circRNF20 aggravates the malignancy of retinoblastoma depending on the regulation of miR-132-3p/PAX6 axis

Abstract Circular RNAs (circRNAs) serve as essential players in diverse human cancers, including retinoblastoma (RB). In this study, the function of circRNA Ring Finger Protein 20 (circRNF20) in RB progression was investigated. Quantitative real-time polymerase chain reaction, western blot assay or immunohistochemistry assay was performed to determine the expression of circRNF20, miR-132-3p and Paired Box 6 (PAX6). Dual-luciferase reporter assay, RNA immunoprecipitation assay and RNA pull-down assay were utilized to verify the relationships among circRNF20, miR-132-3p and PAX6. In vivo experiment was done for circRNF20 function in tumor formation. It was found that ircRNF20 level was increased in RB tissues and linked to advanced tumor, nodes, metastases (TNM) stage and poor overall survival rate. Deficiency of circRNF20 suppressed cell proliferation, migration and invasion and induced apoptosis in vitro, as well as blocked tumor growth in vivo. circRNF20 directly targeted miR-132-3p and miR-132-3p overexpression inhibited RB cell progression. PAX6 was the target gene of miR-132-3p. Moreover, miR-132-3p inhibition or PAX6 overexpression reversed circRNF20 deficiency-mediated effects on RB cell malignant behaviors. In addition, exosomal circRNF20 was able to promote RB cell progression. Thus, we concluded that circRNF20 served as an oncogene in RB progression through the circRNF20/miR-132-3p/PAX6 pathway.


Introduction
Retinoblastoma (RB) is a common ocular malignancy in children [1,2]. RB has attracted extensive attention in the medical field because of its unique genetic law, multidirectional differentiation potential and high degeneration rate [3]. Early effective treatment can preserve some visual functions and prolong the survival of children [4]. However, if not diagnosed and treated on time and the optimal treatment time is missed, it will lead to poor prognosis in RB patients [5]. Therefore, it is urgent to clarify the basic mechanism of the development of RB, find useful biomarkers and explore novel targets for RB therapy.
Circular RNAs (circRNAs) represent a class of endogenous RNAs which are featured by closed-loop structures [6]. circRNAs can modulate the biological functions via the competitive endogenous RNA mechanism, that is, functioning as microRNA (miRNA) sponges, thereby regulating gene expression [7]. Recently, the contributions of circRNAs in cancer progression have been widely reported. For example, circ_0069094 played an oncogenic role in breast cancer development via altering miR-661 and HMGA1 expression [8]. Circ_0000629 decelerated the carcinogenesis of bladder cancer by the miR-1290/CDC73 pathway [9]. In RB, some circRNAs, such as circTET1 [10], circ_0000034 [11] and circ_0001649 [12], were reported to be involved in RB advancement. CircRNA Ring Finger Protein 20 (circRNF20, hsa_circ_0087784) served as a tumor accelerator in non-small-cell lung carcinoma (NSCLC) and breast cancer [13,14]. Nonetheless, the relationship between circRNF20 and RB development is not clear.
miRNAs are short noncoding RNAs that can influence gene expression by combining with the 3′-UTR of target mRNAs [15]. Increasing evidence has documented the involvement of miRNAs in RB. Li et al. reported that miR-218-5p blocked the malignancy of RB through altering NACC1 expression and AKT/mTOR pathway [16]. Zhang and Wu claimed that miR-378a-3p exerted an anticancer effect on RB by binding to FOXG1 [17]. As for miR-132-3p, Han et al. declared its inhibitory effect on RB progression [18]. However, the exact roles of miR-132-3p in RB still need further research.
Paired Box 6 (PAX6) belongs to the PAX gene family and is a regulator of retinal formation and the development of other eye tissues [19,20]. Furthermore, PAX6 was reported to serve as the target of some miRNAs to participate in RB development [21][22][23]. Even so, whether PAX6 can be targeted by miR-132-3p is unknown.

Tissue and serum collection
A total of 52 RB tissues were collected from RB patients and 22 normal tissues were collected from patients with ophthalmorrhexis treated with enucleation at Lianyungang Maternal and Child Health Hospital. The serum samples were also collected from 38 RB patients and 20 normal patients. None of the patients received chemo-and radiotherapy before the surgery. The tumor and serum samples were saved at −80°C till use. The work was authorized by the Ethics Committee of Lianyungang Maternal and Child Health Hospital and written informed consents were offered by the participants. After the above treatment, circRNF20 and RNF20 mRNA levels were examined through qRT-PCR assay.

5-Ethynyl-2′-deoxyuridine (EdU) detection
By using an EdU assay kit (Beyotime), cell proliferation ability was examined. In short, cells were grown in 24-well plates (5 × 10 3 cells per well) and then kept with EdU for 2 h. Next, the cells were fixed with 4% paraformaldehyde (Sigma-Aldrich) for 0.5 h and permeabilized with 0.5% Triton X-100 (Sigma-Aldrich) for 20 min. Afterward, the cells were washed with PBS and dyed with Apollo and DAPI. The EDU-positive cells were quantified under a fluorescence microscope.

Colony formation analysis
Y79 and WERI-Rb-1 cells were cultured in 6-well plates for about 12 days. Then, the colonies were interacted with 0.1% crystal violet (Sigma-Aldrich) for staining. The number of colonies was counted.

Flow cytometry analysis
The apoptosis of Y79 and WERI-Rb-1 cells was analyzed as per the instructions of an Annexin V-fluorescein isothiocyanate/propidium iodide Apoptosis Detection Kit (Beyotime) and estimated by FACScan ® flow cytometry (BD Biosciences, San Jose, CA, USA).

Wound-healing assay
The migration capacity of Y79 and WERI-Rb-1 cells was tested by this assay. In short, the cells were housed in 12-well plates and allowed to achieve 90% confluence. Next, the scratches were made using a pipette tip. The scratch width was observed at 0 and 24 h.

Transwell assay
The transwell chambers (BD Biosciences) pre-covered with Matrigel (BD Bioscience) were utilized for cell invasion analysis. In short, RB cell suspension (1 × 10 4 cells per mL) was added into the top compartment and the complete culture medium was added into the lower compartment. The invaded cells were dyed with 0.1% crystal violet (Sigma-Aldrich) after 24 h and quantified under a microscope.

Subcellular fraction analysis
PARIS Kit (Life Technologies, Austin, Texas, USA) was used to separate the cytoplasm and nucleus from Y79 and WERI-Rb-1 cells with GAPDH and U6 as the controls for cytoplasm and nucleus, respectively.

In vivo experiment
The BALB/c nude mice were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China) and divided into two groups (n = 5 mice/group). Sh-NC or sh-circRNF20#1 transfected Y79 cells were introduced into the mice. The volume of formed tumors was detected every 7 days and evaluated with the formula: volume = 0.5 × length × width 2 . On day 35, the mice were euthanized and tumors were removed and preserved at −80°C until further use. The in vivo study was approved by the Ethics Committee of Animal Research of Lianyungang Maternal and Child Health Hospital.

Immunohistochemistry (IHC) assay
After the tumor tissues were sectioned at 5 μm, IHC assay was utilized to examine the levels of ki-67 and PAX6, as previously reported [24]. The primary antibodies ki-67 (ab15580) and PAX6 (ab195045) and goat anti-rabbit HRP secondary antibody (ab205718) were obtained from Abcam.

Isolation and identification of exosomes
The isolation of exosomes was executed by using an ExoQuick precipitation kit (System Biosciences, Mountain View, CA, USA). In short, the samples were centrifuged at 3,000×g for 20 min. After filtration with a 0.22 mm PVDF filter, the samples were added with precipitant, mixed, shaked and then stored in the refrigerator at 4°C for 30 min. Thereafter, the samples were centrifuged at 1,500×g for 15 min. The supernatant was discarded and exosomes were suspended in PBS and placed at −80°C until use. The exosome morphology was observed under transmission electron microscopy according to the manufacturers' instructions.
To identify the exosomes, exosomal proteins HSP70 and TSG101 were measured. The related primary antibodies (ab2787; ab30871) were offered by Abcam.

Statistical analysis
All experiments were conducted triple times and exhibited as mean ± standard deviation. Data analysis was conducted using GraphPad Prism 7. The survival curve was obtained through Kaplan-Meier plot and estimated by log-rank test. Spearman's correlation coefficient was performed to analyze the correlation between the circRNF20 level and miR-132-3p level. Student's t-test or one-way analysis of variance was used for difference analysis. P < 0.001 was considered to be significantly different.

CircRNF20 was upregulated in RB tissues and cells
Initially, the expression level of circRNF20 in RB tissues (n = 52) and normal tissues (n = 22) was determined by qRT-PCR assay. The results showed that circRNF20 was markedly upregulated in RB tissues compared to normal tissues ( Figure 1a). Compared to tumors at TNM I/II stages (n = 28), tumors at TNM III (n = 24) stages showed higher expression of circRNF20 expression (Figure 1b). Moreover, the RB patients were divided into two groups (cir-cRNF20 low and circRNF20 high ) according to the median of circRNF20 expression in 52 RB tissues. Our results showed that the overall survival rate of RB patients in the circRNF20 high group was lower than patients in the circRNF20 low group (Figure 1c). qRT-PCR assay indicated that circRNF20 was markedly increased in RB cell lines (including Y79, WERI-Rb-1 and SO-RB50) compared to normal cell line (ARPE-19) (Figure 1d). Actinomycin D assay showed that the half-life of RNF20 was lower than circRNF20 (Figure 1e). RNase R assay indicated that circRNF20 was resistant to RNase R treatment, whereas RNF20 was markedly digested by RNase R (Figure 1f). These results indicated that circRNF20 was stable in RB cells and its abnormal expression might be involved in the progression of RB.

CircRNF20 knockdown suppressed RB cell proliferation, migration and invasion and promoted apoptosis
To investigate the functions of circRNF20 in RB progression, Y79 and WERI-Rb-1 cells were transfected with sh-circRNF20#1, sh-circRNF20#2 or sh-circRNF20#3 to silence circRNF20 expression. qRT-PCR assay showed that sh-circRNF20#1, sh-circRNF20#2 or sh-circRNF20#1 transfection led to a notable reduction in circRNF20 expression in both Y79 and WERI-Rb-1 cells in comparison with sh-NC control groups ( Figure 2a). As indicated by CCK-8 assay, the Y79 and WERI-Rb-1 cells with circRNF20 knockdown showed the inhibited cell viability compared to cells with sh-NC transfection (Figure 2b). EDU assay showed that circRNF20 knockdown repressed the proliferation ability of Y79 and WERI-Rb-1 cells (Figure 2c). The results of colony formation assay exhibited that circRNF20 deficiency restrained the colony formation ability of Y79 and WERI-Rb-1 cells The expression of circRNF20 in RB tissues at different TNM stages was detected by qRT-PCR assay. (c) The overall survival rate of RB patients in circRNF20 low and circRNF20 high groups was analyzed. (d) The level of circRNF20 in ARPE-19, Y79, WERI-Rb-1 and SO-RB50 cells was detected by qRT-PCR assay. (e) After Y79 and WERI-Rb-1 cells were treated with actinomycin D for indicated times, the levels of circRNF20 and RNF20 mRNA were detected by qRT-PCR assay. (f) The levels of circRNF20 and RNF20 mRNA in Y79 and WERI-Rb-1 cells treated with or without RNase R were detected by qRT-PCR assay. ***P < 0.001.
compared to sh-NC control groups (Figure 2d). Flow cytometry analysis indicated that circRNF20 silencing promoted the apoptosis rate of Y79 and WERI-Rb-1 cells relative to sh-NC control groups ( Figure 2e). As illustrated by woundhealing assay and transwell assay, circRNF20 silencing suppressed Y79 and WERI-Rb-1 cells to migrate and invade in comparison with sh-NC groups (Figure 2f and g). Moreover, the levels of proliferation-related protein (PCNA), apoptosis-related protein (cleaved-caspase 3) and epithelial-mesenchymal transition (EMT) markers (E-cadherin, Ncadherin and vimentin) in sh-circRNF20#1 transfected Y79 and WERI-Rb-1 cells were measured by western blot assay. The results showed that circRNF20 knockdown reduced the protein levels of PCNA, N-cadherin and vimentin and elevated the protein levels of cleaved-caspase 3 and E-cadherin in Y79 and WERI-Rb-1 cells (Figure 2h). Collectively, circRNF20 deficiency repressed the malignant phenotypes of RB cells.

Overexpression of miR-132-3p repressed RB cell proliferation, migration and invasion and promoted apoptosis
Next, the roles of miR-132-3p in RB development were explored. As suggested by CCK-8 assay, EDU assay and colony formation assay, miR-132-3p overexpression hampered the viability, proliferation and colony formation of Y79 and WERI-Rb-1 cells compared to miR-NC control groups (Figure 4a-c). Flow cytometry analysis indicated that the apoptosis of Y79 and WERI-Rb-1 cells was induced by increasing miR-132-3p compared to miR-NC control groups (Figure 4d). The results of wound-healing assay and transwell assay indicated that miR-132-3p elevation The relationship between circRNF20 and miR-132-3p was investigated by dual-luciferase reporter assay, RIP assay and RNA pull-down assay. ***P < 0.001.
markedly suppressed the migration and invasion of Y79 and WERI-Rb-1 cells in comparison with miR-NC control groups (Figure 4e and f). Additionally, miR-132-3p upregulation decreased the protein levels of PCNA, N-cadherin and vimentin and increased the protein levels of cleavedcaspase 3 and E-cadherin in Y79 and WERI-Rb-1 cells (Figure 4g). Taken together, miR-132-3p played a suppressive role in RB cell progression.

circRNF20 knockdown restrained tumor formation in vivo
Afterward, the role of circRNF20 in tumor growth in vivo was investigated. As a result, circRNF20 silencing restrained tumor growth (including tumor volume and tumor weight) (Figure 7a and b). IHC assay showed that ki-67 and PAX6 levels were reduced in the xenograft tumors in sh-circRNF20#1 groups compared to sh-NC groups (Figure 7c). The tumors in sh-circRNF20#1 groups also exhibited reduced circRNF20 expression and elevated miR-132-3p expression relative to sh-NC groups (Figure 7d). In addition, the protein levels of PCNA, PAX6, N-cadherin and vimentin were decreased and the protein levels of cleaved-caspase 3 and E-cadherin were increased in the xenograft tumors in sh-circRNF20#1 groups compared to sh-NC control groups (Figure 7e). All these results suggested that circRNF20 played an antitumor effect on RB progression.

Exosomal circRNF20 promoted RB cell growth, migration and invasion and repressed apoptosis
At last, the exosomes were isolated from the serums of RB patients and the morphology is presented in Figure 8a. The exosomal markers HSP70 and TSG101 could be detected in the obtained exosomes (Figure 8b). qRT-PCR assay showed that circRNF20 was upregulated in the exosomes derived from RB patients' serums ( Figure 8c). Then, the exosomes were incubated with Y79 and WERI-Rb-1 cells and then the circRNF20 expression level was detected. The results exhibited that the circRNF20 level was increased in Y79 and WERI-Rb-1 cells incubated with RB patients' serum-derived exosomes (Figure 8d). Moreover, exosome incubation promoted the viability, proliferation and colony formation of Y79 and WERI-Rb-1 cells (Figure 8e-g). Flow cytometry analysis showed that exosome incubation repressed the apoptosis of Y79 and WERI-Rb-1 cells (Figure 8h). The results of wound-healing assay and transwell assay indicated that the migration and invasion of Y79 and WERI-Rb-1 cells were inhibited after exosome incubation (Figure 8i and j). In addition, PCNA, N-cadherin and vimentin levels were increased and cleavedcaspase 3 and E-cadherin levels were decreased in Y79 and WERI-Rb-1 cells incubated with exosomes ( Figure 8k). These results indicated that exosomal circRNF20 might promote RB cell progression.

Discussion
Currently, with the development of bioinformatics technology and sequencing technology, RB-related circRNAs have been gradually discovered [25,26]. Even so, limited circRNAs are reported to be linked to RB. Herein, we aimed to clarify the functions of circRNF20 in RB malignancy. It was found that circRNF20 served as an oncogene in RB through miR-132-3p/PAX6 axis. Cao et al. reported that the raised circRNF20 level was related to poor clinical outcome in breast cancer and The level of circRNF20 in the exosomes derived from the serums of RB patients and normal controls was detected by qRT-PCR assay. (d) The expression of circRNF20 in Y79 and WERI-Rb-1 cells incubated with RB patients serums' isolated exosomes was detected by qRT-PCR assay. (e-j) After Y79 and WERI-Rb-1 cells were incubated with the exosomes derived from RB patient serums, cell viability, proliferation, apoptosis, migration and invasion were analyzed. (k) The protein levels of PCNA, cleaved-caspase 3, E-cadherin, N-cadherin and vimentin were measured through western blot assay. ***P < 0.001. accelerated cancer progression by the miR-487a/HIF-1α/ HK2 pathway [14]. Wang et al. unraveled that circRNF20 triggered the proliferation of NSCLC [13]. These reports suggested that circRNF20 played a vital role in human cancer development. In this work, the abnormal elevation of circRNF20 was discovered in RB. circRNF20 elevation was linked to advanced TNM stage and worse overall survival rate. Rapid growth and metastasis are the characteristics of RB [27]. Herein, our results exhibited that circRNF20 silencing restrained RB cell growth, migration and invasion and triggered apoptosis. EMT refers to epithelial to mesenchymal cell transformation, which gives cells the ability to migrate and invade [28]. Thus, in this study, we determined the expression of EMTrelated markers. It was found that circRNF20 deficiency reduced E-cadherin level and increased N-cadherin and vimentin levels in RB cells, indicating the repression of EMT process. Besides, the tumor-suppressive effect of circRNF20 silencing was demonstrated via in vivo experiments.
Afterward, miR-132-3p was demonstrated to be the target of circRNF20 and circRNF20 negatively regulated miR-132-3p expression by sponging miR-132-3p. Previous studies showed that miR-132-3p could be targeted by circ-sirt1 or circ_DOCK1 to influence tumor progression [29,30]. Nevertheless, the relationship between circRNF20 and miR-132-3p was first verified. Moreover, miR-132-3p elevation restrained RB cell growth and invasion [18]. In the present research, miR-132-3p overexpression curbed cell proliferation and motility and stimulated apoptosis in RB cells. Remarkably, miR-132-3p suppression could abrogate circRNF20 silencing-mediated impacts on RB cell malignant phenotypes. In addition, PAX6 was identified as the target for miR-132-3p. Previous studies showed that PAX6 aggravated RB cell growth, migration and invasion and inhibited apoptosis [20][21][22]. In this study, PAX6 enhancement rescued the impact of circRNF20 silencing on RB cell progression, indicating the promotional effect of circRNF20 on RB development.
Exosomes are extracellular vesicles with 40-100 nm, which can be secreted by almost all cells [31,32]. Exosomes exert physiological roles in cell-to-cell communication by secreting proteins, DNA, lipids, circRNAs, miRNAs, mRNAs and other biomolecules [31]. Exosomal circRNAs can regulate the biological process of tumor cells and function as important biomarkers in human cancers [33][34][35]. In this study, our results presented that circRNF20 was upregulated in RB patients' serum-derived exosomes. Moreover, exosomal circRNF20 boosted cell proliferation and metastasis and repressed apoptosis in RB cells. These results suggested that exosomal circRNF20 promoted RB progression and might be a biomarker for RB diagnosis.
Taken together, circRNF20 aggravated the progression of RB by influencing miR-132-3p/PAX6 axis. These findings might offer a promising strategy for RB therapy. Moreover, exosomal circRNF20 level in serums might be a diagnostic marker for RB.