Circular RNA circ_001842 plays an oncogenic role in renal cell carcinoma by disrupting microRNA‐502‐5p‐mediated inhibition of SLC39A14

Abstract Renal cell carcinoma (RCC) is a common urologic malignancy, and up to 30% of RCC patients present with locally advanced or metastatic disease at the time of initial diagnosis. Increasing evidence suggests that circular RNAs (circRNAs) serve as genomic regulatory molecules in various human cancers. Our initial in silico microarray‐based analysis identified that circRNA circ_001842 was highly expressed in RCC. Such up‐regulation of circ_001842 in RCC was experimentally validated in tissues and cell lines using RT‐qPCR. Thereafter, we attempted to identify the role of circ_001842 in the pathogenesis of RCC. Through a series of gain‐ and loss‐of function assays, cell biological functions were examined using colony formation assay, Transwell assay, annexin V‐FITC/PI‐labelled flow cytometry and scratch test. A high expression of circ_001842 in tissues was observed as associated with poor prognosis of RCC patients. circ_001842 was found to elevate SLC39A14 expression by binding to miR‐502‐5p, consequently resulting in augmented RCC cell proliferation, migration and invasion, as well as EMT in vitro and tumour growth in vivo. These observations imply the involvement of circ_001842 in RCC pathogenesis through a miR‐502‐5p‐dependent SLC39A14 mechanism, suggesting circ_001842 is a potential target for RCC treatment.

systemic treatment. 4 Since the 2000s, significant advances in treating RCC have been achieved, which include drugs targeting vascular endothelial growth factor (VEGF) and mammalian target of rapamycin (mTOR) pathways. 5 Despite progress in the therapeutics, RCC-associated deaths are increasing in most developed countries, 6 indicating a substantial need to expand therapeutic options.
To identify putative drug targets, further understanding of the molecular underpinnings of tumorigenesis in RCC becomes essential.
Recently, a class of newly discovered non-coding RNAs with covalent loop structure, named circular RNAs (circRNAs), have emerged as regulators in different biological processes. 7,8 Aberrations in circRNAs have been related to varied diseases such as human cancers, kidney diseases and cardiovascular diseases. 9 In RCC, circPCNXL2 was found to advance cancer progression by up-regulating ZEB2 through decoying miR-153. 10 Additionally, another study found an increased expression of circ-ZNF609 in RCC and demonstrated that up-regulated circ-ZNF609 promoted RCC cell proliferation and invasion ability. 11 Accumulating evidence has revealed that some circRNAs can regulate microRNAs (miRNAs) by functioning as miRNA sponges and play a significant role in transcriptional control. 12 Using the CircInteractome database, in silico analysis predicted that circ_001842 can bind to some miRNAs including miR-502-5p. In general, miRNAs, small non-coding RNAs with length of 18-25 nucleotides, are implicated in the alteration and reprogramming of somatic cells, with the potential to target a number of molecules and regulate protein output. 13 MiR-502-5p, in particular, has been reported as a protective miR in several human diseases, including osteoarthritis, 14 breast cancer 15 and colon cancer. 16 Several web-available databases exploring miRNA-mRNA interactions show that the gene 'solute carrier family 39 member 14' (SLC39A14) is a potential target of miR-502-5p. SLC39A14 is a member of the SLC39A transmembrane metal transporter family. 17 SLC39A14 has been found up-regulated in gastric cancer and thus has been considered as a prognostic biomarker. 18 In contrast, in prostate cancer, low expression of SLC39A14 has been associated with the aggressiveness of malignant tumour and tumour progression. 19 Considering that such opposing findings suggest SLC39A14 may exhibit variable expression patterns and effects on different human cancers, its role in RCC needs to be elucidated. The present study aims to determine the role played by the circ_001842/miR-502-5p/SLC39A14 regulatory network in RCC, in a bid to enhance the understanding of the mechanisms underlying RCC carcinogenesis.

| RNase R tolerance test
Primers were designed by Yunxu Biotechnology Co., Ltd. (Shanghai, China) and synthesized by Beijing Genomics Institute (Beijing, China).
The specificity of the PCR products was verified by RT-qPCR and PCR gel electrophoresis. RNase R was used to treat the total RNA, and RT-qPCR was used to determine the relative expression levels of circ_001842 before and after treatment.

| Colony formation assay
Caki-1 cells were seeded in 6-well plates at a gradient density of 50, 100 and 200 cells per dish. The medium was placed in a 5% CO 2 incubator at 37°C for 2-3 weeks. When colonies were visible without a microscope, they were fixed with 5 mL 4% paraformaldehyde and   and transferred to polyvinylidene fluoride (PVDF) membranes.

| Immunoblotting
The membranes were blocked with 5% milk powder for 1 hours and incubated with the primary antibody against SLC39A14

| Fluorescence in situ hybridization (FISH) assays
The subcellular localization of circ_001842 and miR-502-5p in

| RNA pull-down assay
The biotin probe was designed to bind the junction region of

| Dual-luciferase reporter assay
The predicted binding site fragment and mutant fragment of circ_001842 and miR-502-5p were each inserted into the respective vector, namely reporter plasmid circ_001842-wild type (WT) and circ_001842-mutant type (MUT). In order to determine whether circ_001842 could bind to miR-502-5p, NC and miR-502-5p mimic were cotransfected into 293T System (Promega, Madison, WI, USA) was used to perform the assay.

cells (Oulu
Relative luciferase activity was quantified as the ratio of firefly luciferase relative light unit (RLU) to Renilla luciferase RLU.
The predicted binding site fragment and mutant fragment of 3'-untranslated region (3'-UTR) and miR-502-5p of SLC39A14 were each inserted into the respective vector, namely reporter plasmids SLC39A14-WT and SLC39A14-MUT. NC and miR-502-5p mimic and SLC39A14 mRNA luciferase reporter plasmid were cotransfected to determine whether SLC39A14 bound to miR-502-5p. The specific steps were the same as described earlier.

| Transwell invasion assay
The membrane of the Transwell apical chamber was coated with precooled Matrigel. Caki-1 cells were resuspended in a serum-free Opti-MEM I medium (Invitrogen Inc, CA, USA) and adjusted to a density of 3 × 10 4 cells/mL. Then, 100 μL cell suspension was inoculated, while 600 μL Roswell Park Memorial Institute (RPMI) 1640 medium with 10% foetal bovine serum (FBS) was added into the lower chambers.
After 24 hours of incubation, the cells were fixed with 4% paraformaldehyde and then stained with 0.05% gentian violet for 5 minutes.
Migrated cells were photographed and counted using randomly selected 5 visual fields under an inverted microscope.

| Scratch test
Caki-1 cells were seeded into 6-well plates at a density of 5 × Data (mean ± SD) from three independent experiments; comparisons between two groups were analysed using t test and those between multiple groups were analysed using one-way ANOVA, followed by Tukey's post hoc test

| Immunohistochemistry
The paraffin-embedded tumour tissues were sliced into 5-μm sections, followed by dewaxing with xylene and dehydration with gradient alcohol. After microwave-stimulated antigen retrieval, the sections were blocked with 1% bovine serum albumin (BSA) sealing reagent. Next, the sections were incubated with the following pri-   between circ_001842 and miR-502-5p detected by dual-luciferase reporter assay. Data (mean ± SD) from three cell experiments were compared between two paired groups using paired t test, and data from multiple groups were analysed using one-way ANOVA, followed by Tukey's post hoc test. n = 97. *P < .05 vs the adjacent tissues or normal tissues   sh-circ_001842, sh-NC and miR-502-5p inhibitor and sh-circ_001842 and miR-502-5p inhibitor, respectively, determined by RT-qPCR and Western blot analysis. Data (mean ± SD) from three independent experiments were compared between two groups using unpaired t test and data from multiple groups were compared using one-way ANOVA, followed by Tukey's post hoc test. *P < .05 vs the cells without treatment

| circ_001842 is highly expressed in RCC
The gene expression data set GSE10 0186 was downloaded from the Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/ geo/). Differential expression analysis of circRNAs between RCC tissues (n = 4) and normal tissues (n = 4) was performed using the 'Limma' package in the R statistical environment. A heatmap ( Figure 1A) showing the expression of the first 10 differentially expressed circRNAs revealed that circ_001842 was elevated in RCC tissues as compared with normal tissues. Next, we found that circ_001842 was an alias of circ_0000328 in the circBase database. The circ_001842 primer was subjected to PCR for gel extraction and sequencing, and the resulting Sanger sequence is shown in Figure S1. This sequence was our research object circ_001842. RT-qPCR demonstrated that the expression of circ_001842 did not decrease significantly after RNase R treatment, whereas linear circ_001842 expression normalized to GAPDH was decreased significantly after RNase R treatment ( Figure S2), suggesting that circ_001842 was resistant to RNase R.
The expression of circ_001842 in 97 RCC and normal tissues was further analysed by RT-qPCR, and the results showed that expression of circ_001842 in RCC tissues was significantly increased (P < .05) ( Figure 1B). Then, RT-qPCR was used to de-  Figure 1C).
Therefore, Caki-1 cells were selected for subsequent experiments.
In addition, circ_001842 was found to be expressed in the cytoplasm of Caki-1 cells using FISH ( Figure 1D). Ninety-seven patients with RCC were divided into high circ_001842 expression (n = 48) and low circ_001842 expression (n = 49) based on a cut-off using the median of circ_001842 expression level ( Figure 1E). The patient survival rate among the group with high circ_001842 expression was found to be lower than those with low circ_001842 expression level (P < .05) (Figure 1F), indicating a positive correlation between circ_001842 and the degree of RCC. These data suggested that circ_001842 expression was increased in commercial renal cancer lines and RCC tissues, and thus, it might play a relevant regulatory role in RCC.

| Silencing circ_001842 impedes RCC cell proliferation, migration and invasion in vitro
To further clarify the possible effect of circ_001842 on RCC, we constructed the overexpression and silencing sequences of circ_001842 and transfected them into Caki-1 cells. Firstly, RT-qPCR results showed that circ_001842 expression was increased after transfection of oe-circ_001842 (P < .05, Figure 2A), whereas after transfection of sh-circ_001842, the expression of circ_001842 was significantly decreased (P < .05, Figure 2B).
Compared with NC, overexpression of circ_001842 in Caki-1 cells notably enhanced cell proliferation ability ( Figure 2C and D), reduced the apoptosis ( Figure 2F and G) and increased the migration and invasion of Caki-1 cells (P < .05, Figure 2I, J, L, and M).
Versus controls, silencing circ_001842 in Caki-1 cells reduced the cell proliferation, migration and invasion ability ( Figure 2C, E, I, K, L, and N), but increased the level of apoptosis ( Figure 2F

| circ_001842 competitively binds to miR-502-5p in RCC cells
Previous FISH results indicated that circ_001842 was localized in the cytoplasm. To further explore the mechanism of circ_001842 in RCC, the CircInteractome database (https://circi ntera ctome.nia.nih. gov/) was utilized to predict miRNAs that may bind to circ_001842.
Thereafter, bioinformatics analysis predicted the existence of a binding site between circ_001842 and miR-502-5p, which was further confirmed by dual-luciferase reporter assay. The results showed that

| circ_001842 up-regulates the expression of SLC39A14 by binding to miR-502-5p
The potential target genes of miR-502-5p were predicted using the  Figure 4A). There was only one overlapping gene SLC39A14, which was highly expressed in RCC tissues in the GSE10 0666 data set ( Figure 4B). It was thus speculated that miR-502-5p RCC and normal tissues analysed by RT-qPCR was found to be significantly enhanced in RCC (P < .05, Figure 4C), whereas the expression of miR-502-5p showed an opposite trend (P < .05, Figure 4D).
The subsequent results revealed that in cells cotransfected with miR-502-5p mimic and SLC39A14-WT, the luciferase activity was decreased (P < .05) ( Figure 4E), indicating that miR-502-5p bound to SLC39A14. Caki-1 cells were then transfected with oe-circ_001842, miR-502-5p mimic or both. RT-qPCR and immunoblotting showed that versus oe-NC transfection, SLC39A14 mRNA and protein expression was increased following transfection with oe-circ_001842, but it was decreased upon transfection with oe-NC/miR-502-5p mimic (all P < .05). SLC39A14 mRNA and protein expression in cells cotransfected with oe-circ_001842 and miR-502-5p mimic was higher than that in cells cotransfected with oe-NC and miR-502-5p mimic, while lower than that in cells cotransfected with oe-circ_001842 ( Figure 4F and G). Caki-1 cells were transfected with sh-circ_001842, miR-502-5p inhibitor or both. The results of RT-qPCR and immunoblotting demonstrated that SLC39A14 mRNA and protein expression was reduced in the absence of circ_001842, but increased upon miR-502-5p inhibitor treatment (all P < .05). It was suggested that SLC39A14 expression in cells cotransfected with sh-circ_001842 and miR-502-5p inhibitor was higher than that in cells treated with sh-circ_001842, yet lower than that in cells treated with miR-502-5p inhibitor ( Figure 4H and I).
These results indicated that circ_001842 up-regulated the expression of SLC39A14 in tumour cells by binding to miR-502-5p.

| circ_001842 silencing inhibits the growth of RCC tumours in vivo
Nude mice were injected with Caki-1 cells transfected with sh-NC and sh-circ_001842, and 7 nude mice were randomly selected from each group every 7 days. Tumours were extracted and weighed, and the tumour volume was measured. Compared with the mice treated with sh-NC, the tumorigenesis ability of the mice treated with sh-circ_001842 treatment was lowered (n = 7, P < .05), and the number of tumours formed was reduced (n = 7, P < .05, Figure 6A

| D ISCUSS I ON
RCC is one of the most fatal urinary cancers, and as its early symptoms are usually unnoticed, patients typically present with metastatic disease at the initial diagnosis. 21 Despite some therapeutics like IMA901, the first therapeutic vaccine applied in RCC treatment that yielded fruitful outcome, the development of potential therapeutic strategies for RCC remains challenging. 22,23 Therefore, it is necessary to elucidate molecular mechanism underlying RCC.
Circular RNAs have been found to serve as critical regulators in various human cancers. 24 The role of circ_001842 in RCC is currently unclear. Hence, our study explored how circ_001842 impacted tumorigenesis in RCC. We discovered that silencing circ_001842 suppressed the development of RCC by disturbing miR-502-5p-mediated inhibition of SLC39A14.
Firstly, our study has provided evidence that circ_001842 expression is increased in RCC tissues and cells, and down-regulation of circ_001842 inhibits RCC cell proliferation, migration and invasion abilities. In a similar finding, a recent study found high expression of circ-ZNF609 in RCC cell lines, which accelerate cancer cell invasion and proliferation. 11 circ_101882 is also found overexpressed in gastric cancer tissues and cell lines, while poorly expressed circ_101882 has an antitumour effect, inhibiting cancer cell growth. 25 Another investigation revealed a high expression of circABCC2 in hepatocellular cancer, while a reduction of circABCC2 represses hepatocellular cancer cell invasion and proliferation, but advances apoptosis, 26 a finding that was similar to our result. Our findings verified that circ_001842 promoted tumorigenic ability and lymph node metastasis in vivo. Taken together, our findings provide evidence for the first time that circ_001842 is an oncogenic non-coding RNA.
In addition, we provided evidence that circ_001842 upgraded RCC cell proliferation and metastasis, specifically by increasing SLC39A14 expression. SLC39A14 has been found as involved in tumorigenesis of several cancers and has been identified as a novel biomarker for a variety of cancers, 19 which is in agreement with our finding. SLC39A14 is known to be up-regulated in gastric cancer. 18 Another recent study demonstrated that the expression level of SLC39A14 is amplified in hepatocellular carcinoma cells and tissues. 27 Aligned with these studies, our study revealed that a high expression of SLC39A14 advanced the evolution of RCC.
Moreover, miR-502-5p induced SLC39A14 expression and thus promoted RCC cell proliferation and metastasis. In sum, overexpressed circ_001842 promoted the biological function of RCC cells via the up-regulation of SLC39A14 by binding to miR-502-5p.
The marker E-cadherin, in particular, has been related to lymph node metastasis in laryngeal carcinoma. 28  our findings is that circ_001842 promotes tumorigenic ability and lymph node metastasis in vivo.
In conclusion, we ascertained the carcinogenic role of circ_001842 in RCC and found circ_001842 increased the expression of SLC39A14 by competitively binding to miR-502-5p, thereby promoting invasion, metastasis and inflammation of RCC cells ( Figure 7). Therefore, the circ_001842/miR-502-5p/SLC39A14 axis may be considered as a new biomarker for prognosis and a promising therapeutic target for RCC treatment. Nonetheless, the current study only presents a theoretical basis for the mechanism of circ_001842 involvement in RCC. Further research is required to explore the specific mechanisms involving various other molecular pathways involved in RCC carcinogenesis.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets generated and/or analysed during the current study are available from the corresponding author upon reasonable request.