Downregulation of Circular RNA circPSD3 Promotes Metastasis by Modulating FBXW7 Expression in Clear Cell Renal Cell Carcinoma

Circular RNAs (circRNAs), a novel class of noncoding RNAs, have been shown to play critical regulatory roles in clear cell renal cell carcinoma (ccRCC). Metastasis is the main contributor to the poor prognosis of patients with ccRCC. However, the role of circRNAs in ccRCC metastasis has not been fully elucidated. In this study, microarray and RNA-seq analyses revealed that circPSD3 (hsa_circ_0002111) was dramatically downregulated in ccRCC tissues compared to adjacent nontumor tissues. A qRT-PCR analysis performed on our ccRCC cohorts confirmed the downregulation of circPSD3 in ccRCC tissues and further suggested that a low level of circPSD3 expression was associated with tumor metastasis in patients with ccRCC. Based on the results of functional studies, circPSD3 significantly inhibited cell migration, invasion, and the epithelial-mesenchymal transition (EMT) in vitro and blocked pulmonary metastasis in vivo. Mechanistically, circPSD3 functioned as a competing endogenous RNA for microRNA 25-3p (miR-25-3p) to regulate F-box and WD repeat domain-containing 7 (FBXW7) expression. Further verification indicated that circPSD3 overexpression restrained an EMT-like phenotype in cells, while miR-25-3p partially rescued these effects. In summary, circPSD3 inhibits tumor metastasis by repressing the miR-25-3p/FBXW7-EMT axis and might be developed as a potential diagnostic and therapeutic target for ccRCC.


Introduction
Renal cell carcinoma (RCC) has the highest lethality rate among urological tumors [1], and its incidence and mortality rates are increasing yearly. Clear cell renal cell carcinoma (ccRCC) is the most common type of RCC, accounting for 70-75% of all RCC cases [2]. e ccRCC is insidious, and approximately 25-30% of patients with ccRCC already have distant metastasis at the first diagnosis [3]. Surgery and other treatments for ccRCC have evolved in recent years [4]. However, the therapeutic effects on metastatic ccRCC are far from satisfactory, and the 5-year survival rate is extremely low [5]. Hence, the mechanisms underlying ccRCC metastasis must be elucidated to promote the further development of novel diagnostic and therapeutic strategies.
Circular RNAs (circRNAs) are characterized by the absence of a 5′-cap and a 3′poly-A tail [6]. us, circRNAs are easily resistant to RNase R digestion and are highly stable. CircRNAs were initially considered transcriptional errors but were later found to be widespread with the development of highthroughput sequencing [7,8]. Currently, circRNAs are known to take part in the development of a host of diseases, such as immune diseases, cardiovascular diseases, neurological diseases, and tumors [9]. e biological functions and roles of circRNAs have been reported in all kinds of different cancers, suggesting the potential of circRNAs as biomarkers and therapeutic targets [10,11]. Numerous studies have reported various functions of circRNAs, which serve as microRNA (miRNA) sponges by directly binding miRNAs [12]. For example, circASAP is upregulated in hepatocellular carcinoma (HCC) and accelerates HCC cell metastasis by combining miR-326 and miR-532-5p, thereby regulating the expressions of MAPK1 and CSF-1 [13]. According to previous studies, circ-AKT3 inhibits ccRCC cell metastasis by competitively sponging miR-296-3p and upregulating E-cadherin [14]. Additionally, circRNAs work by interacting with proteins, and some circRNAs are even translated into proteins [15,16]. Recently, mitochondria-specific circRNA (SCAR) was reported as a therapeutic target for NASH by reducing mitochondrial reactive oxygen species (MROS) output, fibroblast activation, and inflammatory response [17]. However, many mechanisms and functions of circRNAs in ccRCC remain unclear.
In this study, circPSD3 was downregulated in ccRCC tissues from two public circRNA datasets. Lower circPSD3 expression correlated with more ccRCC metastases. Gainof-function assays revealed that circPSD3 suppresses the metastasis of ccRCC cells in vitro and in vivo. Mechanistic studies indicated that circPSD3 inhibits F-box and WD repeat domain-containing 7 (FBXW7) degradation by sponging miR-25-3p and guiding the epithelial-mesenchymal transition (EMT). erefore, circPSD3 is a potential diagnostic and therapeutic target for ccRCC.

Patient Samples.
Eighty-one paired primary ccRCC tumor tissues and adjacent nontumor tissues were obtained from patients who had undergone radical nephrectomy or partial nephrectomy without neoadjuvant chemotherapy or radiotherapy at the First Affiliated Hospital of Jinan University (Guangzhou, China). All the samples were independently identified by two pathologists and stored at −80°C until use. All the samples were approved by the Ethics Committee of the First Affiliated Hospital of Jinan University, and informed consent was obtained from each patient.

Cell Culture and Transfection.
All the cell lines were purchased from the ATCC. Each cell line was cultured in a medium containing 10% FBS and 1% double antibody (100 U/mL of penicillin + 100 mg/mL of streptomycin) and was placed in an incubator containing 5% CO 2 at 37°C.
Routine tests were performed using Hoechst DNA staining to ensure no mycoplasma contamination every six months. CircPSD3 small interfering RNAs (siRNAs) were complexed by RiboBio (Guangzhou, China). miR-25-3p mimics/inhibitors were obtained from GenePharma (Suzhou, China), and Lipofectamine 3000 (Invitrogen, USA) was used to transfect cells. Cells were transfected with a circPSD3 overexpression lentivirus using polybrene, and then 2 μg/mL of puromycin (GeneChem, China) was used for 1 week to establish stably transfected cells. ese sequences are shown in Supplementary Table 1.

RNA Extraction and Quantitative Real-Time PCR.
Total RNA was extracted using TRIzol reagent (Invitrogen, USA). RNA was extracted from the nucleus and cytoplasmic components using a PARIS kit ( ermo Scientific, USA). e cDNA was extracted using a reverse transcription kit (Vazyme, Nanjing, China). qRT-PCR was enforced with SYBR Green Real-Time PCR Master Mix (Vazyme, Nanjing, China). GAPDH or U6 was used as a standard control. e relative RNA expression levels were calculated using the 2 −ΔΔCT method. e specific primers used are shown in Supplementary Table 1.

Cell Proliferation
Assays. Cells were adjusted to 4 × 10 4 cells/mL, and 100 μL/well was inoculated into a 96well plate, followed by the culture at 37°C for 24 h. Ten microliters of CCK-8 solution (Dojindo, Kumamoto, Japan) was added, and the absorbance was measured at 450 nm. e proliferation capacity of cells is proportional to the absorbance of cells.

Transwell Assays.
Transwell assays were performed in twenty-four-well chambers (Corning, USA) with or without Matrigel (BD Biosciences) to assess cell motility. ACHN or Caki-1 cells (8 × 10 4 cells/well) were inoculated in the upper chamber. After incubation for 24 h-48 h, the cells were recorded and imaged under a microscope.
2.6. Wound Healing Assays. Cells were wounded using a 20-μL pipette, and images were acquired after washing with PBS (0 h time point). A serum-free medium was used to culture the cells. After one day, images were captured again (24 h time point). e wound healing rate � area of 24 h wound/ area of 0 h wound × 100%.

Luciferase Reporter Assays.
Briefly, cells were inoculated into 96-well plates and cotransfected with the corresponding pSICHECK2 vector and miR-25-3p mimics. e luciferase activity of fireflies and Renilla was measured using a dualluciferase reporter kit (Promega, WI, USA). e luciferase activity is shown as the ratio of firefly luciferase activity to sea kidney luciferase activity. Independent experiments were conducted in triplicate.

RNA Pull-Down Assays.
In total, 1 × 10 7 cells were lysed in 50 μL of lysis buffer ( ermo Scientific, USA), and then, 10 μg of the biotin-labeled probe was added and incubated for 1 h. Washed streptavidin magnetic beads (Invitrogen, USA) were added to each binding reaction and incubated for 1 h. e beads were briefly washed six times with RIP washing buffer. Finally, the obtained RNA was analyzed by qRT-PCR.

Animal Study.
Four-to six-week-old male BALB/c nude mice were purchased from the Guangdong Medical Laboratory Animal Center. For the subcutaneous xenografting of tumors, 5 × 10 6 cells were injected into the right flank of each mouse. Tumor growth was recorded weekly, and the tumor volume (V) was calculated as follows: V � (W 2 × L)/2. In the lung metastasis model, 6 × 10 6 cells were injected into the tail vein of nude mice. After 6 weeks, the nude mice with lung metastases were observed by IVIS, their lungs were removed, and the number of metastases was photographed and recorded. All the animal studies were regulated in conformity to the Jinan University Guidelines for Animal Care and Use.

Hematoxylin and Eosin (H&E) Staining and Immunohistochemistry (IHC).
Paraffin sections were deacylated in xylene and rehydrated with ethanol. e slices were subjected to microwave heating in EDTA to extract the antigens and then incubated with goat serum at 37°C for 1 h and the indicated primary antibody at 4°C overnight. Subsequently, a horseradish peroxidase detection system was applied to quantify indicator protein expression (DAKO, Glostrup, Denmark).

Western Blot Analysis.
Radioimmunoprecipitation assay lysis buffer (RIPA, Beyotime, China) was added to each group of cells to fully lyse and extract the total protein from the cells. SDS-PAGE (8-15%) was performed using 20 μg of protein per lane. e protein isolates were transferred to PVDF membranes (Millipore, USA) and sealed in 5% skim milk powder solution at room temperature for 1 h. Rabbit polyclonal antibodies against FBXW7 (Abcam, USA), vimentin (Cell Signaling Technology, USA), E-cadherin (Cell Signaling Technology, USA), N-cadherin (Cell Signaling Technology, USA), and GAPDH (Cell Signaling Technology, USA) were added, followed by incubation at 4°C overnight. An appropriate secondary antibody (1 : 5000; Cell Signaling Technology, USA) was incubated in a 37°C shaker for 1 h, and the proteins were visualized using ECL chemiluminescence (Millipore, Germany). Image Lab software was used for data analysis.

Statistical
Analysis. SPSS 26.0 and GraphPad Prism 8 software were used for statistical analysis. e measurement data were tested for normality and homogeneity of variance, and the data were expressed as mean ± SD. Student's t-test and one-way ANOVA were performed to evaluate significant differences between two or more groups. P < 0.05 indicated that the difference was statistically significant.

e Downregulation and Characterization of circPSD3 in ccRCC.
First, we analyzed human ccRCC tissue samples from two datasets (GEO; GSE1000186 and own dataset) to examine the role of circRNAs in ccRCC advancement. Finally, 14 circRNAs that overlapped between the two datasets were extracted (Figure 1(a)), and heatmaps of their expression levels in the two datasets were plotted (Figure 1(b)). Significant downregulation of hsa_circ_0002111 (termed circPSD3) was observed in both cohorts; thus, circPSD3 was selected for further study. Next, circPSD3 was validated to have lower expression in ccRCC tissues than in normal renal tissues using qRT-PCR ( Figure 1(c)). e circPSD3 is derived from exons 5-9 of the PSD3 gene with a length of 538 nt. e back-splice junction site of circPSD3 was validated by Sanger sequencing. e identified sequence was the same as that in the circBase database (https://www.circbase.org/) (Figure 1(d)). We measured the half-life of circPSD3 in the indicated cells by treating them with actinomycin D to evaluate the stability of circPSD3, and circPSD3 was more stable than PSD3 mRNA (Figure 1(e)). We assessed circPSD3 expression in cells by qRT-PCR and found that circPSD3 was mainly distributed in the cytoplasm (Figure 1(f )). Similar results were obtained using the FISH assay ( Figure 1(g)).

CircPSD3 Downregulation Is Associated with ccRCC
Metastasis. Next, we analyzed the relationship between the circPSD3 levels and clinical characteristics and revealed that circPSD3 expression in ccRCC tissues was associated with metastasis (Table 1). e patients were diagnosed with metastatic and nonmetastatic ccRCC based on H&E staining and metastasis-related IHC markers (CD10) (Figure 2(a)).   Journal of Oncology CD10 is considered a metastatic marker for ccRCC [19]. e levels of circPSD3 were significantly downregulated in metastatic ccRCC tissues compared with those in nonmetastatic ccRCC tissues (P < 0.001) (Figure 2(b)). Additionally, we analyzed the level of circPSD3 in 81 human metastatic and nonmetastatic specimens and showed that low circPSD3 expression was notably correlated with metastasis in ccRCC tissues (Figure 2(c)). Decreased levels of circPSD3 were dramatically associated with an advanced Fuhrman grade (Figure 2(d)). Next, we established a receiver operating characteristic (ROC) curve (68 tissues from patients without metastatic ccRCC were used as a control) and found that circPSD3 was highly correlated with ccRCC metastasis. e area under the ROC curve was 0.824 (P < 0.001) (Figure 2(e)). Furthermore, we investigated the expression of circPSD3 in ccRCC cell lines, revealing     Figure S1).

CircPSD3 Inhibits the Migration and Invasion of ccRCC
Cells. We used a lentivirus to overexpress circPSD3 in ACHN and Caki-1 cells and explored the functions of circPSD3 in ccRCC cells. In addition, we knocked down circPSD3 in ACHN and Caki-1 cells via transfecting a circPSD3-specific siRNA. e expression levels of circPSD3 and PSD3 in the cells with circPSD3 overexpression or knockdown were validated by qRT-PCR (Figures 3(a) and S2A). e CCK-8, scratch wound healing, and Transwell assays revealed that the proliferation, migration, and invasion abilities of ACHN and Caki-1 cells with circPSD3 overexpression were dramatically inhibited compared with those of the control groups (Figures 3(b)-3(f )). By contrast, knocking down circPSD3 promoted the malignant functions of ccRCC cells (Figures S2B-S2D).
EMT is a biological process in which tumor cells acquire the ability to migrate and invade. erefore, we examined the relationship between circPSD3 and EMT. Notably, after upregulating the circPSD3 levels, the mRNA level of the epithelial marker E-cadherin was reregulated, and those of the mesenchymal markers, namely, N-cadherin and vimentin, were decreased (Figure 3(g)). We observed similar findings by Western blotting (Figure 3(h)).

CircPSD3 Inhibits ccRCC Metastasis by Mediating the EMT In Vivo.
To evaluate the biological function of circPSD3 that affects metastasis in vivo, we established a  Journal of Oncology lung metastasis model by injecting circPSD3-overexpressing cells (Figure 4(a)). e in vivo imaging system (IVIS) and H&E staining showed that mice with circPSD3 overexpression had significantly fewer lung metastases than the controls (Figures 4(b)-4(c)). Subsequently, we built a subcutaneous transplant model with circPSD3-overexpressing Caki-1 cells. e circPSD3 overexpression dramatically inhibited the growth of xenograft tumors ( Figure S3A) and resulted in an apparent decrease in the xenograft weight ( Figure S3B). Next, the xenograft tumors were sectioned to determine whether the alteration in circPSD3 expression also influenced EMT in vivo. e IHC results showed that the changes in the EMT markers were consistent with the in vitro findings described above (Figure 4(d)). Based on these findings, circPSD3 plays an essential role in regulating EMT in ccRCC. 25-3p. CircRNAs located in the cytoplasm function as miRNA sponges. us, we used starBase and miRanda target prediction tools to identify miRNAs that might bind to circPSD3. In total, 6 miRNAs (miR-25-3p, miR-92b-3p, miR-363-3p, miR-32-5p, miR-367-3p, and miR-92a-3p) were taken for possible targets in the two databases ( Figure 5(a)). RNA pull-down assays were conducted to appraise the mutual effects between these miRNAs and circPSD3. e efficiency of the biotin-labeled circPSD3 probe was performed by qRT-PCR ( Figure 5(b)). Next, we extracted the circPSD3-specific pull-down miRNAs.
ese findings conformed to the competing endogenous RNA (ceRNA) mechanism of circRNAs and miRNAs. Overall, our results showed that circPSD3 functions as a sponge for miR-25-3p.
us, FBXW7 is a potential target of miR-25-3p in ccRCC cells. Next, we first identified the binding sites of miR-25-3p and FBXW7 using the TargetScanHuman website and then performed a luciferase reporter assay, which showed that the relative luciferase activity of the FBXW7 wild-type group was decreased by the miR-25-3p mimics relative to that of the FBXW7 mutant group (Figure 6(c)). qRT-PCR and Western blotting showed that the mRNA and protein levels of FBXW7 were decreased or increased after transfection with the miR-25-3p mimics or inhibitor (Figures 6(d), 6(e), 6SA, and 6SB).
Our clinical ccRCC samples exhibited lower FBXW7 levels in ccRCC tissues than in control tissues, and miR-25-3p expression was inversely linked with FBXW7 expression in our ccRCC cohort (Figures 6(h) and 6(i)). Taken together, our results demonstrated that miR-25-3p promotes the metastasis of ccRCC by targeting FBXW7.

Discussion
Although circRNAs were considered to result from RNA splicing errors [23], the advancement of high-throughput sequencing techniques has led to the clear identification of various circRNAs that are stable and specifically expressed in mammalian cells [8]. Studies have revealed important roles for circRNAs in the tumorigenesis and progression of multiple tumor types, such as lung cancer [24], breast cancer [25], bladder cancer [26], HCC [27], and other types of carcinomas [28][29][30]. Nevertheless, the function of circRNAs in ccRCC, particularly in metastatic ccRCC, remains largely unknown. In this study, we screened the circRNAs that were differentially expressed by RNA-seq or microarrays. Following filtering, we identified a novel circRNA and circPSD3    and investigated their effect on ccRCC metastasis in vitro and in vivo. e ceRNA hypothesis suggests that mRNAs and circRNAs share miRNA response elements, compete for the binding of miRNAs, and regulate the expression of each other, leading to the construction of a complex post-transcriptional regulatory network [31]. Most circRNAs in the cytoplasm contain one or more miRNA-binding elements [32], suggesting that they function as miRNA sponges to regulate the expression of downstream genes [33]. Notably, circPSD3 is more highly expressed in the cytoplasm of ccRCC cells. erefore, we first explored the ability of circPSD3 to act as a miRNA sponge in ccRCC. In this study, we used bioinformatics analysis, luciferase reporter assays, and biotin pull-down assays to testify miR-25-3p as a target of circPSD3 in ccRCC cells. On the side, miR-25-3p expression was high and negatively correlated with that of circPSD3 in ccRCC tissues. According to previous studies, miR-25-3p facilitates the development of multiple tumors, including colorectal cancer [34], breast cancer [35], and pancreatic cancer [36]. However, the functions of miR-25-3p in ccRCC have not been reported. In our study, we confirmed that miR-25-3p promoted the migration and invasion of ccRCC cells, and this phenomenon was reversed by circPSD3.
FBXW7, a critical suppressor of human cancers [37], is a direct target of miR-25-3p, as determined by bioinformatics analysis and the luciferase reporter assay. Previous studies have indicated that the downregulation of FBXW7 facilitates tumor cell migration and invasion through inducing EMT in RCC [21,22]. Consistent with previous studies, we discovered that FBXW7 was downregulated in our ccRCC tissues. Furthermore, FBXW7 mediated the ability of circPSD3 to suppress ccRCC cell migration and invasion activities that are regulated by miR-25-3p. EMT is the transformation of epithelial phenotypic cells into mesenchymal phenotypic cells [38,39]. In cancer, EMT is the process by which epithelial-derived tumor cells acquire the ability to migrate and invade [40,41]. Some studies have reported that circRNAs are associated with EMT [42]. In our study, circPSD3 was closely linked to EMT and inhibited cell invasion and metastasis through the miR-25-3p/FBXW7 axis.
In this study, we probed the expression and biological function of circPSD3 in ccRCC tissues and cells. However, we did not estimate circPSD3 expression in the blood serum, urine, or vesicles of patients with ccRCC because of the lack of specimens. Further studies are needed to explore the level of circPSD3 in ccRCC liquid specimens. Additionally, research on the molecular mechanism of circRNAs, including our study, has focused on miRNA sponges because of the cytoplasmic localization of circRNAs. However, this effect cannot be generally applied. Some studies have demonstrated that circRNAs directly interact with proteins, although they are mainly distributed in the cytoplasm. e circRNA-SORE directly binds to the oncogenic protein YBX1 to mediate sorafenib resistance in the cytoplasm of HCC [43]. erefore, circPSD3 may directly bind some specific proteins in ccRCC cells. Inspired by these findings, we plan to study the other mechanisms of circPSD3 in the future.

Conclusions
In conclusion, our study provides the first evidence that circPSD3 is downregulated in patients with ccRCC and inhibits ccRCC cell metastasis by modulating EMT in vivo and in vitro. Mechanistically, the circPSD3/miR-25-3p/ FBXW7 axis inhibits the EMT and metastasis of ccRCC cells (Figure 7(h)). erefore, circPSD3 may be a potential diagnostic and therapeutic target in patients with ccRCC metastasis.
Data Availability e data presented in this study are available on request from the corresponding author.

Ethical Approval
All samples were obtained by protocols approved by the Ethics Committee of the First Affiliated Hospital of Jinan University. e animal experimental protocol was reviewed and approved by the Institutional Animal Ethical Committee, Experimental Animal Center of Jinan University, and followed the Guide for the Care and Use of Laboratory Animals by the US National Institutes of Health.

Consent
Informed consent was obtained from each patient.

Conflicts of Interest
e authors declare that they have no conflicts of interest.

Authors' Contributions
C.L. and N.Y. conceptualized and designed the study; X.X. and H.L. contributed to formal analysis of data; C.L. N.Y., and B.H. acquired funding; X.X., H.L., Y.L., J.L., and Z.C. investigated the study; Z.C. contributed to data curation; N.Y. supervised the study; B.H. validated the study; N.Y. visualized the study; C.L. was responsible for project administration; C.L. provided resources; X.X. wrote the original draft; N.Y. reviewed and edited the manuscript. Xuexia Xie and Haomin Li equally contributed to this work. 16 Journal of Oncology