Circ‐LAMP1 contributes to the growth and metastasis of cholangiocarcinoma via miR‐556‐5p and miR‐567 mediated YY1 activation

Abstract Dysregulation of circular RNAs (circRNAs) executes important regulatory roles in carcinogenesis. Nonetheless, few studies focused on the mechanisms of circRNAs in cholangiocarcinoma (CCA). qRT‐PCR was applied to verify the dysregulated circRNAs in CCA. Fisher's exact test, Kaplan‐Meier analysis and Cox regression model were utilized to investigate the clinical implications of circ‐LAMP1 in the patients with CCA. The viability, apoptosis, migration and invasion of CCA cells were detected after silencing/overexpression of circ‐LAMP1. Xenograft and lung metastasis assays were performed to verify the in vitro results. The regulatory networks of circ‐LAMP1 were unveiled by bioinformatic analysis, RNA immunoprecipitation (RIP), RNA pulldown and luciferase reporter assays. Up‐regulation of circ‐LAMP1 was found in CCA tissue samples and cell lines. Enhanced level of circ‐LAMP1 was linked to clinical severity, high post‐operative recurrence and poor prognosis for the patients with CCA. Gain/loss‐of‐function assays confirmed the oncogenic role of circ‐LAMP1 in mediating cell growth, apoptosis, migration and invasion. Nevertheless, the level of circ‐LAMP1 had no effect on normal biliary epithelium proliferation and apoptosis. Animal study further verified the in vitro data. Mechanistically, circ‐LAMP1 directly sponged miR‐556‐5p and miR‐567, thereby releasing their suppression on YY1 at post‐transcriptional level. Rescue assay indicated that the oncogenic role of circ‐LAMP1 is partially dependent on its modulation of YY1 in CCA. In summary, this study suggested that circ‐LAMP1 might be used as a promising biomarker/therapeutic target for CCA.


| INTRODUC TI ON
Cholangiocarcinoma (CCA) is an aggressive malignancy in hepatobiliary system. A considerable portion of patients are diagnosed at advanced stages, which lack effective therapies, leading to poor treatment outcome. 1 The pathogenesis of CCA is a complex process, which was affected by various factors and oncogenes. Tumour biomarkers are of great clinical value and crucial significance for the early detection, targeted treatment and therapeutic effect evaluation. 2,3 Therefore, it is enormously crucial to discover novel and reliable treatment strategy for CCA. Circular RNAs (circRNAs) are a series of ncRNAs with limited protein-coding capacity. 4,5 Up to now, lots of circRNAs are discovered from mammalian cells via high throughput sequencing and bioinformatics analysis. 5 Growing findings imply that circRNAs are tightly implicated in the dysregulation of gene transcription and translation, suggesting that circRNAs may go for the development of human diseases. [6][7][8] Currently, circRNAs are known to have the following roles: (a) they act as "miRNA sponges" to modulate miRNAs activity 4 ; (b) they interact with RNA-binding proteins 9 ; and (c) they are translated to specific proteins. 10 CircRNAs, especially those with miRNA response elements, bind to miRNAs, thereby acting as competing endogenous RNAs (ceRNAs). 11 For instance, ciRS-7 reduces miR-7 activity and thus increases several oncogenic factors expression. 12,13 The evidence indicates that the circRNA-miRNA-mRNA axis is essential to the mechanisms that determine the progress of cancers.
The current study identifies a novel CCA-related circRNA, circ-LAMP1, which is elevated in CCA tissues analysed by cir-cRNA microarray. Circ-LAMP1 (hsa_circRNA_101303; circBase ID: hsa_circ_0030998; circBank ID: hsa_circLAMP1_010) is located on chr13:113963957-113964177. The spliced sequence length of circ-LAMP1 is 220 nt. The genomic structure illustrates that circ-LAMP1 is looped by the exon-3 of LAMP1 gene. In the study, up-regulated circ-LAMP1 correlates with the clinical severity and prognosis of CCA patients. In vitro and in vivo assays revealed its effect in promoting the growth and metastasis of CCA cells. Mechanistically, we identified that circ-LAMP1 sponges miR-556-5p and miR-567 to up-regulate YY1 at post-transcriptional level. Furthermore, we demonstrated that circ-LAMP1 exerts oncogenic properties via miR-556-5p/miR-567/YY1 axis.

| Clinical samples
A total of 216 individuals were enrolled from the Second Affiliated Hospital of Harbin Medical University. The use of patient specimens was authorized by the Ethics Committee of our hospital. Each patient signed the written informed consent prior to sample donation.
None of the patients received pre-operative anti-tumour treatment before specimen collection. Immediately after surgical resection, the specimens were stored at −80°C.

| Cell lines and culture
Human CCA cells (HCCC-9810 and RBE) were acquired from Chinese Academy of Sciences (Shanghai). Huh-28, QBC939, HuCCT1, CCLP1 and the normal cell line (HIBEC) were acquired from Professor Lianxin Liu (Changjiang Scholar, University of Science and Technology of China) as a gift. The cells were maintained in complete medium containing 90% RPMI-1640 (Hyclone) and 10% foetal bovine serum (FBS; Gibco). All the cells were placed in a 5% CO 2 humidity condition at 37°C. The solution was changed every 3 days cell passage. Cells in all of the following experiments were taken in the logarithmic growth phase. All cell lines were passaged for no more than 6 months.

| qRT-PCR and cell transfection
Total RNA from tissues and cells was isolated by utilizing TRIzol in accordance with manufacturer's descriptions. First Strand cDNA Synthesis Kit (Roche) was recruited to generate cDNA from harvested RNA. Based on this, the equal cDNA was mixed with RNase free water, primers and reagents of SYBR-Green PCR kit (QIAGEN).
Relative gene expression was calculated by the 2 −ΔΔCt method.

| Cell apoptosis determination
Cells at logarithmic growth phase were harvested by trypsin. After resuspend in 400 μl of 1× binding buffer, the cells were then stained by 5 μl of FITC-Annexin V and 5 μl of PI (Beyotime, Haimen, China).
A flow cytometer was utilized to evaluate apoptotic cells (BD Biosciences).

| Wound healing assay
KMBC and RBE cells were planted in 6-well dishes overnight. The cell monolayer was wounded with a pipette tip to create a scratch.
Afterwards, the cells were washed with PBS and the serum-free medium was added. Images were captured at 24/36 h following the initial scratch to evaluate cell migration rate.

| Transwell experiments
After being dispersed with 0.25% trypsin, the CCA cells (1 × 10 5 for cell invasion and 5 × 10 4 for cell migration) were centrifuged, resuspended and dispersed in the top compartment of transwell unit (Corning, Beijing, China). Matrigel (BD Biosciences) was used in invasion experiment, but not in migration experiment. At the same time, the RPMI-1640 containing 10% FBS was supplied into the lower compartment.
After culturing for 36 h, the cells in the top membranes were abandoned, and the cells in the lower surface of the upper chamber were fixed and stained. Lastly, the migrated/invaded cells were visualized and counted microscopically.

| RNA Immunoprecipitation (RIP) assay
RIP was carried out using the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore) in agreement with the manufacturer's instruction. KMBC and RBE cell lysates were obtained and incubated with RIP buffer containing magnetic beads conjugated with human anti-Ago2 antibody or normal mouse IgG (control). RNA was extracted from immunoprecipitate and analysed by qRT-PCR.

| RNA pulldown assay
The biotin-labelled circ-LAMP1 probe targeting the junction sequence of circ-LAMP1 was in vitro synthesized by Genecreate (Wuhan, China), followed by incubation with cell lysate at 4°C overnight. Then, the above complex was incubated with streptavidinconjugated magnetic beads at room temperature for 2 h. Lastly, interacted RNAs were purified and evaluated by qRT-PCR.

| In vivoexperiments
The animal assay was authorized by the Institutional Animal Care and Use Committee of our hospital. KMBC cells (6 × 10 6 ) were injected into the left flank of nude mice (n = 4/group, six-week-old) subcutaneously. The volume of the formed tumour was measured every 3 days at 6 days post-injection. Tumour volume was calculated following the formula: volume = length × width 2 × 0.5. After injection for 27 days, the mice were killed, and the subcutaneous xenografts were harvested and weighed.
For cell metastasis detection, the transfected KMBC cells (2 × 10 6 ) were injected into the tail vein to construct the lung metastatic model (n = 4/group). Four week later, the mice were killed and their lungs were harvested and subjected to H&E staining.

| Data analysis
Data analyses of the project were accomplished by SPSS 22.0 software (IBM). For group comparison, the differences of the two-group and multiple-groups were analysed via Student's t test and one-way analysis of variance with Tukey's test, respectively.
Correlation test was carried out to examine the correlation between miR-556-5p/miR-567, circ-LAMP1 and YY1. Survival curves were estimated applying Kaplan-Meier method with log-rank test.
Multivariate cox regression analysis was used to determine the independent factors that affect overall survival (OS) and diseasefree survival (DFS) for CCA patients. P <.05 meant the existence of statistical significance.

| Circ-LAMP1 is up-regulated in CCA and correlates with unfavourable prognosis
We previously performed circRNA microarray by using four pairs of CCA tissues and adjacent normal samples. Among the 224 differentially expressed circRNAs (fold change > 2, P < .05), 97 were up-regulated, and 127 were down-regulated in CCA tissues in comparison with those in normal tissues. We randomly selected five elevated circRNAs (hsa_circRNA_101303, hsa_cir-cRNA_100641, hsa_circRNA_100989, hsa_circRNA_104168 and hsa_circRNA_100017) for further study. As a result, hsa_cir-cRNA_101303 was the most up-regulated one among the recruited circRNAs ( Figure 1A). Hsa_circRNA_101303 originates from exon-3 of a protein-coding gene locus, LAMP1 ( Figure 1B). In addition, the half-life of circular form of LAMP1 (circ-LAMP1) is remarkably longer than that of linear LAMP1 mRNA ( Figure 1C). To explore the expression status of circ-LAMP1, we evaluated the level of circ-LAMP1 in 216 pairs of CCA tissues compared with that of normal specimens.
qRT-PCR analysis convinced that circ-LAMP1 was apparently augmented in CCA tissues ( Figure 1D). The recruited patients were divided into two groups followed by the median value of circ-LAMP1 expression. Notably, circ-LAMP1 in cancerous tissues was linked to number of tumours (P =.003) and TNM stages (P =.013), both significantly (Supplementary file 1, Table S1). Kaplan-Meier analysis appraised possible significance for circ-LAMP1 degree in predicting OS and DFS for CCA cases. In accordance with the relevant findings, CCA cases harbouring high circ-LAMP1 degrees experienced a worse OS than low ones (P < .001, Figure 1E). Additionally, circ-LAMP1 expression in CCA tissue samples was remarkably linked to high post-operative recurrence for the patients with CCA (P < .001, Figure 1F). Multivariate analysis evaluated whether circ-LAMP1 expression level or other clinicopathological characteristics were independent prognostic markers for CCA cases. Consequently, high circ-LAMP1 level is a valuable indicator for predicting OS (P = .001, Table S2). Furthermore, higher expression of circ-LAMP1 was tightly correlated with patients' DFS (P = .001, Supplementary file 3, Table S3). Circ-LAMP1 expression in six CCA cells and HIBEC was further investigated. As exhibited in Figure 1G, circ-LAMP1 was overexpressed in almost all the recruited CCA cells than HIBEC.

| Circ-LAMP1 accelerates CCA cell progression in vitro
The results described above indicate that circ-LAMP1 was elevated in CCA and thus may be an oncogene. Hence, we deter- The results showed more apoptotic cells in circ-LAMP1 depletion group relative to the control group ( Figure 2G). Moreover, apoptotic cells were found to be remarkably decreased in circ-LAMP1 up-regulated cells ( Figure 2H). Cell migration of CCA cells were detected after circ-LAMP1 expression was knocked down/ overexpressed. The data of wound healing and transwell migration tests verified that cell migration of KMBC cells was inhibited by transfected with si-circ-LAMP1-1 ( Figure 2I). Elevated circ-LAMP1 induced the opposite effect in RBE cells ( Figure 2J). To explore whether circ-LAMP1 affects cell metastatic properties, Transwell invasion assay was conducted. As a result, silenced circ-LAMP1 induced down-regulation of cell invasive potential ( Figure 2K). By contrast, elevated circ-LAMP1 enhanced cell invasion in RBE cells ( Figure 2L).

| Circ-LAMP1 contributes to CCA progression by sponging miR-556-5p and miR-567
After isolation of nuclear and cytoplasmic fractions, qRT-PCR results indicated the predominant cytoplasmic distribution of circ-LAMP1 ( Figure 3A, B). We then asked whether circ-LAMP1 may interact with certain miRNAs in CCA cells. We found that circ-LAMP1 was obviously enriched in anti-Ago2 than that in anti-IgG and was less enriched after circ-LAMP1 silenced ( Figure 3C).
A biotin-labelled circ-LAMP1 probe was applied to pull down circ-LAMP1. As Figure 3D  Therefore, miR-615-5p was also included in the subsequent study.
We then isolated miRNAs after pulldown assay and measured the binding sites for miR-556-5p/567, were established ( Figure 3I).
The results showed that miR-556-5p/567 mimics remarkably suppressed the luciferase intensity in wild-type circ-LAMP1 reporter gene, but not the mutant-type ( Figure 3J, K).

| Circ-LAMP1 can not affect HIBEC proliferation and apoptosis
To investigate whether circ-LAMP1 may have side effect on normal biliary epithelium, we knocked down and overexpressed circ-LAMP1 expression in HIBEC. As Figure 7A demonstrated, the two selected siRNAs had effective knockdown efficiencies. Then, ectopic expression of circ-LAMP1 in HIBEC was also achieved ( Figure 7B). For the part of functional assay, cell viability was not altered after circ-LAMP1 knockdown/overexpression ( Figure 7C). Furthermore, cell apoptotic rate could not be affected by circ-LAMP1 in HIBEC cells ( Figure 7D).

| Circ-LAMP1 deletion resulted in the decrease of tumour growth and metastasis in vivo
In view of the foregoing description, we attempted to explore the role of circ-LAMP1 in vivo. KMBC cells were transfected with shcirc-LAMP1-1. The transfected cells were inoculated into nude mice. As a result, the tumour volume ( Figure 8A, B) and weight ( Figure 8C) were sharply diminished via circ-LAMP1 absence.
Subsequently, we focused on the level of YY1 in xenograft tumours. Immunohistochemistry determined that YY1 was decreased F I G U R E 6 Circ-LAMP1 regulates CCA cell growth, apoptosis and invasion via up-regulating YY1 expression. A, B, The protein level of YY1 was measured by Western blotting after transfection in KMBC and RBE cells. C, D, CCK-8 assay was performed to analyse the proliferation of KMBC and RBE cells after transfection. E, F, Flow cytometric assay was performed to detect the apoptosis of KMBC and RBE cells after transfection. G, H, Transwell assay was performed to detect the invasion of KMBC and RBE cells after transfection. *P < .05, **P < .01 in the tumours formed from circ-LAMP1 depleted KMBC cells ( Figure 8D). For lung metastasis model, a weaker fluorescence intensity of lungs was observed in the mice injected with circ-LAMP1 decreased KMBC cells ( Figure 6E). The lungs of the nude mice were harvested after four weeks inoculation. Decreased circ-LAMP1 led to less lung metastases ( Figure 6F, G).

| D ISCUSS I ON
Although therapy has been standardized, the rate of morbidity and mortality for CCA remains high at present. 15 In recent years, the role of circRNAs in CCA already has made significant progress in basic medical research. 16,17 Recently, diagnosis and therapy from the perspective of circRNA has gradually become a new approach in cancer research. Therefore, exploring the CCA-related circRNAs and find the ideal molecular markers for targeted therapy of CCA is the direction to further improve the prognosis of CCA. 18,19 Nevertheless, the functions and exact mechanisms of novel circRNAs in CCA are still not fully understood.
The current work identifies a novel circRNA, circ-LAMP1, which is enhanced in CCA specimens screened by circRNA microarray. In the current work, we investigated its clinical significance and found the patients with high expression of circ-LAMP1 was linked to more than one tumour, and higher TNM stages. Additionally, analysis data suggested that circ-LAMP1 might be an independent biomarker for CCA prognosis/recurrence. Furthermore, in vitro and in vivo experiments revealed its effect in promoting the growth and metastasis of CCA cells, whereas we found circ-LAMP1 expression did not affect the proliferation and apoptosis of normal biliary epithelium (HIBEC), indicating its limited side effect on normal cells.
CircRNA-miRNA-mRNA regulatory axis has been widely concerned on account of their rising role in human cancers. [20][21][22] The tumour progression-promoting effect of circ-LAMP1 was also confirmed in T-cell lymphoblastic lymphoma previously. 14 Deng et al unveiled that circ-LAMP1 contributes to cell proliferation through playing as a ceRNA for miR-615-5p to up-regulate DDR2 level. 14 However, we found miR-615-5p could not interacted with circ-LAMP1 in our study, suggesting the tissue specific mechanism of circ-LAMP1 in human cancers. In this study, up-regulation of circ-LAMP1 induced cell proliferation, metastasis and hampered cell apoptosis in CCA cells by absorbing miR-556-5p and miR-567. MiR-556-5p is a newly identified tumour suppressive miRNA in human cancers. Recently, the study illustrated that miR-556-5p, negatively regulated by long non-coding RNA SNHG1, suppressed meningioma progression through Wnt signalling pathway. 23 In this work, it is the first time to demonstrate its tumour suppressing role in CCA.
The role of miR-567 in the facilitation or inhibition of tumour progression is still controversial. In osteosarcoma, miR-567 decreases cell growth and metastatic properties through modulating FGF5. 24 Similarly, decreased miR-567 in breast cancer cells contributes to carcinogenesis. 25 However, up-regulation of circ-cMras suppressed lung adenocarcinoma tumorigenesis by targeting miR-567/PTPRG signalling, suggesting the oncogenic function of miR-567 in lung adenocarcinoma. 26 In the current study, the data agreed with the studies in osteosarcoma and breast cancer. 24,25 The downstream target of miR-556-5p and miR-567 was further unveiled. YY1 belongs to the GLI-Kruppel family which could activate or inactivate gene expression depending on interacting partners, promoter context and chromatin structure. 27,28 YY1 has been identified up-regulated and functioned as an oncogene in several kind of cancers. 29 Consistent with our expectation, we identified that the oncogenic properties induced by circ-LAMP1 are partly dependent on its up-regulation of YY1 in CCA. However, there are still several limitations within the current study. For example, the downstream target of YY1 was not evaluated and needs further study.
To sum up, the current study revealed that circ-LAMP1 was overexpressed in CCA tissue specimens and cells. Circ-LAMP1 functions as a ceRNA for miR-556-5p and miR-567 to up-regulate YY1 expression. The data clarified the potential mechanisms related to circ-LAMP1 in regulating CCA cell growth and metastasis. This study suggests that circ-LAMP1 might be a potential treatment target for CCA.

E TH I C S A PPROVA L A N D CO N S E NT TO PA RTI CI PATE
This study was approved by the Ethics Committee of the Second Affiliated Hospital of Harbin Medical University.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data obtained during this research are available within the paper or available from the authors as needed.