Transcription factor FOXC2 regulates miR-145/ADAMTS5 axis to inhibit angiogenesis in hepatocellular carcinoma via circular RNA 0002898

Abstract

Background

Circular RNAs (circRNAs), a novel class of single-stranded, covalently closed non-coding RNA molecules, have been increasingly recognized for their roles in the cellular landscape over the past decade. These molecules are predominantly localized within the cytoplasm of eukaryotic cells and have been implicated in the pathogenesis of a spectrum of cancers. Despite this, the specific contributions of circRNAs to hepatocellular carcinoma (HCC) have yet to be fully delineated.

Methods

Our research has identified that cyclic RNA0002898 is downregulated in HCC, with its reduced expression correlating with advanced tumor grade and diminished patient survival outcomes. Furthermore, we have demonstrated that the transcription factor forkhead box C2 (FOXC2) regulates the expression of cyclic RNA0002898, and diminished levels of cyclic RNA0002898 in HCC are associated with enhanced tumor cell proliferation, migration, and invasion in vitro. Mechanistic insights revealed that cyclic RNA0002898 could modulate the expression levels of the tumor suppressor a disintegrin and metalloproteinase with thrombospondin motif 5 (ADAMTS5) by antagonizing miR-145, thereby inhibiting angiogenesis and suppressing the growth of HCC cells.

Results

Our findings collectively elucidate the regulatory role, functional significance, and underlying mechanism of cyclic RNA0002898 in HCC, a previously uncharted relationship. The potential prognostic implications of cyclic RNA0002898 and its therapeutic potential as a target in HCC warrant further investigation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-15171-z.

Introduction

Circular RNAs (circRNAs) represent a distinct subclass of non-coding RNAs, distinguished by their covalently closed, single-stranded circular conformation, which is devoid of the 5’ cap and 3’ polyadenylated tail that are hallmarks of linear RNA molecules [1]. These entities are pervasively detected within the cytoplasm of eukaryotic cells [2]. The circular architecture of circRNAs confers a heightened stability, shielding them from degradation by ribonucleases present in the blood [3]. CircRNAs are implicated in a spectrum of biological processes, including the modulation of splicing and transcription, the sequestration of proteins, and the facilitation of RNA transport [4]. An in-depth analysis of the fundamental roles of circRNAs, in conjunction with their dynamic interplay during physiological and pathological progressions, reveals a significant correlation between these molecules and cellular attributes such as proliferative capacity, cell cycle regulation, apoptotic thresholds, and autophagic processes [5]. This association implicates circRNAs as potentially pivotal players in the etiology and pathogenesis of various diseases, suggesting their potential as biomarkers or therapeutic targets in medicine.

The circRNAs often participate in biological processes through Competing Endogenous RNA (ceRNA) networks. These networks primarily encompass circRNAs, long non-coding RNAs (lncRNAs), pseudogenes, and protein-coding RNAs, which can serve as microRNA Response Elements (MREs) [6]. These molecular entities engage in competitive binding interactions with a shared pool of microRNAs (miRNAs), modulating their expression profiles. Among them, as crucial components of the ceRNA network, circRNAs can act as miRNA sponges, sequestering many miRNAs that correspond to them, and thus post-transcriptionally regulate the expression of target genes, modulating the initiation and progression of biological programs [7].

In recent years, the significance of circRNAs in the etiology of malignant tumors has garnered considerable interest [8]. Many circRNAs exhibit aberrant expression profiles within neoplastic tissues, implicating their role in the genesis and progression of various cancers [9–11]. This includes, but is not limited to, gastric, pulmonary, pancreatic, and colorectal malignancies, etc. Hepatocellular carcinoma (HCC), among the most prevalent neoplasms, has been the focal point of numerous investigations. A wealth of evidence has demonstrated that circRNAs are intricately woven into the complex tapestry of HCC pathogenesis, exerting a profound influence on critical oncogenic processes such as cell proliferation, migration, and invasion [12, 13]. Circular RNA cSMARCA5 promotes TIMP3 expression and suppresses HCC proliferation and metastasis via miR-17-3p and miR-181b-5p sequestration [14]. CircRHOT1, by recruiting TIP60 to the NR2F6 promoter, enhances the transcription of NR2F6, thereby augmenting the proliferation, migration, and invasive capabilities of HCC [15].

Circular RNAs exert regulatory functions in HCC. However, their roles, regulations, and underlying mechanisms remain largely elusive. HCC ranks sixth among the most common malignant tumors worldwide, with approximately 745,000 deaths annually, making it the second leading cause of cancer-related mortality globally [16]. Therefore, there is an urgent need to elucidate the mechanisms underlying HCC progression. In this study, we discovered a significant downregulation of the circular RNA0002898 in HCC, where diminished expression levels were significantly associated with advanced tumor stages and an unfavorable clinical outcome. Our research revealed that the expression of circular RNA0002898 is modulated by the transcription factor forkhead box C2 (FOXC2), which exhibits a similar downregulation pattern in HCC. Furthermore, we demonstrated that circular RNA0002898 could interact with miR-145, thereby enhancing the expression of the tumor suppressor a disintegrin and metalloproteinase with thrombospondin motif 5 (ADAMTS5). This interaction is crucial as it leads to the suppression of tumor angiogenesis, a critical process in HCC progression, culminating in inhibiting HCC tumorigenesis. These findings suggest that circular RNA0002898 may suppress HCC through the miR-145/ADAMTS5 axis, indicating its potential as a therapeutic target for HCC treatment.

Materials and methods

Ethical approval

All human tissues and clinical data used in this study were approved by the Zhejiang Provincial People’s Hospital Ethics Review Committee (Zhejiang Provincial People’s Hospital Ethics Review No. 247, 2024), and written consent was obtained from the patients. Patients’ names were anonymized according to ethical and legal standards.

The protocols related to animal testing conducted in this research have been reviewed and approved by the Animal Management and Use Committee of Zhejiang Provincial People’s Hospital. They comply with the “Regulations on the Administration of Laboratory Animals,” the “Guiding Opinions on the Humane Treatment of Laboratory Animals,” the “Ethical Review Guidelines for Laboratory Animal Welfare,” as well as the relevant regulations concerning laboratory animals at Zhejiang Provincial People’s Hospital and are permitted for implementation.

Clinical material and patient inclusion criteria

HCC and para-cancer tissues of 80 patients who underwent curative hepatectomy or ultrasound-guided targeted biopsy at Zhejiang Provincial People’s Hospital from August 2014 to April 2017. The inclusion criteria were as follows: (1) Pathological results confirmed the patient’s diagnosis as HCC. (2) Patients had not received systemic therapy for HCC. (3) Liver tumor tissues and corresponding non-tumorous liver tissues should be paired. The exclusion criteria were as follows: (1) The patient has a history of liver surgery. (2) The pathological results and Gleason score are incomplete. (3) There is severe inflammation or necrosis present in the tissue samples. (4) The patient has other significant systemic diseases.

Cell culture

SNU-182 and SNU-449 cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco, America, 11875093) containing 10% fetal bovine serum (Gibco, America, 10100147) and Pen-Strep (Beyotime, China, C0222). HL-7702, PLC/PRF/5, MHCC97H, HepG2, Hep3B, and Huh7 cells were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM, Biological Industries, Israel, 01–055−1 A) with 10% fetal bovine serum and Pen-Strep. HUVEC cells are cultured in a specialized medium (Pricella, China, CM-0122). All cells were incubated at 37 °C in 5% CO₂.

Construction of vectors

The entire coding sequences of human ADAMTS5 and FOXC2 genes were meticulously synthesized and cloned into the pcDNA3.1 vector by Tsingke Biotechnology (Wuhan, China). Additionally, Tsingke Biotechnology synthesized the wild-type (WT) or mutated (Mut) sequences of the 3’ untranslated region (UTR) of circ0002898 and ADAMTS5, which correspond to the binding sites for miR-145. The BMP-Luc reporter plasmid, containing the wild-type full-length DNAJC16 promoter, was procured from Tsingke Biotechnology. Furthermore, the QuikChange Site-Directed Mutagenesis Kit (Agilent, America, 200519) was adeptly employed to introduce mutations within the FOXC2 binding site of the full-length DNAJC16 promoter region.

Construction of cell lines

We utilized a lentiviral approach to generate cell lines with knockdown and overexpression of circ0002898, FOXC2, and ADAMTS5. Complementary DNA nucleotides of circ0002898, FOXC2, and scrambled control siRNA were synthesized, annealed, and cloned into the pLKO.1 plasmid will be used to construct the hairpin shRNA expression cassette by Tsingke Biotechnology. The circ0002898 sequence was synthesized by Tsingke Biotechnology and inserted into the pLV-circ-puro plasmid. Concurrently, the transcription factor FOXC2 and ADAMTS5 sequences were inserted into the FUGW plasmid. We employed 293 T cells to produce lentiviral particles [17]. Briefly, when the cells reached approximately 80% confluence in 2-cm dishes, 293 T cells were cotransfected with 1 µg of pMD2.G, 1.5 µg of psPAX2, and 2 µg of the target plasmid using Lipofectamine 3000 (Invitrogen, America, L3000015) according to the manufacturer’s protocol. The culture medium was refreshed after an 8-hour incubation, and the supernatants containing viral particles were collected by centrifugation following a 48-hour cultivation period. The sh-RNA sequences used in this paper are presented in Supplemental Table 1.

Polymerase chain reaction

We employed PCR Master Mix (Thermo, America, K0171) to perform polymerase chain reaction. The PCR primers used throughout the article are presented in Supplemental Table 1.

Quantitative real-time PCR

Total RNA was extracted employing an RNA isolation kit (Promega, USA, LS1040). Following the manufacturer’s protocol, 1 µg of the isolated RNA was reverse transcribed into complementary DNA (cDNA) utilizing the SuperScript III First-Strand Synthesis System (Thermo, America, 18080051). Quantitative real-time polymerase chain reaction (qRT-PCR) analyses were executed with the AceQ qPCR SYBR Green Master Mix (Vazyme, China, Q111–02). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and U6 small nuclear RNA were employed as endogenous controls for circRNA and miRNA expression levels, respectively. The relative expression levels were determined by the 2^−ΔΔCt comparative threshold cycle method. The primer sequences utilized in the qRT-PCR are delineated in Supplementary Table 1.

Cell proliferation assay

1 × 10⁴ cells, resuspended in a 100 µL aliquot of culture medium, were seeded into individual wells of a 96-well microplate. Cell proliferation was assessed using the Cell Counting Kit-8 assay (CCK-8) (Beyotime, China, C0038).

Colony formation assay

Cells were inoculated into a 2 cm diameter culture plate at a seeding density of 1,000 cells per well and cultivated at 37 °C in an atmosphere containing 5% CO2 for 13 days. Subsequently, the cells were rinsed with phosphate-buffered saline (PBS) (Beyotime, China, C0221A), immobilized with a 4% paraformaldehyde solution (Beyotime, China, P0099-100 mL) for 20 min, and stained with a 0.1% solution of crystal violet (Beyotime, China, C0121-100 mL) for an additional 20 min.

Cell migration and invasion assays

The invasive and migratory capabilities of the cells were assessed using the Transwell chamber method [18]. The experiment utilized 24-well chambers with a pore size of 8.0 μm (SPL Life Sciences, Korea, 36224). The presence of a Matrigel layer (BD, America, 356234) indicated an invasion assay, while its absence denoted a migration assay. For the migration assay, 2 × 10⁵ cells were seeded in the upper compartment of the chamber, whereas for the invasion assay, 3 × 10⁵ cells were seeded. The lower chamber was filled with a medium containing 20% fetal bovine serum. The cells were subjected to a 36-hour incubation period for the migration assay and 48 h for the invasion assay. Subsequently, the cells adhering to the upper aspect of the membrane were carefully wiped off using a cotton-tipped applicator. The cells on the underside of the membrane were then treated with a 4% paraformaldehyde solution for fixation, followed by staining with a 0.1% solution of crystal violet. Images were captured using a microscope. The cells that had migrated or invaded were counted in three independent experiments.

Chromatin Immunoprecipitation (ChIP) assay

The Chromatin immunoprecipitation (ChIP) assay was conducted as previously reported via a ChIP assay kit (Bersin Bio, China, Bes5001) [19]. Briefly, 2 × 10⁷ cells were rinsed twice using phosphate-buffered saline (PBS) and centrifuged to pellet the cells. Crosslinking of DNA and proteins was performed using a 1% formaldehyde solution at ambient temperature. After that, the cells were subjected to washing and lysis using a buffer solution composed of 50 mM HEPES-KOH at pH 7.5, 140 mM sodium chloride, 1 mM EDTA at pH 8, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% sodium dodecyl sulfate, and a cocktail of protease inhibitors to facilitate extraction of nuclear contents. Chromosomal DNA was fragmented through sonication and immunoprecipitated using a specific anti-FOXC2 antibody from Abcam (Britain, ab5060). The crosslinking was reversed by incubating the samples overnight at 65 °C. Ultimately, the immunoprecipitated DNA was isolated, and its quantity was determined by employing quantitative real-time polymerase chain reaction (qRT-PCR) and PCR methodologies. The oligonucleotide primer sequences utilized are detailed in Supplementary Table 1.

RNA binding protein Immunoprecipitation (RIP) assay

We used the Magna RNA-Binding Protein Immunoprecipitation (RIP) Kit (Sigma, USA, 17–700) to perform RIP assays according to the manufacturer’s protocol [20]. Cells were lysed in RIP lysis buffer supplemented with protease and RNase inhibitors. The lysate was incubated with magnetic beads conjugated with AGO2 (CST, USA, 2897) or IgG (CST, USA, 2729 S) antibodies at 4 °C. After washing, proteinase K treatment was applied to digest proteins. RNA was subsequently isolated and analyzed by qRT-PCR and agarose gel electrophoresis.

Biotin-coupled probe pull-down assay

We performed a biotinylated RNA pull-down assay, utilizing a protocol detailed in the kit [21] (Beyotime, China, P0654S). To summarize, we prepared 5’-biotinylated miR-145, a biotinylated DNA oligonucleotide that is antisense to the splice junction of circular RNA0002898, and 5’-biotinylated non-specific single-stranded RNA or DNA as a negative control, all procured from Tsingke Biotechnology, China. These probes were reconstituted in a 500 µL volume of binding buffer and incubated with streptavidin-coated agarose beads at 25 °C for 2 h. Subsequently, the beads, now coated with probes, were incubated with cell lysate (Beyotime, China, P0013C) for 2 h and then subjected to a series of washes using a washing buffer. The RNA complexes bound to the beads were eluted and extracted using TRIzol reagent (Invitrogen, USA, 15596026). The isolated RNA was employed as a template for quantitatively assessing miR-145 and circular RNA0002898 expression levels using quantitative real-time PCR (qRT-PCR). The sequences of the oligonucleotides are provided in Supplementary Table 1.

Dual-luciferase reporter assay

Cells were inoculated into 24-well plates at a seeding density of 3 × 10⁴ cells per well and allowed to grow for 8 h, after which they were subjected to transfection with a luciferase reporter plasmid. After a 48-hour interval, the cells were collected to assess luciferase activity. The manufacturer’s instructions present further information and specific procedural details [22] (Promega, USA, E1910).

RNA in situ hybridization (RNA-ISH)

Outdo Biotech (Shanghai, China) performed the RNA-ISH detection for circ0002898 in liver cancer and the corresponding para-carcinoma tissues. The reagents required for this experiment are included in the custom-made kit (Boster, China, MK11303). To summarize, after deparaffinization and hydration, the tissue sections were treated with enzymatic digestion using proteinase K, followed by post-fixation in a 4% paraformaldehyde solution. After that, the sections were hybridized with a specific probe overnight. Then, the sections were incubated with an anti-digoxin monoclonal antibody (mAb) at 4 °C overnight. The immunoreactivity was then visualized using a diaminobenzidine (DAB) chromogenic substrate. The intensity of staining was quantified and categorized as an expression level, detected by multiplying the product of the staining intensity score (0 for no staining, 0.5–1 for weak, and 1.5–2 for strong staining) and the percentage of positively stained cells (0 for none, 1 for 1–24%, 2 for 25–49%, 3 for 50–74%, and 4 for 75–100% stained cells). The in-situ hybridization (ISH) scores below 5 are low expression, whereas ISH scores of 5 or above are considered high expression.

Endothelial cells tube formation assay

We utilized Matrigel (BD, USA, 356234) to facilitate the formation of tubular structures by endothelial cells. Herein, we provide a brief description of the methodology [23]. 50 µL of Matrigel was added to 48-well plates and incubated at 37 °C for 30 min to solidify the Matrigel layer. The supernatant from Hep3B or Huh7 cell cultures (after a 24-hour incubation) was collected and mixed with HUVEC-specific culture medium (Biospes, China, BK1147) to achieve concentrations of 100%, 50%, or 25%. Depending on experimental requirements, 20 ng/mL of VEGF protein (Abcam, USA, Catalog No. ab259412) was added to the mentioned culture medium. Each well was seeded with 4 × 10⁴ cells and incubated at 37 °C for 12 h. Images were captured using a microscope (Thermo, USA, EVOS M5000). Quantification of the lumen formation assay was performed using the sum of tubular structure lengths by ImageJ software. The total length in the control samples was normalized to 100% to evaluate the effects of the treatment groups on lumen formation.

Apoptosis assay

Collect the supernatant from Hep3B or Huh7 cell culture medium (after a 24-hour incubation) and mix it with HUVEC-specific culture medium (Biospes, China, BK1147) to create concentrations of 100%, 50%, or 25%. According to experimental requirements, Z-VAD-FMK (Sigma, USA, Catalog No. 627610) is added. HUVEC cells are seeded at 1 × 10⁴ cells per well in a Matrigel-coated 96-well plate and cultured for 24 h. Apoptosis is detected using an ELISA kit (Roche, Switzerland, 11544675001).

In vivo tumor growth assays

In brief, 2 × 10⁶ cells were subcutaneously implanted into six-week-old male BALB/c nude mice. Tumor sizes were measured on the designated dates of the experiment. At the final stage of the experiment, or when the tumor volume exceeded 2000 mm³, to ensure the welfare of the mice, we resorted to carbon dioxide asphyxiation for humane euthanasia. Before this method, mice were administered appropriate anesthetics to ensure a pain-free procedure.

Statistical analysis

Statistical analyses were conducted utilizing GraphPad Prism version 9 software. Group-wise disparities were statistically evaluated by employing Student’s t-test for pairwise comparisons, one-way or two-way analysis of variance (ANOVA) was performed, followed by Tukey-Kramer tests for multiple group comparisons. The chi-square test was used for categorical data. Associations among different groups were measured using Pearson’s correlation coefficient. The diagnostic performance was evaluated by analyzing the receiver operating characteristic (ROC) curve. Survival data were analyzed using the Kaplan-Meier estimator, with the log-rank test employed to determine the significance of observed differences in survival distributions. Results are presented as the mean ± standard deviation (SD). A p-value of less than 0.05 was considered to denote statistical significance.

Results

Down-regulation of circ0002898 in HCC is associated with poor prognosis and survival

To ascertain the impact of circ0002898 on HCC, we examined the expression levels of circ0002898 in HL7702, SNU-182, SNU-449, PLC/PRF/5, MHCC97H, HepG2, Hep3B, and Huh7 cell lines. Our findings indicated that the expression of circ0002898 was higher in the normal liver cell line HL7702 compared to the HCC lines SNU-182, SNU-449, PLC/PRF/5, MHCC97H, HepG2, Hep3B, and Huh7 (Fig. 1A). Subsequently, we analyzed 80 HCC patients’ cancerous tissues and their adjacent normal tissues using qRT-PCR, revealing that circ0002898 was highly expressed in the adjacent normal tissues and lowly expressed in the cancerous tissues (Fig. 1B). Additionally, we employed RNA in situ hybridization (RNA-ISH) to examine the expression patterns within malignant and adjacent non-malignant tissues extracted from HCC patients. The findings substantiated that the relative abundance of circ0002898 was elevated in the non-malignant tissues compared to the malignant counterparts (Fig. 1C). Based on the RNA-ISH staining scores from the cancer tissues of HCC patients, we stratified the patients into two groups: one with high expression of circ0002898 and another with low expression (Fig. 1D). Survival curve analysis demonstrated that patients with high circ0002898 expression had a more favorable prognosis than those with low expression (Fig. 1E). The circ0002898 expression levels and their correlation with clinical pathological characteristics are detailed in Supplementary Table 2. Moreover, in HCC patients, those with a lower tumor stage (Fig. 1F) and Gleason score (Fig. 1G) displayed a higher expression of circ0002898 within their malignant tissues. In contrast, patients with an advanced tumor stage (Fig. 1F) and a higher Gleason score (Fig. 1G) exhibited a reduced expression of circ0002898. The receiver operating characteristic (ROC) curve results indicated that the expression levels of circ0002898 have diagnostic value in HCC screening, with an area under the curve (AUC) of 0.741 (Fig. 1H). In summary, the low expression of circ0002898 could serve as a biomarker for a poor prognosis in HCC patients.

Fig. 1.

The downregulation of circ0002898 correlates with higher tumor grades and is indicative of a poorer prognosis for survival. A The expression levels of the circular RNA circ0002898 were examined in the designated cell types using qRT-PCR. B A quantitative assessment of circ0002898 expression was conducted in HCC tissues and their corresponding adjacent non-tumorous tissues employing qRT-PCR (N = 80). C Representative images depict the in-situ hybridization (ISH) staining of circ0002898 in HCC tissues and their adjacent non-tumorous tissues. D The ISH staining scores for HCC tissues were evaluated. A staining score of less than 5 was categorized as indicative of low circ0002898 expression, and a score of 5 or higher was considered to represent high expression, with a total of n = 80 samples analyzed. E Kaplan-Meier survival curves were generated to analyze the prognostic significance of circ0002898 expression in HCC patients. F and G The ISH staining scores of circ0002898 were stratified according to the clinical stages and Gleason scores of HCC. H Receiver operating characteristic (ROC) curve analysis was conducted to evaluate the diagnostic accuracy of circ0002898 expression levels in HCC patients. *p < 0.05, **p < 0.01, ***p < 0.001

The molecular attributes and biological properties of circ0002898

The circ0002898 is localized to exon 4 of the DNAJC16 gene on human chromosome 1, with a total length of 340 nucleotides (Supplementary Fig. 1A). To verify the existence of circ0002898 in HCC cells, we employed a divergent primer strategy to amplify circ0002898 from cDNA derived from Hep3B and Huh7 cells, whereas amplification was not from gDNA (Supplementary Fig. 1B). Furthermore, we utilized Sanger sequencing to confirm the sequence of the circular junction of circ0002898 in Hep3B and Huh7 cells (Supplementary Fig. 1C). Compared to the linear form of 0002898 (DNAJC16) mRNA, circ0002898 exhibits resistance to RNase R and stability upon actinomycin D treatment in Hep3B (Supplementary Fig. 2A and 2B) and Huh7 (Supplementary Fig. 2D and 2E) cells. Nuclear-cytoplasmic fractionation experiments demonstrated that circ0002898 is predominantly localized in the cytoplasm of Hep3B (Supplementary Fig. 2C) and Huh7 (Supplementary Fig. 2F) cells.

Transcription factor FOXC2 regulates the expression of circ0002898

The regulatory mechanism of circ0002898 remains largely unknown. However, most reports support the notion that the expression of circRNAs is regulated by transcription factors [24, 25]. Subsequently, we investigated potential transcriptional regulators of circ0002898 (DNAJC16) by querying the JASPAR database and discerned a conserved binding motif for the transcription factor FOXC2 within the promoter sequence of circ0002898 (−523 TAAGTAAACAAA − 512). To substantiate this finding, we engineered plasmids harboring either the wild-type (WT) or mutant (Mut) versions of the circ0002898 (DNAJC16) promoter, employing a luciferase-based reporter assay system (Fig. 2A). The results indicated that overexpression of FOXC2 in Hep3B and Huh7 cells increased the luciferase activity of the WT circ0002898 (DNAJC16) promoter (Fig. 2C). Subsequent ChIP assays demonstrated that the circ0002898 (DNAJC16) promoter was significantly enriched in the precipitate pulled down by FOXC2 antibodies in Hep3B and Huh7 cells (Fig. 2B and D). Overexpression of FOXC2 in Hep3B and Huh7 cells increased the circ0002898 expression levels and the DNAJC16 gene (Fig. 2E). In contrast, the knockdown of FOXC2 in Hep3B and Huh7 cells reduced the circ0002898 expression levels and the DNAJC16 gene (Fig. 2F). These data suggest that FOXC2 is a transcription factor for circ0002898 (DNAJC16). Furthermore, we found that the expression level of FOXC2 in HCC tissue is lower than in the corresponding adjacent non-cancerous tissue (Fig. 2G). Additionally, FOXC2 expression levels positively correlate with circ0002898 in HCC samples (Fig. 2H). Collectively, the downregulation of circ0002898 is attributed to the low expression of FOXC2.

Fig. 2.

Transcription factor FOXC2 regulates circ0002898 expression. A A schematic representation is presented, depicting the binding sites of the transcription factor FOXC2 on the promoter region of wild-type (WT) and mutated (Mut) DNAJC16 (circ0002898). B Agarose gel electrophoresis presented the enrichment of the DNAJC16 promoter via ChIP assays in Hep3B and Huh7 cell lines. C Luciferase reporter assays were conducted to measure the activity in Hep3B and Huh7 cells by co-transfected with the wild-type or mutated DNAJC16 promoter-reporter constructs and a FOXC2 expression vector. D ChIP-qPCR was employed to ascertain the enrichment of the DNAJC16 promoter in Hep3B and Huh7 cells. E The overexpression FOXC2 was shown to upregulate the expression of circ0002898 and DNAJC16 in Hep3B and Huh7 cells. F The suppression of FOXC2 expression via knockdown was observed to downregulate the mRNA levels of circ0002898 and DNAJC16 in Hep3B and Huh7 cells. G qRT-PCR was performed to analyze the expression levels of FOXC2 in HCC tissues and their corresponding adjacent non-tumorous tissues. H A Pearson correlation analysis was conducted to evaluate the relationship between the expression levels of circ0002898 and FOXC2 in HCC samples (n = 80). *p < 0.05, **p < 0.01, ***p < 0.001

Circ0002898 inhibits the proliferation and invasion of HCC cells in vivo and in vitro

To ascertain the functionality of Circ0002898 in HCC, we established Hep3B and Huh7 cell lines with knockdown and overexpression of Circ0002898 (Fig. 3A). Cell proliferation assays indicated that the knockdown of Circ0002898 enhanced the proliferation of Hep3B (Fig. 3B) and Huh7 (Fig. 3C) cells, whereas overexpression of Circ0002898 inhibited their proliferation. Scratch assays demonstrated that the knockdown of Circ0002898 intensified the migratory movement and reparative capabilities of Hep3B (Fig. 3D and F) and Huh7 (Fig. 3E and G) cells, while overexpression of Circ0002898 attenuated these capabilities. Similarly, the knockdown of Circ0002898 elevated the migration and invasion of Hep3B (Fig. 3H and J) and Huh7 (Fig. 3I and K) cells, whereas overexpression of Circ0002898 reduced these parameters. Furthermore, colony formation assays confirmed that the knockdown of Circ0002898 augmented the unlimited proliferative potential of Hep3B and Huh7 cells, while overexpression of Circ0002898 diminished this potential (Fig. 3L and M). To validate whether the findings above are functionally consistent in vivo, we conducted in vivo experiments using Hep3B and Huh7 cells. The results indicated that the knockdown of Circ0002898 enhanced the heterotopic growth of Hep3B (Fig. 3N and P) and Huh7 (Fig. 3O and Q) cells in nude mice, whereas overexpression of Circ0002898 reduced this growth. We also provided data on tumor weight from xenograft models with Circ0002898 knockdown or overexpression in Hep3B and Huh7 cells (Supplementary Fig. 3A). Overexpression of Circ0002898 significantly prolonged the survival of nude mice bearing Hep3B (Supplementary Fig. 3B) or Huh7 (Supplementary Fig. 3C) tumors, whereas knockdown of Circ0002898 shortened survival. Furthermore, immunohistochemical staining revealed that Circ0002898 overexpression reduced the expression levels of CD31 (Supplementary Fig. 3D) and Ki67 (Supplementary Fig. 3E) in tumor tissues, whereas Circ0002898 silencing enhanced their expression. Collectively, both in vitro and in vivo, Circ0002898 exerts an inhibitory function in HCC.

Fig. 3.

Circ0002898 inhibits the proliferation, migration, and invasion capabilities of HCC cells. A The knockdown and overexpression of circ0002898 were verified by qPCR in Hep3B and Huh7 cells. The knockdown of circ0002898 enhanced the proliferation, while overexpression of circ0002898 inhibited the proliferation of Hep3B (B) and Huh7 (C) cells. Cell scratch experiments demonstrated that circ0002898 knockdown enhanced migration movement and repair ability, while overexpression of circ0002898 decreased migration movement and repair ability in Hep3B (D and F) and Huh7 (E and G) cells. circ0002898 knockdown promoted cell migration and invasion, while circ0002898 overexpression inhibited cell migration and invasion in Hep3B (H and J) and Huh7 (I and K) cells. L Colony formation assays were performed, and (M) colony numbers were counted with circ0002898 knockdown and overexpression in Hep3B and Huh7. The knockdown of circ0002898 promoted the tumor growth of Hep3B (N and P) and Huh7 (O and Q) cells in nude mice, while the overexpression of circ0002898 inhibited the tumor growth (n = 8). *p < 0.05, **p < 0.01, ***p < 0.001

Circ0002898 exerts an inhibitory effect on HCC by acting as a MiRNA sponge to sequester miR-145

The prevailing consensus within the scientific reports is that circular RNAs (circRNAs), localized within the cytoplasm, act as miRNA sponges, thereby specifically regulating the functionality of their interacting miRNAs [26]. To ascertain whether Circ0002898 binds to miRNAs, we employed PCR to amplify Circ0002898 from immunoprecipitation of AGO2 antibody (Supplementary Fig. 4). This result indicates an interaction between miRNAs and Circ0002898. Subsequently, we mined six potential miRNAs that may bind to Circ0002898 from a database (https://rnasysu.com/encori/), which are miR-127-5p, miR-1279, miR-145, miR-490-5p, miR-548c-3p, and miR-766. Knockdown of Circ0002898 in Hep3B (Fig. 4A) and Huh7 (Fig. 4B) cells resulted in a significant increase in the expression level of miR-145. Conversely, overexpression of Circ0002898 in Hep3B and Huh7 cells led to a marked decrease in the expression level of miR-145 (Fig. 4C). Furthermore, we utilized RNA pull-down assays to demonstrate the interaction between Circ0002898 and miR-145. The results showed that Circ0002898 probes could pull down miR-145 (Fig. 4D), and miR-145 probes could pull down Circ0002898 (Fig. 4E) in Hep3B and Huh7 cells. Additionally, we identified a potential binding site for miR-145 within Circ0002898 (Supplementary Fig. 5). We constructed a dual-luciferase reporter system containing WT and Mut Circ0002898 sequences to verify this binding site. The results indicated that in Hep3B and Huh7 cells, miR-145 mimics significantly repressed the activity of the WT reporter gene while having negligible effects on the Mut reporter gene activity (Fig. 4F). Moreover, the expression level of miR-145 in HCC tissues was significantly higher than in adjacent normal tissues of HCC (Fig. 4G). In HCC tissues, the expression level of miR-145 exhibited a negative correlation with the expression level of Circ0002898 (Fig. 4H). In conclusion, Circ0002898 can act as a sponge for miR-145, thereby regulating the biological activities associated with miR-145.

Fig. 4.

Circ0002898 functions as a molecular sponge for miR-145. The downregulation of circ0002898 led to an elevation in the levels of miR-145 within the Hep3B (A) and Huh7 (B) cell lines. C The overexpression of circ0002898 resulted in a significant reduction of miR-145 levels in Hep3B and Huh7 cells. D The circ0002898 probe effectively facilitated the pull-down of miR-145. E The miR-145 probe was utilized to pull down circ0002898. F The impact of miR-145 mimics on the luciferase activities of reporters harboring either wild-type (WT) or mutated (Mut) miR-145 binding sites was assessed. G qRT-PCR was conducted to analyze miR-145 expression levels in HCC and their corresponding adjacent non-tumorous tissues (n = 80). H A Pearson correlation analysis determined the relationship between circ0002898 and miR-145 expression levels in HCC samples. The expression of miR-145 mimics enhanced the proliferative capabilities of Hep3B (I) and Huh7 (J) cells. The miR-145 inhibitor reduced the proliferative capabilities of Hep3B (K) and Huh7 (L) cells. M and N The expression of miR-145 mimics enhanced the clonogenic capabilities of Hep3B and Huh7 cells, whereas the miR-145 inhibitor reduced cell colony formation. The miR-145 mimics facilitated migratory and invasive behaviours, while the miR-145 inhibitor curtailed these processes in Hep3B (O and Q) and Huh7 (R and P) cells. *p < 0.05, **p < 0.01, ***p < 0.001

miR-145 enhances the proliferation, migration, and invasive capabilities of HCC cells

Currently, the role of miR-145 in HCC remains elusive. Our data indicate that miR-145 mimics enhance the proliferation of Hep3B (Fig. 4I) and Huh7 (Fig. 4J) cells, whereas miR-145 inhibitors diminish their proliferation (Fig. 4K and L). Further, colony formation assays corroborate these findings (Fig. 4M and N). Subsequently, we observed that miR-145 mimics increase the migratory and invasive capabilities of Hep3B (Fig. 4O and Q) and Huh7 (Fig. 4P and R) cells while miR-145 inhibitors attenuate these capabilities. Finally, scratch assays demonstrate that the presence of miR-145 mimics enhances the migratory movement and reparative abilities of Hep3B (Supplementary Fig. 6A, 6 C, and 6D) and Huh7 (Supplementary Fig. 6B, 6E, and 6 F) cells. In contrast, the presence of miR-145 inhibitors reduces these abilities. Collectively, these results suggest that miR-145 exerts a promoting effect on HCC.

miR-145 directly targets the 3’ untranslated region (3’ UTR) of ADAMTS5 mRNA

MicroRNAs (miRNAs) typically exert regulatory functions by binding to downstream target genes [27]. Utilizing the bioinformatics tool TargetScan, we predicted seven potential target genes for miR-145, which include ADAMTS5, FSCN1, ABHD17C, CSRNP2, PPP3CA, YTHDF2, and ERG. Our results indicated that miR-145 mimics specifically reduced the expression of ADAMTS5 mRNA in Hep3B and Huh7 cells (Fig. 5A), with no effect on the mRNA expression levels of FSCN1 (Supplementary Fig. 7A), ABHD17C (Supplementary Fig. 7B), CSRNP2 (Supplementary Fig. 7C), PPP3CA (Supplementary Fig. 7D), YTHDF2 (Supplementary Fig. 7E), and ERG (Supplementary Fig. 7F). Concurrently, miR-145 inhibitors specifically increased the expression of ADAMTS5 mRNA in Hep3B and Huh7 cells (Fig. 5A) without affecting the mRNA expression levels of the genes mentioned above (Supplementary Fig. 7). These findings suggest that ADAMTS5 may be a target gene of miR-145. Furthermore, we examined the impact of miR-145 mimics and inhibitors on the protein levels of ADAMTS5, and the results were consistent with the mRNA data. That is, miR-145 mimics downregulated the expression of ADAMTS5 protein in Hep3B and Huh7 cells, while miR-145 inhibitors upregulated its expression (Fig. 5B and C and Supplementary Fig. 9A). We identified two binding sites for miR-145 in the 3’-UTR of ADAMTS5 mRNA. Consequently, we constructed luciferase reporter vectors containing either wild-type (WT1, WT2, WT1 + WT2) or mutant (Mut1, Mut2, Mut1 + Mut2) ADAMTS5 3’-UTR sequences (Supplementary Fig. 8). Experimental results demonstrated that miR-145 mimics inhibited the luciferase activity of the WT reporter in Hep3B (Fig. 5D) and Huh7 (Fig. 5E) cells, while the activity of the Mut reporter was negligibly affected. These data support the notion that ADAMTS5 mRNA is a target gene of miR-145.

Fig. 5.

MiR-145 has been identified as a regulatory molecule for ADAMTS5. The expression levels of ADAMTS5 mRNA (A) and proteins (B and C) were quantified in Hep3B and Huh7 cells following the transfection with miR-145 mimics or miR-145 inhibitors. Luciferase activity assays were conducted in Hep3B (D) and Huh7 (E) cells co-transfected with miR-145 mimics, and either the ADAMTS5-3′UTR-WT or ADAMTS5-3′UTR-Mut reporter constructs. F and G The depletion of circ0002898 resulted in a reduction of ADAMTS5 expression levels. H and I The overexpression of miR-145 inhibitor mitigated the suppressive effect of circ0002898 knockdown on ADAMTS5 expression. J ADAMTS5 expression was reduced in HCC tissues compared to adjacent normal tissues. Pearson correlation analyses revealed an inverse relationship between ADAMTS5 mRNA levels and miR-145 expression (K) and a positive correlation between ADAMTS5 mRNA levels and circ0002898 expression (L) in HCC samples (n = 80). *p < 0.05, **p < 0.01, ***p < 0.001

Circ0002898 suppresses the proliferation, migration, and invasion of HCC by targeting the miR-145/ADAMTS5 axis

The suppression of circ0002898 leads to decreased ADAMTS5 protein expression in Hep3B and Huh7 cells (Fig. 5F and G, and Supplementary Fig. 9B). This phenomenon can be reversed by miR-145 inhibitors added (Fig. 5H and I, and Supplementary Fig. 9C). The mRNA expression level of ADAMTS5 is significantly lower in HCC samples compared to the adjacent non-cancerous tissue of HCC (Fig. 5J). Furthermore, there is a negative correlation between the expression levels of miR-145 and ADAMTS5 mRNA (Fig. 5K), while a positive correlation exists between Circ0002898 and ADAMTS5 mRNA expression levels (Fig. 5L). Knockdown of Circ0002898 enhances the proliferation (Fig. 6A and B), migration, invasion (Fig. 6C, D and E, and 6F), unlimited replicative potential (Supplementary Fig. 10A), migratory movement, and repair capabilities (Supplementary Fig. 11A, 11B, 11 C and 11D) in Hep3B and Huh7 cells. However, these effects can be blocked by miR-145 inhibitors. Further, the expression of miR-145 mimics increases the proliferation (Fig. 6G and H), migration, invasion (Fig. 6I, J and K, and 6L), and unlimited replicative potential (Supplementary Fig. 10B), as well as the migratory movement and repair capabilities (Supplementary Fig. 11E, 11 F, 11G and 11 H) of Hep3B and Huh7 cells, which can be reversed by co-expression of ADAMTS5. Co-expression of ADAMTS5 also reverses the high proliferation (Fig. 6M and N), migration, invasion (Fig. 6O, P and Q, and 6R), unlimited replicative potential (Supplementary Fig. 10C), and migratory movement and repair capabilities (Supplementary Fig. 11I, 11 J, 11 K and 11 L) induced by Circ0002898 knockdown in Hep3B and Huh7 cells. In conclusion, circ0002898 can inhibit HCC cells through the miR-145/ADAMTS5 axis.

Fig. 6.

Circ0002898 exerts its inhibitory effects on HCC through the miR-145/ADAMTS5 axis. miR-145 inhibitors were added to Hep3B (A) and Huh7 (B) cells with circ0002898 knockdown, and their proliferation was evaluated over a specified period of time. The migratory and invasive capabilities of Hep3B (C and E) and Huh7 (D and F) cells with circ0002898 knockdown transfected with miR-145 inhibitor were evaluated. Hep3B (G) and Huh7 (H) cells transfected with miR-145 mimics or co-transfected with miR-145 mimics and ADAMTS5 were seeded in 96-well plates, and their proliferation was measured within the specified time. The migration and invasion of Hep3B (I and K) and Huh7 (J and L) cells transfected with miR-145 mimics or co-transfected with miR-145 mimics and ADAMTS5 were determined. Hep3B (M) and Huh7 (N) cells with circ0002898 knockdown transfected with ADAMTS5 were cultured in 96-well plates, and their proliferation was assessed within the specified time. The migration and invasion of Hep3B (O and Q) and Huh7 (P and R) cells with circ0002898 knockdown transfected with ADAMTS5 were evaluated. *p < 0.05, **p < 0.01, ***p < 0.001

ADAMTS5 overexpression inhibits tumor angiogenesis in vitro

ADAMTS5 is a metalloproteinase that can be secreted extracellularly, and its role in cartilage degeneration and arthritis is well-documented [28, 29]. It has been reported that ADAMTS5 can inhibit angiogenesis in melanoma [23, 30], but its impact on HCC remains unexplored. To elucidate the function of ADAMTS5 in HCC, we constructed Hep3B and Huh7 cell lines overexpressing ADAMTS5 (Fig. 7A and B). Our results indicate that conditioned media extracted from Hep3B and Huh7 cells overexpressing ADAMTS5 inhibit the angiogenic capacity of HUVEC cells in vitro (Fig. 7C and D). Angiogenesis is fundamentally propelled by the proliferation of endothelial cells, with VEGF serving as a potent stimulant for endothelial cell division [31, 32]. Our findings demonstrate that conditioned media extracted from Hep3B and Huh7 cells overexpressing ADAMTS5 significantly and dose-dependently inhibit VEGF-induced proliferation of HUVEC cells (Fig. 7E and F). Many angiogenesis inhibitors operate by disrupting endothelial cell survival and inducing apoptosis. VEGF is well-known as an endothelial cell survival factor that protects endothelial cells from apoptosis [33, 34]. Our results show that, under conditions of VEGF presence, conditioned media extracted from Hep3B and Huh7 cells overexpressing ADAMTS5 induce apoptosis in HUVEC cells in a dose-dependent manner (Fig. 7G). Significantly, the induction of apoptosis is caspase-dependent. Pretreatment of HUVEC cells with the pan-caspase inhibitor Z-VAD-FMK significantly diminishes the apoptotic effect induced by conditioned media from Hep3B and Huh7 cells engineered to overexpress ADAMTS5 (Fig. 7H). We hypothesized that ADAMTS5 overexpression in HCC cells suppresses VEGF downstream signaling in HUVECs. To test this, we performed VEGF salvage experiments. Administration of recombinant VEGF reversed the inhibition of angiogenesis elicited by ADAMTS5 overexpression (Fig. 7I and J). Collectively, ADAMTS5 overexpression in HCC cells disrupts VEGF downstream signaling in endothelial cells, thereby effectively suppressing tumor associated angiogenesis.

Fig. 7.

ADAMTS5 overexpression has an anti-angiogenic effect. A qPCR and (B) western blotting verified ADAMTS5 overexpression in Hep3B and Huh7 cells. C Imaging and (D) quantitative data demonstrated that conditioned media extracted from Hep3B and Huh7 cells overexpressing ADAMTS5 were found to inhibit the formation of tubular structures in HUVEC cells. The Mock group was treated with the medium containing 2% fetal bovine serum. The Control vector group was derived from conditioned media extracted from Hep3B and Huh7 cells transfected with an empty vector. The ADAMTS5 overexpression group was obtained from conditioned media extracted from Hep3B and Huh7 cells overexpressing ADAMTS5. E Imaging and (F) quantitative data demonstrated a dose effect of inhibiting tubular structure formation in HUVEC cells by conditioned media derived from Hep3B and Huh7 cells overexpressing ADAMTS5. 20%, 50%, and 100% indicate the dose of conditioned medium added, extracted from Hep3B and Huh7 cells expressing ADAMTS5. G Conditioned medium derived from Hep3B and Huh7 cells overexpressing ADAMTS5 inhibited VEGF-mediated proliferation of HUVEC cells in a dose-dependent manner. H Under conditions of VEGF presence, conditioned media extracted from Hep3B and Huh7 cells overexpressing ADAMTS5 induced apoptosis in HUVEC cells in a dose-dependent manner. Pre-incubation with the pan-caspase inhibitor Z-VAD-FMK abolished the apoptotic effect of conditioned media extracted from Hep3B and Huh7 cells overexpressing ADAMTS5 on HUVEC cells. I Imaging and (J) quantitative results demonstrate that the addition of VEGF protein rescued the inhibition of angiogenesis induced by ADAMTS5 overexpression. *p < 0.05, **p < 0.01, ***p < 0.001

Collectively, our data suggest that the FOXC2/Cyclic RNA0002898/miR-145/ADAMTS5 axis can inhibit angiogenesis in HCC (Fig. 8). These key molecules are potential targets for HCC therapy.

Fig. 8.

Mechanistic diagram of FOXC2/Cyclic RNA0002898/miR-145/ADAMTS5 axis inhibition of HCC angiogenesis

Discussion

In this article, we demonstrate for the first time the downregulation of circ0002898 expression in hepatocyte carcinoma (HCC) and establish a correlation between decreased circ0002898 expression and higher tumor grades, as well as shorter patient survival. Circ0002898 induces the expression of the tumor suppressor ADAMTS5 by binding to miR-145 in the cytoplasm, thereby inhibiting tumor angiogenesis. Our research findings reveal previously unknown functions and potential mechanisms of circ0002898 and elucidate the molecular mechanisms of Circ0002898 expression regulation in HCC.

Sequencing data indicate that circular RNAs (circRNAs) exhibit aberrant expression in various cancers [35–37], yet the underlying molecular mechanisms remain elusive. The expression levels of circRNAs are generally influenced by the expression levels of their precursor transcripts [38, 39]. In this study, we identified FOXC2 as a transcription factor for DNAJC16 (the precursor of circ0002898), which regulates the expression of circ0002898. In human hepatocyte carcinoma (HCC) samples, we observed a downregulation of FOXC2 expression, which correlates positively with the downregulation of circ0002898. The data above support the notion that the downregulation of FOXC2 is a crucial factor leading to the downregulation of circ0002898. Furthermore, the literature suggests that the aberrant expression of FOXC2 is closely associated with the occurrence and progression of multiple cancers [40–42]. Consequently, strategies to enhance the expression levels of FOXC2 represent a potential therapeutic avenue for the treatment of HCC.

CircRNAs are generally recognized as miRNA sponges, sequestering miRNAs and mitigating their suppressive influence on target mRNA transcripts [43]. Several molecules have been identified as miR-145 sponges, including circPVT1 [44], lncRNA CACS15 [45], LINC00707 [46], lncRNA TUG1 [47], lncRNA MALAT1 [48], and lncRNA ROR [49]. These molecules can block downstream PI3K/AKT, Wnt/β-catenin signaling pathways, and target genes of miR-145, thereby promoting or inhibiting the occurrence and development of tumors. Although miR-145 is frequently reported as a tumor suppressor in HCC, inhibiting progression by targeting multiple oncogenes [50, 51], our study reveals an oncogenic role for miR-145 through its suppression of ADAMTS5, a key angiogenesis inhibitor. This apparent discrepancy likely stems from context-specific differences in its targetome. While relevant literature associates elevated FOXC2 expression in HCC with the promotion of EMT/invasion and poor prognosis [52], our findings present an opposing perspective. The ongoing debate concerning FOXC2’s function in HCC fundamentally reflects the complexity of tumor heterogeneity and microenvironmental regulation. The FOXC2 low-expression/tumor-suppressive phenotype identified in our study may predominate in early-to-mid-stage HCC, whereas microenvironmental remodeling during metastatic progression might trigger a functional reversal. Future studies integrating single-cell sequencing technologies are warranted to elucidate the dynamic role of FOXC2 in the spatiotemporal evolution of HCC.

ADAMTS5, as a secreted metalloproteinase [29], exerts potent anti-angiogenic effects by disrupting VEGF-mediated signaling and inducing caspase-dependent apoptosis in endothelial cells. This creates a nutrient-deprived microenvironment that restricts tumor expansion. Crucially, the downregulation of ADAMTS5 in HCC tissues corresponds with increased vascular density, thereby liberating tumor cells from metabolic constraints and enhancing their proliferative and migratory capacities. This mechanism explains the observed correlation between ADAMTS5 deficiency and aggressive tumor phenotypes. As a direct target of miR-145, we speculate that secreted ADAMTS5 may impair HCC cell migration and proliferation by degrading pro-migratory extracellular matrix components such as aggrecan [29]. The consistent downregulation of ADAMTS5 in HCC creates a permissive ECM environment for invasion, migration and proliferation. Our functional rescue experiments substantiate this mechanism: Reconstitution of the FOXC2/circ0002898/miR-145/ADAMTS5 axis through genetic manipulation (knockdown/overexpression) directly reversed migration and proliferation phenotypes, confirming the axis’s cell-autonomous regulatory role.

In conclusion, our study presents novel evidence that circ0002898 inhibits HCC via the miR-145/ADAMTS5 signaling axis. This discovery implies that circ0002898 may function as a tumor suppressor in HCC, highlighting its potential as a therapeutic target. Moreover, we observed an inverse relationship between circ0002898 expression levels and tumor progression, coupled with a positive association with patient survival rates. These correlations suggest that circ0002898 could be a candidate for serving as both a diagnostic and prognostic biomarker in HCC. It is widely acknowledged that circRNAs are vectored through exosomes from the cellular interior to the extracellular environment, where they can be detected within various biofluids [53]. Consequently, it would be intriguing to ascertain the circulation of circ0002898 levels in the plasma of HCC patients, which could offer significant insights into disease progression. Furthermore, by correlating the levels of circ0002898 in the plasma with patient survival rates, we may establish its potential as a prognostic biomarker for HCC.

Abbreviations

HCC

Hepatocellular Carcinoma

CircRNAs

Circular RNAs

ROC

Receiver Operating Characteristic

ChIP

Chromatin immunoprecipitation

AUC

Area Under Curve

RIP

RNA immunoprecipitation

ISH

In Situ Hybridization

Authors’ contributions

Jiangbei Yuan, Qiang He, and Hongying Pan initiated and designed the experiments. Jiangbei Yuan performed experiments and wrote the manuscript. Fei Lv, Yue Zhao, Zixiang Pan, Wei Zheng, Qiaoqiao Yin, LanJie Wu, Jianli Yu, and Cheng’an Xu analysed the data.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.