (Figure 4e,h). The results showed that when circ_0001946 was overexpressed, the GBM cells displayed strong apoptosis and weakened proliferation, migration, and invasion, which was the opposite of the results observed with the overexpression of miR‐671‐5p. In addition, circ_0001946 could reduce the impact of miR‐671‐5p.

3.6. miR‐671‐5p had an inhibitory effect on CDR1

We verified a CDR1‐overexpressing cell line by qRT‐PCR (Figure 5a) and western blotting (Figure 5b,c) and then investigated the effects of CDR1 and CDR1+miR‐671‐5p in GBM. The experimental content included colony formation (Figure 6a,b), flow cytometry (Figure 6c,d) and transwell migration and invasion experiments (Figure 6e–g). The results indicated that CDR1 inhibited cell colony formation, migration, and invasion and promoted apoptosis, but miR‐145‐5p could inhibit the function of CDR1.

3.7. Circ_0001946 inhibited glioma progression in a xenograft mouse model

The expression levels of circ_0001946, miR‐671‐5p, and CDR1 in vivo were detected by qRT‐PCR. The results indicated that the expression of circ_0001946 and CDR1 was promoted and that of miR‐671‐5p was suppressed in the tumor tissue compared with the adjacent nontumor tissue (Figure 7a,c). The RNA expression levels showed the correlations between circ_0001946/miR‐671‐5p/CDR1 as well. The correlation between circ_0001946 expression and CDR1 expression was positive, whereas the circ_0001946 and CDR1 levels were both negatively correlated with the miR‐671‐5p level (Figure 7d–f). The expression levels suggested that the regulation described above functioned not only in vitro but also in vivo.

In addition, we established a xenograft ovarian cancer mouse model and examined the effects in the control, si‐circ_0001946, circ_0001946, circ_0001946+miR‐671‐5p, and circ_0001946+CDR1 groups on the tumor (Figure 8a). The results showed that the circ_0001946 group exhibited a significant increase in the survival time of the mice and a decrease in the volume and weight of tumors, and these results were even more obvious in the circ_0001946+CDR1 group (Figure 8b,c). However, the results from the circ_0001946+miR‐671‐5p group were basically identical to those of the control group. The immunohistochemistry results in the tumor tissue also indicated that circ_0001946 could significantly promote the expression of Ki67 and offset the effect of miR‐671‐5p (Figure 8d–e).