Circular RNA hsa_circ_101555 promotes hepatocellular carcinoma cell proliferation and migration by sponging miR-145-5p and regulating CDCA3 expression

Circular RNAs have been reported to play significant roles in regulating pathophysiological processes while also guiding clinical diagnosis and treatment of hepatocellular carcinoma (HCC). However, only a few circRNAs have been identified thus far. Herein, we investigated the role of a specific closed-loop structure of hsa_circ_101555 that was generated by back-splicing of the host gene casein kinase 1 gamma 1 (CSNK1G1) in the development and proliferation of HCC. We investigated the expression of Hsa_circ_101555 in HCC and normal tissues using bioinformatics. The expression level of hsa_circ_101555 was further detected by fluorescence in situ hybridization and qRT-PCR in ten HCC patients. Transwell, migration, WST-1 assays, and colony formation assays were used to evaluate the role of hsa_circ_101555 in HCC development and proliferation. The regulatory mechanisms of hsa_circ_101555 in miR-145-5p and CDCA3 were determined by dual luciferase reporter assay. A mouse xenograft model was also used to determine the effect of hsa_circ_101555 on HCC growth in vivo. hsa_circ_101555 showed greater stability than the linear RNA; while in vitro and in vivo results demonstrated that hsa_circ_101555 silencing significantly suppressed cell proliferation, migration, and invasion of HCC cells. Rescue experiments further demonstrated that suppression of miR-145-5p significantly attenuated the biological effects of hsa_circ_101555 knockdown in HCC cells. We also identified a putative oncogene CDCA3 as a potential miR-145-5p target. Thus, our results demonstrated that hsa_circ_101555 might function as a competing endogenous RNA of miR-145-5p to upregulate CDCA3 expression in HCC. These findings suggest that hsa_circ_101555 may be a potential therapeutic target for patients with HCC.


Introduction
Hepatocellular carcinoma (HCC), the most prevalent form of primary liver cancer, represents one of the most common malignant tumors globally 1 . HCC is an aggressive disease with dismal prognosis, constituting the third leading cause of cancer-related deaths worldwide 2 . Despite advances in the clinical understanding of the underlying mechanisms in HCC development and progression, its 5-year survival rate remains low 3 . Additionally, the molecular pathogenesis and therapeutic targets in HCC remain largely unknown. Therefore, understanding the pathogenic process of HCC and its regulatory mechanisms would significantly aid its management.
Circular RNAs (circRNAs) are a newly discovered noncoding RNA (ncRNA) that ubiquitously exist in several species 4,5 . Unlike canonical linear RNAs, circRNAs form a covalently closed continuous loop structure that lack 5′ caps and 3′ polyadenylated tails, making them more stable than linear RNAs 6 . circRNAs reportedly function as molecular sponges for microRNAs (miR-NAs) 7,8 and RNA-binding proteins 9 , while also serving as vital regulators of gene transcription and expression 10,11 . Furthermore, circRNAs are highly conserved across multiple species and exhibit tissue-specific and development stage-dependent expression patterns 12,13 . These features imply that circRNAs possess significant functions in biological and pathological processes.
Recent studies have reported that circRNAs are differentially expressed in HCC and serve a central role in its carcinogenesis and progression [14][15][16][17][18][19][20] . Thus, circRNAs may represent promising diagnostic markers and therapeutic targets for HCC. However, compared with other ncRNAs, such as miRNAs and long noncoding RNAs, the research on circRNA in HCC remains in its infancy. To date, only a few functional circRNAs have been discovered and characterized in HCC 21 , while a large number remain to be explored or identified.
In the present study, we analyzed the expression profiles of three circRNA in human HCC tissues and identified hsa_circ_101555 to be conserved and significantly upregulated. Therefore, we focused on investigating the role of hsa_circ_101555 in the development and proliferation of HCC with respect to the miR-145-5p/CDCA3 signaling axis.

hsa_circ_101555 is upregulated in HCC
To identify unique circRNAs involved in HCC, we analyzed the microarray data of GSE7852, GSE94508, and GSE97322 datasets downloaded from the Gene Expression Omnibus (GEO) database. We then visualized the differentially expressed circRNAs (DEcircRNAs) in HCC and normal tissue samples using the "limma" package of R software. A false discovery rate < 0.05 and |log2foldchange > 1 were set as the cutoff criteria for screening the DEcircRNAs (Fig. 1A-C). Among the top 25 upregulated DEcircRNAs, only hsa_circ_101555 appeared in all three GSE datasets (Fig. 1D). Thus, we focused on investigating its role in HCC progression. Subsequently, according to circBase (http://www.circbase.org), we found that hsa_-circ101555 was derived from the host gene casein kinase 1 gamma 1 (CSNK1G1), consisting of six exons (exon 1-6, 815 nt) cyclized by the head-to-tail splicing of exon 1 and exon 6. The existence of a back-spliced junction was confirmed by Sanger sequencing (Fig. 1E). We further verified the expression of CSNK1G1 in HCC tissues, and found it to be upregulated ( Supplementary Fig. 1A). Moreover, analysis of the overall survival and CSNK1G1 expression using GEPIA (http://gepia.cancer-pku.cn/) revealed that high CSNK1G1 expression in HCC was not associated with overall survival (Supplementary Fig. 1B-C).

hsa_circ_101555 is primarily localized in HCC cell cytoplasm
In general, the subcellular localization of circRNA determines its primary mode of action. Fluorescent in situ hybridization (FISH) analysis revealed that hsa_-circ_101555 expression was higher in tumor tissues than in matched nontumor sections ( Fig. 2A). Furthermore, qRT-PCR examination confirmed that hsa_circ_101555 expression was markedly higher in HCC tissues than in adjacent normal tissues (Fig. 2B). High hsa_circ_101555 expression in HCC was associated with overall survival (Supplementary Fig. 2A).
We also analyzed hsa_circ_101555 expression within the serum of HCC patients and healthy controls and found the expression to be significantly higher in HCC patients compared to healthy controls ( Supplementary Fig. 3A). Meanwhile, we also observed that hsa_circ_101555 was upregulated in III-IV stage HCC compared to I-II stage patients, based on TNM stage ( Supplementary Fig. 3B). Additionally, the expression of hsa_circ_101555 was higher in all HCC cell lines (HepG2, SK-Hep-1, Huh-7, and HCCLM3) compared to normal liver LO2 cells, with a significantly higher expression in HepG2 and HCCLM3 cells than in Huh-7 and SK-Hep-1 cells (Fig. 2C).
The circular structure of hsa_circ_101555 was confirmed using RNase R. As shown in Fig. 2D, amplified linear transcripts of CSNK1G1 were clearly degraded by RNase R, while hsa_circ_101555 was resistant to degradation. These data demonstrated both the presence and circular structure of hsa_circ_101555 in HCC cells (HepG2 and HCCLM3). Moreover, hsa_circ_101555 was predominantly located in the cytoplasm (Fig. 2E-F).

hsa_circ_101555 acts as an oncogene in HCC
To further investigate the regulatory role of hsa_circ_ 101555, we designed three hsa_circ_101555 small interfering RNAs (siRNAs) to specifically target different binding sites on the back-splice junction sequence of hsa_circ_101555. As si-hsa_circ_101555_001 and si-hsa_circ_101555_003 effectively silenced hsa_circ_101555 expression in HCCLM3 and HepG2 cell lines, they were used for subsequent experiments (Fig. 2G). We also generated stable hsa_circ_101555 overexpressing Huh-7 and SK-Hep-1 cells through lentiviral transduction and observed that lv-hsa_circ_101555 treatment caused marked upregulation of hsa_circ_101555 in these cells (Fig. 2H).
Subsequently, we detected the effect of hsa_circ_101555 knockdown and overexpression on HCC tumor progression in vitro. WST-1 assay results showed that hsa_circ_101555 silencing reduced HCCLM3 and HepG2 cell proliferation ( Fig. 3A and B), whereas its overexpression had the opposite effect in these cells ( Fig. 3C and D). Colony formation assays further revealed that hsa_circ_101555 was positively associated with HCCLM3 Fig. 1 circRNA expression profiles in HCC and paired nontumorous samples. A-C Volcano and scatter plots show variation in circRNA expression between HCC and nontumor liver tissues from the GSE7852 (A), GSE94508 (B), and GSE97322 (C) datasets. D Venn diagram. E hsa_circ_101555 consists of six exons (exon 1-6, 815 nt) and is derived from the host gene CSNK1G1. The existence of the back-splice junction was confirmed by Sanger sequencing. and HepG2 cell proliferation (Fig. 3E), whereas hsa_-circ_101555 overexpression had the opposite effect on Huh-7 and SK-Hep-1 cells (Fig. 3F). Notably, proliferating cell nuclear antigen levels in these cells were initially elevated, however, became attenuated following transfection with the hsa_circ_101555 back-splice junctionspecific siRNA (Supplementary Fig. 4A-B).
HCCLM3 and HepG2 cell migration and invasion were also suppressed by hsa_circ_101555 silencing (Fig. 3G and I), whereas these properties were enhanced by hsa_circ_101555 overexpression in Huh-7 and SK-Hep1 cells ( Fig. 3H and J). We further evaluated the effect that hsa_circ_101555 has on the expression of migration markers in HCC cell lines. Western blot analysis revealed that hsa_circ_101555 silencing inhibited the expression of fibronectin and vimentin, while that of Ecadherin was increased in HCCLM3 and HepG2 cells ( Supplementary Fig. 5A-B). Moreover, Hsa_circ_101555 overexpression significantly decreased the relative expression of E-cadherin, while increasing that of vimentin and fibronectin in both Huh-7 and SK-HepG1 cells ( Supplementary Fig. 5C-D). Taken together, these results suggest that hsa_circ_101555 was required to sustain the proliferation, migration, and invasion of HCC cells in vitro.
The RNA-binding protein EIF4A3 regulates hsa_circ_101555 expression EIF4A3, a core component of the exon junction complex, has been shown to play an essential role in pre-mRNA splicing 22 , while also inducing circRNA expression via binding to the upstream or downstream regions of host gene mRNA and inducing circular RNA formation [23][24][25] . In fact, we discovered four putative binding sites for EIF4A3 in the upstream and downstream region of the CSNK1G1 mRNA transcript (CSNK1G1 pre-mRNA) via CircInteractome (https:// circinteractome.nia.nih.gov/index.html) (Fig. 4A).
We conduced pull-down assay using CSNK1G1 mRNA and confirmed the enrichment of EIF4A3 protein in the pull-down of CSNK1G1 mRNA rather than control (Fig.  4B). The data from an RIP (RNA-binding protein immunoprecipitation) assay using anti-eIF4A3 antibody further indicated that eIF4A3 can bind CSNK1G1 mRNA through these four putative binding sites, which we designated as, a, b, c, and d, however, did not bind hsa_circ_101555 (designated e) in the corresponding RNA-protein complex (Fig. 4C). We then knocked down eIF4A3 and observed reduced hsa_circ_101555 expression and increased CSNK1G1 expression ( Fig. 4C and Supplementary Fig. 6A). Meanwhile, overexpression of eIF4A3 increased hsa_circ_101555 and decreased host gene CSNK1G1 expression ( Fig. 4D and Supplementary  Fig. 6B). In contrast, silencing hsa_circ_101555 did not affect EIF4A3 expression (Fig. 4E). Taken together, EIF4A3 was observed to increase hsa_circ_101555 expression, potentially by combining with the flanking sequences.

Identification of miRNAs that binding hsa_circ_101555
Given that circRNAs may act as competing endogenous RNAs (ceRNA) for miRNAs and regulate mRNA expression, we assessed the potential hsa_circ_101555 targets via a ceRNA-dependent mechanism. First, we determined the expression profiles of miRNAs from GSE115016 and GSE4187 datasets in HCC and normal tissue samples using miRNA microarray (Fig. 5A). Based on the target prediction tool, miRNAs were found to contain binding sites for hsa_circ_101555. Subsequent Venn analysis indicated the possible involvement of miR-145-5p in HCC (Fig. 5B). Therefore, we analyzed its expression in ten paired HCC and adjacent normal tissues by qRT-PCR and found it to be significantly downregulated in HCC tissues. Moreover, patients with low miR-145-5p level had a poorer 10-year overall survival (OV), compared to those with high miR-145-5p levels ( Fig. 5C and D). Pearson correlation analysis further revealed a negative correlation (p < 0.0001 r = −0.7364) between hsa_circ_101555 and miR-145-5p expression levels (Fig. 5E). Additionally, luciferase reporter assays using either a wild-type hsa_circ_101555 sequence or a mutated miR-145-5p binding sites, inserted into the 3′ untranslated region of Renilla luciferase, showed that miR-145-5p overexpression significantly reduced the luciferase activity of the wild-type reporter without impacting that the mutant sequence-bound reporter (see figure on previous page) Fig. 2 hsa_circ_101555 localizes in cytoplasm. A hsa_circ_101555 in HCC tumor and adjacent nontumor tissues detected by FISH. B Expression of hsa_circ_101555 in ten HCC tumor and adjacent normal tissues measured by qRT-PCR (***p < 0.001). C Expression of hsa_circ_101555 in multiple HCC cell lines (*P < 0.05; **P < 0.01; n = 4). D Linear 101555 is readily digested by RNase R, while hsa_circ_101555 is not (***P < 0.001; n = 4). E, F FISH confirmed that hsa_circ_101555 is predominantly located in cytoplasm. Nuclei stained with DAPI. U6, 18 S, and hsa_circ_101555 labeled with cy3 (n = 4). G Diagrammatic representation of siRNA targeting the back-splice junction (si-hsa_circ_101555-#1, si-hsa_circ_101555-#2, and si-hsa_circ_101555-#3); qRT-PCR results for hsa_circ_101555 expression in HCCLM3 and HepG2 cells treated, with or without siRNA. si-NC, control oligonucleotides with scramble sequence; si-hsa_circ_101555-#1, si-hsa_circ_101555-#2, and si-hsa_circ_101555-#3, oligonucleotides targeting the back-splice junction (***P < 0.001; n = 4). H qRT-PCR results for hsa_circ_101555 in Huh-7 and SK-Hep-1 cells treated with lv-NC or lv-hsa_circ_101555 (**P < 0.01; n = 4).

hsa_circ_101555 modulates HCC cell proliferation, migration, and invasion via miR-145-5p
We further investigated whether hsa_circ_101555 affects the function of HCC cells via miR-145-5p by determining its expression levels in vitro. qRT-PCR revealed that hsa_circ_101555 silencing increased miR-145-5p levels in HepG2 and HCCLM3 cells (Fig. 5H). We also found that the proliferation (via WST-1 assay) and colony-forming ability of HCC cells co-transfected with si-hsa_circ_101555_003 and miR-145-5p inhibitor were higher than those of HCC cells transfected with only siRNA-hsa_circ_101555, suggesting that the downregulation of miR-145-5p could partially reverse the proliferation-inhibitory effect of siRNA-hsa_circ_101555 ( Fig. 5I-L). Moreover, the decrease in HepG2 and HCCLM3 cell migration and invasion caused by hsa_circ_101555 silencing could be reversed by miR-145-5p inhibition ( Fig. 5M-N). Collectively, these results demonstrate that hsa_circ_101555 was required to sustain HCC cell progression and partly functions by impairing the tumor suppressor miR-145-5p.
We further detected CDCA3 mRNA expression in ten paired HCC and adjacent normal tissues by qRT-PCR to better understand its role and found it to be significantly upregulated in HCC tumors (Fig. 6H). Additionally, western blot and immunohistochemistry (IHC) analyses revealed that the protein expression of CDCA3 in HCC tissues was significantly upregulated as compared to their adjacent normal tissues ( Fig. 6I and J).

miR-145-5p directly targets the CDCA3 3ʹ-UTR
The miR-145-5p-CDCA3 interaction was confirmed via luciferase reporter assays, where in miR-145-5p significantly reduced the activity of the luciferase reporter compared to the negative control (wild-type CDCA3 sequence). However, such reduction was not observed following mutation of the miR-145-5p binding sites (Fig. 7A). These results indicate that miR-145-5p negatively regulated the expression of CDCA3.

hsa_circ_101555 is required to sustain HCC tumor growth in vivo
Considering that study demonstrated that hsa_circ_ 101555 knockdown suppressed the viability of HepG2 and HCCLM3 cells, we further examined its role in HCC tumorigenesis in vivo. Figure 8A provides a schematic diagram of tumor xenograft generation and subsequent treatment with cholesterol-modified hsa_circ_101555 shRNA or negative control shRNA. Body weight and tumor volume measurements, taken every 3 days for 33 days, revealed that mice receiving hsa_circ_101555 shRNA exhibited a marked decrease in tumor volume and weight compared to those receiving control shRNA (Fig. 8B-F). This suggested that hsa_circ_101555 promoted HCC cell growth in vivo. These results further supported a role for hsa_circ_101555 in HCC tumorigenesis and development.

Discussion
In the present study we identified 36 circRNAs with differential and consistent expression in GSE94508, GSE97322, and GSE78520 datasets. As hsa_circ_101555 was the only upregulated circRNA appearing in the three GSE datasets, it became the focus of further analyses. We characterized hsa_circ_101555 in HCC cells, and found that it consisted of six exons (exon 1-6, 815 nt) and was derived from the host gene CSNK1G1. Moreover, it was cyclized with the head-to-tail splicing of exon 1 and exon 6 and resisted RNase R treatment, demonstrating its stability compared to linear RNA.
Previous studies have shown that circRNAs play an essential role in cell cycle progression and proliferation [26][27][28][29] . hsa_circ_101555 was specifically reported as upregulated in tumor tissues and as associated with the prognosis of colorectal cancer patients, while its silencing significantly suppresses cell proliferation, induces apoptosis, and impairs the DNA repair capacity of CRC cells 30 . Herein, we found that hsa_circ_101555 was highly expressed in HCC cell lines (most markedly in HCCLM3) as well as patient tissues compared to adjacent nontumor tissues. More importantly, in a murine xenograft model, hsa_circ_101555 silencing significantly reduced HCC growth. We also provided evidence that the ectopic expression of hsa_circ_101555 is likely required to sustain cell proliferation. Meanwhile, we previously demonstrated that circMAST1 silencing inhibits HCC cell migration and invasion, which are important determinants of tumor metastasis 31 . Our results are consistent with those of previous studies that showed a regulatory role for circRNAs in cancer proliferation, migration, and invasion [32][33][34][35][36] . Thus, our research confirmed the stable role of hsa_circ_101555 in promoting HCC progression. Although the mechanisms through which circRNAs regulate carcinogenesis and cancer progression have not been fully elucidated, the "circRNA-miRNA-mRNA" axis, also known as the "miRNA sponge," is frequently cited 37 . In the present study, we confirmed that hsa_circ_101555 is an miR-145-5p sponge, evidenced by the significant increase in miR-145-5p expression following silencing of hsa_circ_101555, which in turn inhibited the proliferation, migration, and invasion of HCC cell lines. We also confirmed a direct correlation between miR-145-5p and hsa_circ_101555 expression. Consistent with our findings, several other studies have shown that circRNAs act as miRNA sponges during the development and progression of HCC. Hu et al. reported that circASAP1 acts as a ceRNA for miR-326 and miR-532-5p, which are tumor suppressors that regulating cancer cell proliferation, colony formation, migration, and invasion 38 . Further, many miRNAs have been shown to play critical roles in HCC initiation, development, and progression [39][40][41][42][43] . miR-145-5p is a tumor suppressor that is downregulated in several cancer types including glioma 44 , upper tract urothelial carcinoma 45 , and gastric cancer 46 . Consistent with these reports, our findings indicate that miR-145-5p serves as a tumor suppressor miRNA in HCC. As hsa_circ_101555 sponges miR-145-5p, the increased expression of hsa_circ_101555 in HCC cells leads to a (see figure on previous page) Fig. 6 Identification of CDCA3 as an miRNA target oncogenes. A Heatmap analyses demonstrating differentially expressed mRNAs in HCC with | log2fold-change | > 1.0 and P < 0.05. B, C Venn analysis between predicted mRNA genes targeted by miR-145-5p and differentially expressed mRNAs identified four mRNAs, NTN4, CDCA3, SLC25A25, and SLC1A2 involved in HCC. D-G CDCA3 is significantly upregulated in HCC tumors and is associated with poorer OS. H Expression of CDCA3 between ten HCC tumor tissues and adjacent normal tissues measured by qRT-PCR (***P < 0.001, n = 10). I Expression of CDCA3 between seven HCC tumor and adjacent normal tissues measured by western blot (**P < 0.01; n = 7). J CDCA3 protein expression in HCC and corresponding adjacent normal tissues detected by IHC (n = 3). K Pearson correlation analysis demonstrating correlation between miR-145-5p and CDCA3 expression (n = 20). L Pearson correlation analysis showing the correlation between hsa-circ-101555 and CDCA3 expression (n = 20).
The role of miRNA sponging in tumor progression has previously been described 47 . We confirmed, via bioinformatics and luciferase reporter gene analyses, that hsa_circ_101555-miR-145-5p targets the oncogene CDCA3. Although several studies have implicated CDCA3 in the regulation of cancer development and progression [48][49][50] , to our knowledge, this is the first study to confirm CDCA3 expression in liver cancer tissues and report its positive correlation with hsa_circ_101555 expression and negative correlation with miR-145-5p expression. Meanwhile, a previous study in colorectal cancer demonstrated that miR-145-5p suppressed proliferation, metastasis, and epithelial-mesenchymal transition by targeting CDCA3 51 . Our results were consistent with these reports. In addition, we found that the miR-145-5p/CDCA3 axis is regulated by hsa_circ_101555 via a sponging mechanism. Further, we demonstrated that miR-145-5p suppression promotes CDCA3 expression, which in turn increases the proliferation, migration, and invasion of HCC cells. The simultaneous inhibition of miR-145-5p and CDCA3 attenuates the tumorigenic features of HCC cells to a greater extent than miR-145-5p inhibition alone. To our knowledge, our study is the first to demonstrate hsa_circ_101555 involvement in regulation of CDCA3 expression. Moreover, we demonstrated that hsa_circ_101555 silencing significantly reduces CDCA3 expression in vivo. These findings suggest that hsa_circ_101555 protects CDCA3 from miR-145-5pmediated degradation in a ceRNA-mediated manner.
The biogenesis of circRNAs is regulated by specific cisacting elements and trans-acting factors. It has been shown that certain RNA-binding proteins promote cir-cRNA expression 52 . RNA-binding protein EIF4A3 is the core component of exon junction complex (EJC), which is considered as an important regulator of posttranscriptional regulation processes including mRNA splicing, transport, translation, and surveillance 53 . Through bioinformatic analysis and experiments, we predicted and screened that EIF4A3 could bind to a flanking sequence of hsa_circ_101555. Our research reveals that EIF4A3-mediated reverse splicing of exons can be a potential mechanism to induce high expression of hsa_circ_101555.
We acknowledge that our research has certain limitations. Although this study clarifies that hsa_circ_101555 functions as a sponge of miR-145-5p to promote CDCA3induced HCC cancer cell proliferation and invasion, cir-cRNAs may regulate the development and progression of HCC via other mechanisms. For example, circRNAs have been shown to regulate parental gene expression and the expression of peptides/proteins in other cancers 36,54 . Thus, additional research is required to further explore the role of hsa_circ_101555 in HCC. In terms of clinical diagnosis and treatment, we additional studies are required, including expanding the sample size and expression stability of hsa_circ_101555 in the peripheral blood of HCC patients, as well as evaluating the initiation of its high expression among the stages of HCC.

Conclusions
Our study demonstrated that hsa_circ_101555 is upregulated in HCC cell lines and patient tissues, and its high expression is associated with HCC progression. Moreover, hsa_circ_101555 functions as a tumor promotor and is required to sustain the proliferation and invasion of HCC by directly binding to miR-145-5p and impeding its suppression of CDCA3. Furthermore, we also demonstrate that EIF4A3 could mediate the biogenesis of hsa_circ_101555, (see figure on previous page) Fig. 8 hsa_circ_101555 promotes HCC tumor growth in vivo. A Schematic diagram shows that HCCLM3 cells were injected subcutaneously into BALB/c nude mice for 15 days to create tumor xenografts. After 15 days, 10 nmol cholesterol-modified hsa_circ_101555 shRNA or negative control shRNA was subcutaneously injected at the tumor site every 3 days for 18 days. B Representative images of HCC tumor bearing BALB/c nude mice. C Body weight of xenograft nude mice. D Body weight measured every 3 days. E Tumor volumes measured every 3 days (*P < 0.05; **P < 0.01; ***P < 0.001; n = 4). F Relative weights of tumors (***P < 0.001; n = 4). G IHC staining and IHC analysis of Ki-67 expression in subcutaneous xenograft tumors. Knockdown of hsa_circ_101555 downregulates Ki-67 expression (*P < 0.05; n = 4). H-J Expression of hsa_circ_101555, miR-145-5p and CDCA3 in subcutaneous xenograft tumors (**P < 0.01; ***P < 0.001; n = 4). K Protein expression level of CDCA3 in seven paired HCC and normal tissues (**P < 0.01; n = 4). L IHC staining and IHC analysis of CDCA3 expression in subcutaneous xenograft tumors. Knockdown of hsa_circ_101555 downregulates CDCA3 expression (**P < 0.01; n = 4). M Schematic diagram shows that hsa_circ_101555 promotes HCC cells proliferation, migration, and invasion through the miR-145-5p/CDCA3 axis. but the detailed mechanism needs further study. Based on its role in regulating the miR-145-5p/CDCA3 axis, our findings suggest that hsa_circ_101555 may represent a potential novel biomarker and therapeutic target for HCC.

Cell lines and clinical tissues
A total of 38 HCC samples were obtained from the clinical sample bank of the First Affiliated Hospital, Harbin medical university. The collection of human specimens was approved by the Biomedical Ethics Committee of the Harbin medical university First Affiliated Hospital and written informed consent was obtained from each patient (HMUDQ20200010202). Inclusion criteria for patient selection was curative hepatectomy performed between 2017 and 2018. All patients were pathologically diagnosed with hepatocellular carcinoma and liver specimens were evaluated by pathologists to determine clinical staging according to the TNM classification. HCC patients with the following conditions were excluded: (1) patients ≤18 or ≥70 years of age or without full civil capacity; (2) patients with a history of preoperative anticancer radiotherapy or chemotherapy, biological, immune, or traditional Chinese medicine administration; (3) patients with incomplete postoperative follow-up data; (4) patients with a history of another organ malignancy, or systemic immune disease. All specimens were collected within 15 min of removal from the body and were immediately snap-frozen in liquid nitrogen before storage at −80°C. Ten paired samples were used to compare the expression levels of the genes of interest between HCC and paired nontumorous tissues. The detailed clinicopathological features, as well as the correlations between hsa_circ_101555 expression and the clinical characteristics, are described in Additional files 8-9: Tables S1-S2.
Cell lines used in this study (HepG2, HCCLM3, and SK-Hep1) were purchased from the Cell Bank of Type Culture Collection (Chinese Academy of Sciences, Shanghai, China). Huh-7 and L02 were purchased from the Procell (Wuhan, China). All cells were cultured in DMEM/high glucose medium (Hyclone, Logan, UT, USA) supplemented with 10% fetal bovine serum (PAN-Biotek, Aidenbach, Bavaria) and 1% penicillin-streptomycin (Hyclone) in a humidified atmosphere at 37°C containing 5% CO 2 .

Xenograft nude mouse model
Six-week-old male BALB/C nude mice purchased from Vital river (Beijing, China) were maintained under specific pathogen-free conditions with a 12-h light/dark cycle. All animal experiments were performed in accordance with a statement of compliance with ethical regulations and approved by the Biomedical Ethics Committee of the Harbin medical university. Animals are grouped randomly during the experiment. HCCLM3 cells were subcutaneously injected into the right upper back of the nude mice (1 × 10 6 cells per mouse) for 15 days, after subcutaneous incubation of HCC tumor mass. Next, 10 nmol cholesterol-modified hsa_circ_101555 shRNA or control shRNA RiboBio (Guangzhou, China) were intratumorally injected every 3 days for 18 days. the last day of the injections, which would account for day 33 of the experiment, the mice were sacrificed and tumor tissues were collected for examination.

Circular structure confirmation
The circular structure of hsa_circ_101555 was confirmed by Sanger sequencing, divergent primer PCR and RNase R treatment. PCR products, amplified by divergent primers of hsa_circ_101555, were inserted into the T vector and delivered to SinoGENE (Beijing, China) for Sanger sequencing. The results were crosschecked with the back-spliced region of hsa_circ_101555 supplied by circBASE 55 . Total RNA was extracted from HCC tissue using Trizol solution. For RNase R treatment, 3 μg of total RNA extracted from HCC was incubated with 10 U RNase R (20 U/μL, Epicentre, Madison, WI, USA) in a 10 μL total volume at 37°C for 45 min, followed by incubation at 70°C for 10 min to deactivate RNase R. The treated RNAs were used for qRT-PCR 27 .
Quantitative real-time polymerase chain reaction, western blotting, and immunofluorescence analyses Total RNA was extracted from HCC cell lines and tissue using Trizol solution, and complement DNA was generated using the Golden 1st cDNA Synthesis kit (Haigene, China) following RNA quantification. qRT-PCR assays were performed using Power SYBB Green PCR Master Mix (Life Technologies, Carlsbad, CA, USA). The cir-cRNA and gene expression levels were normalized to that of GADPH, while miRNA expression levels were normalized to that of U6. Each sample was tested in triplicate. The relative expression was analyzed by the comparative cycle threshold (Ct) method, according to the equation 2 −ΔΔCt [Δ Ct = Ct-Ct (GAPDH)]. The primer sequences of hsa_circ_101555 and U6 were designed by RiboBio (Guangzhou, China). The linear101555, CDCA3, miR-145-5p, and GAPDH sequences were designed by Genscript (Nanjing, China). All experiments were performed in triplicate. The primers used in this study are listed in Additional file 10: Table S3. RNase R digestion was performed using RNase R purchased from Epicentre Biotechnologies (Madison, WI, USA) as per the manufacturer's instructions. Briefly, 1 µg of extracted RNA was treated with 3 U of RNase R (Epicenter, USA) and incubated at 37°C for 1 h, after which the RNA was purified by phenol-chloroform extraction and subjected to qRT-PCR.
For western blotting, the total protein extracts from cells were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred onto polyvinylidene difluoride membranes, and incubated with the corresponding antibodies. The membranes were developed using the enhanced chemiluminescence method (Haigene, China). All antibodies used in this study are listed in Additional file 11: Table S4.
Fresh samples were cut to an appropriate size and fixed in 4% paraformaldehyde for 24 h. The fixed specimens were dehydrated in a graded series of ethanol solutions, embedded in paraffin and cut at a thickness of 4 μm. The sections were dewaxed and rehydrated using xylene and ethanol, and high-pressure heat was applied for antigen retrieval. The sections were incubated with the primary antibodies overnight at 4°C. Finally, all sections were dehydrated, cleared, mounted, and visualized with a diaminobenzidine-based colorimetric method. The antibodies used in this study are listed in Additional file 11: Table S4.

Fluorescence in situ hybridization
In situ hybridization was performed with a FISH kit (RiboBio, Guangzhou, China). Cells, frozen HCC sections, and paired adjacent liver tissues were briefly rinsed in PBS and fixed in 4% formaldehyde for 10 min. The cells were then permeabilized in PBS containing 0.5% Triton X 100 at 4°C for 5 min, washed with PBS three times for 5 min, and prehybridized at 37°C for 30 min before hybridization. Next, an anti-hsa_circ_101555, anti-U6, or anti-18S oligodeoxynucleotide probe (RiboBio, Guangzhou, China) was used in the hybridization solution at 37°C overnight in the dark. The next day, cells were counterstained with DAPI and imaged using a NA1.4 inverted Leica DMI6000 microscope (Leica, Heidelberg, Germany). Images were captured using a Hamamatsu ORCA-R2 camera (Hamamatsu, Japan) and recorded using LAS AF software (Leica). The experiments were conducted in triplicate.

Cell proliferation, migration, and Matrigel invasion assay
Cell proliferation was assessed using the WST-1 assay (Beyotime Biotechnology, Nantong, China). Cells (2 × 10 3 ) were seeded into each well of 96-well plates and 10 μL of WST-1 solution was added to each well at four timepoints (0, 24, 48, and 72 h). After 4 h of incubation at 37°C, the absorbance at 450 nM was measured using a Spectra Max 250 spectrophotometer (Molecular Devices, Sunnyvale, CA, USA). Experiments were independently performed in triplicate.
For the colony formation assays, cells (1 × 10 2 ) were suspended and plated into each well of 6-well plates. After 14 days incubation at 37°C in a chamber with an atmosphere of 5% CO 2 , colonies were fixed with 1 mL of 4% paraformaldehyde (Solarbio, Beijing, China) for 30 min and stained with crystal violet (Beyotime Biotech-nology, Nantong, China) for 25 min. Colonies were then counted after photographing the sample (Nikon, Tokyo, Japan).
Cell migration and invasion were measured using a transwell migration assay and a Matrigel invasion assay. For the transwell migration assay, 2-4 × 10 5 cells were suspended in 200 µL of DMEM without serum and placed in the cell culture insert (8 µm pore size; BD Falcon, San Jose, CA) of a companion plate (BD Falcon) with prewarmed culture medium containing 20% fetal bovine serum. The cells were incubated for 24 h at 37°C in 5% CO 2 and subsequently fixed with 4% paraformaldehyde in PBS.
For the Matrigel invasion assay, 2-4 × 10 5 cells were suspended in 200 µL of DMEM without serum and were placed in the cell culture insert precoated with 50 µL Matrigel (BD Biosciences, San Jose, CA, USA). A prewarmed culture medium containing 20% fetal bovine serum was added to the well. The cells were incubated for 48 h at 37°C in 5% CO 2 and were then fixed with 4% paraformaldehyde in PBS. The nonmigrated or invaded cells on the top of the membrane were gently removed with a cotton swab. Cell migration or invasion was determined by staining cells with 0.1% crystal violet (Sigma, St Louis, MO) and the cells were counted under a light microscope (×200 magnification) in eight randomly selected areas.
Transfection experiment siRNA specific for hsa_circ_101555 was synthesized by RiboBio (Guangzhou, China), while the inhibitor of the miR-145-5p mimics and negative control, miRNA-145-5p, siRNA of CDCA3 were synthesized by Gene Pharma (Shanghai, China). HepG2 and HCCLM3 cells were transfected with siRNA of hsa_circ_101555 or Huh-7 and SK-hep1 cells were transfected with lv-hsa_circ_101555 using the Lipofectamine 2000® siRNA transfection reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer's protocol. The target sequences of siRNAs are listed in Additional file 12: Table S5.

Dual luciferase assay
Targeted binding of hsa_circ_101555 to miR-145-5p was predicted using bioinformatics websites, including Cir-cInteractome, and miRanda; whereas targeted binding of miR-145-5p to CDCA3 was predicted using TargetScan, and miRDB. The full-length sequences of hsa_circ_101555, with and without mutated predicted miR-145-5p binding sites, were subcloned into pmirGLO reporter vector (Promega, WI, USA). The full-length sequences of CDCA3, with and without mutated predicted miR-145-5p binding sites, were subcloned into pmirGLO reporter vector (Promega, WI, USA). Lipo2000 was then used for transfection of the vectors into 293 T cells. Finally, luciferase activity was measured using the dual luciferase assay kit.

RNA pull-down
RNA pull-down technology used desulphurized biotinlabeled RNA of CSNK1G1 mRNA and streptavidinlabeled magnetic beads to efficiently enrich and identify the RNA-binding protein, EIF4A3. RNA probes were labeled with biotin by in vitro transcription, and then incubated with cytoplasmic protein extract to form RNAprotein complexes. This complex was then allowed to bind streptavidin-labeled magnetic beads to separate them from the other components in the solution. After elution, western blotting was used to detect the binding proteins of EIF4A3 that interacted with CSNK1G1 mRNA.

RIP assay
RIP (RNA-binding protein immunoprecipitation) assay was performed with the BersinBio™ RNA Immunoprecipitation kit (Guangzhou, China). Cells were collected and cultivated in RIP lysis buffer, followed by immunoprecipitation with EIF4A3 antibody (Abcam, MA, USA). The final retrieved RNA was subjected to quantitative realtime PCR analysis. Normal mouse immunoglobulin G (IgG) served as negative controls.

Statistical analyses
All statistical analyses were performed using SPSS version 21.0 (IBM SPSS Inc., Chicago, IL, USA) and GraphPad Prism version 6.0 (GraphPad Software, La Jolla, CA, USA) software. Categorical variables are expressed as a count or percentage and tested using Chi-squared or Fisher's exact tests, as appropriate. Continuous data are reported as mean ± standard deviation (SD) and compared using Student's t-test, one-way analysis of variance (ANOVA) test, or Mann-Whitney U-test, as appropriate. Correlations were calculated using Pearson correlation analysis. P < 0.05 was considered statistically significant.