Circular RNA circSDHC serves as a sponge for miR-127-3p to promote the proliferation and metastasis of renal cell carcinoma via the CDKN3/E2F1 axis

Human RCC cell lines (A498, 786-O, 769P and Caki-1), a human renal proximal tubular epithelial cell line (HK2), and a human embryonic kidney cell line (HEK-293 T), were purchased from the Chinese Academy of Sciences. 786-O, and 769P were cultured in RPMI 1640 (Gibco, China) supplemented with 10% FBS (PAN-Seratech, Germany). A498, Caki-1, HK2, and HEK-293 T were cultured in DMEM (Gibco, China) supplemented with 10% FBS (PAN-Seratech, Germany). The incubation environment was at 37 °C with 5% CO₂. Cells were routinely checked for mycoplasma infection during cell culture (Beyotime, China).

A total of 140 patients with ccRCC, who underwent radical nephrectomy without neoadjuvant chemotherapy or radiotherapy between 2002 and 2012, at the First Affiliated Hospital of Sun Yat-sen University (Guangzhou, China), were recruited into the study cohort. Patients were followed up regularly, with a median follow-up time of 99.0 months. Overall survival (OS) was defined as the duration from the date of surgery to the date of patient’s death for any reason. Formalin-fixed, paraffin-embedded (FFPE) samples of both tumor and adjacent normal tissue, were collected from these patients for analysis of RNA expression. Total RNA was extracted from tissue specimens using a nucleic acid isolation kit for FFPE (ThermoFisher, USA). Samples used in this study were approved by the Medical Ethics Committee of the First Affiliated Hospital of Sun Yat-sen University.

Microarray datasets were obtained from GEO by searching using the keywords “circRNA’” and “renal cell carcinoma”. Two datasets were retrieved: GSE100186, consisting of 4 tumors and matched adjacent normal tissue; and GSE137836, consisting of 3 primary tumors and 3 metastatic tumors. Arraystar Human circRNA microarray V2 chip info was downloaded for further analysis reference (https://​www.​ncbi.​nlm.​nih.​gov/​geo/​query/​acc.​cgi?​acc=​GPl21825). The Cancer Genome Atlas (TCGA) clear cell renal cell carcinoma (ccRCC) sequencing and clinical data were downloaded from Firebrowse (http://​firebrowse.​org/​). R (version 3.4.3) (https://​www.​r-project.​org/​) was used for subsequent data analysis.

Total RNA samples were extracted from cells using Trizol (Invitrogen, USA) according to the manufacturer’s instructions. Genomic DNA (gDNA) was extracted using a genomic DNA isolation kit (Sangon Biotech, China).

Aliquots of total RNA (2 μg) were incubated with or without 3 U/μg RNase R (Epicenter Technologies, USA) for 30 min at 37 °C and the product RNAs purified using an RNeasy MinElute cleaning Kit (Qiagen, Germany). Isolated RNA was first reverse-transcribed to cDNA using the PrimeScript RT Master Mix (Takara, China) containing random and oligo (dT) primers. Then, PCR was performed using GoTaq Green Master Mix (Promega, USA), according to the manufacturer’s instructions. Primer sequences are provided in Additional file 1: Table S1. PCR products were subjected to electrophoresis on a 2% agarose gels and visualized using Safe Green (Biosharp, China).

For qRT-PCR assays, 2X SYBR Green Pro Taq HS Premix II (AGbio, China) was used and reactions were conducted on QuantStudio 5 real-time-PCR instruments (ThermoFisher, USA). Primer sequences are provided in Additional file 1: Table S1. CircRNA and mRNA levels were normalized to those of GAPDH, while miRNA levels were normalized to those of small nuclear U6. The 2^−ΔΔCt method was used to calculate relative expression levels.

Cells were cultured with or without 2 μg/ml actinomycin D (Sigma, USA) in medium. Then, the cells were harvested at different time points, followed by RNA extraction and qRT-PCR detection of RNA stability, as described above.

Cy3-labeled circSDHC and FAM-labeled miRNA-127-3p probes were synthesized by RiboBio (China). A Fluorescent in Situ Hybridization Kit (RiboBio, China) was used to hybridize the probes to cells. Images were captured on a confocal laser scanning microscope (FV1000; Olympus, Japan).

Cells were harvested and lysed on ice in RIPA buffer (ThermoFisher, USA) containing proteinase inhibitor (Beyotime, China). Then, lysates were incubated on ice for 15 min before being centrifuged for 15 min (13,000 RPM, 4 °C). Supernatants were collected and protein concentration measured using a BCA protein assay kit (ThermoFisher, USA). Protein samples (20 μg) were loaded in each lane of SDS-PAGE gels. After electrophoresis, proteins were transferred onto PVDF membranes, which were blocked in non-fat milk. Then, membranes were incubated overnight at 4 °C with primary antibody, and subsequently with secondary antibody for 1 h at room temperature. Hybridizations were detected using a western blot substrate kit (Tanon, China) on a FluorChem E System (ProteinSimple, USA). Antibodies used in western blots were as follows: CKDN3 (1:1000 dilution, Abcam, USA), E2F1 (1:1000 dilution, Cell signaling, USA), GAPDH (1:1000 dilution, Cell signaling, USA), CDK1 (1:1000 dilution, ThermoFisher, USA), CDK2 (1:1000 dilution, ThermoFisher, USA), HRP-conjugated goat anti-mouse (1:5000 dilution, Proteintech, China), and HRP-conjugated goat anti-rabbit antibody (15,000 dilution, Proteintech, China).

For circSDHC over-expression plasmids, human circSDHC cDNA was synthesized and cloned into a pLVX-cir vector (Genomeditech, China); empty vector was used as the negative control. For siRNA assays, two targeting siRNAs and one scrambled siRNA (negative control) were synthesized by RiboBio (China) (Additional file 1: Table S1). Both overexpression plasmid and siRNAs were transfected using Lipofectamine 3000 (Invitrogen, USA), according to the manufacturer’s instructions. Functional assays were carried out 48 h after transfection. Protein and RNA were harvested 48 h after transfection.

A biotinylated probe targeting the junction area of circSDHC was synthesized (RiboBio, China); an oligo probe served as the negative control. Briefly, to generate probe-coated beads, probes were incubated with streptavidin magnetic beads (Invitrogen, USA) at room temperature for 2 h. Then, the probe-coated beads were mixed with cell lysates overnight at 4 °C. After centrifuging to wash the beads, pulled down miRNAs were extracted using Trizol (Invitrogen, USA) and subjected to qRT-PCR.

Biotinylated miR-127-3p and scrambled negative control miRNA were synthesized (RiboBio, China). Biotinylated miRNAs were transfected into cells using Lipofectamine 3000 (Invitrogen, USA), then cell lysates collected 48 h after transfection and incubated with streptavidin magnetic beads (Invitrogen, USA) at room temperature for 2 h. After centrifuging to wash the beads, pulled down circRNAs were extracted using Trizol (Invitrogen, USA) and subjected to qRT-PCR.

Twenty-four hours before transfection, 3 × 10³ HEK-293 T cells per well were seeded in 96-well plates. A mixture of 50 ng luciferase reporter vectors, 5 ng Renilla luciferase reporter vectors (pRL-TK), and different miRNA mimics were co-transfected into the cells. After 48 h of incubation, luciferase activity was measured using a dual luciferase reporter assay kit (Promega, USA) on a Varioskan LUX machine (Thermo, USA). Luciferase values were normalized to those of corresponding Renilla luciferase, and fold-changes in luciferase values calculated.

Transwell assays were used to evaluate the invasion and migration ability of cells in vitro. Prior to assays, cells were starved by culture in serum-free medium for 8 h. Then, cells were collected, adjusted to a concentration of 1 × 10⁵ in 100 μl of serum-free medium, and added to transwell inserts (Corning, USA), which were coated with (in the invasion assay) and without (in the migration assay) 2% Matrigel (Corning, USA). Medium supplemented with 10% FBS was added to the lower chamber as a nutritional attractant. After incubation (8 h for migration assay and 16 h for invasion assay), transwell inserts were collected, fixed with 4% polyformaldehyde (Beyotime, China), and stained with 0.4% crystal violet (Beyotime, China) for 20 min. Cells on the upper surface were wiped out with a cotton swab, and invaded/migrated cells calculated by capturing five random fields under an Olympus IX83 inverted microscope (Japan).

For MTT assays, 1 × 10³ cells were seeded in each well of a 96-well plate and incubated for a specific period of time. At the harvesting time point, 10 μl of 5 mg/ml MTT (Beyotime, China) was added to each well and incubated for 2 h. Then, the medium was aspirated and 100 μl of DMSO (MP Biomedicals, USA) added to each well, followed by brief shaking to dissolve the crystals, and measurement of absorbance at a wavelength of 490 nm.

These procedures were performed as described previously [15, 16]. Antibodies used for IHC were as follows: CDKN3 (1:100 dilution, ThermoFisher, USA) and E2F1 (1:100 dilution, ThermoFisher, USA). Paraffin sections (thickness, 5 μm) were used for staining. Images were captured with an Olympus IX83 inverted microscope (Japan). Pathological samples were evaluated and scored separately by two qualified pathologists. The IHC scoring is as follows: 0 for no staining, 1+, 2+, 3+ and 4+ for 1–24, 25–49%, 50–74% and over 75% staining intensity, respectively.

All animal care and experimental procedures were conducted according to the guidelines of the National Institutes of Health, and were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University. BALB/c nude mice (3–4 weeks old) were purchased from Vital River Laboratory Animal Technology (China). Stable transfection of control vector and circSDHC shRNA was performed in the 786-O cell line, which was subsequently used for animal studies. For the metastasis experiment, there were eight mice per group. Cells (1 × 10⁶) transfected with control vector or shRNA were injected via tail veins. Mouse body weights were measured weekly. All mice were euthanized after 8 weeks and lung tissues collected and subjected to HE staining. Images were captured using an Olympus IX83 inverted microscope (Japan) and lung metastatic foci were counted in each sample. For the tumor growth study, eight mice were used in each group. Cells (5 × 10⁶) transfected with control vector or shRNA were inoculated subcutaneously into the left side of the body. Tumor size was measured weekly. All mice were euthanized after 4 weeks and subcutaneous tumors were dissected and collected. Final tumor weights were measured and tumor samples were subjected to HE and IHC staining.

All statistical analyses were conducted using GraphPad Prism version 7.0 and R (version 3.4.3) (https://​www.​r-project.​org/​). The student’s t-test was used for comparisons between two experimental groups. Pearson coefficients were calculated to assess correlations. Survival analysis was performed using Kaplan–Meier curves, with application of the logrank test to calculate statistical significance. Univariate and multivariate hazard ratios (HRs) were calculated using Cox regression. All quantitative experimental data are from at least three repeated experiments and are presented as mean ± SD. All p values < 0.05 were considered significant (*P < 0.05; **P < 0.001; ***P < 0.0001).

Two GEO datasets (GSE137836 and GSE100186) from circRNA microarray chips analyzing human tissue samples were used to investigate the role of circRNA in RCC development and progression. The GSE137836 dataset was from three primary and three metastatic tumors, while GSE100186 includes data from four tumors and matched adjacent normal tissue. After analysis of differentially expressed genes, circRNAs with log2 fold-change > 1 or < − 1, and p < 0.05 were selected, illustrated in the volcano plot (Fig. 1a). Commonly altered circRNAs that overlapped across the two datasets, representing a consistent regulatory pattern in RCC, were extracted, yielding a subset of 434 circRNAs (Fig. 1b). To focus on circRNAs with the most oncogenic potential, those located on chromosomes regions frequently amplified in RCC were identified (Additional file 2: Table S2). Subsequently, the 20 most significantly up-regulated circRNAs (log2 fold-change > 2) were selected (Additional file 3: Table S3) and heatmaps of the expression of these circRNAs in the two datasets were plotted (Fig. 1c).

Next, the selected 20 most up-regulated circRNAs were validated in our patient cohort by evaluation of their expression patterns by qRT-PCR. The results for circSDHC which had higher expression in the tumor compared to adjacent normal tissues, were the most consistent across our cohort (Additional file 4: Fig. S1A), with up-regulation in patients who eventually developed metastasis (Additional file 4: Fig. S1B), and an association with poor patient overall survival (Additional file 4: Fig. S1C). Correlation analysis demonstrated that circSDHC expression was associated with clinicopathological parameter TNM stage (Table 1). Further, both univariate and multivariate Cox analysis indicated that higher circSDHC expression was associated with unfavorable survival outcomes (univariate HR: 3.3, 95% CI: 1.8–6.2; multivariate HR: 2.9, 95% CI: 1.5–5.6) (Table 2).

CircSDHC (Arraystar ID: hsa_circRNA_100372, circBase ID: hsa_circ_0015004) is derived from the SDHC gene on chromosome 1, resulting from back-splicing of exons 2, 3, 4, and 5 (385 bp). We conducted Sanger sequencing to confirm the back-splicing junctions of circSDHC (Fig. 1d). Compared with the normal kidney epithelial cell line HK2, four RCC cell lines (Caki-1, A498, 786-O, 769P) showed elevated circSDHC expression (Fig. 1e). PCR analysis confirmed that divergent primers could amplify the circSDHC from reverse-transcribed RNA (cDNA), but not from gDNA (Fig. 1f). Furthermore, circSDHC was more resistant to RNase R treatment, while SDHC mRNA was significantly degraded by this treatment (Fig. 1g). CircSDHC also had longer half-life than SDHC mRNA, as confirmed by actinomycin D treatment (Fig. 1h). In addition, the subcellular localization of circSDHC in 786-O cells was analyzed using a FISH assay, demonstrating that the majority of circSDHC localized to the cytoplasm (Fig. 1i).

To evaluate the biological roles of circSDHC in RCC, we conducted functional assays. To manipulate circSDHC expression, siRNAs targeting the back-splicing junction were constructed (Fig. 2a). Using the siRNAs, expression of circSDHC was successfully knocked down in two RCC cell lines (786-O and A498) with the highest circSDHC levels, without altering expression of the linear form of SDHC mRNA (Fig. 2b). Knockdown of circSDHC significantly decreased the migratory and invasive ability of 786-O and A498 cells (Fig. 2c, e). Further, proliferation of these cells was also suppressed on knockdown of circSDHC (Fig. 2d, f). To further investigate its function, a circSDHC overexpression vector was constructed, which significantly up-regulated circSDHC levels in another RCC cell line, 769P, which had relatively lower circSDHC abundance among the RCC cell lines (Fig. 2g). Consistent with findings from knockdown experiments, the migration, invasion, and proliferation rates of 769P cells were increased on overexpression of circSDHC (Fig. 2h, i). Additionally, we also tested the effect of circSDHC overexpression in normal kidney proximal tubular epithelial cell line HK2. The result showed that circSDHC could also promote malignancy in normal kidney cells, evidenced by increased growth, migration and invasion ability in HK2 after overexpression of circSDHC (Additional file 4: Fig. S1D-F).

A well-recognized function of circRNAs in cytoplasm is as sponges that regulate miRNA expression [17]. As our data indicated that circSDHC localizes to the cytoplasm, we attempted to determine whether it could function as a miRNA sponge. First, downstream miRNAs that may be regulated by circSDHC were predicted using two databases: the Cancer-specific circRNAs database (CSCD, http://​gb.​whu.​edu.​cn/​CSCD/​) and the Encyclopedia of RNA Interactomes (ENCORI, http://​starbase.​sysu.​edu.​cn/​index.​php). Four miRNAs, including miR-3612, miR-650, miR-519a-3p, and miR-127-3p, were identified as potential targets of circSDHC in both databases (Fig. 3a). To further evaluate the relationship between circSDHC and miRNAs, a biotin-labeled circSDHC probe was constructed, and qRT-PCR confirmed that the probe could specifically pull down the circSDHC, compared with a control oligo probe (Fig. 3b, c). Next, miRNAs binding to circSDHC in 786-O and A498 cells were pulled down using this probe. Subsequent qRT-PCR confirmed that miR-127-3p could bind to circSDHC in both 786-O and A498 cells (Fig. 3d). Next, we conducted luciferase reporter assays to validate the binding experiments, and the results confirmed the previous findings (Fig. 3e). Further, biotin-labeled miR-127-3p captured more circSDHC than a negative control probe (Fig. 3f) and FISH analysis in 786-O cells showed that circSDHC and miR-127-3p were co-localized in the cytoplasm (Fig. 3g). Furthermore, analysis of our ccRCC patient data showed a negative correlation between miR-127-3p and circSDHC (Fig. 3h). These results demonstrate that circSDHC can bind to miR-127-3p and act as a sponge for miR-127-3p.

According to TCGA ccRCC database, miR-127-3p expression is lower in tumor tissues of ccRCC, compared to normal tissue (Additional file 5: Fig. S2A). Moreover, it has previously been reported that lower expression of miR-127-3p correlates with early relapse in patients with RCC following nephrectomy [18]. These data suggested that miR-127-3p may have an inhibitory role in RCC development and progression. We also had similar observations in our patient cohort, where lower miR-127-3p expression was detected in tumor tissues than in adjacent normal adjacent tissues (Additional file 5: Fig. S2B). Further, patients who eventually developed distal metastasis had lower miR-127-3p expression in the tumor (Additional file 5: Fig. S2C). Moreover, lower miR-127-3p was correlated with unfavorable OS (Additional file 5: Fig. S2D). Consistent with these findings, overexpression of miR-127-3p in 786-O and A498 cells resulted in decreases in both aggression and proliferation (Fig. 3i-l).

Next, the ENCORI database was used to determine target genes and downstream pathways regulated by miR-127-3p. Cyclin dependent kinase inhibitor 3 (CDKN3) was identified as a potential candidate gene regulated by miR-127-3p, with the binding classified as 7mer-m8 (Fig. 4a). Additionally, we conducted dual luciferase reporter assays, where vector containing wild-type CDKN3 sequence was co-transfected with miR-127-3p mimics, resulting in a reduction of luciferase activity by more than 50%. In contrast, transfection of a vector containing a mutant CDKN3 sequence had no impact on luciferase activity (Fig. 4b). Western blot assays confirmed this regulation (Fig. 4c). Moreover, in our patient cohort, we detected a negative relationship between miR-127-3p and CDKN3 expression levels (Additional file 6: Fig. S3A). These results indicate that miR-127-3p can inhibit RCC progression through downregulation of CDKN3.

According to the Gene Expression Profiling Interactive Analysis (GEPIA) database (http://​gepia.​cancer-pku.​cn/​index.​html) [19], CDKN3 is expressed at higher levels in tumor samples than in normal tissue, and its levels were positively correlated with pathologic stage, and could predict unfavorable OS and disease-free survival outcomes in the ccRCC TCGA dataset (Additional file 6: Fig. S3B-E). CDKN3 knockdown in RCC cells led to a lower proliferation rate and diminished migration and invasion abilities (Fig. 4d-g). To predict the pathway involved downstream of CDKN3, we used Gene Set Enrichment Analysis (https://​www.​gsea-msigdb.​org/​gsea/​index.​jsp) [20, 21]. The results suggested E2F1 as a potentially involved pathway (Fig. 4h). E2F1 is a potent transcription regulator that participates in the development of many different types of cancer [22‐25], and a previous paper reported that E2F1 is also involved in ccRCC progression [26]. GEPIA prediction indicated a positive correlation between CDKN3 and E2F1 levels (Fig. 4i) and western blot analysis following CDKN3 knockdown in 786-O and A498 cells confirmed this relationship (Fig. 4j-k).

To investigative whether circSDHC promotes RCC progression through its sponge effect on miR-127-3p, an miR-127-3p inhibitor and mimic were used to perform rescue experiments following circSDHC depletion or overexpression. Knockdown of circSDHC inhibited the aggression and proliferation of 786-O cells; however, the inhibition could be rescued by application of an miR-127-3p inhibitor (Fig. 5a, c). Similar effects were observed in 769P cells overexpressing circSDHC, which exhibited increased aggression and proliferation that could be attenuated by miR-127-3p mimic (Fig. 5b, d). Western blot assays also showed results consistent with the observed functional changes, where the CDKN3 and E2F1 pathway was less activated upon transfection of circSDHC siRNA, and could be rescued by administration of miR-127-3p inhibitor in 786-O cells (Fig. 5e). In contrast, the CDKN3/E2F1 pathway could be activated by transfection with the circSDHC overexpression vector and this effect was diminished by introduction of miR-127-3p mimic into 769P cells (Fig. 5f). Additionally, since CDKN3 is an important regulator of CDK1 and CDK2, we also tested the effect of circSDHC on CDK1 and CDK2. Our result proved that knockdown of circSDHC caused downregulation of CDKN3 and further decreased the expression of CDK1 and CDK2 (Additional file 6: Fig. S3F, G).

To evaluate the oncogenic effect of circSDHC in vivo, 786-O cells were stably transfected with an shRNA targeting circSDHC, while cells transfected with a scrambled vector were used as a negative control (Fig. 6a). Mice injected with circSDHC knockdown cancer cells demonstrated less cachexia than the control group, as represented by changes in mouse body weight (Fig. 6b). After 8 weeks, mice in the experimental and control groups were euthanized and lung tissues collected. On both gross and microscope examination, lungs from mice injected with circSDHC knockdown cancer cells had less metastatic foci than controls (Fig. 6c). Further, evaluation of subcutaneous tumor formation demonstrated that the circSDHC knockdown group had significantly lower overall mean tumor volume (Fig. 6d) and tumor weight (Fig. 6e) than the control group. IHC analysis of these subcutaneous tumors revealed that CDKN3 and E2F1 expression levels were decreased in the circSDHC knockdown group (Fig. 6f), and the IHC scores of CDKN3 and E2F1 were significant between two groups (Fig. 6g). These results demonstrate that knockdown of circSDHC inhibits RCC tumor progression in vivo.