Circular RNA Circ-0006302 Promotes the Growth, Migration, and Invasion of Cholangiocarcinoma Cells by Regulating the Mir-1299/PD-L1 Axis as a Competing Endogenous RNA


 Background: Cholangiocarcinoma (CCA) is an aggressive malignancy with a poor prognosis, with no effective therapy other than surgical resection. Circular RNAs (circRNAs) serve as a brand-new class of transcription products among abundant cancer processes. Nevertheless, the mechanisms account for their modification in CCA remain unknown. Methods: First, microarray sequencing was applied to detect the difference of circRNAs expression between CCA and corresponding non-tumor tissues. We utilized qRT-PCR to measure circ-0006302 levels in CCA cells and specimens. Gain/loss of-function assays and animal model of CCA were performed for the purpose of revealing the functions of circ-0006302 on the invasion, migration, and proliferation of CCA. We performed dual luciferase reporter, RNA-FISH and rescue assays for clarifying the mechanism behind. Results: In CCA tissues and cell lines circ-0006302 was highly expressed relatively. In vitro, overexpression of circ-0006302 intensifies the epithelial-to-mesenchymal transition (EMT) and the invasion, migration, and growth of CCA cells; and intensifies the growth as well as metastasis of tumors in a CCA mouse model. Furthermore, it was elucidated that circ-0006302 sponged miR-1299 to upregulate PD‐L1 expression. Through the process above, circ-0006302 binds to miR-1299 and emancipates PD-L1, facilitating the invasion, migration, and proliferation in CCA cells. Momentously, the results obtained revealed that circ-0006302 silencing elevated the expression of interferon (IFN)‐γ, and interleukin (IL)‐4 but diminished the IL-10 expression, while these effects could be reversed by miR-1299 inhibitor.Conclusion: circ-0006302 silence blocked the CCA progression via intensifying miR‐1299‐targeted downregulation of PD‐L1. Our conclusion provides novel therapeutic tactics for treating this fatal disease.

PD-1, a receptor of the Ig superfamily which can interact with the speci c ligands (PD-L), is capable of negatively regulating T cell antigen receptor signaling, consequently playing a role in maintaining self-tolerance [15]. For the purpose of escaping from the antitumor immune response, tumor cells alter the T cell activity, which contributes to the survival of tumor cells [16,17]. It is worth noting that tumor cells can interact with CD8 + T cells and induce their apoptosis, thus promoting tumor growth and metastasis [18,19]. Moreover, plenty of studies have offered convincing evidence that neutralizing PD-L1 or PD-1 to block the PD-1/PD-L1 was able to activate CD8 + T cells and attenuate the immune evasion of tumor cells, which indicates that antibody against PD-1 and PD-L1 may be a potential treatment for cancers [20,21]. In CCA patients, PD-L1 overexpression was reported, and the expression of PD-L1 was a separate prognostic element of CCA patients [22,23].
In this study, we found circ-0006302 displayed high expression in CCA tissues. Also, we identi ed a new mechanism by which circ-0006302 sponged miR-1299 to upregulate PD-L1 expression, so that to intensify the migration, proliferation, and invasion of CCA.

Patients and clinical sample collection
From 2017 to 2019, we harvested CCA and paired non-tumor tissue samples from 60 patients without receiving chemotherapy or radiotherapy before the study. After retrieval we placed samples in liquid nitrogen instantly. We provided informed consent for the involved, and our research gained approval by the Ethics Committee of Lishui Hospital of Zhejiang University.

Cells and cell culture
We purchased human CCA cell lines (TFK-1, SNU-869, RBE, HuCCT1 and HuH28), normal human intrahepatic biliary cell (HIBEC), and HEK293T cells from the Chinese Academy of Scien ces (Shanghai, China). We maintained cells in a humidi ed atmosphere at 37 °C and 5% CO 2 in RPMI-1640 (Gibco, Grand Island, NY, USA), which contains 10% fetal bovine serum (Invitrogen Life Technologies, Carlsbad, CA, USA). We cultivated HEK293T cells in 10% FBS DMEM/high glucose medium. All cell lines were passaged for less than 6 months.

circRNAs Microarray and Data Analysis
With Arraystar Human circRNA Array V2 we analyzed three pairs of CCA and non-tumor tissue samples. With the NanoDrop ND-1000 we quanti ed total RNA from each sample. In conformity with the Arraystar's standard protocols we conducted the microarray hybridization and sample preparation. In brief, we digested total RNAs with Rnase R (Epicentre Technologies, Madison, WI, USA) for removing linear RNAs and enriching circular RNAs.
Afterward, with a random priming method (Arraystar Super RNA Labeling Kit; Arraystar) we ampli ed the enriched circular RNAs and transcribed them into uorescent cRNA. We hybridized the labeled cRNAs onto the Arraystar Human circRNA Array V2 (8x15K, Arraystar). With the Agilent Scanner G2505C we scanned the arrays after washing the slides. For analyzing obtained array images, we used Agilent Feature Extraction software (version 11.0.1.1). We performed subsequent data processing and quantile normalization with the R software limma package. We performed hierarchical clustering for demonstrating the difference of circRNAs expression pattern in the midst of the samples.

Cell transfection
We purchased pLVX-EF1α, PLKO.1-puro and plasmid vectors from BioVector NTCC Inc., Guangzhou, China. A shRNA sequence that targeted circ-0006302 and a negative shRNA control sequence were designed and synthesized, and we cloned them into PLKO.1-puro. The sequences which encode circ-0006302, PD-L1, and a negative control were also synthesized, and cloned into pLVX-EF1α. We purchased the miR-1299 mimics and miR-1299 inhibitor from RIBOBIO, Guangzhou, China. We cultured cells for 24 hours, and transfected them with plasmids via Lipofectamine 3000 Transfection Reagent (Invitrogen, Carlsbad, CA, USA), per manufacturer's instructions. We harvested cells to conduct RNA extraction after 48 hours. We performed the experiments in triplicate.

RNA extraction and Quantitative real-time polymerase chain reaction (qRT-PCR)
Centrifugation was performed at 4 °C. From the upper aqueous phase, we obtained the isopropanol precipitates at room temperature (20-25 ºC), then based on the TRIzol total RNA manual (Invitrogen, Carlsbad, CA, USA) we rinsed and dried them. Consequently, DEPC-treated water was added and for each sample we calculated the RNA concentration. At -80 °C we stored the RNA. Via the OneStep PrimeScript® miRNA cDNA Synthesis Kit (Takara) we generated cDNA, per manufacturer's instructions. SYBR Green I uorescence method was applied to conduct RT and we carried out PCR detection. We utilized primer sequences listed in Table 1  RNase R digestion 3 units of RNase R (Epicentre Biotechnologies) per 1 μg circ-0006302 were added and at 37 °C we incubated the mixture for 15 minutes. Consequently, qRT-PCR was performed to assess the levels of GAPDH and circ-0006302.
Cell counting kit-8 assay We utilized cell Counting Kit-8 (Beyotime Inst Biotech, China) to assess cell proliferation. In brief, into 96-well plates we seeded 5 × 10 3 cells/well. The cells were transfected with appropriate plasmids and controls after 24 hours. To assess proliferation a microplate reader (Bio-Rad, Hercules, CA, USA) (absorbance wavelength 450 nm) was used. We performed the experiments in triplicate.

Colony formation assay
In 6-well plates we seeded cells and in media containing 10% FBS we cultured them. The cells were xed with methanol after 14 days, and with 0.1% crystal violet (Sigma-Aldrich) we stained them, then we counted clones.
We performed the experiments in triplicate.

Ethynyl-2-deoxyuridine (EdU) incorporation assay
Ethynyl-2-deoxyuridine incorporation assay (EdU Apollo DNA in vitro kit, RIBOBIO, Guangzhou, China) was utilized to identify cell proliferation, per manufacturer's instructions. In brief, we added 100 μl of 50 μM EdU/well after transfecting cells with plasmids, and incubated the cells at 37 °C for 2 hours. Fluorescence microscopy was used for determining proliferation. We performed the experiments in triplicate.

Wound healing and transwell invasion assays
We cultured the different groups of CCA cells (1 × 10 6 cells/well) up to 90% con uency and with a sterile pipette tip (100 μl) we scratched the monolayer of cells in individual wells. We cultured the cells continually for 24 h and imaged them at 0 and 24 h post scratching with a digital camera (Leica, Heerburg, Germany). By the distance of migration into the denuded area we assessed the extent of wound healing.
Transwell TM chambers coated with Matrigel were utilized. We resuspended cells (5 × 10 4 /L) in 200 μL serum-free medium and seeded them into the upper chambers. To the lower chambers 600 μL complete medium were added. We wiped cells that remained on the upper lter surface away after 48 hours at 37 °C. With formaldehyde we xed cells migrated to the bottom of the lter and with crystal violet we stained them. Under an Olympus uorescence microscope (Tokyo, Japan) we counted the cella.

RNA uorescent in situ hybridization (FISH)
In short, we placed the cover glasses on the bottom of a 24-well plate, and cultured the cells with 6 × 10 4 cells per well. We waited until the cell con uence reached about 60%-70%, then we xed them in 4% paraformaldehyde and with the precooled permeation reagent (1 ml/well) we permeabilized them. With the prehybridization solution (20 μl/well) we sealed the cells and hybridized them with the Stellaris RNA FISH solution (Biosearch Technologies, Petaluma, CA), which contains the probe for circ-0006302. Then we rinsed the cells by lotion I, lotion II, lotion III, and 1× PBS ultimately. Consequently, we stained the cells for 10 min with FAM uorescent dye liquor. Finally, with a mounting agent (like nail polish) we xed the cover glasses on the glass slides, and under a uorescence microscope (Olympus) we viewed them for uorescence detection.

Dual-luciferase reporter assays
We established and ampli ed circ-0006302 wild type or PD-L1 wild type, which contained mutant sequences with target sites deletion or designed miR-1299 binding sites before they were cloned into the pRL-TK plasmid (Promega) vector. After that, HEK293T cells were seeded (2×10 4 cells/well) into 96-well plates. We co-transfected the cells with the luciferase plasmids (0.1 μg/well) and miR-1299 mimics or controls. We assessed re y and renilla luciferase activity using the Dual-Luciferase Reporter Assay System (Promega) after two days as mentioned above.

Results
Circ-0006302 expression is up-regulated in CCA tissues and cell lines Via microarray sequencing we detected the circRNAs expression in CCA tissues and corresponding non-tumor tissues, which were harvested from three CCA patients. After analyzing microarray results, we found circ-0006302 showed high expression in the CCA tissues. Based on the criteria of log2 fold change > 2 or < -2 and P value < 0.05, we identi ed different expression of circRNAs. In the heatmap we showed the differentially expressed circRNAs including circ-0006302 (Fig. 1A). Then, circ-0006302 was veri ed by Sanger sequencing and RNase R treatment. The circ-0006302 sequence ampli ed by the primer was con rmed to be identical to the sequence in Circbase, which was derived by Sanger sequencing (Fig. 1B). We then validated the circular structure of circ-0006302, which presented more stable resistant to RNase R (Fig. 1C). Compared with the corresponding non-tumor controls, circ-0006302 expression was remarkably higher in tumor tissues (Fig. 1D). Expression of circ-0006302 was remarkably higher in the CCA cell lines (TFK-1, RBE, SNU-869, HuH28 and HuCCT1) than in HIBEC, which was consistent with data above (Fig. 1E).
Circ-0006302 promotes the growth of CCA cells Whether circ-0006302 regulates the growth of CCA cells was then investigated. Initially, HuCCT1 cell line was cocultured with sh-circ-0006302, and RBE cell line was co-cultured with circ-0006302 overexpression vector, respectively ( Fig. 2A). The proliferation of CCA cells curves was identi ed via CCK-8 assays, EdU assay, and colony-formation assays. Cell growth and colony-forming ability were inhibited in HuCCT1 cells by circ-0006302 knockdown (Fig. 2B-D). The proliferation and colony-forming ability of RBE cells were intensi ed by circ-0006302 overexpression (Fig. 2B-D). With these data we suggest that in vitro circ-0006302 intensi es the growth of CCA cells.
Circ-0006302 stimulates cell migration, invasion, and the EMT of CCA cells Whether circ-0006302 impacted on the invasion and migration of CCA cells was evaluated. Wound healing assay showed that knockdown of circ-0006302 inhibited the migration of HuCCT1 cells and overexpression of circ-0006302 enhanced the migration of RBE cells (Fig. 3A). Via transwell invasion assays we observed a similar pattern of data in HuCCT1 and RBE cells (Fig. 3B).
Whether circ-0006302 modulated the EMT of CCA cells was further investigated. Western blotting and qRT-PCR were utilized to assess the expression of involved mRNAs and proteins in the EMT. In HuCCT1 cells, circ-0006302 knockout diminished vimentin and snail expression yet stimulated E-cadherin expression. While in RBE cells circ-0006302 overexpression stimulated vimentin and snail expression yet diminished E-cadherin expression (Fig. 3C, D). With these results we suggested that in CCA cells circ-0006302 intensi es cell invasion, migration, and the EMT.
Circ-0006302 promotes PD-L1 expression by binding to miR-1299 Potential molecular roles and biological function signi cantly affected by the subcellular localization of circRNAs. Initially, we utilized RNA-FISH to explore the subcellular localization of circ-0006302. It was showed that majority of the positives located in the cytoplasm, while in the nucleus minority remained (Fig. 4A). Previous studies showed that PD-L1 drives metastasis and initiation of CCA [24][25][26]. We performed qRT-PCR for investigating the mechanisms under circ-0006302-mediated biological processes. It was showed that there was a positive correlation between circ-0006302 expression levels and PD-L1 expression levels in CCA (Fig. 4B). Compared with HIBEC cell line, PD-L1 was increased in CCA cell lines (Fig. 4C). Also, knockdown of circ-0006302 decreased PD-L1 expression in CCA cells (Fig. 4D). We speculated that circ-0006302 may regulate PD-L1 expression through competing endogenouse RNA mechanism. We then searched online bioinformatics database (TargetScan and CircInteractome) and performed bio-information analysis, which predicted that both miR-1299 and miR-876-3p have common putative binding sites with circ-0006302 and PD-L1 (Fig. 4E). Then, the level of two miRNAs were tested in CCA and HIBEC cell line, compared with HIBEC cell line only miR-1299 showed low expression level in CCA cell lines (Fig. 4F). As expected, compared with paired non-tumor tissue miR-1299 expression displayed remarkably lower levels in CCA tissues (Fig. 4G). Moreover, as bioinformatics analysis showed in Fig. 4H, the seed sequence of miR-1299 and the 3'UTR sequence of PD-L1 was complementary. According to dual-luciferase reporter assay results, circ-0006302-Wt and miR-1299 mimics co-transfection remarkably attenuated luciferase activity, while circ-0006302-Mut and miR-1299 mimics co-transfection couldn't affect luciferase activity. Moreover, the similar results were observed in the luciferase activity of cells transfected with PD-L1-Wt (Fig. 4H). Whether circ-0006302 modulated the expression of PD-L1 via miR-1299-dependent manner in CCA cells was further determined. Results showed that miR-1299 overexpression diminished PD-L1 expression in CCA cells (Fig. 4I). Besides, the PD-L1 expression inhibition induced by silencing circ-0006302 in CCA cells could be reversed by miR-1299 knockdown (Fig. 4J). Overexpression of miR-1299 reversed the increase in PD-L1 expression of CCA cells induced by up-regulated of circ-0006302 (Fig. 4K). The results indicated that in CCA cells circ-0006302 positively regulates PD-L1 expression through sponging miR-1299.
Circ-0006302 knockdown reverses the PD-L1 overexpressioninduced increase in proliferation and invasion of CCA cells.
Hence, we explored if circ-0006302 impacted on the invasion and proliferation of CCA cells via a PD-L1dependent mechanism. HuCCT1 cells co-transfected with PD-L1 overexpression vectors (with or without sh-circ-0006302). QRT-PCR and western blotting assays showed that PD-L1 overexpression vectors signi cantly increased the RNA and protein level of PD-L1, while, the increase of PD-L1 can be partially reversed by cotransfection with sh-circ-0006302 (Fig. 5A). Besides, it was found that circ-0006302 silencing remarkably reversed the improvement of growth and colony-forming ability induced by PD-L1 overexpression in HuCCT1 cells (Fig. 5B, C). Similarly, circ-0006302 silencing signi cantly reversed the improvement of invasive ability induced by PD-L1 overexpression in HuCCT1 cells (Fig. 5D). Therefore, we indicate that in a PD-L1-dependent manner circ-0006302 impacts on the invasion, migration, and growth of CCA cells.
Overexpression of circ-0006302 promotes the growth and metastasis of tumors in a mouse model of CCA It was found that upregulation of circ-0006302 promoted tumor growth in a CCA mouse model (Fig. 6A-C). In Fig.  6A we displayed tumors dissected from the mice. The tumors in the circ-0006302 group were bigger (Fig. 6B) and heavier (Fig. 6C) compared with the vector treatment group. Besides, the expression of Ki-67 was lower in vector group than in circ-0006302 group (Fig. 6D). In Fig. 6E pulmonary metastasis slices are showed. With these results we indicate circ-0006302 intensi es the growth and metastasis of CCA tumors in vivo.
Circ-0006302 competitively binds to miR-1299 to up-regulate PD-L1, thereby inhibiting the activation of CD8 + T cells The sh-NC, sh-circ-0006302, vector, circ-0006302 overexpression vector (circ-0006302), inhibitor NC (inh-NC), and miR-1299 inhibitor (miR-1299 inh) were delivered into the HuCCT1 cells respectively, then HuCCT1 cells and CD8 + T cells were cocultured. We utilized ELISA to determine the levels of antitumor cytokines IFN-γ and IL-4 as well as pro-tumor cytokine IL-10 (Fig. 7A). We found that after the delivery of sh-NC, inh-NC, vector or miR-1299 inh + sh-circ-0006302 signi cant changes didn't emerge in the levels of IFN-γ, IL-10, and IL-4 in the cocultured CD8 + T and HUCCT1 cells. However, sh-circ-0006302 transfection exhibited a decreased level of IL-10 yet increased levels of IL-4 and IFN-γ, whereas circ-0006302 transfection led to a higher IL-10 level yet lower levels of IL-4 and IFN-γ. Via ow cytometry we identi ed no obvious change emerged in the apoptosis of CD8 + T cells transfected with inh-NC, vector, sh-NC or miR-1299 inh + sh-circ-0006302, however, the apoptosis of CD8 + T cells was notablely inhibited by transfection of sh-circ-0006302 whereas intensi ed by circ-0006302 overexpression (Fig.  7B). After the transfection of inh-NC, vector, sh-NC or miR-1299 inh + sh-circ-0006302, there is no noticeable change in the expression of PD-L1 and PD-1. Nonetheless, the transfection of sh-circ-0006302 could signi cantly reduce the expression of PD-1 and PD-L1 yet circ-0006302 overexpression could markedly increase the expression of PD-1 and PD-L1 (Fig. 7C). In a whole, circ-0006302 sponges miR-1299 and upregulates the expression of PD-L1 to suppresse CD8 + T cell activation.

Discussion
Recently, circRNAs have emerged as crucial factors and in uential regulators of gene expression in multiple cancers [27]. For example, circHIPK3 acts as a tumor-suppressive function in ovarian cancer [28]. Via activating JAK2/STAT3 signaling pathway, hepatocellular carcinoma progression was driven by circSOD2 induced epigenetic alteration [29] Circ-0000284 modulates the biological functions of peripheral cells via cellular communication and arouses malignant phenotype of CCA cells [30]. In this study, we observed circ-0006302 expression levels in CCA tissues are distinctly higher than that in non-tumorous tissues. Circ-0006302 knockdown attenuated the growth, invasion, and migration of CCA cells. In CCA, a positive correlation between the expression of circ-0006302 and PD-L1 was identi ed, and circ-0006302 knockdown resulted in a synchronous drop in PD-L1 expression. Besides, both PD-L1 and circ-0006302 were found to have the same putative binding sites in their 3'-UTRs, which was for the seed sequence of miR-1299.
The reduced expression and tumor-suppressive role of miR-1299 have been found in prostate cancer [31], ovarian cancer [32] and gastric cancer [33]. Meanwhile, through EGFR/PI3K/AKT signaling pathway, miR-1299 impedes the progression of NSCLC [34]. What's more, Xu and his colleagues found that miR-1299 serves as a tumor suppressor, thus inhibiting the metastasis and proliferation of CCA [35]. In this study, result of RNA-FISH exhibited that circ-0006302 was mainly localized in the cytoplasm. QRT-PCR showed the expression of miR-1299 in CCA was lower level than in non-cancerous tissue. Also, miR-1299 overexpression decreased the expression of PD-L1 and could signi cantly reversed the increase of PD-L1 induced by overexpression of circ-0006302. Thus, circ-0006302 may act as a ceRNA of miR-1299, to regulate PD-L1.
In subsequent experiments, circ-0006302 silencing partly rescued the promotion of proliferation, migration, and invasion of CCA cells induced by upregulation of PD-L1. Importantly, we also determined that circ-0006302 could suppress CD8 + T cell activation by sponging miR-1299 and upregulating the PD-L1 expression subsequently.

Conclusion
In summary, we drew a conclusion that circ-0006302 sequester miR-1299 by acting as a miRNA sponge. Circ-0006302 binds with miR-1299, thus recruiting more PD-L1 to participate in other interactions that promote the migration, growth, and invasion of CCA cells. In addition, circ-0006302 upregulates the expression of PD-L1 through sponging miR-1299 to suppress the cell activation of CD8 + T, thus helping CCA cells escape from the antitumor immune response (Fig. 8). Hence, our ndings provide a new base for studying how CCA develops and progresses. Besides, our conclusion manifests that circ-0006302 is a promising biomarker for CCA diagnosis and a potential target for its therapeutics.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.    overexpression. Data represent the mean ± SD of 3 independent experiments; *p < 0.05; **p < 0.01; ***p < 0.001.   response to circ-0006302 overexpression or silencing, or inhibition of miR-1299, as determined by RT-qPCR. All data were measurement data and expressed as mean ± standard deviation; *p < 0.05; **p < 0.01.