Circ_000829 Plays an Anticancer Role in Renal Cell Carcinoma by Suppressing SRSF1-Mediated Alternative Splicing of SLC39A14

Background Covalently closed circular RNAs (circRNAs) play critical oncogenic or anticancer roles in various cancers including renal cell carcinoma (RCC), pointing to their regulation as a promising strategy against development of RCC. We, thus, studied the tumor-suppressive role of circ_000829 in RCC through in vitro and in vivo experiments. Methods The expression of circ_000829 was validated in clinical RCC tissues and RCC cell lines. Based on ectopic expression and knockdown experiments, we examined the interactions among circ_000829, serine and arginine rich splicing factor 1 (SRSF1), and solute carrier family 39 member 14 (SLC39A14, zinc transporter). Then, the effects of circ_000829, SRSF1, and SLC39A14 on cell cycle distribution and proliferation in vitro and on tumor growth in vivo were evaluated in RCC cells. Results Circ_000829 was poorly expressed in RCC tissues and cells, while SRSF1 was highly expressed. Restoration of circ_000829 reduced the levels of SRSF1 and SLC39A14B, thereby repressing the RCC cell proliferation in vitro and tumor growth in vivo. Meanwhile, overexpression of SRSF1 and SLC39A14B promoted the proliferation and cell cycle entry of RCC cells. Mechanistically, circ_000829 directly bound to SRSF1, and SRSF1 enhanced the expression of SLC39A14B by mediating the alternative splicing of SLC39A14. SLC39A14B upregulation negated the effect of SLC39A14 knockdown on RCC cell proliferation. Conclusion Hence, this study suggests the antiproliferative role of circ_000829 in RCC growth and further elucidates the underlying mechanism involving the inhibited SRSF1-mediated alternative splicing of SLC39A14 mRNA.


Introduction
Renal cell carcinoma (RCC) originates from the epithelial cells of the renal tubules [1,2]. Previous evidence has shown that the combination of cabozantinib, nivolumab/ ipilimumab, and atezolizumab/bevacizumab yields better efficacy than other treatments in the clinical treatment of RCC [3], but the three all have disadvantages including narrow range of use, large differences in efficacy, or high treatment costs. At present, in-depth study of the carcinogenic mechanism of RCC has led to the clinical approval of several therapeutic drugs targeting vascular endothelial growth factor or mTOR pathway. However, the therapeutic responses in patients with advanced RCC are still unsatisfactory [4], thus calling for still more effective targeted therapies.
Serine and arginine rich splicing factor 1 (SRSF1) is highly expressed in a variety of tumor cells and plays a role through alternative splicing of genes related to cancers [5]. Indeed, during development, over 95% human genes undergo alternative splicing, and the misregulation of alternative splicing often leads to diseases including cancers [6]. For instance, SRSF1 is involved in the alternative splicing of solute carrier family 39 member 14 (SLC39A14, zinc transporter) through the Wnt pathway in colorectal cancer cells [7]. SLC39A14, also known as ZIP14, has two alternatively spliced products, SLC39A14A (ZIP14A) and SLC39A14B (ZIP14B). Although there is no difference in the total expression of ZIP14, the ratio of ZIP14A to ZIP14B in tumor cells may also be significantly lower than that in normal tissues. Therefore, ZIP14 splicing variant was differentially expressed in cancer tissues [8]. Moreover, evidence has been presented demonstrating the involvement of SRSF1 in the malignant biological processes of RCC cells, which may indicate unfavorable survival RCC patients [9,10]. However, the specific mechanisms of SRSF1 in RCC still need to be further studied.
Circular RNAs (circRNAs) are regulators of various biological mechanisms that can inhibit tumorigenesis [7,11]. There is a paucity of data have confirmed the regulatory effect of circRNAs in the development of RCC, pointing to the importance of circRNAs in developing new targets for RCC therapy [12][13][14]. Of note, evidence exists indicating that circRNAs modulate alternate splicing of mRNAs by binding to SRSF1. For instance, circRPAP2 was suggested to bind to SRSF1 and subsequently repressed SRSF1mediated alternate splicing of PTK2, resulting in the suppression of breast carcinogenesis [15]. However, the mechanistic actions between circRNAs and SRSF1 in RCC have been poorly explored. Herein, we first determined the abnormal expression of SRSF1-related circRNAs in RCC by bioinformatics analysis and identified circ_000829 as a key circRNA that might bind to SRSF1. Then, we explored the role of circ_000829 in RCC and elaborated the underlying mechanisms.

Material and Methods
2.1. Ethics Statement. The current study was approved by the Ethics Committee of Mianyang Central Hospital (S2018085, P2020030) and performed in strict accordance with the Declaration of Helsinki. All participants or legal guardians had signed the informed consent documentation prior to sample collection. Animal experiments were approved by the Animal Ethics Committee of Mianyang Central Hospital (P2020030).

Bioinformatics Analysis.
The expression matrix of the RCC-related circRNA microarray dataset GSE100186 and its comment probe file was downloaded from the Gene Expression Omnibus database. Differentially expressed cir-cRNAs were screened from the GSE100186 dataset ðjlogFC j > 2, p < 0:05Þ using the R language "limma" package. GSE100186 included four RCC samples and four normal samples. A circRNA-RNA binding protein interaction anal-ysis was performed to predict the circRNAs that likely interact with SRSF1 through the Circinteractome website. The expression of SLC39A14B in RCC tissues and normal tissues was compared by the GEPIA site, an interactive web server of cancer expression profiles.

Study Cohort and Sample Collection.
A total of 67 paired RCC tissues and adjacent normal tissues were obtained from RCC patients, including 43 males and 24 females aged 35-76 (56:7 ± 12:0) years, who underwent partial or radical nephrectomy with nephron preservation in the Urology Department of Mianyang Central Hospital from June 2017 to June 2019. The patients were pathologically diagnosed with RCC after surgery, but had not received preoperative neoadjuvant chemotherapy or immunotherapy [16]. The demographic and clinical characteristics are presented in Supplementary Table 1. The tissue samples were stored at -80°C for gene expression determination. The clinical samples were classified by two pathologists according to the WHO/ISUP classification. Grade I (13 cases) and grade II (26 cases) were further defined as the low-level group, and grade III (28 cases) was defined as the high-level group for analysis.

Northern Blot.
As previously reported, digoxigeninlabeled antisense RNA probes (200 nt) were used [17]. Northern blot assay was performed using NorthernMax Kits (AM1940, Thermo Fisher, Waltham, MA) to evaluate the expression of circ_000829 in RCC tissues and adjacent normal tissues and that in A498 and 786-O cell lines in the circ_ 000829 knockdown presence. This assay was also employed to determine the SLC39A14 expression in RCC tissues and adjacent normal tissues and that in A498 and 786-O cell lines in response to SLC39A14 overexpression or knockdown. The blots were visualized using a ChemiDoc XRS+ imaging system (Bio-Rad, Hercules, CA) [18].
2.5. Immunohistochemical (IHC) Staining. Tissue specimens were paraffin-embedded and sectioned. The sections were treated with 3% methanol-H 2 O 2 and then subjected to antigen retrieval. Next, the sections were blocked with normal goat serum and then incubated with primary rabbit antibodies (Abcam, Cambridge, UK) against SRSF1 (ab38017), Ki67 (ab15580), and UBE2C (ab252940) overnight at 4°C. The next day, the sections were reprobed with the goat anti-rabbit IgG (1 : 1000, ab6785, Abcam) at 37°C for 20 min and incubated with HRP-labeled streptavidin protein working solution. After DAB development and hematoxylin counterstaining, the images in five randomly selected highpower fields from each section were observed with a microscope. There were 67 samples in each group when IHC staining was performed to analyze the expression of SRSF1 in RCC tissues and adjacent normal tissues. There were 8 samples in each group when IHC staining was performed to analyze the expression of Ki67, SRSF1, and UBE2C in the tumors of nude mice. PBS-rinsed cells were detached with trypsin and then centrifuged at 800 g to remove the supernatant. The pelleted cells were fixed with 75% ethanol and incubated with 5 μL PI staining solution. A flow cytometer (BD FACSCalibur, San Jose, CA) was adopted to evaluate the cell cycle distribution. Cell apoptosis was detected using FITC apoptosis detection kits (AP101, Multisciences, China). Cells were incubated with 500 μL buffer, 5 μL FITC-labeled Annexin V, and 10 μL PI. The stained cells were evaluated using a LSRFortessa flow cytometer (BD Biosciences) [19].
2.11. EdU Assay. The EdU assay was performed according to the EdU cell proliferation kit (C10310, Guangzhou Ribo Biotechnology Co., Ltd., Guangzhou, China) [20]. Cells were seeded in a 24-well plate with an initial density of 1 × 10 6 cells/mL and seeding volume of 200 μL and incubated for 2 h with the culture medium containing EdU at the concentration to 10 μmol/L. Next, cells were incubated with 100 μL staining solution per well for 30 min in the dark at room temperature and then stained with Hoechst. Cells then were observed under a fluorescence microscope (FM-600, Shanghai Pudan Optical Instrument Co., Ltd., Shanghai, China).

RNA Binding Protein Immunoprecipitation (RIP)
Assay. The cells were lysed with 1 mL RIPA lysis buffer (P0013B, Beyotime Biotechnology Co., Shanghai, China) in the presence of protease inhibitor in an ice bath for 1 h. The binding of circ_000829 to SRSF1 and that of SRSF1 to SLC39A14 mRNA subtype were detected using the RIP kit (Merck Millipore, Billerica, MA). An aliquot of 10 μg cell extract was collected as input. One portion of 200 μg cell lysate was incubated with flag antibody, and another portion of 200 μg was incubated with isotype control antibody IgG. Cell lysates were incubated with RIP buffer containing magnetic beads MA1-91878 (Thermo Fisher Scientific). The beads were added with RNase inhibitor and incubated overnight in a 4°C refrigerator. Magnetic bead products were incubated at 65°C for 45 min with 117 μL RIP buffer, 15 μL 10% SDS, and 18 μL proteinase K. The isolated RNA was purified and stored at -80°C. The purified RNA product was used for subsequent qPCR detection. Antibodies used in this assay included rabbit anti-Flag (1 : 100, AE063, ABclonal, China) and rabbit antihuman IgG (1 : 100, ab109489, Abcam, serving as a NC). The Protein A/G magnetic beads (HY-K0202, MCE, Sovizzo Vicenza, Italia) used were 100 μL/reaction, and the concentration was 1 mg/mL [21].
2.13. RNA Pull-Down Assay. The biotinylated circ_000829 probe was labeled by transcription in vitro, and 100 pmol of biotinylated circ_000829 was incubated with 200 μg cytoplasmic protein extract to form an RNA-protein complex. Part of the cell extract was taken as input, and the rest was 3 Oxidative Medicine and Cellular Longevity incubated with 50 μL M-280 streptavidin magnetic beads (Sigma-Aldrich, St. Louis MO) precoated with RNase-free tRNA (Sigma-Aldrich) at 4°C for 3 h. Total protein was extracted to detect SRSF1 expression using western blot.
2.14. Xenograft Tumor Model. Specific pathogen free BALB/ c nude mice (age: 6-8 weeks) were housed in separate cages with humidity of 60-65% and temperature of 22-25°C, with free access to food and water under a 12 h/12 h light-dark cycle. A total of 1 × 10 6 A498 cells transduced either with lentivirus containing NC or circ_000829 vector were inoculated to the dorsal region of mice (n = 8). The mice were euthanized 30 days later and the excised tumors weighed. The tumor volumes (width × length 2 × 0:5) were measured every 6 days (days 0, 6, 12, 18, 24, and 30).  Table 3. 2.16. Statistical Analysis. All data were analyzed using SPSS 21.0 statistical software (IBM Corp. Armonk, NY). The measurement data are shown as mean ± standard deviation. Cell experiments were repeated in triplicate. Paired t-test was used for comparison between the RCC tissues and adjacent normal tissues, and unpaired t-test was used for comparison between the other two groups. One-way analysis of variance (ANOVA) and Tukey's post hoc test were used for data comparison among multiple groups, and repeated measures ANOVA and Bonferroni's post hoc test were used for data comparison among multiple groups at different time points. Pearson correlation was used to analyze the correlation between two factors. A p value < 0.05 indicated statistically significant difference.

Overexpression of Circ_000829 Inhibits RCC Cell
Proliferation while Increasing Cell Apoptosis. Expression of the oncoprotein SF2/SRSF1 is closely related to poor survival rate of RCC patients [7]. In order to identify the circRNAs that may regulate SRSF1, we first predicted the expression of regulatory circRNAs of SRSF1 through the bioinformatics analysis website Circinteractome. Poorly expressed cir-cRNAs were screened in the GSE100186 dataset and Circinteractome, which yielded only hsa_circ_000829 at the intersection of potential binding circRNAs and poorly expressed circRNAs (Figure 1(a)). We then analyzed the circ_000829 expression level in the GSE100186 dataset and found that circ_000829 was downregulated in RCC samples (Figure 1(b)).
To verify the role of circ_000829 in the development of RCC, the expression of circ_000829 was determined in clinical RCC tissues and adjacent normal tissues by RT-qPCR (Figure 1(c)) and northern blot (Figure 1(d)). It was demonstrated that the expression of circ_000829 was indeed significantly reduced in RCC tissues (Figures 1(c) and 1(d)). In addition, the overexpression efficiency of circ_000829 vector in A498 and 786-O cells was confirmed by RT-qPCR (Figure 1(e)). Moreover, overexpression of circ_000829 increased the number of G0/G1 phase-arrested cells, but decreased that of S phase-arrested cells, as demonstrated by flow cytometry (Figure 1(f) and Supplementary Figure 1A). Cell proliferation was also suppressed in the presence of circ_000829 overexpression (Figure 1(g) Figure 2A). However, transduction with sh-circ_000829 resulted in a decline in the circ_000829 expression as determined by RT-qPCR (Figure 1(h)) and northern blot (Figure 1(i)) and a reduced proportion of G0/G1 phase-arrested cells, yet enhanced cell proliferation (Figure 1(k) and Supplementary Figure 2B) and increased number of cells in the S phase (Figure 1(j) and Supplementary Figure 1B). Furthermore, knockdown of circ_000829 in A498 and 786-O cells inhibited cell apoptosis, while overexpression of circ_000829 promoted cell apoptosis (Figure 1(l)).
Thus, overexpression of circ_000829 inhibited the proliferation of RCC cells and promoted their apoptosis in vitro, while knockdown of circ_000829 negated the results.

Circ_000829 Binds to SRSF1 and Inhibits Its Expression.
Initial bioinformatics analysis suggested that circ_000829 might bind to SRSF1, and we then sought to elucidate the role of circ_000829 binding to SRSF1 in RCC. RIP experiment showed that SRSF1 antibody could pull down a large amount of circ_000829 as compared to IgG antibody (Figure 2(a)). Meanwhile, when biotinylated circ_000829 was used for RNA pull down assay, the results showed a significant increase in the binding of circ_000829 to SRSF1 (Figure 2(b)), suggesting that circ_000829 could directly bind to SRSF1.
Furthermore, RT-qPCR analysis results revealed higher mRNA expression of SRSF1 in RCC tissues than in adjacent normal tissues (Figure 2(c)). Similar results were yielded for the positive expression of SRSF1 in RCC tissues by IHC staining (Figure 2(d)). Pearson analysis demonstrated that circ_000829 and SRSF1 levels were negatively correlated in RCC tissues (Figure 2(e)).
The above results indicate that circ_000829 could bind to SRSF1 and reduce the expression of SRSF1.

Circ_000829 Inhibits Proliferation of RCC Cells and
Promotes Their Apoptosis by Targeting SRSF1. To further clarify the inhibitory mechanism of circ_000829 in A498 and 786-O cells, the two cell lines were transduced with circ_000829 vector alone or combined with SRSF1 vector. Western blot results showed that circ_000829 overexpression suppressed the SRSF1 expression (Figure 3(a)). The EdU results clearly showed that upregulation of circ_ 000829 markedly inhibited cell proliferation, while overexpression of SRSF1 promoted cell proliferation (Figure 3  3.4. SRSF1 Promotes Alternative Splicing of SLC39A14 mRNA to Produce the SLC39A14B mRNA. SRSF1 has been implicated in the alternative splicing of SLC39A14 in colorectal cancer cells [7]. In order to elucidate the regulatory  (Figure 4(a)). The results of RT-qPCR indicated that SRSF1 knockdown caused an increase in SLC39A14A mRNA expression, but a decrease in SLC39A14B mRNA (Figure 4 To further determine whether SRSF1 could bind to SLC39A14A and SLC39A14B, we overexpressed FLAGtagged SRSF1 in A498 cells. RIP and RT-qPCR results showed that SRSF1 pulled down SLC39A14B (Figure 4(f)), indicating a higher affinity between SRSF1 and SLC39A14B.
Next, to verify whether SRSF1 upregulated SLC39A14 pre-mRNA expression or favored its alternate splicing to B isoform over A, we estimated the expression of SLC39A14 pre-mRNA by RT-qPCR. The results revealed that neither overexpression nor knockdown of SRSF1 had any significant effect on SLC39A14 pre-mRNA (Figure 4(g)).
Together, SRSF1 only regulated "B splicing" instead of "A splicing," and the total level of A + B remained unchanged.

SLC39A14B Abrogates the Antiproliferation and Proapoptosis Effects of SLC39A14 Knockdown on RCC Cells.
To clarify the effect of SLC39A14B on RCC, the expression of SLC39A14 protein in RCC samples was analyzed using the GEPIA database. The results showed that the expression of SLC39A14 in RCC samples was increased ( Figure 5(a)). RT-qPCR data further confirmed that the mRNA expression of SLC39A14B was increased in RCC tissues compared to adjacent normal tissues ( Figure 5(b)). Furthermore, north-ern blot analysis results revealed that SLC39A14B was upregulated in RCC tissues ( Figure 5(c)).
SLC39A14 expression was knocked down in A498 and 786-O cells, which were then transduced with SLC39A14A vector or SLC39A14B vector. Combined transduction of sh-SLC39A14 and SLC39A14A vector increased the expression of SLC39A14A, while combined transduction of sh-SLC39A14 and SLC39A14B vector increased the expression of SLC39A14B (Figure 5(d)).
Furthermore, the EdU results displayed that the proliferation of A-498 and 786-O cells was inhibited in the SLC39A14 silencing presence, which could be reversed by overexpression of SLC39A14B but not by overexpression of SLC39A14A ( Figure 5(e) and Supplementary Figure 2D). Flow cytometry detection indicated that the number of S phase-arrested cells was reduced following knockdown of SLC39A14, which could be reversed by overexpression of SLC39A14B ( Figure 5(f) and Supplementary Figure 1E, 1F). In addition, overexpression of SLC39A14B abolished the proapoptotic effect of SLC39A14 knockdown in A-498 and 786-O cells ( Figure 5(g)).
Therefore, overexpression of SLC39A14B can reverse the antiproliferation and proapoptosis effects of SLC39A14 knockdown on RCC cells.

Circ_000829 Inhibits the Tumorigenesis of RCC Cells In
Vivo. We subcutaneously inoculated A498 cells overexpressing circ_000829 into nude mice to further verify the above findings in vivo. It was demonstrated that overexpression of circ_000829 reduced the tumor volume and tumor weight (Figures 6(a) and 6(b)). IHC analysis results revealed that circ_000829 overexpression reduced Ki-67 positive cells in tumor tissues, and that overexpression of circ_000829 diminished the expression of SRSF1 and UBE2C (proliferation-and invasion-related marker) in the tumor tissues

Discussion
Recently, circRNAs, a new type of endogenous RNAs, have been reported as regulators in many biological processes such as transcription, cell cycle control, and carcinogenesis [22,23]. In this study, we investigated the effect of circ_ 000829 on RCC development. Our results showed an inhibitory role of circ_000829 in the progression of RCC by targeting SRSF1 to impede alternative splicing of SLC39A14. We found in this study that circ_000829 was downregulated in RCC and confirmed its suppressive function in RCC. Accumulating evidence reveals that circRNAs play distinct roles as suppressors or oncogenes in various cancers [14,[24][25][26]. However, the role of circ_000829 in cancers has not hitherto been reported. A prior study did show that circ_001842 disrupted miRNA-502-5p-induced inhibition of SLC39A14 to exert an oncogenic effect on RCC [26], which is partially consistent with our above findings. We found that circ_000829 could bind to SRSF1 directly. SRSF1 was overexpressed in RCC cells, and circ_000829 inhibited the proliferation of RCC cells by targeting SRSF1. Martinez-Terroba et al. have proposed that SRSF1, an RNA-binding protein that has substantial effect on human pathogenesis, is implicated in splicing and functions as a proto-oncogene, and that SRSF1 overexpression enhanced the expression of oncogenic isoforms of multiple genes to promote cancer progression [27]. Moreover, as a pivotal member of serine and arginine rich protein family, SRSF1 could regulate alternative splicing of some proteins and is implicated in many biological events such as translation and senescence [28,29]. These results partially supported our

12
Oxidative Medicine and Cellular Longevity finding that circ_000829 suppressed the growth of RCC cells via targeting SRSF1. In addition, our results indicated that SRSF1 promoted the alternative splicing of SLC39A14 to SLC39A14B, which contributed to the proliferation of RCC cells. Consistently, the study of Thorsen et al. has suggested that SRSF1 was involved in alternative splicing of SLC39A14 to drive the proliferation of colorectal cancer cells [7]. Alternative splicing is considered an important factor related to physiological complexity, which occurs in 95% or more of human multi-exon genes [30]. However, alternative splicing usually occurs in the exons and introns of circRNA, making it difficult to determine its internal structure; therefore, circRNA splicing may be cancer-specific [31]. The data of Song et al. have also reported that changes in splicing patterns could alter the function of proteins and showed that other splicing events could be used as promising prognostic biomarkers for clear cell RCC [32]. Therefore, we suppose that SRSF1-mediated alternative splicing may play an important role in RCC. In addition, as a glycosylated protein located on the plasma membrane, SLC39A14 acts as a metal ion transporter for iron, zinc, manganese, and cadmium [8,33]. SLC39A14, also known as ZIP14, has two different isoforms, namely, SLC39A14A (ZIP14A) and SLC39A14B (ZIP14B), of which the latter one has a higher affinity for Cd 2+ than does SLC39A14A [8]. The affinity difference between SLC39A14A and SLC39A14B for Cd 2+ is a result of alternative splicing [10]. Cd 2+ can promote cell migration and invasiveness and is reported to promote the process of RCC [34], indicating that SLC39A14B is a promoter of RCC. In short, SRSF1 promoted the alternative splicing of SLC39A14 into SLC39A14B, which aggravated RCC. We also performed in vivo experiments in mice, which revealed that the overexpression of circ_000829 not only downregulated the expression of SRSF1 and SLC39A14B but also reduced the volume and weight of the tumor, all of which further confirmed the above findings.

Conclusion
In summary, circ_000829 had antitumor effects in RCC. Circ_000829 suppressed the alternative splicing of SLC39A14 by targeting SRSF1, thereby impeding the progression of RCC (Figure 7), which may offer a new therapeutic biomarker or target for RCC. However, this study is still limited by the lack of detection on the concentration of metal ions in clinical samples and cell function experiments. As a metal transporter, how SLC39A14 is affected by the concentration of metal ions in this study remains unclear and needs future investigations.