Hsa_circ_0003945 promotes progression of hepatocellular carcinoma by mediating miR‐34c‐5p/LGR4/β‐catenin axis activity

Abstract Accumulating evidence suggests that circular RNAs (circRNAs) play essential roles in regulating cancer progression, but many circRNAs in hepatocellular carcinoma (HCC) remain unknown. Dysregulated circRNAs in HCC were identified through bioinformatics analysis of Gene Expression Omnibus data sets. Quantitative real‐time PCR (qRT‐PCR), Sanger sequencing, RNase R digestion and actinomycin D treatment were conducted to confirm the characterization of circRNAs. CCK‐8, wound‐healing and Transwell assays were performed to assess the functional roles of Hsa_circ_0003945 (Circ_0003945) in HCC cell lines. Subcellular fractionation and fluorescence in situ hybridization (FISH) were performed to locate Circ_0003945 in HCC cells. Dual‐luciferase reporter assay was executed to verify the binding of Circ_0003945 to microRNAs (miRNAs) or the miRNAs to their target genes. In this study, we found that Circ_0003945 was upregulated in HCC tissue, and higher Circ_0003945 expression was positively correlated with tumour size and tumour stage. Furthermore, high plasma levels of circulating Circ_0003945 were confirmed in HCC patients compared with those in non‐HCC groups. The functional experiments revealed that overexpression or knockdown of Circ_0003945 promoted or attenuated tumour growth and migration, respectively. Mechanistically, Circ_0003945 might exert as a miR‐34c‐5p sponge to upregulate the expression of leucine‐rich repeat‐containing G protein‐coupled receptor 4 (LGR4), activating the β‐catenin pathway, and finally facilitating HCC progression. Additionally, a β‐catenin activator could reverse the effect of Circ_0003945 knockdown. In conclusion, Circ_0003945 exerts a tumour‐promoting role in HCC cells by regulating the miR‐34c‐5p/LGR4/β‐catenin axis, which may be a potential target for HCC therapy.


| INTRODUC TI ON
Hepatocellular carcinoma (HCC) is one of the most common malignant tumours globally. 1 Surgery remains the most effective treatment strategy, with curative potential in well-selected candidates. 2 Despite improved surveillance and treatment strategies in recent years, the clinical outcome of HCC remains dismal due to the high rates of relapse and metastasis, which result in a poor 5-year survival rate. 3,4 Thus, a thorough investigation into the mechanisms underlying HCC progression and metastasis is urgently needed to develop new therapeutic approaches and improve the clinical outcome of HCC patients.
CircRNAs are generated via proactive back-splicing of pre-mRNA and are endowed with neither 5' to 3' polarity nor a polyadenylated tail, but they are characterized by covalently closed-loop structures and show resistance to ribonuclease R (RNase R) digestion. [5][6][7] Accumulating evidence indicates that circRNAs are abnormally expressed in many types of cancers and are closely related to cancer progression and prognosis. [8][9][10] Furthermore, circRNAs might mediate pathological processes of cancer cells, 11,12 such as acting as a miRNA sponge to mediate the biological function of their downstream targets or regulate their parental gene expression. 9,13 Although some aberrant circRNAs play important roles in HCC tumorigenesis and progression, [14][15][16] the overall biological and molecular contributions of most circRNAs to HCC progress remain elusive.
In the present study, dysregulated circRNAs in HCC were screened by analysing Gene Expression Omnibus (GEO) datasets followed by verification in clinical samples. We further characterized one significantly overexpressed circRNA derived from exons 11 and 12 of the ubiquitin-associated protein 2 (UBAP2) and named Hsa_circ_0003945 (Circ_0003945) in the circBase database. The function and mechanism of Circ_0003945 in HCC progression were also investigated.

| Methods for screening circRNA profiles of HCC tissue
To show the dysregulated circRNAs, GEO databases (https://www. ncbi.nlm.nih.gov/geo/; GSE78520, GSE94508 and GSE97332) were selected based on inclusion of results of circRNA microarray analysis of HCC tissue and paired healthy adjacent liver tissue. 17,18 The GEO2R tool was used to screen the dysregulated circRNAs in HCC tissue, with a p value <0.05 used as a cut-off threshold.

| Clinical specimens
All human HCC tissue and paired adjacent normal liver tissue (n = 50) and plasma samples (n = 151, including 41

| Animal experiments
Male BALB/c nude mice (5-6 weeks old) were purchased from the Department of Experimental Animals of the Chinese Academy protein-coupled receptor 4 (LGR4), activating the β-catenin pathway, and finally facilitating HCC progression. Additionally, a β-catenin activator could reverse the effect of Circ_0003945 knockdown. In conclusion, Circ_0003945 exerts a tumour-promoting role in HCC cells by regulating the miR-34c-5p/LGR4/β-catenin axis, which may be a potential target for HCC therapy.

K E Y W O R D S
Circ_0003945, hepatocellular carcinoma, leucine-rich repeat-containing G protein-coupled receptor 4, miR-34c-5p, tumour progression of Sciences (Shanghai, China). HCCLM3 cells transfected with shCirc_0003945 or shNC (5 × 10 6 cells/mouse) were injected subcutaneously into the right dorsum to generate subcutaneous tumours (n = 6/group). Mouse body weight and tumour size were measured every week. Four weeks later, mice were sacrificed and tumours were weighed and processed for histological analysis.
Tumour volume was calculated according to the formula: volume (mm 3 ) = width 2 × length/2. All animal care and procedures were performed following guidelines approved by the Institutional Animal Care and Use Committee at Zhongshan Hospital, Fudan University.

| Statistical analysis
Statistical analysis was performed using Statistical Program for Social Sciences (SPSS) 24.0 Software (SPSS) and GraphPad Prism 8.0 (GraphPad Software). The distribution of each group was determined by the Kolmogorov-Smirnov test. Student's t-test (two-tailed) was used to assess statistical significance between the two groups.
The paired t-test was used to analyse the statistical significance of circRNAs, miRNAs, or mRNAs between HCC and adjacent normal liver tissue. Chi-square test was used to analyse the correlation between circRNA level and clinicopathological features of HCC patients. Pearson's correlation or Spearman correlation was used to assess the correlation between circRNA and miRNA or mRNA. All experiments were repeated at least three times. Data are presented as mean ±SD. p < 0.05 was considered statistically significant.

| Characteristics and expression of Circ_0003945 in HCC tissue and cell lines
To screen dysregulated circRNA expression profiles in HCC tissue, we re-analysed the datasets from the GEO database ( Figure S1A). Ten significant candidate circRNAs were selected for further validation according to the p value (p < 0.05), |Log 2 FC| > 1.5 (Table S1). Finally, Hsa_ circ_0001955 (Circ_0001955), Hsa_circ_0005397 (Circ_0005397),

Hsa_circ_0027478
(Circ_0027478) and Hsa_Circ_0003945 (Circ_0003945) were successfully verified. The back-spliced regions of these circRNAs were confirmed by Sanger sequencing, and all were in agreement with circBase (Figure and Figure S1B). Furthermore, RNase R digestion revealed that candidate circRNAs were more resistant than associated mRNAs ( Figure 1B and Figure S1C), and actinomycin D treatment clearly attenuated the half-life of parental mRNA levels in HCC cells, but had little effect on circRNAs ( Figure 1C and Figure S1D).
These results confirmed the circular structure of circRNAs, showing that circRNAs were more stable than their linear forms.
Then, the expression level of circRNAs was measured by qPCR in HCC samples. We found that Circ_0001955, Circ_0005397 and Circ_0003945 were significantly increased in HCC tissue compared with adjacent normal liver tissue ( Figure 1D and Figure S2A-2B), with Circ_0003945 showing the most significant increase of the differentially expressed circRNAs (p < 0.001, Figure 1D); no significant difference was found in the transcription of Circ_0027478 ( Figure S2C). Further analysis showed that higher expression level of Circ_0003945 in HCC tissue was closely associated with tumour size and China liver cancer staging (CNLC) stage, implying a possible role of Circ_0003945 in HCC progression (Table 1). Moreover, circulating Circ_0003945 in plasma was also significantly increased in HCC patients compared with non-HCC groups (HD, CHB and LC; Figure 1E), indicating its potential diagnostic role.

| Circ_0003945 promotes the proliferation and migration of HCC cells in vitro
To investigate the functional role of Circ_0003945 in HCC, the Circ_0003945 expression level in HCC cell lines was measured and

| Circ_0003945 acts as a miR-34c-5p sponge in HCC cells
To explore the underlying mechanism of Circ_0003945 in HCC progression, subcellular fractionation analysis of Circ_0003945 was first performed to determine whether circRNAs play different roles depending on their localization in cells. 5,20 This analysis revealed that Circ_0003945 was mostly located in the cytoplasm of HCC cells ( Figure 3A). Further FISH analysis also verified this ( Figure 3B), implying that Circ_0003945 may act as a miRNA sponge to participate in HCC progression.
The regulatory effects of miRNAs on their target genes are dependent on RNA-induced silencing complex (RISC), which contains the protein AGO2. 21,22 Therefore, RIP analysis was conducted using anti-AGO2 antibody in HCC cells. We found that the anti-AGO2 antibody significantly bound Circ_0003945, but not circANRIL (a circular RNA that does not bind AGO2 and was used as a negative control 23,24 ), indicating that Circ_0003945 might serve as a platform for AGO2 and miRNAs ( Figure 3C).
Next, the association between Circ_0003945 and miR-34c-5p was addressed. The miR-34c-5p was expressed at a lower level in HCC tissue than in adjacent normal tissue (Figure 3F), and a negative correlation was found between Circ_0003945 and miR-34c-5p in tissue samples ( Figure 3G). Further, knockdown of Circ_0003945 upregulated miR-34c-5p expression level ( Figure 3H).

| Transfection of Circ_0003945 antagonizes the inhibitory effects of miR-34c-5p on HCC progression in vitro
Next, functional analysis was carried out to evaluate the effect of Circ_0003945 sponging miR-34c-5p. Results showed that inhibiting miR-34c-5p expression reversed the reduced cell viability induced by Circ_0003945 knockdown ( Figure S4C; Figure 4A), and the miR-34c-5p mimics attenuated the promotional effects of Circ_0003945 overexpression on proliferation ( Figure S4D; Figure 4B). By analogy, suppressing miR-34c-5p expression enhanced the role of Circ_0003945 in maintaining the migratory ability of HCC cells ( Figure 4C). Meanwhile, miR-34c-5p mimics dramatically inhibited the migration of HCC cells induced by Circ_0003945 overexpression ( Figure 4D). A similar effect was observed in Transwell assays ( Figure 4E-F). These results collectively illustrate that Circ_0003945 is necessary to maintain HCC cell progression in part by absorbing miR-34c-5p.

| Circ_0003945 contributes to HCC progression by promoting LGR4 expression by sponging miR-34c-5p
The miRNAs have been reported to regulate their downstream target genes by recognizing the guide sequence. The downstream targets of miR-34c-5p were predicted according to the TargetScan and miRDB databases. Based on the comprehensive score, immunoglobulin superfamily member 1 (IGSF1), CUE domain-containing protein 1 (CUEDC1), cell death-inducing p53 target 1 (CDIP1), small G protein signalling modulator 2 (SGSM2), and leucine-rich repeatcontaining G protein-coupled receptor 4 (LGR4) were finally selected for further verification (Table S3). The qPCR results showed that only LGR4 expression was dramatically decreased by transfection of miR-34c-5p mimics compared with miR-NC, and LGR4 was upregulated by suppressing miR-34c-5p expression ( Figure S5A).
Western blot analysis also confirmed this ( Figure S5B). Further, luciferase reporters were constructed to verify this interaction ( Figure S5C). Luciferase assay showed that transfection of miR-34c-5p mimics dramatically decreased the activity of the luciferase reporter carrying the wild-type LGR4 3' untranslated region (3'-UTR) compared with the miR-NC group. In contrast, the mutated luciferase reporter showed no significant change with miR-34c-5p overexpression ( Figure S5D). Additionally, LGR4 was expressed at remarkably high level in HCC tissue compared with adjacent normal liver specimens ( Figure S5E). A negative correlation was also found between the level of miR-34c-5p and LGR4 expression in tissue samples ( Figure S5F).
As Circ_0003945 may function as a miR-34c-5p sponge to regulate LGR4 expression, the relationship between Circ_000345 and LGR4 was also addressed. Western blot analysis revealed that LGR4 expression significantly decreased with decreasing level of Circ_0003945 ( Figure 5A and Figure S5G), and the suppressed LGR4 level in Circ_00039450-knockdown HCC cells was promoted by miR-34c-5p inhibition ( Figure 5B and Figure S5H). Clinically, we also found that Circ_0003945 level was positively correlated with LGR4 expression ( Figure 5C).
We further explored the biological role of LGR4 induced by Circ_0003945. Functional analysis showed that repressed proliferation of Circ_0003945-knockdown HCC cells was rescued by LGR4  Figure 5D). Meanwhile, enhanced proliferation of Circ_0003945-overexpressing HCC cells was suppressed by suppressing LGR4 expression ( Figure 5E). The migration assay revealed similar results ( Figure 5F-5I).

| Circ_0003945/miR-34c-5p/LGR4 promotes HCC progression via affecting β-catenin pathway
To identify the signalling pathway underlying the effects of the Circ_0003945/miR-34c-5p/LGR4 axis on HCC cells, KEGG analysis (Entry:map 04310) and previous studies showed that LGR4 is the key gene in the Wnt/β-catenin pathway. 25,26 Therefore, we investigated whether β-catenin pathway was influenced by the Circ_0003945/ miR-34c-5p/LGR4 axis. It revealed that stable knockdown of Circ_0003945 contribute to the phosphorylation of β-catenin and miR-34c-5p reversed this effect ( Figure 6A and Figure S6A), which may induce accumulation of β-catenin. Nuclear and cytoplasmic protein assays also confirmed that the accumulation of βcatenin in the nucleus in overexpressing-Circ_0003945 HCC cells ( Figure S6B). The representative downstream target genes of the βcatenin pathway (c-Myc and Cyclin D1) were attenuated by silencing Circ_0003945 (Figure S6C,D) and miR-34c-5p also participated in this process.
To further determine whether the β-catenin pathway is regulated by Circ_0003945, CHIR-99021 27 (an activator of Wnt/β-catenin) was used. We found that CHIR-99021 suppressed the phosphorylation of β-catenin in Circ_0003945-knockdown MHCC97H cells ( Figure 6B; Figure S6E). In addition, CCK-8 assay showed that CHIR-99021 reversed the reduced cell viability of Circ_0003945 knockdown ( Figure 6C). Transwell assays also indicated a similar role of CHIR-99021 in migration and invasion of HCC cells ( Figure 6D,E).
Moreover, β-catenin was knockdown in Circ_0003945overexpressing HCC cells, and we found that Circ_0003945 enhanced cell proliferation and migration, while intervening β-catenin could inhibit the enhanced role of Circ_0003945 ( Figure S7A-C).

| Circ_0003945 promotes the growth of HCC tumours in vivo
To in Circ_0003945-knockdown groups ( Figure 7D). In conclusion, Circ_0003945 acts as a miR-34c-5p sponge to upregulate LGR4, activating the β-catenin axis and promoting HCC progression ( Figure 7E).  In summary, we propose that Circ_0003945 is significantly upregulated in HCC cells and is related to HCC progression. Functionally and mechanistically, Circ_0003945 might directly target miR-34c-5p to release its inhibition of LGR4 expression, finally activating the βcatenin pathway, and thus, enhancing malignant properties of HCC cell. Therefore, Circ_0003945 might serve as a novel biomarker and therapeutic target for anti-HCC therapy.

E TH I C S S TATEM ENT
The present study was approved by the Medical Ethics Committee of Zhongshan Hospital, Fudan University. All participants in this study provided written informed consent in accordance with the Declaration of Helsinki.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets analysed for this study can be found in the Gene Expression Omnibus (GSE78520, GSE94508 and GSE97332).