circGFRA1 and GFRA1 act as ceRNAs in triple negative breast cancer by regulating miR-34a

Tumor and paired adjacent normal mammal tissues from TNBC patients who received treatment at Sun Yat-Sen University Cancer Center were collected and immediately cut and stored in RNAlater (Ambion) and subjected to quantitative real-time PCR (qRT-PCR) analysis. None of these patients received neoadjuvant therapy. This study was approved by the Ethics Committee of Sun Yat-Sen University Cancer Centre Health Authority and in accordance with the ethical standards formulated in the Declaration of Helsinki. All participants provided written informed consent.

All the cell lines were obtained from American Type Culture Collection (Manassas, USA), including human mammary epithelial (HME) cell lines (MCF10A and 184A1) and breast cancer cell lines (SKBR3, T47D, BT474, MCF-7, BT-483, BT-20, BT549, MDA-MB-468 and MDA-MB-231). All the cell lines were passaged in our laboratory for less than six months and maintained according to the supplier’s instructions. The cell lines were found to be free of mycoplasma infection and authenticity verified by DNA fingerprinting before use.

Three TNBC cell lines (MDA-MB-231, BT549 and MDA-MB-468) and normal mammary epithelial cell line (MCF-10A) were analyzed by Arraystar Human circRNA Array V2. Total RNA was quantified using NanoDrop ND-1000. The sample preparation and microarray hybridization were performed based on the Arraystar’s standard protocols. Briefly, total RNAs were digested with Rnase R (Epicentre Technologies, USA) to remove linear RNAs and enrich circular RNAs. Then, the enriched circRNAs were amplified and transcribed into fluorescent cRNA utilizing a random priming method (Arraystar Super RNA Labeling Kit). The labeled cRNAs were hybridized onto the Arraystar Human circRNA Array V2 (8x15K, Arraystar). After washing the slides, the arrays were scanned by the Agilent Scanner G2505C. Agilent Feature Extraction software (version 11.0.1.1) was used to analyze the acquired array images. Quantile normalization and subsequent data processing was performed using the R software limma package. Differentially expressed circRNAs were identified through Fold Change filtering. Hierarchical Clustering was performed to show the distinguishable circRNAs expression pattern among samples.

Total RNA was isolated using TRIzol reagent (Life Technologies, USA). The nuclear and cytoplasmic fractions were isolated using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific). Complementary DNA was synthesized using the PrimeScript RT reagent kit (Takara Bio Inc., China), and RT-PCR was performed using SYBR Premix Ex Taq (Takara Bio Inc.). The primers for circGFRA1 are F: 5′-CCTCCGGGTTAAGAACAAGC-3′, R: 5′-CTGGCTGGCAGTTGGTAAAA-3′. The primers for GFRA1 are F: 5′-CCAAAGGGAACAACTGCCTG-3′, R: 5′-CGGTTGCAGACATCGTTGGA-3′. The threshold cycle (CT) value for circGFRA1 or GFRA1 was normalized against the CT value for control β-actin, while U6 snRNA was used as an internal control for the relative amount of miR-34a. The relative fold-change in expression with respect to a control sample was calculated by the 2-ΔΔCt method. All the real-time PCR assays were performed with the Bio-Rad IQTM5 Multicolour Real-Time PCR Detection System (USA).

Cell proliferation was assessed by Cell Counting Kit-8 assay (Dojindo Laboratories, Japan). Cells (1 × 10³) were seeded into 96-well plates and incubated at 37 °C for 24 h before transfection. CCK-8 solution(10 μl) was added to each well 48 h after transfection. After 2 h of incubation at 37 °C, the absorbance at 450 nM was measured using Spectra Max 250 spectrophotometer (Molecular Devices, USA). Triplicate independent experiments were performed.

Six-well plates were covered with a layer of 0.6% agar in medium supplemented with 20% fetal bovine serum. A total of 1000 cells were prepared in 0.3% agar and cultured for 2 weeks at 37 °C. The numbers of colonies per well were fixed and stained with crystal violet, imaged and counted. Triplicate independent experiments were performed.

For apoptosis assay, cells were stained by propidium iodine/Annexin V-FITC staining (BD Biosciences) then analyzed by flow cytometry FACS Calibur instrument (BD Biosciences) according to the manufacturer’s instructions.

All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) of Sun Yat-Sen University Cancer Center. Standard animal care and laboratory guidelines were followed according to the IACUC protocol. 4-week old female BALB/c nude mice were injected subcutaneously with 2 × 10⁶ cancer cells (5 mice per group) and treated with intratumoral injection (40 μL si-NC or si-circGFRA1). Tumor volumes were measured every 4 days for 28 days and calculated by the formula: volume = length × (width/2)².

Cytoplasmic and nuclear RNA Isolation were performed with PARIS™ Kit (Invitrogen, USA) following the manufacturer’s instruction. Briefly, the cells were lysed with cell fractionation buffer and centrifuged to separate the nuclear and cytoplasmic cell fractions. The supernatant was transferred to a fresh RNase-free tube. The remaining lysate was washed with cell fractionation buffer and centrifuged. Cell disruption buffer was added to lyse the nuclei. Mix the lysate and the supernatant above with a 2× lysis/binding solution and add equal volume of ethanol Draw the mixture through a filter cartridge then wash the sample with wash solution. The RNA of cytoplasmic and nuclear was eluted with elution solution.

Luciferase reporter vector with the full length of the 3′-UTR of circGFRA1 or GFRA1 and the mutant version were constructed. Luciferase reporter vector with miR-34a mimics or miR-34a inhibitors was transfected into MDA-MB-231 cells. After 48 h of incubation, the firefly and Renilla luciferase activities were quantified with a dual-luciferase reporter assay (Promega, USA).

The MS2bp-MS2bs based RIP assay was performed as follows. Briefly, cells were co-transfected with MS2bs-circGFRA1 vector, MS2bs-circGFRA1mt vector (the miR-34a complementary sites in circGFRA1 was mutated by deletions to remove complementarity to miR-34a) or blank control vector MS2bs-Rluc together with MS2bp-GFP using Lipofectamine 2000 (Invitrogen, USA). After 48 h, cells were used to perform RIP using a GFP antibody (Roach, USA) and the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, USA) according to the manufacturer’s instructions. The complexes of RNA were then treated with Trizol (Life Technologies, USA) for further purification and the miR-34a level was analyzed by qRT-PCR.

Western blot analysis was performed using standard procedures. Briefly, total proteins were extracted and separated by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore, USA). To block nonspecific binding, the membranes were incubated with 5% skim milk powder at room temperature for one hour. The membrane was then incubated with primary antibody against GFRA1 (1:1000, CST, USA), followed by horseradish peroxidase (HRP)-labeled secondary antibody (Santa Cruz) and detected by chemiluminescence. An anti-β-actin antibody (1:1000, Affinity, USA) was used as a protein loading control.

Comparisons between groups were analyzed with t tests and χ² tests. Overall survival (OS) and disease-free survival (DFS) curves were plotted using the Kaplan-Meier method and survival differences were evaluated using a log-rank test. Pairwise expression correlation was analyzed by Pearson correlation tests. P < 0.05 was considered as statistically significant. The statistical analysis was performed using SPSS 19.0 software.

To investigate the potential involvement of circRNAs in TNBC, we performed high throughput circRNAs microarray assay on three TNBC cell lines (MDA-MB-231, MDA-MB-468 and BT549) and one HME cell line (MCF-10A). Hierarchical clustering showed the circRNAs expression pattern (Fig. 1a) and the variation of circRNAs expression was revealed in the scatter plots (Fig. 1b). The result showed that 174 circRNAs were downregulated and 93 circRNAs were upregulated with fold change greater than 3, P < 0.05 and FDR < 0.05.

The top 10 upregulated circRNAs were listed in Fig. 1c and qRT-PCR was performed to verify the expression of the top eight upregulated circRNAs (Fig. 1d). The expression of hsa_circ_005239 was the top upregulated circRNA in TNBC cell lines. According to human reference genome (GRCh37/hg19), we further assumed that hsa_circ_005239, located at chr10:117,849,251–117,856,275, is derived from gene GFRA1 (GDNF family receptor alpha 1), which is located on chromosome 10q25. Thus, we termed hsa_circ_005239 as “circGFRA1”.

We further confirmed the expression level of circGFRA1 in 11 breast cell lines, including nine breast cancer cell lines and two HME cell lines. As Fig. 2a showed, circGFRA1 was up-regulated especially in TNBC cell lines (including BT-20, BT549, MDA-MB-468 and MDA-MB-231). Moreover, the expression level of circGFRA1 was evaluated in 51 TNBC tissues and their paired adjacent normal tissues (Fig. 2b). The results showed that circGFRA1 was significantly up-regulated in TNBC tissues.

Next, we explored the potential clinicopathological implications of circGFRA1 in TNBC. The tissues of 222 TNBC patients were used for qRT-PCR analysis of circGFRA1 expression. Patients were divided into high or low expression groups based on circGFRA1 expression level. Patients with circGFRA1 expression levels equal to or greater than the average circGFRA1 expression level were defined as “circGFRA1 high” group. We found that circGFRA1 expression level was positively correlated with tumor size, TNM staging, lymph node metastasis and histological grade but not with age or menopause (Table 1). These results revealed that circGFRA1 might play a vital role in tumor progression of TNBC.

To explore the significance of circGFRA1 in clinical prognosis, we used Kaplan-Meier survival analysis to make the OS and DFS curves. The results showed that patients with high circGFRA1 expression level had shorter OS and DFS than patients with low circGFRA1 expression level (Fig. 2c and d). These results demonstrated that high circGFRA1 expression level was significantly associated with poor TNBC outcomes.

Next, we used RNA interference to knock down the expression of circGFRA1 to evaluate its biological functions in TNBC. qRT-PCR analysis demonstrated that the inhibition was successful (Fig. 3a). Subsequent cell proliferation assay revealed that downregulation of circGFRA1 significantly suppressed the growth of TNBC cells (Fig. 3b). Furthermore, 5-ethynyl-20 -deoxyuridine (EdU) assay showed that the proliferation potential of TNBC cells was impaired upon downregulation of circGFRA1 (Fig. 3c). In colony formation assay, colony-forming ability of TNBC cells were significantly reduced after downregulation of circGFRA1 (Fig. 3d). Apoptosis assay showed that downregulation of circGFRA1 markedly promoted the apoptosis of TNBC cells (Fig. 3e). To further investigate the biological significance of circGFRA1 in tumor growth in vivo, xenograft experiments were performed. And we found that inhibition of circGFRA1 led to a significant decrease in tumor growth (Fig. 3f-3i). All these findings suggest that circGFRA1 could promote proliferation and inhibit apoptosis in TNBC.

Next, we characterized the intracellular location of circGFRA1 in TNBC cell lines. Nuclear and cytoplasmic fractions were separated from cells and the levels of the nuclear control transcript (U6) and cytoplasmic control transcript (GAPDH mRNA) were detected by qRT-PCR, respectively. The results revealed that circGFRA1 mostly distributed in the cytoplasm of TNBC cells (Fig. 4a). As circGFRA1 was predominantly localized in the cytoplasm, it might function as a ceRNA to sequester miRNAs, leading to the liberation of corresponding miRNA-targeted transcripts. Subsequently, we explored whether circGFRA1 could act as a miRNA sponge. The circRNA/miRNA interaction was predicted with Arraystars’s home-made miRNA target prediction software based on TargetScan and miRanda. Potential binding sites of miR-34a were found within the circGFRA1 sequence (Fig. 4b). Thus, we performed luciferase reporter assays to determine whether miR-34a could directly target the 3′UTR of circGFRA1. qRT-PCR analysis demonstrated that the transfection was successful (Fig. 4c). The luciferase intensity was reduced by more than 40% when cells were co-transfection with luciferase reporters and miR-34a mimics (Fig. 4d). Moreover, the luciferase intensity increased when cells were co-transfection with luciferase reporters and miR-34a LNA (Fig. 4d). To confirm the direct interaction between miR-34a and circGFRA1, the MREs of miR-34a in the luciferase reporter were mutated. We found that co-transfection of the mutated luciferase reporter and miR-34a mimics or miR-34a LNA had no significant effect on luciferase activity (Fig. 4d). qRT-PCR analysis further confirmed that miR-34a mimics could inhibit the expression of circGFRA1 while miR-34a LNA increased circGFRA1 expression (Fig. 4e). And the expression of miR-34a was upregulated after inhibition of circGFRA1 (Fig. 4F). To validate the direct binding between circGFRA1 and miR-34a, we performed RIP assay with MS2-binding protein (MS2bp), which specifically binds RNA containing MS2-binding sequences (MS2bs), to pull down endogenous miRNAs associated with circGFRA1. We generated a construct containing circGFRA1 transcript combined with MS2bs elements (named MS2bs-circGFRA1) and cotransfected it into TNBC cells with a construct containing MS2bp-GFP. The immunoprecipitation was then performed using GFP antibody and IgG was used as a negative control. And miR-34a expression was analyzed using qRT-PCR. Figure 4g showed that miR-34a was significantly enriched in RNAs retrieved from MS2bs-circGFRA1 compared with that from control MS2bs-Renilla luciferase (named MS2bs-Rluc) and MRE-mutated MS2bs-circGFRA1 (named MS2bs-circGFRA1mt). No significant difference was shown in miR-34a enrichment between MS2bs-circGFRA1mt group and control MS2bs-Rluc group, indicating the specific interaction between circGFRA1 and miR-34a. In line with luciferase assays, our data confirmed that circGFRA1 functionally interacts with miR-34a and serves as a sponge for miR-34a.

To explore whether circGFRA1 acts as a ceRNA to sequester miR-34a and liberate the expression of GFRA1, we continued to detect the expression of GFRA1 in breast cancer cell lines and tissues. The results showed that GFRA1 was also up-regulated in TNBC cell lines (Fig. 5a) and tissues (Fig. 5b). Furthermore, we used TargetScan to identify the putative target genes of miR-34a and GFRA1 was predicted (Fig. 5c). To confirm this finding, we constructed a luciferase reporter vector with the full length of GFRA1 3′-UTR containing target sites for miR-34a and a mutant version. Luciferase reporter vector, with miR-34a mimics, was transfected into MDA-MB-231 cells. A significant decrease of luciferase activity was observed when co-transfected with miR-34a mimics, but not with scrambled oligonucleotide (Fig. 5d). However, there was not a significant difference in luciferase activity when co-transfected with mutant luciferase reporter. On the contrary, a significant increase of luciferase activity was observed when co-transfected with miR-34a inhibitors (Fig. 5d). qRT-PCR and Western blot analyses validated that the expression of GFRA1 was reduced by miR-34a (Fig. 5e and e). Moreover, the expression of miR-34a was upregulated by GFRA1 inhibition (Fig. 5g). These results indicated that GFRA1 is a direct target of miR-34a and could also sequester miR-34a.

Next, we detected the mRNA levels of GFRA1 after knockdown of circGFRA1 and found a repression of GFRA1 (Fig. 5h). To prove whether circGFRA1 functions as a ceRNA, we co-transfected miR-34a inhibitor and si-circGFRA1 into TNBC cell lines, and observed that the repression was reversed (Fig. 5h). We further detected the expression levels of circGFRA1 after knockdown of GFRA1. The result showed a decreased expression of circGFRA1. And it was also reversed through co-transfection with miR-34a inhibitors and si-GFRA1 (Fig. 5i). These results suggest that circGFRA1 and GFRA1 serve as ceRNAs by harboring miR-34a to modulate the progress of TNBC.

Since circGFRA1 could serve as a ceRNA to regulate the progress of TNBC by sequestering miR-34a, we further explored whether miR-34a and knock down of circGFRA1 could inhibit the progress of TNBC synergistically. Subsequent cell proliferation assay and apoptosis assay revealed that miR-34a combined with downregulation of circGFRA1 further suppressed the growth and promoted the apoptosis of TNBC cells (Fig. 5j & 5k).