circRNA circ‐CCND1 promotes the proliferation of laryngeal squamous cell carcinoma through elevating CCND1 expression via interacting with HuR and miR‐646

Abstract Cyclin D1 (CCND1) is a well‐known proliferation promoter that accelerates G1/S transition in cancer. However, the underlying mechanism by which CCND1 is regulated is still largely unknown. In this study, we identified a novel circular RNA (circRNA) derived from CCND1 (circ‐CCND1, hsa_circ_0023303) as a key regulator for CCND1. circ‐CCND1 was found to be markedly up‐regulated in laryngeal squamous cell carcinoma (LSCC) and closely associated with aggressive clinical features and adverse prognosis. Depletion of circ‐CCND1 significantly inhibited LSCC cell proliferation in vitro and retarded tumour growth in vivo. Regarding the mechanism, circ‐CCND1 physically bound to human antigen R (HuR) protein to enhance CCND1 mRNA stability; on the other hand, circ‐CCND1 could act as an effective sponge for miR‐646 to alleviate the repression of miR‐646 on CCND1 mRNA. As a result, circ‐CCND1 post‐transcriptionally elevated CCND1 expression via coordinated avoidance of CCND1 mRNA decay, thereby promoting LSCC tumorigenesis. Taken together, our findings uncover the essential proliferation‐promoting role of circ‐CCND1 through regulation of the stability of CCND1 mRNA in LSCC. Targeting circ‐CCND1 may be a promising treatment for LSCC patients.

has been well documented that CCND1 is a proto-oncogene that is frequently overexpressed in various human cancers, including LSCC. 6 Dysregulation of CCND1 can lead to uncontrolled cell proliferation and malignancy, which enables it to be a feasible and effective therapeutic target for cancer. 7 Although CCND1 is closely linked to tumorigenesis and aggressiveness, the underlying mechanism regulating the function of CCND1 in LSCC still remains largely elusive.
As a special class of non-coding RNA, circular RNA (circRNA), characterized by the covalently closed ring structure, is attracting considerable attention. 8 circRNA is highly stable and conservative with regulatory potency. 9 A large number of studies have shown that circRNA is involved in the development and progression of human diseases, especially in cancer. 10 Emerging evidence suggests that cir-cRNA can modulate its parental gene expression in a cis-or trans-acting manner. For instance, circ-ITGA7 up-regulated its linear isoform ITGA7 via the miR-370-3p/neurofibromin 1/Ras pathway in colorectal cancer. 11 circ-FLI1 was capable of directly interacting with TET1 and DNMT1 to coordinately increase FLI1 expression in breast cancer. 12 circ-PABPN1 was reported to inhibit the translation of PABPN1 by reducing the binding of HuR and PABPN1 mRNA. 13 These studies imply that there is a crosstalk between circRNA and its host gene.
In the present study, we screened CCND1-derived circRNAs and identified that circ-CCND1 (hsa_circ_0023303) was highly expressed and could elevate CCND1 in LSCC. The potential mechanism by which circ-CCND1 regulated its host gene was further elucidated.

| Ethical standards
This study was approved by the Ethics Committee of People's Hospital of Henan Province and complied with the principles of the Helsinki Declaration. Handwritten informed consent was provided by each enrolled patient. Besides, the animal study was carried out with approval of the Animal Care Committee of People's Hospital of Henan Province.

| LSCC tissues and cell lines
A total of 101 pairs of carcinoma and noncancerous tissues were collected from patients who were diagnosed as LSCC and underwent surgical resection. Among them, 25 matched tissues were used to determine the aberrantly expressed in LSCC. All surgically removed tissues were stored in liquid nitrogen before use. We routinely followed up patients every three months. The detailed clinicopathological features of patients are described in Table 1

| Cell transfection
All oligomers used in this study including siRNA targeting circ-CCND1

| Colony formation and cell cycle
For colony formation assay, 500 Hep-2 and TU212 cells transfected with si-NC or si-circ-CCND1 were plated into six-well plates and routinely grown for 10 days. Then, the cells were washed, fixed with methanol and stained with crystal violet. For cell cycle analysis, the treated cells were fixed in 75% pre-cooled ethanol and then incubated with propidium iodide solution, followed by calculation for the cell proportion in G0/G1, S and G2/M phases by a flow cytometer.

| Xenograft model
A total of 5 × 10 6 Hep-2 cells were subcutaneously injected into nude mice (n = 8 in each group). After one week, the subcutaneous tumours were visible and then intratumorally injected with cholesterol-coupled si-NC or si-circ-CCND1 (GenePharma) three times weekly for two weeks. Two weeks after the last injection of cholesterol-coupled oligomers, all mice were killed and the tumours were carefully dissected, followed by weighing and photography.

| RNA pull-down and RIP assays
For RNA pull-down assay, the biotin-labelled probes were de-

| Luciferase reporter assay
The full-length sequence of circ-CCND1 or CCND1-3′-UTR with wildtype or mutant miR-646 binding site was cloned into psiCHECK2
org/), we found that a total of four circRNAs were generated from CCND1. Among them, three (circ-CCND1, hsa_circ_0023304 and hsa_circ_0023305) could be detected in human 293T cells.
We then tested the expression levels of these three circRNAs in 25 LSCC and adjacent normal tissues ( Figure 1A). The qRT-PCR results showed that only circ-CCND1 ( Figure 1B), but not hsa_circ_0023304 and hsa_circ_0023305 ( Figure 1C,D), was significantly aberrantly expressed in LSCC when compared to normal tissues. To further confirm this finding, we additionally collected 76 matched LSCC and normal tissues (101 cases in total) to assess circ-CCND1 expression, and the results showed that circ-CCND1 was markedly up-regulated in LSCC tissues in comparison with precancerous tissues ( Figure 1E). Importantly, high circ-CCND1 was positively correlated with larger tumour size (P < .001), poor differentiation (P = .007) and advanced TNM stage (P = .004; Table. 1). Moreover, Kaplan-Meier plotter showed that LSCC patients with high circ-CCND1 expression had shorter overall survival time than those with low circ-CCND1 expression ( Figure 1F). In all, these data suggest that circ-CCND1 is a dysregulated circRNA in LSCC and may be used as a promising biomarker.

| Depletion of circ-CCND1 delays LSCC cell growth both in vivo and in vitro
To explore the function of circ-CCND1, we first tested its expression in LSCC cell lines. Likewise, up-regulated circ-CCND1 expression was observed in seven LSCC cells as compared to normal NP-69 cells, especially in Hep-2 and TU212 cells (Figure 2A). We then designed three siRNAs targeting the junction sequence of circ-CCND1 to knock down circ-CCND1 ( Figure 2B), and the results showed that si-circ-CCND1#1 and #2, but not si-circ-CCND1#3, could effectively reduce circ-CCND1 expression by about 70% in both Hep-2 and TU212 cells ( Figure 2B). The decreased number of cells as well as clones was observed after circ-CCND1 silencing ( Figure 2C,D). In addition, cell cycle analysis displayed that depletion of circ-CCND1 resulted in more cell arrest in G0/G1 phase ( Figure 2E). Further, we established the xenograft model to explore the function of circ-CCND1 in vivo. As expected, the volume and weight of subcutaneous tumours in the circ-CCND1-depleted group were less than those in the control group (n = 8 in each group) ( Figure 2F). Taken together, these functional experiments reveal that circ-CCND1 is a positive regulator for LSCC cell proliferation.
As shown in Figure 3A,  Figure 3D). Besides, we found that CCND1 mRNA expression was significantly up-regulated in LSCC tissues ( Figure 3E) and strongly positively correlated with circ-CCND1 expression ( Figure 3F).

Moreover, the decreased proliferative abilities of Hep-2 and TU212
cells caused by circ-CCND1 depletion were completely rescued after exogenous expression of CCND1 ( Figure 3G). Altogether, these results indicate that circ-CCND1 is a stabilizer for CCND1 in LSCC.

| circ-CCND1 physically interacts with HuR to enhance the binding of HuR to CCND1 mRNA
To determine how circ-CCND1 increased CCND1 mRNA stability, we first detected the subcellular localization of circ-CCND1. As shown in Figure 4A, circ-CCND1 was predominately localized in the cytoplasm, which is consistent with its post-transcriptional regulation of CCND1. By the online CircInteractome tool, we found that circ-CCND1 can interact with some RNA-binding proteins, in which human antigen R (HuR) attracted our attention mainly due to its function of RNA stabilization. 14 RNA pull-down results showed that a large amount of HuR protein was enriched by circ-CCND1 probe, but not control probe, and it was diminished when circ-CCND1 was depleted ( Figure 4B). To further confirm this interaction, we performed RIP assay, and the results showed that more circ-CCND1 was precipitated by HuR antibody in comparison with IgG negative control antibody in both Hep-2 and TU212 cells ( Figure 4C).
Through online catRAPID algorithm, HuR was predicted to likely bind to the 83-134 bp of circ-CCND1 ( Figure 4D). To test this prediction, we conducted RNA pull-down assay and found that mutation of this region of circ-CCND1 blocked the interaction between circ-CCND1 and HuR ( Figure 4E). More importantly, we found that depletion of circ-CCND1 dramatically decreased the binding of HuR to CCND1 3′-UTR ( Figure 4F). In addition, ectopic expression of HuR could partially rescue the reduced CCND1 mRNA level caused by circ-CCND1 knockdown ( Figure 4G). Overall, these data suggest that HuR mediates the regulation of circ-CCND1 on CCND1 mRNA stability.

| circ-CCND1 serves as a sponge for miR-646
As CircInteractome tool, we found that several miRNAs had circ-CCND1 binding sites, of which miR-646 had the largest number of binding sites ( Figure 5A,B). And CCND1 mRNA 3′-UTR was also predicted to be bound by miR-646 using the miRanda and TargetScan databases. We then performed luciferase reporter assay to evaluate the relationship between circ-CCND1, miR-646 and CCND1. The results showed that overexpression of miR-646 dramatically reduced the luciferase activity of circ-CCND1 or CCND1 vector with wild-type miR-646 binding site, whereas it had no effect on the mutant one ( Figure 5C,D). To test whether they interacted with each other, we conducted RNA pull-down assay in Hep-2 and TU212 cells. As shown in Figure 5E, compared with control probe, circ-CCND1 probe could abundantly pulled down miR-466. Likewise, circ-CCND1 and CCND1 3′-UTR were substantially enriched by wild-type miR-646 probe, but not by the mutated probe ( Figure 5F). Depletion of circ-CCND1 significantly up-regulated miR-646 expression ( Figure 5G). Besides, the diminished CCND1 mRNA and proliferative abilities of Hep-2 and TU212 cells caused by circ-CCND1 silencing were partly rescued by miR-646 silencing and were wholly rescued after miR-646 knockdown combined with HuR overexpression (Figure 5H, In all, these above results demonstrate that circ-CCND1 can stabilize the mRNA of its host gene via HuR and miR-646 in LSCC ( Figure 5J).

F I G U R E 4 circ-CCND1 interacts with
HuR to increase CCND1 mRNA stability. A, qRT-PCR analysis for circ-CCND1 expression in the cytoplasmic and nuclear fractions. B, RNA pull-down assay in control and circ-CCND1-silenced Hep-2 and TU212 cells, followed by detection of HuR expression using Western blot assay. C, RIP assay in control and circ-CCND1-silenced Hep-2 and TU212 cells, followed by qRT-PCR analysis for circ-CCND1 expression. D, The interaction profile between HuR and circ-CCND1 predicted by online catRAPID algorithm. E, RNA pull-down assay in Hep-2 and TU212 cells with indicated wild-type or mutated probe, followed by Western blot analysis for HuR protein expression. F, RIP assay in control and circ-CCND1-silenced Hep-2 and TU212 cells with HuR or IgG antibody, followed by qRT-PCR analysis for CCND1 3′-UTR enrichment. G, qRT-PCR analysis for CCND1 mRNA expression in control and circ-CCND1-silenced Hep-2 and TU212 cells transfected with HuR expression vector. *P < .05, **P < .01, ***P < .001  implying that miR-646 may be also an anti-cancer gene in LSCC.

| D ISCUSS I ON
Therefore, these data suggest that circ-CCND1 partially regulates the stability of its host gene via functioning as a ceRNA in LSCC.
On the other hand, circRNA is capable of physically interacting with RNA-binding protein to regulate gene expression. 25 Herein, we found that circ-CCND1 could directly bind to HuR to potentiate CCND1 mRNA stability. HuR is a well-known RNA-binding protein that prevents mRNA decay via targeting the uridylate-rich element (ARE, (A/U)UUU(A/U)) on the 3′-UTR. 26 A wealth of studies have demonstrated that HuR is ubiquitously expressed in eukaryotic organization and frequently elevated in human cancers, including LSCC. 27 CCND1 is one of the best-known HuR targets, in which HuR increases the half-life of CCND1 by the ARE motif on CCND1 3′-UTR. 28 In our study, circ-CCND1 was shown to interact with HuR and facilitate its enrichment on CCND1 3′-UTR, thereby increasing CCND1 mRNA stability. Through sequence analysis, we found that there are two ARE motifs on the 3′-UTR of CCND1 mRNA, which do not overlap with the binding site of miR-646 (GCUGCU). Although some evidence shows that the interaction of HuR with the 3′-UTR alters the conformation of mRNA, thereby competitively blocking the binding of miRNA to mRNA 3′-UTR, 29 our data displayed that exogenous HuR expression did not affect the interaction between miR-646 and CCND1 mRNA 3′-UTR (data not shown), suggesting that

The work was funded by Henan Province Science and Technology
Tackling Project (201702196).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
BLW designed the study. YZZ, JL and YT performed the experiments and analysed the data. YZZ and BLW wrote and revised the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets and supporting materials generated and/or analysed during the current study are available from the corresponding author on reasonable request.