FOXA1-induced circOSBPL10 potentiates cervical cancer cell proliferation and migration through miR-1179/UBE2Q1 axis

Human normal cervical epithelial cells (H8) and human CC cells (C33A, CaSki, HeLa and SiHa) were bought from Chinese Academy of Sciences (Shanghai, China). Cells were cultured with Dulbecco’s Modified Eagle Medium (DMEM; Invitrogen, Carlsbad, CA, USA) adding 10% fetal bovine serum (FBS; Invitrogen), 1% penicillin/streptomycin (Sigma-Aldrich, Milan, Italy), and then incubated in an incubator of 5% CO₂ at 37 °C.

HeLa and SiHa cells were transfected with specific short hairpin RNAs (shRNAs) against circOSBPL10 (sh-circOSBPL10#1#2), FOXA1 (sh-FOXA1#1#2) and their corresponding negative control (NC) sh-NC, as well as pcDNA3.1/circOSBPL10, pcDNA3.1/UBE2Q1, pcDNA3.1/FOXA1 and empty pcDNA3.1 (±) circRNA Mini vector, empty pcDNA3.1 vector, severally. The miR-1179 mimics, miR-1179 inhibitor, NC mimics and NC inhibitor were synthesized by GenePharma (Shanghai, China). Transfection experiments were executed by Lipofectamine 2000 (Invitrogen).

Total RNA of cells was isolated using TRIzol reagent, followed by cDNA (complementary DNA) synthesis with Reverse Transcription Kit (Invitrogen). RT-qPCR was measured by SYBR-Green Real-Time PCR Kit (Takara, Tokyo, Japan) operated on Bio-Rad CFX96 system (Takara). Relative expression level was calculated utilizing 2^−ΔΔCt method with normalization to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or U6. The sequences of primers were presented in Additional file 1: Table S1.

In short, 1 × 10³ cells were plated into a 96-well plate. After incubation for specific times (24, 48, 72, 96 h), cells were processed with 10 μL of CCK-8 reagent for additional 4 h. Absorbance at 450 nm was measured via a microplate reader (Olympus, Tokyo, Japan).

Transfected cells (1 × 10³) were first coated into 6-well plates. After 2 weeks of incubation, cells were rinsed with phosphate buffer saline (PBS; Sigma-Aldrich, San Francisco, USA), fixed in methanol (Sigma-Aldrich) and dyed using crystal violet (Sigma-Aldrich). The visible colonies were counted manually.

To block transcription, 2 mg/ml Actinomycin D (ActD; Sigma-Aldrich) or dimethylsulphoxide (DMSO; Sigma-Aldrich) was added into culture medium. Total RNA was cultivated with or without 3 U/μg of RNase R (Epicentre Technologies, Madison, WI, USA) for 30 min. After treatment with ActD or RNase R, RT-qPCR was applied for determining the expression levels of circOSBPL10 and OSBPL10 mRNA.

Convergent primers and divergent primers were designed to amplify OSBPL10 mRNA or circOSBPL10. The level of circOSBPL10 in PCR products from cDNA and genomic DNA was examined by agarose gels with TE (Tris-ethylene diamine tetraacetic acid) buffer from Thermo Scientific (Waltham, MA, USA).

Apoptosis transfected SiHa and HeLa cells were assessed utilizing TUNEL Apoptosis Kit (Invitrogen). 4′,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich) was employed to dye above cells. The percentage of positive stained cells was observed by fluorescence microscopy (Olympus) and then analyzed.

Cell apoptosis analysis was performed via Cell Apoptosis Analysis Kit (Takara). After incubation in 6-well plates, SiHa and HeLa cells were rinsed with PBS and resuspended in binding buffer. Followed by fixation with 70% ice-cold ethanol (Sigma-Aldrich), cells were double-stained by Annexin V-fluorescein isothiocyanate and propidium iodide. Last, cell apoptosis rate was detected by Flow Cytometer (Becton–Dickinson, MA, USA).

Cell migration abilities were examined using transwell chambers (Corning, NY, USA). Transfected cells with serum-free medium were placed into top compartment, while medium with 10% FBS was added into the lower compartment. 48 h later, cells in the lower chamber were immobilized and dyed in methanol and crystal violet, separately. Then migratory cells were then counted in five random chosen fields under a microscope (Olympus).

SiHa and HeLa cells were added in 6-well plates for cultivation. When cell confluence was 80%, scratches were produced in cell layer using sterile pipette tip. Afterward, cells were cleaned using PBS and incubated in a culture medium for 24 h. Images of migrating cells were detected at last.

Total protein was reaped in Radio Immunoprecipitation Assay (RIPA) lysis buffer (Beyotime, Shanghai, China). Bicinchoninic acid (BCA) kit (Beyotime) determined protein concentrations. Proteins were separated through sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and moved onto polyvinylidene fluoride (PVDF) membranes. The membranes were sealed with 5% skimmed milk and cultivated with primary antibodies against UBE2Q1 (orb77618, Biorbyt, San Francisco, California, USA) and GAPDH (ab8245, Abcam, Cambridge, UK). Secondary antibody was added for incubating for 1 h. GAPDH was an internal control. Proteins quantities were evaluated via chemiluminescence detection system.

The wild-type (WT) and mutant (Mut) binding sites of miR-1179 in circOSBPL10 or UBE2Q1 3′UTR was sub-cloned into pmirGLO dual-luciferase vector to construct circOSBPL10-WT/Mut or UBE2Q1-WT/Mut. And plasmids were co-transfected with miR-1179 mimics or NC mimics into SiHa and HeLa cells, respectively. The pGL3-OSBPL10 promoter was co-transfected with pcDNA3.1/FOXA1 or pcDNA3.1 into cells. Dual-Luciferase Reporter Assay System (Promega, USA) detected the luciferase activity.

Briefly, cell lysates were incubated with biotinylated RNA including Bio-miR-1179-WT, Bio-miR-1179-Mut and Bio-NC. Moreover, M-280 streptavidin magnetic beads (Sigma-Aldrich) were added to co-culture for 48 h. The relative enrichment of RNAs pulled down in each group were assayed by RT-qPCR.

RIP assays were progressed with Magna RIPTM RNA-Binding Protein Immunoprecipitation Kit (Millipore, Bedford, USA). SiHa and HeLa cells were lysed with RIP lysis buffer, followed by incubation with magnetic beads conjugated with anti-Ago2 (Millipore) or anti-IgG (Millipore). The RT-qPCR was performed to evaluate the expression levels of circOSBPL10, miR-1179 and UBE2Q1 in the precipitates.

Via Magna ChIP Kit (Millipore), ChIP experiment was achieved. DNA in cell lysates was interrupted into 200-300-bp chromatin fragments by ultrasound. After that, lysates were subjected to immunoprecipitation with anti-FOXA1 or anti-IgG (negative control group). The precipitated DNA fragments were detected by RT-qPCR.

To study the cellular function of circOSBPL10 in CC, we first applied RT-qPCR analysis and unveiled a marked elevation of circOSBPL10 expression in CC cell lines compared with H8 cells (Fig. 1a). Then, nucleic acid electrophoresis manifested that in SiHa and HeLa cells, divergent primers could produce circOSBPL10 from cDNA but not from genomic DNA (gDNA), while convergent primers amplified linear OSBPL10 from both cDNA and gDNA (Fig. 1b). Besides, OSBPL10 mRNA was greatly degraded by ActD whereas circOSBPL10 exhibited as resistant to ActD (Fig. 1c). Additionally, OSBPL10 expression was dramatically reduced whereas circOSBPL10 expression demonstrated no evident change after SiHa and HeLa cells were treated with RNase R (Fig. 1d). Then, we verified that circOSBPL10 expression was lowered in two CC cells after transfection with sh-circOSBPL10#1/2, while those with sh-circOSBPL10#1 showed higher silencing efficiency (Fig. 1e). Subsequently, cell proliferation assays depicted a notably weakened proliferation ability of SiHa and HeLa cells under circOSBPL10 silence (Fig. 1f, g). Moreover, cell apoptosis capability was proved to be facilitated after silencing circOSBPL10 in SiHa and HeLa cells (Fig. 1h, i). Furthermore, it was uncovered that circOSBPL10 deficiency gave rise to attenuated migration ability of SiHa and HeLa cells (Fig. 1j, k). Taken together, circOSBPL10 is expressed at high levels in CC and knockdown of it impairs malignant behaviors in CC cells.

For the purpose of investigating the molecular mechanism of circOSBPL10 in regulating CC, we first detected its cellular sublocalization in SiHa and HeLa cells via subcellular fractionation. As illustrated in Fig. 2a, circOSBPL10 was majorly distributed in cytoplasm. Thus, we speculated that circOSBPL10 might affect CC via serving as a sponge of specific miRNA. After searching starBase (http://​starbase.​sysu.​edu.​cn/​) with certain condition (CLIP Data: strict stringency ≥ 5, Degradome Data: low stringency ≥ 1), three miRNAs (miR-1179, miR-27a-3p and miR-27b-3p) were revealed to have binding potentials with circOSBPL10 (Fig. 2b). Then, we discovered a significant downregulation of miR-1179, whereas no apparent changes on the levels of miR-27a-3p and miR-27b-3p, in CC cell lines compared to normal H8 cells (Fig. 2c). Therefore, miR-1179 was chosen for further analysis. Subsequently, circOSBPL10 and miR-1179 were presented to be conspicuously concentrated in anti-Ago2 group (Fig. 2d). Afterwards, two binding sites between circOSBPL10 and miR-1179 were predicted via starBase (Fig. 2e). Moreover, we validated that miR-1179 bound with circOSBPL10 at site 1 (Fig. 2f). To further test whether circOSBPL10 promoted CC progression via its interaction with miR-1179, we mutated the sequence of circOSBPL10 recognized by miR-1179. As displayed in Fig. 2g, the expression of circOSBPL10 was observably elevated in C33A and CaSki cells after overexpressing circOSBPL10 and circOSBPL10 (mut). Interestingly, it seemed that cell proliferation and migration in C33A and CaSki cells could be fortified by overexpression of full-length circOSBPL10 but not by upregulation of circOSBPL10 with mutated miR-1179 binding sites (Fig. 2h, i), indicating that the function of circOSBPL10 in CC depended on its binding to miR-1179. In conclusion, circOSBPL10 interacts with miR-1179 to drive CC progression.

To further investigate the downstream mechanism of circOSBPL10 in CC, starBase was utilized. As predicted under certain circumstances (CLIP Data: strict stringency ≥ 3, Degradome Data: low stringency ≥ 1), four candidates (UBE2Q1, PRPF38B, CACUL1 and LUC7L3) were found as the targets of miR-1179 (Fig. 3a). The following RNA pull-down assay demonstrated the distinct enrichment of UBE2Q1 whereas limited harvest of other three mRNAs in Bio-miR-1179-WT group (Fig. 3b). More importantly, UBE2Q1 level was cut down in SiHa and HeLa cells by circOSBPL10 knockdown as well as by augmented miR-1179 expression (Fig. 3c, and Additional file 2). Besides, RT-qPCR obtained a conspicuous elevation on UBE2Q1 expression in CC cell lines relative to H8 cells, at both mRNA and protein levels (Fig. 3e and Additional file 2). Afterwards, the binding sites between UBE2Q1 and miR-1179 were obtained through starBase prediction (Fig. 3f). Importantly, we observed declined luciferase activity of UBE2Q1-WT owing to miR-1179 upregulation through luciferase reporter assay. However, no overt changes on that of UBE2Q1-Mut were noted between two groups (Fig. 3g). More importantly, a significant enrichment of circOSBPL10, miR-1179 and UBE2Q1 in Anti-Ago2 groups was observed via RIP analysis (Fig. 3h). Further, UBE2Q1 expression was visibly upregulated by circOSBPL10 overexpression, whereas enhanced expression of mutated circOSBPL10 had no such influence on UBE2Q1 expression in these two CC cells (Fig. 3i and Additional file 2). In sum, circOSBPL10 facilitates UBE2Q1 expression by sponging miR-1179 in CC.

Given that circOSBPL10 sponged miR-1179 to upregulate UBE2Q1 expression in CC, we intended to detect whether this mechanism contributed to CC progression. Herein, the efficacy of elevating UBE2Q1 expression or lowering miR-1179 level was analyzed at first by RT-qPCR and the outcome appeared to be satisfactory in SiHa cells (Fig. 4a). Afterwards, we testified that UBE2Q1 upregulation or miR-1179 inhibition could offset circOSBPL10 depletion-mediated suppressive effect on cell proliferation (Fig. 4b, c). In addition, cell apoptosis facilitated by circOSBPL10 knockdown was counteracted by UBE2Q1 overexpression or miR-1179 suppression (Fig. 4d, e). Moreover, the attenuated cell migration capability in circOSBPL10-depleted cells was restored by upregulation of UBE2Q1 or inhibition of miR-1179 (Fig. 4f, g). To conclude, circOSBPL10 promotes CC progression via miR-1179/UBE2Q1 axis.

Since the downstream signaling responsible for the regulation of circOSBPL10 in CC had been studied, we subsequently focused on figuring out its possible upstream mechanism. Through utilizing UCSC (University of California, Santa Cruz: http://​genome.​ucsc.​edu/​), FOXA1 seemed to be a probable transcription factor of OSBPL10 (the host gene of circOSBPL10). Prior to testify the influence of FOXA1 on circOSBPL10 expression in CC, we first silenced or overexpressed FOXA1 with satisfactory efficacies in SiHa and HeLa cells (Fig. 5a). Of interest, we discovered that the expression of circOSBPL10 was distinctly declined by FOXA1 depletion whereas overtly augmented by FOXA1 upregulation (Fig. 5b). Later on, we employed JASPAR database (http://​jaspar.​genereg.​net/​) and obtained the DNA motif of FOXA1 (Fig. 5c). Seen from Fig. 5d, the sequences of OSBPL10 promoter were fragmented into five parts (P1-5). Interestingly, it was verified by ChIP assay that FOXA1 mainly bound to OSBPL10 promoter at P4 region (Fig. 5e). Moreover, when overexpressing FOXA1 in SiHa and HeLa cells, the luciferase activity of OSBPL10 promoter-WT was observably increased while there were no evident changes on that of OSBPL10 promoter-Mut (with mutated FOXA1 binding sites predicted in P4 region) (Fig. 5f). In a word, FOXA1 activates OSBPL10 transcription and thereby facilitates circOSBPL10 expression in CC.