circPOLR1C Promotes the Development of Esophageal Cancer by Adsorbing miR-361-3p and Regulating Cancer Cell Apoptosis and Metastasis

Background The effect of circular RNA-RNA polymerase I and III subunit C (circPOLR1C) on esophageal cancer (EC) has not been reported. Herein, this study is designed to unveil the effect and the regulatory mechanism of circPOLR1C on EC. Methods The expression of circPOLR1C in EC tissues and cells was detected by qRT-PCR. Circular structure, stability, and cell localization of circPOLR1C were confirmed by qRT-PCR, RNase R, actinomycin D, and fluorescence in situ hybridization (FISH) assay. Cell function experiments, nude mouse xenograft, lung transplant model, and HE staining were performed to evaluate the effects of CircPOLR1C on EC in vitro and in vivo. A regulatory relationship between miR-361-3p and circPOLR1C was confirmed by qRT-PCR, circRNA in vivo precipitation, RIP, FISH, CircInteractome database, dual-luciferase reporter assay, and immunohistochemistry. Rescue experiments were applied to assess the effects of miR-361-3p and circPOLR1C on EC cells and tissues. Apoptosis- and epithelial-mesenchymal transformation (EMT)-related gene expressions were quantified by qRT-PCR and Western blot. Results Highly expressed circPOLR1C in EC was related to tumor differentiation and invasion. circPOLR1C, which mainly exists in the cytoplasm, is a stable circular RNA. circPOLR1C silencing inhibited circPOLR1C expression and EC cell malignant function, while circPOLR1C overexpression promoted the growth of transplanted tumors and lung metastasis. The enrichment of miR-361-3p was higher than that of other targeted miRNAs. circPOLR1C adsorbed miR-361-3p to regulate apoptosis- and EMT-related genes and partially reversed the tumor suppressive effect of miR-361-3p, which was lowly expressed in EC tissues. Silencing the target genes of miR-361-3p also inhibited the malignant development of EC cells. Conclusion circPOLR1C adsorbs miR-361-3p and regulates apoptosis- and EMT-related gene expressions to promote the development of EC.


Introduction
Esophageal cancer (EC) occupies the 8th place in the cancer incidence worldwide and ranks 6th in the global cause of cancer death [1]. Terefore, EC has become a major public health problem, greatly threatening the life of people. According to the tumor registration data in China, the incidence of EC ranks 5th in the country's total cancer incidence [2]. Besides, the 2012 survey data showed that China has the largest number of new cases of EC in the world [3]. Te histological types of EC are mainly divided into esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma. Since the 1970s, the incidence of ESCC in many Western countries has shown a downward trend, whereas the incidence of esophageal adenocarcinoma has rapidly increased and become one of the fastest growing malignant tumors [4,5]. Terefore, fathoming the pathogenesis of EC, fnding targets for early diagnosis and intervention, and probing into prognostic biomarkers are of great signifcance for the prevention and treatment of EC.
Similar to most tumors, EC is a complex disease that is involved in multiple factors, multiple genes, and multiple stages of development. Te molecular mechanism of its occurrence and development has not been fully elucidated [6]. In exploring the pathogenesis of EC, in addition to oncogenes, tumor suppressor genes, cytokines, and other regulatory factors, noncoding RNA has become a new direction for researchers due to its important role in cancer [7]. Circular RNAs (circRNAs) are single-stranded RNAs with a closed-loop structure and are one of the important members of the noncoding RNA family [8]. A few studies have reported that circRNA could afect the development of EC. For instance, through in vivo and in vitro experiments, Sang et al. confrmed that circRNA ciRS-7 promotes the development of ESCC by sponging miR-876-5p to upregulate MAGE-A signals [9]. Since the high expression of circGSK3β has been found to be positively correlated with the development of ESCC and a poor prognosis, Hu et al. recommended that the level of circGSK3β in plasma should be used as a biomarker for the diagnosis and prognosis of ESCC patients [10].
RNA polymerase I and III subunit C (POLR1C) is a gene encoding shared POLR1 and POLR3 subunits [11]. Studies have shown that POLR1C (hsa_circ_0076535) plays a key role in Treacher Collins syndrome and leukocyte dystrophy [12,13]. Joseph et al. proposed that the expression of POLR1C might be related to the prognosis of patients with triple-negative breast cancer [14]. Previous studies have reported that circPOLR1C (hsa_circ_0076535) is highly expressed in EC tissues [15], but the regulatory mechanism of circPOLR1C in EC has not been reported. On the basis of the above research studies, this study further explored the role and mechanism of circPOLR1C formed by linear POLR1C cyclization in EC, so as to supplement the expression profle and function of circRNA in EC.

Ethical Statement.
Since this study involved clinical experiments and animal experiments, the research group obtained the approval of the Ethics Committee of Zhongshan Hospital, Fudan University (No. 201701007XHK), and the Committee of Experimental Animals of Zhongshan Hospital, Fudan University (No. 201810003XHK), respectively. After consultation with patients and family members, the patients signed the informed consent form and agreed to donate samples of EC and adjacent tissues which were limited to scientifc experimental research. Animal experiments strictly followed animal ethical requirements in terms of the necessity of research, the standardization of operations, and the protection of experimental animals.

Tissue Sample Collection.
From January 2017 to December 2018, 46 patients undergoing pathological diagnosis of primary ESCC in the Department of Gastroenterology of our hospital, who had not received radiotherapy or chemotherapy before surgery, were included in this research. EC tissues and adjacent tissues (≤3 cm from the cancer tissue, about 1 cm 3 ) of the patients were collected during esophagectomy and placed in the RNA locker quickly. After the operation, the tissues were stored at −20°C for later use. Later, we also analyzed the correlation of circPOLR1C expression with gender, age, tumor size (cm), diferentiation, tumor invasion (T), lymph node metastasis, TNM stage, and other clinical information of patients.

Extraction of RNA.
Te extraction method of RNA referred to the existing literature reports. Briefy, TRIzol (1 mL, 15596018, Invitrogen, USA) was used to separate RNA and other components from cells. After the addition of trichloromethane (200 μL, T819285, Macklin, China) and subsequent centrifugation, the mixed solution was divided into aqueous (containing RNA) and organic phases. Next, isopropanol (500 μL, I811932, Macklin, China) was added to recover the RNA precipitate. Tereafter, the obtained precipitate was dissolved by an appropriate amount of diethylpyrocarbonate (DEPC) water, after which RNA concentration was detected by using a Qubit ™ 4 fuorometer (Q33226, TermoFisher, USA). Later, the RNA solution was stored in a refrigerator at −80°C. To determine the specifc expression site of circPOLR1C in cells, the cytoplasm and the nucleus were separated, and RNA was extracted by using the Minute ™ cytoplasmic and nuclear separation kit (SC-003, Invent, China).

Ribonuclease R (RNase R) Treatment.
RNase R is a 3′-5′ ribonuclease derived from the E. coli RNR superfamily and has the function of cleaving RNA [16]. RNase R (R0301, Geneseed, China) treatment was required before the analysis and identifcation of circPOLR1C enrichment, given that RNase R could digest almost all linear RNAs, but it was not easy to digest circular RNAs. Concretely, 10 U of RNase R was added into 2.5 μg total RNA, and the resulting mixture was incubated at 37°C for 30 minutes (min).

circRNA In Vivo Precipitation (circRIP).
Te circRIP assay was conducted to purify and analyze RNA or miRNA related to circPOLR1C. Concretely, biotin-labeled DNA probes (Shanghai Biotechnology Co., Ltd.; probe sequences, TACATTTATTTGAGTGGGTTTGTTGTGCCTCGTCCC-GATGATCCTGCGTGCGCGCTCT-biotin), which could specifcally bind circPOLR1C, were transfected into EC cells. 24 h later, the cells were washed with PBS and fxed by 1% formaldehyde for 10 min. Next, the cells were treated by 1 mL of RIPA lysis and extraction bufer (89901, Termo Scientifc ™ , USA) and then placed in an ultrasonic disruption instrument (SM-150D, Shunma Tech, China) for 10 min. Tereafter, the supernatant was collected and then incubated with Dynabeads ™ M-280 streptavidin (11205D, Invitrogen ™ , USA) at 30°C for 12 h. Ultimately, the mixed solution was further processed with Invitrogen ™ proteinase K solution (AM2546, USA), followed by RNA extraction, reverse transcription, and qRT-PCR.
2.11. RNA Immunoprecipitation (RIP). RIP was the use of antibodies against the target protein to precipitate the target protein and the RNA bound to it. After separation and purifcation, RNA in the complex could be detected and analyzed by qRT-PCR. As described in the relevant literature [18], we used the Geneseed RNA immunoprecipitation kit (P0101, China) to complete detection. circANRIL was used as a negative control in the experiment and did not bind to the Argonaute-2 (AGO2) antibody (ab32381, Abcam, UK).

Cell Counting Kit (CCK)-8 Assay. EC cells were digested
and seeded in 96-well plates at a density of 1 × 10 4 cells/well. Next, 10 μL of the CCK-8 reagent was slowly added to the bottom of the 96-well plate containing cancer cells in order to prevent air bubbles. Ten, the cells were incubated at 37°C with 5% CO 2 and subsequently detected by using a SpectraMax ® Paradigm ® multifunctional microplate reader (Molecular Devices, USA) at a wavelength of 450 nm. Finally, the optical density (OD) values were recorded at 0 h, 24 h, and 48 h.

Cell Clone Formation
Assay. EC cells transfected with sicircPOLR1C or the miR-361-3p mimic were frst digested with Trypsin-EDTA solution (0.25%, 25200072, GIBCO, USA) and then made into cell suspension with a concentration of 3 × 10 4 cells/mL. Tereafter, methanol fxation (15 min) and Giemsa (G4507-5G, Sigma, USA) staining (20 min) were carried out. Eventually, the fully dried cells were placed under a Leica DMi8 microscope (Germany) for observation of the formed cell colonies in order to detect cell proliferation.
2.14. Scratch Test. Predigested EC cells (3 × 10 5 cells/well) were seeded on 12-well plates.After incubation in an MCO-20AIC CO 2 incubator at 37°C for 24 h, a sterile pipette was applied to vertically draw parallel lines (1 cm apart) in the plate. Next, PBS was used to wash foating cells and debris of the plate. Subsequently, cell culture fuid without serum was added to the corresponding cells. After 48 h, the 12-well plates containing the cells were again observed and photographed under a microscope at a magnifcation of 100x.
Ultimately, the migration rate was calculated to assess the efects of diferent intervention conditions on EC cell migration.

Transwell Assay.
According to the manufacturer's instructions, the Transwell chamber with 8 μm pore polycarbonate membrane insert and Matrigel purchased from BD Biosciences (USA) were installed and added, respectively. In a nutshell, EC cells (5 × 10 4 cells/well) were inoculated into the upper chamber of the device covered with Matrigel, and the lower chamber was supplemented with complete medium containing FBS. 48 h later, the cells were harvested, fxed in paraformaldehyde (4%) for 30 min, and stained with crystal violet (0.1%) for 10 min. In the end, the change in cell invasion was observed by using a Leica DMi8 microscope (Germany) under 250 × magnifcation, followed by calculation of the cell invasion rate.

Flow Cytometry.
Apoptosis, as one of the important basic biological functions of cells, was detected by fow cytometry. In brief, 1 × 10 5 EC cells were inoculated into 500 μL diluted 1 × Annexin V binding bufer working solution according to the instructions of the Annexin V-FITC/ PI fuorescence double staining apoptotic detection kit (P-CA-201, Procell, China), followed by being mixed into a cell suspension. Subsequently, 5 μL of Annexin V-FITC and 5 μL of propidium iodide (PI) staining solution were added to the EC cell suspension in sequence. Te mixture was then blended and incubated at room temperature for 20 min. After the reaction, the liquid was transferred to a DxFLEX fow cytometer (Backman Coulter, USA) for further analysis and detection.

Tumor Formation Experiment and Metastasis
Experiment of Nude Mice. 72 male BALB/C nude mice (6 weeks old, weight 18 ± 1 g) used in the experiment were all purchased from the Guangdong Medical Laboratory Animal Center (GDMLAC). Te experimental animals were kept in the SPF-level experimental animal center, and they could eat and drink freely. 48 nude mice were randomly divided into 8 groups (6 mice/group) according to body weight after 3 days of feeding. Mice in the NC + MC group were injected with KYSE-30 or EC9706 cells transfected with the negative control and miR-361-3p mimic control; mice in the circ-POLR1C + MC group were injected with KYSE-30 or EC9706 cells transfected with the circPOLR1C overexpression plasmid and the miR-361-3p mimic control; mice in the NC + M group were injected with KYSE-30 or EC9706 cells transfected with the negative control and the miR-361-3p mimic; mice in the circPOLR1C + M group were injected with KYSE-30 cells or EC9706 cells transfected with the overexpression plasmid of circPOLR1C and the miR-361-3p mimic. 1 × 10 5 cells were injected subcutaneously into the left armpit of each nude mouse. In order to ensure the timeliness of transfected cells, we supplemented the injection once a week. Tumor volume (mm 3 , long diameter-× long diameter × short diameter/2) was measured with a vernier caliper every 5 days. All animals were sacrifced on the 35th day after injection, and after that, the transplanted tumor was removed, photographed, and weighed.
BALB/C nude mice were also used in the lung metastasis experiment. 24 nude mice were randomly divided into 8 groups (3 mice/group) according to the same grouping principle as mentioned before. Te lung transplantation model was constructed by an injection of KYSE-30 or EC9706 cells (1 × 10 5 cells) via the tail vein of mice. After supplementary injection in the same way as mentioned Table 1: Primers for qRT-PCR.

Genes
Forward primer (5′-3′) Journal of Oncology before (once a week), all nude mice were sacrifced in the 8th week after injection and the lung tissue was collected.

Target Prediction and Verifcation.
CircInteractome (https://circinteractome.nia.nih.gov/index.html) was utilized to predict miRNAs that could bind to circPOLR1C and to identify binding sites. Additionally, the starBase database (https://starbase.sysu.edu.cn/) was applied to analyze the target genes with binding sites for miR-361-3p. Ten, the dual-luciferase reporter assay was conducted to detect the specifc sequence binding of transcription factors and their target promoters. Wild-type (WT: circPOLR1C-WT, DUSP2-WT, XIAP-WT, and BCL2-WT) and mutant-type (MUT: circPOLR1C-MUT, DUSP2-MUT, XIAP-MUT, and BCL2-MUT) reporter plasmids were constructed using pmirGLO plasmids (CL414-01, Biomed, China). Tereafter, the abovementioned reporter plasmids were cotransfected with the miR-361-3p mimic (mimic control) or miR-361-3p inhibitor (inhibitor control) into EC cells for 24 h, respectively. Later, 150 μL 1 × passive lysis bufer and 50 μL Luciferase assay reagent II were added to the transfected EC cells in sequence, and after that, the mixed solution was tested twice by Promega GloMax. Specifcally, frefy luciferase activity was detected in the frst part. In the second part, after the former fuorescence reaction was stopped by Stop & Glo, Renilla luciferase activity was determined. Ultimately, the fnal report of plasmid activity was obtained by calculating the frefy/Renilla ratio.

Data
Analysis. Data analysis was conducted using SPSS 22.0 software (USA). Student's two-tailedt-test was applied to compare diferences between two groups, and one-way analysis of variance (ANOVA) was utilized to contrast diferences among multiple groups. Te diference was determined to be statistically signifcant at p < 0.05.

Characteristics of circPOLR1C in EC.
Te structure of circPOLR1C is presented in Figure 1(a). We frst detected the diferential expression of circPOLR1C in EC tissues and cells. Te results showed that the expression of circPOLR1C in EC tissues was signifcantly higher than that in adjacent normal tissues (Figure 1(b), p < 0.001), and circPOLR1C was highly expressed in EC cells compared to that in Het-1A cells (Figure 1(c), p < 0.001). In the study of the correlation between circPOLR1C and the clinical characteristics of patients, we found that circPOLR1C expression was only correlated with diferentiation and tumor invasion ( Table 2, p < 0.01). We further verifed the circular structure of circPOLR1C and analyzed the resistance, stability, and localization of RNase R in circPOLR1C. First, we used random hexamer primers and oligo(dT) 18 primers through qRT-PCR to analyze the expressions of circPOLR1C and linear POLR1C.     Journal of Oncology Subsequently, we discovered that oligo(dT) 18 primers led to obvious downregulation of circPOLR1C but barely afected linear POLR1C expression, compared with random hexamer primers ( Figure 1(d), p < 0.001). Tis result proved that circPOLR1C did not have a poly (A) tail (a characteristic of linear molecules). Experiments on the stability of circ-POLR1C showed that after RNase R treatment, the level of POLR1C markedly decreased, while the level of circPOLR1C remained unchanged (Figure 1(e), p < 0.001). After treatment with actinomycin D, the half-life of circPOLR1C was remarkably longer than that of linear POLR1C (Figure 1(f ),  (Figure 2(a), p < 0.001). In addition, the efects of sicircPOLR1C on the viability, proliferation, migration, invasion, and apoptosis of EC cells were analyzed by the CCK-8 assay, clone formation assay, scratch test, transwell assay, and fow cytometry method. When compared with those in the siNC group, the OD value (Figure 2    recording the tumor volume, we found that overexpressed circPOLR1C increased the tumor volume and promoted the growth of transplanted tumors (Figures 2(k) and 2(l), p < 0.001). Not only that, we observed through HE staining of lung tissues that the vacuole structure notably increased in the circPOLR1C group when compared with that in the vector group, signifying that the lung tissues were obviously damaged after transfection of overexpressed circPOLR1C plasmids (Figure 2(m)). Moreover, the in vivo experimental results showed that overexpressed circPOLR1C promoted the growth of EC xenografts and potentiated lung tissue destruction.

circPOLR1C Could Act as a Sponge for miR-361-3p and
Negatively Regulated miR-361-3p Tat Was Lowly Expressed in EC Tissues. In view of the fact that the reported circRNA could serve as a miRNA sponge, we proved by RIP that circPOLR1C could be prominently enriched by the anti-AGO2 antibody (Figure 3(a), p < 0.001) and could serve as a binding platform for AGO2 and miRNAs. Among the miRNAs predicted by CircInteractome that circPOLR1C might target, we selected the frst 6 miRNAs and analyzed their enrichment by circRIP. As shown in Figure 3(b), we observed that probes specifcally targeting circPOLR1C could enrich circPOLR1C, miR-361-3p, and miR-182, and miR-361-3p was enriched more obviously than miR-182.
To further verify the direct binding of miR-361-3p and circPOLR1C, we compared the diferences in the enrichment of circPOLR1C and circANRIL (negative control) in KYSE-30 and EC9706 cells using biotin-labeled miR-361-3p-WT and biotin-labeled miR-361-3p-MUT. It turned out that biotinlabeled miR-361-3p-WT could enrich circPOLR1C rather than circANRIL (Figure 3(c), p < 0.001). In Figure 3(d), the binding sites of circPOLR1C and miR-361-3p are exhibited. Additionally, through the dual-luciferase reporter assay, we found that the luciferase activity of KYSE-30 and EC9706 cells increased in the circPOLR1C-WT + I group relative to that in the circPOLR1C-WT + IC group, while there was no change in luciferase activity of KYSE-30 and EC9706 cells between the circPOLR1C-MUT + IC group and circPOLR1C-MUT + I group (Figure 3(e), p < 0.01). Besides, the luciferase activity of KYSE-30 and EC9706 cells decreased in the circPOLR1C-WT + M group compared with that in the circPOLR1C-WT + MC group (Figure 3(f), p < 0.01). Tese results proved that miR-361-3p could bind with circPOLR1C. Furthermore, the results of the FISH experiment directly demonstrated the colocalization of circPOLR1C and miR-361-3p (Figure 3(g)).

miR-361-3p Could Target DUSP2, XIAP, and BCL2.
Based on the previous research results, we further verifed the binding of miR-361-3p to DUSP2, XIAP, and BCL2 by the dual-luciferase reporter assay. In cells cotransfected with the wild-type reporter genes and miR-361-3p mimic, the luciferase activity of EC cells was remarkably lower than that of control cells (Figures 5(a)-5(f), p < 0.001). Tese results showed that DUSP2, XIAP, and BCL2 could bind to miR-361-3p. Herein, the in vivo fndings on expression regulation of circPOLR1C and miR-361-3p in transplanted tumor tissues were also consistent with the results of in vitro experiments. As shown in Figures 8(a) and 8(b), the miR-361-3p mimic had no efect on the expression of circPOLR1C, whereas overexpressed circPOLR1C suppressed the mRNA expression of miR-361-3p (p < 0.001). Besides, the expression of Ki-67 in tissues was analyzed by immunohistochemistry. Te results are shown in Figure 8(c). Compared with the NC + MC group, the brown positive expression area in the NC + M group was prominently reduced. After circPOLR1C cotransfection, the positive expression area in the circPOLR1C + M group was explicitly increased compared with that in the NC + M group.

Efects of siDUSP2, siXIAP, and siBCL2 on the Basic
We extracted lung tissues from lung transplantation model mice. Te results of HE staining and lung nodule statistics showed that circPOLR1C overexpression increased the number of lung tissue pathological damage and lung nodules, while the miR-361-3p mimic had the opposite efect (Figures 8(d) and 8(e), p < 0.001). More importantly, circ-POLR1C counteracted the protective efects of the miR-361-3p mimic.

Discussion
EC is a malignant tumor of esophageal origin. Although in the past few decades, great improvement has been made in the treatment of EC after combination of surgical resection, neoadjuvant chemotherapy, radiotherapy, and biological therapy. However, the lack of early diagnosis and treatment still makes the overall survival rate of EC patients low [20]. Te discovery of noncoding RNAs and known biological functions provide a new direction for the prevention and treatment of EC. By dint of its special covalent closed-loop structure, circRNA protects itself from exonucleasemediated degradation, so the expression of circRNA has high stability [8]. In this study, we further detected the expression of circPOLR1C in EC tissues and cells, probed into whether circPOLR1C was circular, and verifed its stability and cell localization. As a result, we proved that circPOLR1C was abundantly expressed in EC cells and mainly located in the cytoplasm of circRNA molecules with high stability. In addition, we demonstrated that circ-POLR1C, which was highly expressed in EC, had a correlation with tumor diferentiation and invasion of EC cells.   Furthermore, knocking down circPOLR1C expression suppressed the proliferation and metastasis yet promoted the apoptosis of EC cells, whereas upregulated circPOLR1C promoted the growth of EC xenografts and metastasis. Notably, these experimental results have not been reported in previous studies.
At present, there are few studies on circRNA intervention in EC. circRNAs, such as circRAD23B [21], ciRS-7 [22], and circRNA_100367 [23], regulate the occurrence and development of EC, whose mechanisms are reported to be mostly based on circRNA acting as an miRNA sponge. With this remarkable feature of circRNA, this study further explored whether circPOLR1C could bind miRNA. Te results showed that circPOLR1C could adsorb miR-361-3p to form a competitive sponge of miR-361-3p, partially reversing the regulatory efect of miR-361-3p on EC.
Abnormal expression of miR-361-3p has been reported to be related to the proliferation and apoptosis of various tumor cells. For instance, the low expression of miR-361-3p attenuates the proliferation and metastasis of non-small-cell lung cancer cells by inhibiting the SH2B1 gene [3]. Besides, miR-361-3p, which is also lowly expressed in retinoblastoma, has been shown to exert a tumor suppressive efect by targeting SHH signaling [24]. Similar to the above research results, the expression of miR-361-3p in EC tissues was found to be decreased in this study, while overexpression of miR-361-3p can suppress the proliferation and metastasis yet promote the apoptosis of EC cells. Tis result suggests that miR-361-3p acts as a tumor suppressor gene in EC. In addition to cancer suppression, miR-361-3p has also been demonstrated to have a cancer-promoting efect. As previously described, miR-361-3p enhances the EMT process of pancreatic ductal adenocarcinoma cells by regulating the DUSP2/ERK axis [25]. Also, it has been reported that the cancer-promoting efect of miR-361-3p with abnormally elevated expression in breast cancer tissue is achieved by inhibiting the E2F1/P73 pathway [26]. Tese fndings show that miR-361-3p can produce diferent stimulation efects in diferent types of cancer. Apart from the regulatory role of miR-361-3p itself, miR-361-3p is the target gene of circRNA, which is also regulated by circRNA and interferes with tumor progression. Chen et al. [27] found that circRNA 100146 binds with miR-361-3p in non-small-cell lung cancer to regulate downstream mRNA. Wang et al. [28] confrmed that as a sponge of miR-361-3p, circ-MYBL2 promotes the growth and metastasis of cervical cancer cells. hsa_-circ_0011385, which also acts as a sponge and negatively regulates miR-361-3p, has a signifcant promoting efect on thyroid cancer [29]. Tese pieces of research evidence strongly support our research results that miR-361-3p, as the target gene of circPOLR1C, is adsorbed by circPOLR1C and negatively regulated. At the same time, this study demonstrated that circPOLR1C regulates EC cell apoptosis and EMT-related genes by adsorbing miR-361-3p.
Cleaved caspase-3, Bax, Bcl-2, DUSP2, and XIAP are the key genes that regulate the process of cell apoptosis. A prior study has shown that the BIR region in the XIAP structure can inhibit caspase activity and is the key to the antiapoptotic efect of XIAP [30]. Besides, overexpression of DUSP2 promotes the apoptosis of ectopic stromal cells [31]. Also, the antiapoptotic gene Bcl-2 can directly adjust the permeability of the mitochondrial membrane to prevent the release of cytochrome C into the cytoplasm, thereby inhibiting the activation of caspase-3 and preventing apoptosis [32]. On the contrary, Bax from the same family as Bcl-2 has a signifcant proapoptotic efect [32]. Notably, in this study, we discovered that overexpression of circ-POLR1C adsorbed miR-361-3p and partially ofset the effects of miR-361-3p on downregulating the levels of antiapoptotic genes (Bcl-2, DUSP2, and XIAP) and upregulating the levels of proapoptotic genes (cleaved caspase-3 and Bax), ultimately inhibiting the apoptosis of EC cells. Tere are accumulating evidence supports that EMTplays an important role in cancer cell migration and invasion, which can be regulated by E-Cadherin, N-Cadherin and Vimentin. For example, Angst et al. [33] proposed that the conversion of E-cadherin to N-cadherin is an important reason for activating the EMT process. In addition, upregulation of vimentin activates the TGF-β pathway and induces the EMT process [34]. Similarly, in this study, we observed that overexpressed circPOLR1C also reversed the regulation of miR-361-3p on E-cadherin, N-cadherin, and vimentin and promoted the migration and invasion of EC cells.

Conclusion
In summary, this study confrmed that circPOLR1C, which is highly expressed in EC tissues and cells, promotes xenograft growth, lung metastasis, proliferation, and metastasis yet inhibits the apoptosis of EC cells. Te specifc mechanism is to regulate the genes related to apoptosis and EMT through adsorption and inhibition of miR-361-3p. To some extent, this study further enriches the basic research arguments of circRNA intervention in EC.

Data Availability
Te analyzed datasets generated during the study are available from the corresponding author on reasonable request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.