YY1 and eIF4A3 are mediators of the cell proliferation, migration and invasion in cholangiocarcinoma promoted by circ-ZNF609 by targeting miR-432-5p to regulate LRRC1

Cholangiocarcinoma is a highly aggressive malignant tumor, and its incidence is increasing all over the world. More and more evidences show that the aberrant expression of circular RNAs play important roles in tumorigenesis and progression. Current studies on the expression and function of circRNAs in cholangiocarcinoma are scarce. In this study, circ-ZNF609 was discovered as a novel circRNA highly expressed in cholangiocarcinoma for the first time. The circ-ZNF609 expression is connected with the advanced TNM stage, lymphatic invasion and survival time in cholangiocarcinoma patients, and can be used as an independent prognostic factor for the patients. Circ-ZNF609 can promote the cholangiocarcinoma cells proliferation, migration and invasion in vitro, it can also catalyze the xenograft growth in vivo. The promoting effect of circ-ZNF609 on cholangiocarcinoma is achieved via oncogene LRRC1 up-regulation through targeting miR-432-5p by endogenous competitive RNA mechanism. In addition, transcription factor YY1 can bind to the promoter of ZNF609 to further facilitate the transcription of circ-ZNF609. RNA binding protein eIF4A3 can bind to the pre-mRNA of circ-ZNF609 which promotes the circ-ZNF609 circular formation. Overall, YY1/eIF4A3/circ-ZNF609/miR-432-5p/LRRC1 have a significant role in progression of cholangiocarcinoma, and circ-ZNF609 is expected to become a novel biomarker for targeted therapy and prognosis evaluation of cholangiocarcinoma.


INTRODUCTION
Among the liver malignancies, the incidence of cholangiocarcinoma (CCA) is second only to hepatocellular carcinoma (HCC), which is a malignant tumor of epithelial origin. The overall incidence of cholangiocarcinoma has increased over the past few decades [1][2][3][4][5][6][7]. Spain and Asia were reported to have the highest incidence of CCA (2.8 to 3.3 per 100,000), while the highest mortality rates for cholangiocarcinoma are found in Asians (1.4 per 100,000) and American Indians (1.3 per 100,000) [8,9]. Currently, early screening and diagnosis of cholangiocarcinoma remains a challenge, as most patients with cholangiocarcinoma are asymptomatic in the early stages, the highly desmoplastic, difficult to obtain anatomic location features, which reduces the sensitivity of cytological and pathological diagnosis methods. As a result, most patients with CCA are in advanced-stage [10]. Consequently, the discovery of new molecular abnormality is essential for targeted therapy and prognostic assessment of this highly aggressive tumor.
Currently, the investigation of circular RNAs (circRNAs) is a hotspot in the field of non-coding RNAs. The circRNAs formed a circular structure by linking free 3'-to 5'-ends. These circRNAs are widely identified in mammalian cells and rarely encode AGING proteins [11,12]. Recent evidences suggest that circRNAs play important roles in the development of several diseases, including osteoarthritis, intervertebral disc degeneration and tumors. It has also been found that circRNAs performed crucial biological functions through numerous mechanisms, including competitive endogenous RNA (ceRNA), RNA binding protein (RBP), gene transcription, and RNA splicing regulatory factors [13][14][15]. Although circRNAs are emerging as a new research direction in many tumors, little is known about their association with cholangiocarcinoma.
A novel circRNA, circ-ZNF609 (ID: hsa_circ_0000615), derived from zinc finger protein 609 (ZNF609) pre-mRNA, located at chr15: 64791491-64792365 [16,17]. Circ-ZNF609 was found to be exceptionally upregulated and carried out diagnostic biomarker roles in breast cancer, renal cancer, colorectal carcinoma, rhabdomyosarcoma, nasopharyngeal carcinoma, gastric cancer and lung adenocarcinoma [18][19][20][21][22][23][24][25]. Nevertheless, the association of circ-ZNF609 with cholangiocarcinoma has not been investigated, so it was necessary to explore the clinical values and biological functions of circ-ZNF609 in cholangiocarcinoma. In this research, qRT-PCR was used to examine the expression of circ-ZNF609 in CCA tissues and adjacent normal tissues. Furthermore, the expression of circ-ZNF609 and the clinicopathological data of the patients were statistically analyzed. In addition, exogenous regulation of circ-ZNF609 expression is then assessed for malignant biological behavior of CCA cells in vitro and in vivo. Mechanically, circ-ZNF609 can directly target miR-432-5p and inhibit the binding of miR-432-5p to LRRC1, leading to increased expression of LRRC1 protein and ultimately contributing to the proliferation, migration and invasion. YY1, a transcription factor, is overexpressed in a diverse range of tumors, including cholangiocarcinoma [26]. The present study confirmed that YY1 binds to the circ-ZNF609 promoter and enhances transcription. Furthermore, the RNA binding protein eIF4A3 has been shown to be an essential component of RNA splicing [27]. The present data indicated that eIF4A3 can conjugate to pre-mRNA, thereby accelerating circ-ZNF609 circular formation. In summary, circular RNA ZNF609 is involved in the cholangiocarcinoma development, suggesting that it may be a prospective target for therapeutic and prognostic evaluation.

RESULTS
Circ-ZNF609 was highly expressed in CCA and was related to the clinicopathological characteristics Circ-ZNF609 was spliced from exon 2 of ZNF609 ( Figure 1A). In Figure 1B, the circ-ZNF609 and ZNF609 expression after using RNase R was detected by qRT-PCR, these results revealed that circ-ZNF609 was more resistant to RNA enzyme digestion than ZNF609. The qRT-PCR data showed that the circ-ZNF609 expression was higher in cholangiocarcinoma tissues than adjacent normal tissues ( Figure 1C). The clinicopathological characteristics of CCA patients with circ-ZNF609 revealed that high expression of circ-ZNF609 was significantly associated with advanced TNM and lymph node invasion ( Figure 1D, 1E). The association of circ-ZNF609 expression with other clinical parameters is shown in Table 1. Additionally, the overall survival of patients with high circ-ZNF609 expression was inferior ( Figure 1F), and circ-ZNF609 was negatively correlated with the survival time of CCA patients ( Figure 1G). In addition, anomalous upregulation of circ-ZNF609, lymph node invasion and advanced TNM stage were independent risk factors for unfavorable prognosis in patients with cholangiocarcinoma ( Table 2). The receiver operating characteristic curve (ROC) was used to evaluate the prognostic efficiency of circ-ZNF609, CA19-9 and circ-ZNF609+CA19-9 expression. The value of area under curve (AUC) was 0.719 (95% CI: 0.626-0.811), 0.772 (95% CI: 0.688-0.855) and 0.786 (95% CI: 0.703-0.869), respectively ( Figure 1H-1J). The results indicated that both circ-ZNF609 and CA19-9 were statistically significant as biomarkers for prognosis evaluation and were more effective predictors when used in combination.

Circ-ZNF609 promoted proliferation, migration and invasion of CCA cells
Circ-ZNF609 was expressed highly in cholangiocarcinoma cells (TFK-1, RBE, CCLP-1, QBC939) compared to normal biliary cell (HIBEC), and follow-up experiments were performed applying CCLP-1 and QBC939 ( Figure 2A). The siRNAs (si-circ-1 and si-circ-2) of circ-ZNF609 both declined the expression of circ-ZNF609 in CCLP-1 and QBC939 cells ( Figure  2B). Then, cell viability assays (CCK-8, EdU, colony formation) displayed that knocking down circ-ZNF609 suppressed the CCLP-1 and QBC939 cells proliferation ( Figure 2C-2E). In Figure 2F, the growth of mice xenografts was restrained by down-regulation of circ-ZNF609 compared with the control group. The results of wound healing assay suggested that transfection with siRNA-circ-ZNF609 evidently inhibited the migratory ability of CCLP-1 and QBC939 cells ( Figure 3A). Transwell assay manifested that the quantity of migrating or invading cells in the lower chamber was considerably reduced after circ-ZNF609 silencing ( Figure 3B). Epithelial-mesenchymal transformation (EMT) performed vital roles in tumor metastasis. In Figure 3C, regulating the circ-ZNF609, the expression of E-cadherin, N-cadherin and vimentin was remarkably AGING influenced. It can be speculated that circ-ZNF609, as a pro-cancer circRNA, participated in the CCA proliferation, migration, invasion and EMT process.

Circ-ZNF609 could sponge miR-432-5p in CCA
The results of subcellular fractionation assay demonstrated that circ-ZNF609 was predominantly distributed in the cytoplasm ( Figure 4A), which indicated that circ-ZNF609 has been involved in post-transcriptional regulation. Extensive studies had confirmed that circRNA can function as a ceRNA to indirectly enhance the expression of downstream target mRNAs by competing with mRNAs to bind miRNAs. On the basis of predictions from bioinformatics databases (targetscan, circinteractome and mirMap)  and qRT-PCR detection, miR-432-5p was chosen as the downstream miRNA of circ-ZNF609 ( Figure 4B, 4C). MiR-432-5p had low expression in the cholangiocarcinoma tissues and cells, and was negatively correlated with circ-ZNF609 expression ( Figure 4D-4F). In Figure 4G, the efficiencies of miR-432-5p mimics and inhibitor identified after transfection into tumor cells were shown. After constructing luciferase reporter plasmid, including wild-type or mutant-type binding sites of circ-ZNF609 and the coprecipitation of RNA with anti-Ago2, the results of dual luciferase reporter assay and RIP assay showed that circ-ZNF609 could directly sponge miR-432-5p ( Figure  4H-4J). Up-regulation of miR-432-5p could prohibit the proliferation, migration and invasion ability of CCA cells, and down-regulation of miR-432-5p promoted the proliferation, migration and invasion ability of CCA cells ( Figure 4K-4M).

YY1 and eIF4A3 could promoted circ-ZNF609 expression
For the purpose of exploring the circ-ZNF609 potential upstream regulatory mechanism, screening was performed using the JASPAR database (http://jaspar.genereg.net). As shown in Figure 7A, YY1 is a transcription factor that may bind to circ-ZNF609 promoter, where E1, E2 and E3 are three transcription factor binding sites. To validate this prediction, qRT-PCR results revealed that exogenous regulation of YY1 could influence the expression of circ-ZNF609 directly ( Figure 7B, 7C). In addition, the ChIP and luciferase reporter assay results indicated that YY1 was predominantly enriched and bound to the E3 fragments ( Figure 7D-7F), which data showed  AGING that YY1 could facilitate the transcription of circ-ZNF609 by binding to the E3 region of the ZNF609 promoter. Furthermore, predicted data on circinteractome (https://circinteractome.nia.nih.gov/index.html) indicated that eIF4A3 might combine with circ-ZNF609 pre-mRNA, as shown in Figure 7G. It was previously suggested that the RNA binding protein eIF4A3 was a core component of the exon junction complex and performed RNA splicing functions. The RNA splicing functions important roles in the process of linear RNA AGING forming circular RNA. The binding of eIF4A3 to circ-ZNF609 pre-mRNA was subsequently substantiated by the RIP experimental results ( Figure 7H). Moreover, upand down-regulation of eIF4A3 could respectively enhance and impair the expression of circ-ZNF609 ( Figure 7I, 7J). The above results suggested that YY1 could bind to the promoter to further facilitate the transcription of circ-ZNF609 and eIF4A3 could contribute to the circ-ZNF609 circular formation by binding to circ-ZNF609 pre-mRNA ( Figure 7K).

DISCUSSION
Cholangiocarcinoma was an aggressive malignancy, the prevalence of cholangiocarcinoma in Asian countries had increased in the last decades. Survival and prognosis of CCA patients were very unsatisfactory due to difficulties in early diagnosis and monotherapy modalities. Consequently, the discovery and application of new biomarkers for cholangiocarcinoma were the important measure to reverse this situation. CircRNAs were class of back-spliced ncRNAs that form loops and rarely encode proteins. As research progresses, circRNA was no longer considered as by-product of transcription and carried out multiple roles in the normal growth and disease development. For example, circRNA interacted with Pol II transcription complex as a transcriptional regulator, sponged miRNA as ceRNA or recruited protein to activate or repress genes expression [28][29][30].
Circ-ZNF609 was up-regulated in breast cancer and promoted cell proliferation, migration and invasion through sponging miR-145-5p [18]. Circ-ZNF609 promoted the development of renal carcinoma by the miR-138-5p/FoxP4 axis [19]. The correlation between circ-ZNF609 and cholangiocarcinoma, on the other hand, had not been investigated. In the present study, circ-ZNF609 expression was found to be significantly higher in cholangiocarcinoma tissues and cells than in normal adjacent tissues and cells for the first time. And the dysregulation of circ-ZNF609 was intimately associated with TNM stage, lymph node invasion and prognosis of patients with cholangiocarcinoma. The results of cell viability assay (CCK-8, EdU, colony formation) indicated that down-regulation of circ-ZNF609 inhibited the cholangiocarcinoma cells proliferation. Wound healing and transwell assays showed that silencing circ-ZNF609 impaired the migration and invasion efficiency of tumor cells. The results of western blot demonstrated that knockdown of circ-ZNF609 affected E-cadherin, N-cadherin, vimentin, which suggested that the process of CCA epithelial mesenchymal transformation was restricted. The results of tumor xenograft assay also suggested that exogenous silencing of circ-ZNF609 inhibited tumor growth. This present study confirmed that circ-ZNF609 had a carcinogenic effect in cholangiocarcinoma.
Extensive studies had established that circRNAs could sponge miRNAs through endogenous competitive RNA mechanism and affect the expression of downstream target genes and thus promoted or inhibited occurrence and development of many diseases, including tumors. To illustrate, circGSE1 exerted oncogenic effects in cervical cancer through sponging miR-138-5p [31]. In this study, it was confirmed that circ-ZNF609 could sponge miR-432-5p directly through ceRNA mechanism, and miR-432-5p could further regulate the proliferation, migration and invasion of CCA. Similarly, it was predicted and evidenced that mRNA LRRC1 could be targeted by miR-432-5p in 3'-UTR regions. High expression of LRRC1 promoted malignant biological behaviors of tumor cells in cholangiocarcinoma. Furthermore, the rescue experiments results further demonstrated that circ-ZNF609 affected the proliferation, migration and invasion of CCA by regulating the miR-432-5p/LRRC1 axis.
For the sake of investigating the upstream regulatory mechanisms of circ-ZNF609 in the development of CCA, the transcription factor YY1 with high binding score to ZNF609 promoter region was predicted by using bioinformatics databases. Also, YY1 had been shown to be involved in various tumors progression and associated with adverse clinical prognosis in tumor patients, for instance, YY1 stimulated LINC00667 transcription that enhanced the proliferation of cholangiocarcinoma [26]. The results of present research indicated that elimination of YY1 significantly limited the expression of circ-ZNF609, whereas overexpression of YY1 promoted the circ-ZNF609 expression, which was significantly enriched in the E3 region of circ-ZNF609 promoter. In addition, eIF4A3, as an RNA-binding protein, was a core component of exon junction complex and was recognized as a vital regulator of post-transcriptional regulation, such as mRNA splicing, transport and translation [27]. RNA splicing was one of the essential causes of the circular RNA formation. In this study, it was confirmed by in vitro experiments that eIF4A3 could bind to the circ-ZNF609 pre-mRNA to its circular formation, and circ-ZNF609 expression was up-or down-regulated after promoting and inhibiting eIF4A3 expression.
Collectively, this study demonstrated for the first time that the circular RNA ZNF609 was up-regulated in cholangiocarcinoma, its expression was intimately associated with the clinicopathological characteristics and prognosis of patients, and could be an independent prognostic factor for CCA patients. It was further experimentally proved that circ-ZNF609 could facilitate the tumor cells proliferation, migration and invasion in cholangiocarcinoma. Mechanically, circ-ZNF609 sponged miR-432-5p through the ceRNA mechanism, which further strengthened the expression of LRRC1, the downstream oncogene of miR-432-5p. In addition, the upstream transcription factor YY1 and RNA binding protein eIF4A3 could facilitate circ-ZNF609 expression by promoting the gene transcription and circRNA circular formation, respectively. These results indicated that circ-ZNF609 might be the novel biomarker of targeted therapy and prognosis evaluation of cholangiocarcinoma.

Patients and clinical specimens
61 pairs of CCA tissues and adjacent normal tissues from patients with CCA treated at The 2nd Affiliated Hospital of Harbin Medical University were obtained after informed consent of the participants and approval of the Ethics Committee of The 2nd Affiliated Hospital of Harbin Medical University. The obtained tissues were rapidly frozen in liquid nitrogen and stored at -80° C for the following studies. All patients did not receive radiotherapy or chemotherapy treatment before operation.

Cell culture
Normal biliary cell line (HIBEC) and the CCA cells (CCLP-1, TFK-1 and QBC939) were stored in our laboratory. RBE was purchased from Type Culture of Chinese Academy of Sciences (Shanghai, China). The above cholangiocarcinoma cell lines and normal biliary cell were cultured in DMEM (Gibco, Grand Island, USA) with 10% FBS (Invitrogen, Carlsbad, USA) and 1% double antibody (Sigma, St. Louis, USA) in an incubator at 37° C with 5% CO 2 . It was passaged every 72 hours, and all cells were cultured for less than 6 months.

Quantitative reverse transcription PCR
Total RNA was extracted from cells and frozen tissue samples using TRIzol reagent (Invitrogen) based on the manufacturer's protocol and complementary DNA (cDNA) was synthesized using the Transcriptor First Strand cDNA Synthesis Kit (Roche, Penzberg, Germany). FastStart Universal SYBR Green aster (Roche) and C1000 Thermal Cycler (Bio-Rad, Hercules, USA) were used for 10 μl volume qRT-PCR. GAPDH and U6 were used as controls. The 2 -ΔΔCt method was used to calculate the transcription level of target genes. RNase R (Epicentre, Madison, USA) was used to degrade linear mRNA. All the primer sequences were listed in Supplementary Table 1.

Cell viability assay
The cells in logarithmic growth phase were inoculated in 96-well plates at a density of 5×10 4 cells per well. The cells were cultured for 0-72 h. Every 24 h, 10 ul of CCK-8 reagent (Dojindo, Kumamoto, Japan) was inserted into the petri dish. After incubation for 2 h at 37° C in a humidified incubator, the optical density (OD) at 450nm wavelength was detected by microplate reader (Tecan, Männedorf, Switzerland). The cell proliferation capacity was also detected by 5-ethynyl-2'-deoxyuridine (EdU) assay. The treated cells were inoculated in 96well plates. After incubation for 24 h, 100 μl EdU solution (Ribobio, Guangzhou, China) was added to each well and stored for 2 h. The treated cells were fixed with paraformaldehyde and then stained with Apollo 567 and Hoechst 33342 in the dark. The EdU-positive cells were detected using a fluorescence microscope (Leica, Wetzlar, Germany). After treatment, cells were diluted and inoculated in 6-well plates at a density of 1×10 3 cells per well. The treated cells were immobilized with paraformaldehyde and stained with crystal violet (Beyotime, Beijing, China) for 30 minutes. When colonies were visible the transfected cells were preserved in DMEM containing 10% FBS, which were incubated at 37° C with 5% CO 2 for 2 weeks. The colonies were then fixed with paraformaldehyde and stained with crystal violet (Beyotime, Beijing, China) for 30 minutes. The plates were photographed, and a large number of stained colonies were counted.

Migration and invasion assays
The CCA cells migration ability was measured by wound healing assay. When the fusion degree of CCA cells plated in 6-well plates reached 80-90%, a straightline wound was scratched on the surface with ruler and pipette. The percentage of cell healing was measured by taking photographs and calculating the average distance of wound cells at 0 h and 48 h. Transwell assay was AGING performed with transwell chambers (Corning, NY, USA). Transfected cells were placed on the upper chamber at a density of 1×10 5 cells per well, coated without Matrigel (BD, San Jose, USA) and contained with DMEM (serum-free). And DMEM with 10% FBS was added to the lower chamber. After 24 h inoculation, upper cells were gently removed, and the migrating cells were immobilized, stained with paraformaldehyde and crystal violet respectively, and then the cells were photographed with an inverted microscope (Leica). The invasion assay resembled the migration assay except that the chambers were precoated with Matrigel. The migration and invasion abilities were counted through respectively counting the migrated and invaded cells.

Western blot assay
The transfected cells were treated with RIPA lysis buffer (Beyotime) and the total proteins were extracted. Protein concentrations were tested with a bicinchoninic acid kit (Beyotime). The protein lysates were split by 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to polyvinylidene fluoride (PVDF) membrane (Millipore, Billerica, USA). The PVDF membranes were then blocked with 5% skimmed milk powder for one hour, and then incubated with the primary antibody (overnight) and the secondary antibody (1.5 h) in turn. Proteins on the membrane were visualized by BeyoECL kit (Beyotime). GAPDH was applied as an endogenous control and all antibodies were purchased from Abcam (Cambridge, USA).

Subcellular fractionation assay
Cytoplasmic and nuclear components were separated by Paris kit (Life Technologies, Carlsbad, USA). After lysis of gelatin cells with cell separation buffer, the upper layer of the gelatin cells was centrifuged to obtain the cytoplasmic contents, and the lower layer of the precipitated part was treated with cell division buffer to extract the nuclear contents. The RNA extracted from cytoplasm and nucleus was detected by qRT-PCR. Cytoplasmic control and nuclear control were expressed by GAPDH and U6, respectively.

RNA immunoprecipitation (RIP) assay
RIP assays were performed using the EZ-Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore) in accordance with the manufacturer's instructions. The RIP lysis buffer containing protease inhibitor cocktail and RNase inhibitor was made to lyse and then collect cells in different groups. Cells were incubated with RIP buffer containing magnetic beads coated with Ago2 antibodies (Millipore). After incubation at 4° C for 2 h, the coprecipitated RNA was eluted and the extracted RNAs was analyzed by qRT-PCR. IgG group was the negative control group.

Chromatin immunoprecipitation (ChIP) assay
The interaction between transcription factor YY1 and circ-ZNF609 promoter was verified by EZ-ChIP kit (Millipore) and producer's recommendations. The cross-linking reaction was performed with 1% formaldehyde, sonication converted the DNA-protein complex into small chromatin fragments. Then IgG and anti-YY1 antibodies (Abcam) were added to form immunoprecipitation. After the cross-link was washed and reversed, qRT-PCR was used to detect the relative enrichment of circ-ZNF609 fragments containing speculated YY1 binding sites in the DNA promoter region.

Tumor xenograft assay
BALB/c nude mice were purchased from Vital River (Beijing, China) and housed at constant humidity (45-50%) and temperature (25-27° C). 6-week-old nude mice were in situ injected with CCA cells in logarithmic phase. The cell suspension 0.2 ml was injected subcutaneously into the groin to construct the model of xenograft of cholangiocarcinoma. 4 weeks later, the mice were sacrificed and the weight of the tumor was weighed by balance. The study was authorized by the Ethics Committee of the 2nd Affiliated Hospital of Harbin Medical University.

Statistical analysis
GraphPad Prism 8.4 software (GraphPad Software, La Jolla, CA) and SPSS 21.0 software (IBM SPSS, Armonk, NY) were utilized in this study. Measurement data were expressed as mean ± standard deviation (SD) and based on at least three independent experiments. Comparison between cholangiocarcinoma tissues and adjacent normal tissues were performed by paired t-test. Data were compared between two groups using an unpaired t-test, and between multiple groups using oneway analysis of variance (ANOVA). Pearson correlation was used to analyze the relationship between the two indicators. The survival rate of the patients were calculated using the Kaplan-Meier and Cox regression analysis. The diagnostic value of circ-ZNF609 expression in cholangiocarcinoma was assessed by receiver operating characteristic (ROC). P value< 0.05 was indicative of statistically significant difference.

AUTHOR CONTRIBUTIONS
Canghai Guang, Lang Liu and Yuqiao Zhao designed and executed the experiments. Xianhe Zhang and Guanglin Liu collected the data. Xiangyu Zhong and Xingming Jiang provided the experimental resources. Haicun Wang and Xin Gao wrote and modified the manuscript. All authors contributed to the article and approved the submitted version.