A hsa_circ_001726 axis regulated by E2F6 contributes to metastasis of hepatocellular carcinoma

The gene expression profile in liver hepatocellular carcinoma (LIHC) were obtained from The Cancer Genome Atlas (TCGA). GSE121714 dataset was download from Gene expression omnibus (GEO) database. GSE121714 dataset contained the circRNA expression profile in non-tumor tissues (Normal group, 10 samples), tumor tissues of HCC patients with metastasis (Metastasis group, 10 samples) and HCC patients without metastasis (Primary group, 10 samples). The differentially expressed circRNAs (DEcircRNAs) between Normal group and Primary group, Primary group and Metastasis group were identified with same R package. The cut-off criteria was |log₂fold change|> 1 and P-value < 0.05. Applying “pheatmap” and “ggplot2” R packages, heatmap and volcanos of the DEcircRNAs were obtained. The Venn diagram was employed to take intersection of DEcircRNAs. CircBase database was used to retrieve the source of hsa_circ_001726. The PROMO database predicted the transcription factors that regulate the CCT2 promoter.

Tumor tissues were obtained from HCC patients with metastasis (40 samples) and HCC patients without metastasis (40 samples). Unpaired adjacent nontumor tissues (40 samples) were collected as control. HCC patients were received hepatectomy in the first affiliated hospital of Nanchang University. The patients did not receive any treatment before surgery. Using pathological section examination, these patients were diagnosed with HCC.

Normal human hepatic cell line LO2 and liver cancer cell lines (Huh7, MHCC97H, Sk-Hep1, Bel7402) were purchased from the Type Culture Collection of Chinese Academy of Sciences (Shanghai, China). All cells were cultured in DMEM or RPMI1640 medium (Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (Gibco) and 1% penicillin–streptomycin (Sangon-Biotech, Shanghai, China). The culture condition was 37 °C and 5% CO₂. MHCC97H and Huh7 cells were treated with RNase R (Sangon-Biotech) at 37 °C for 30 min.

Small interfering RNA (siRNA) specifically targeting E2F6 (si-E2F6: 5’-GGG TAT TCT TGA CTT AAA CAA-3’; 5’-GAA AGC GGA GAG TGT ATG ACA-3’; 5’- GAA GGC CCT GAG GAA GAA GAA-3’) or hsa_circ_001726 (si-hsa_circ_001726: 5’-GGC AAC CTG GAG GCA ATT CAT-3’; 5’-GCA ACC TGG AGG CAA TTC ATA-3’; 5’-GCA GAT TCC TAT TTA GAT GAA-3’), short hairpin RNA (shRNA) specifically targeting hsa_circ_001726 (sh-hsa_circ_001726: 5’-GGC AAC CTG GAG GCA ATT CAT TCA AGA GAT GAA TTG CCT CCA GGT TGC C-3’; 5’-GGC AAC CTG GAG GCA ATT CAT CTC TTG AAT GAA TTG CCT CCA GGT TGC C-3’), miR-671-5p inhibitor (miR-671-5p-In: 5’-CUC CAG CCC CUC CAG GGC UUC CU-3’) were used to silence E2F6, hsa_circ_001726 or miR-671-5p in liver cancer cells. MiR-671-5p mimic (5’-TGC AGG CTA GTC AAC TTA GGC ACG TCA A-3’) and pcDNA3.1 carrying PRMT9 (PRMT9-OE) were applied to overexpress miR-671-5p or PRMT9 in liver cancer cells. The si-NC (5’-TAG CGC TGA GGC GTG CAA GCT GAT TCT TA-3’), sh-NC (5’-GAT CCC CTT CTC CGA ACG TGT CAC GTT TCA AGA GAA CGT GAC ACG TTC GGA GAA TTT TT-3’), mimic NC (5’-TAG CCC AGA CAC TGC AGC GTA GCG ATC CA-3’), inhibitor NC (In-NC: 5’-CAG UAC UUU UGU GUA GUA CAA-3’) and empty pcDNA3.1 vector (Vector) were served as control. The sh-hsa_circ_001726 and sh-NC were packaged into lentiviral particles, generating (LV-sh-hsa_circ_001726 and LV-sh-NC). These vectors were obtained from Genepharm (Shanghai, China). Huh7 and MHCC97H cells were transfected with vectors or oligonucleotides applying Lipofectamine™2000 reagent (Thermo Fisher Scientific, San Jose, CA, USA). Huh7 cells were infected with LV-sh-hsa_circ_001726 or LV-sh-NC applying polybrene (MedChemExpress, Monmouth Junction, NJ, USA).

Applying TRIzol reagent (Thermo Fisher Scientific), total RNA was extracted from cells and tissues. Cytoplasmic and nuclear RNA purification kit (Norgen Biotek, Canada) was utilized to extract cytoplasmic and nuclear RNA from cells. For detection of the structure of CCT2 and hsa_circ_001726, total RNA was digested with RNase R (Sangon-Biotech) at 37 °C for 30 min. Total RNA was served as template to synthesize complementary DNA applying PrimeScript reverse transcriptase reagent kit or Mir-X miRNA First-Strand Synthesis Kit (TaKaRa, Beijing, China). Real‐time PCR was performed on StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) applying TB Green® Premix Ex Taq™ II or Mir-X miRNA qRT-PCR TB Green® Kit (TaKaRa). GAPDH, β-actin and U6 were served as loading control. The primer sequences (5’-3’) were listed as follow: CCT2: forward-TGC GGG CAC AAC ATT ATC CT and reverse-TGC GCA AGA ACA CAA AGA GC; E2F6: forward-TGC CTT CGC CAT GAA TCC TT and reverse-TCG GAC TCC CAG TTT CGT TG; hsa_circ_001726: forward-TGC GGG CAC AAC ATT ATC CT and reverse-AGG TTG CCA GAG CCT TTC AG; miR-617-5p: forward-ACA CTC CAG CTG GGA GGA AGC CCT GG and reverse-CTC AAC TGG TGT CGT GGA GTC GGC AAT; PRMT9: forward-TCC TCT CCC ATT TCA ACC TGG and reverse-TCC TCT CCC ATT TCA ACC TGG; E-cadherin: forward-CCT GGG ACT CCA CCT ACA GA and reverse-AGG AGT TGG GAA ATG TGA GC; N-cadherin: forward-CAA GAT GGG TCA ATG GAA ATA G and reverse-CTC AGG AAT ACG AGC CTT CAC; GAPDH: forward-GCA CCG TCA AGG CTG AGA AC and reverse-TGG TGA AGA CGC CAG TGG A; β-actin: forward-ATC GTG CGT GAC ATT AAG GAG AAG and reverse-AGG AAG GAA GGC TGG AAG AGT G; U6: forward-GCT TCG GCA GCA CAT ATA CTand reverse-GGT GCA GGG TCC GAG GTA T. Data were analyzed by 2^−ΔΔCt method.

The stability of CCT2 and hsa_circ_001726 was measured using Actinomycin D (Sigma-Aldrich, St. Louis, MO, USA) assay. Briefly, Huh7 cells were treated with 5 µg/mL Actinomycin at 37 °C for 24 h. At 0, 4, 8, 12, 24 hours after Actinomycin D treatment, total RNA was isolated from cells. The expression of CCT2 and hsa_circ_001726 was examined by performing qRT-PCR assay.

The dissected tissues from HCC patients were paraffin-embedded and cut into Sects. (4 μm). Following deparaffinization and hydration, sections were subjected to heat-induced antigen retrieval. The sections were incubated with anti-E-cadherin (1: 100 dilution; Abcam, Cambridge, MA, USA) or anti-N-cadherin (1: 100 dilution; Abcam) at 4 °C overnight, and then stained with goat anti-rabbit HRP-conjugated IgG (1: 2000 dilution; Abcam) at room temperature for 1 h. The sections were stained with 3,3'-Diaminobenzidine and then counterstained with Mayer’s hematoxylin.

Tissues and cells were treated with RIPA lysis buffer (Beyotime, Shanghai, China), and total proteins were extracted. The protein samples were separated by performing 10% SDS-PAGE protein electrophoresis, and then transferred to a PVDF membrane (Millipore, Billerica, MA, USA). The membranes were incubated primary antibodies at 4 °C overnight, including anti-E2F6 (1: 1000 dilution; Abcam), anti-E-cadherin (1: 1000 dilution; Abcam), anti-ZO-1 (1: 1000 dilution; Abcam), anti-N-cadherin (1: 5000 dilution; Abcam), anti-Vimentin (1: 500 dilution; Abcam), anti-PRMT9 (1: 500 dilution; Abcam), anti-Notch1 (1: 1000 dilution; Abcam), anti-Hes 1 (1: 1000 dilution; Abcam) at room temperature for 1 h. The membranes were stained with goat anti-rabbit HRP-conjugated IgG (1: 2000 dilution; Abcam) at room temperature for 1 h. β-actin (1: 2000 dilution; Abcam) was served as loading control. The immunoprecipitated bands were developed with ECL reagent and analyzed by ImageJ software.

BeyoClick™ EdU Cell Proliferation Kit with Alexa Fluor 488 (Beyotime) was utilized to assess cell proliferation. MHCC97H and Huh7 cells (5 × 10⁵) were cultured in 6-well plates overnight, and then incubated with 10 μM EdU for 2 h. Cells were fixed with 4% paraformaldehyde for 10 min and permeabilized with 0.3% Triton X-100 for 20 min. Cells were stained with Alexa Fluor 488 azide in darkness for 30 min. Nucleus were stained with DAPI in darkness for 10 min. Finally, the proliferative cells were observed under a fluorescence microscope (Nikon, Tokyo, Japan).

MHCC97H and Huh7 cells (5 × 10⁵) were seeded into 6-well plates and cultured at 37 °C and 5% CO₂ until 100% confluence. Applying a 200-μL pipette tip, an artificial and straight scratch was created. Then, cells were washed with PBS to remove the suspended cells and cultured in serum-free medium. At 0 and 24 hours after cell culture, wound closure images were captured.

A 24-well Transwell plate chamber with 8 μm pore size (Corning Costar, Cambridge, MA, USA) were used to examine cell invasion ability. Briefly, MHCC97H and Huh7 cells (1 × 10⁶ cells/ml) were seeded into the upper chamber coated with Matrigel. The lower chamber was added to the culture medium. After 24 hours incubation, the cells on the upper chamber were wiped off with a cotton swab, fixed with methanol for 10 min and stained with 0.1% crystal violet. The invasive cells were observed under an optical microscope.

Apoptosis of HCC cells was evaluated using One Step TUNEL Apoptosis Assay Kit (Beyotime). Following 24 hours of incubation, MHCC97H and Huh7 cells were subjected to fixation with 4% paraformaldehyde and permeabilization with 0.3% Triton X-100. Cells were stained with 50 μL TUNEL detection reagent at 37 °C in darkness for 2 h. Finally, apoptotic cells were observed under a fluorescence microscope.

Huh7 cells (1 × 10⁴ cells/well) were cultured in 48-well plate and fixed in 4% paraformaldehyde. Cells were permeabilized with 0.5% Triton X-100 for 15 min at 4 °C. Cy3-labeled hsa_circ_001726 probe were designed and synthesized by GenePharma. Cells were incubated with digoxigenin-labeled hsa_circ_001726 probe at 55 °C for 4 h, and then incubated with HRP-conjugated anti-DIG antibody (Jackson, West Grove, PA, USA). DAPI was used to stain nuclei. The fluorescence was observed under a confocal laser scanning microscope (Olympus, Tokyo, Japan).

Pierce Magnetic ChIP Kit (Thermo Fisher Scientific) was utilized to verify the interaction between E2F6 and CCT2 in Huh7 and MHCC97H cells following the protocol of manufacturer. Briefly, cells were cultured in serum-free medium overnight for serum starvation. Cells were treated with 1% formaldehyde for 10 min. Following cell lysis, the chromatin was sonicated to obtain DNA fragment with a size of 500–1000 bp. Cell lysate was incubated with anti-E2F6 (Abcam) or isotype anti-IgG (Abcam) at 4 °C overnight. Then, cell lysate was treated with Protein A/G Magnetic Beads to obtain chromatin-antibody complexes. DNA was eluted from the chromatin-antibody complexes, and then analyzed by qRT-PCR.

To evaluate the influence of E2F6 on the activity of CCT2 promoter, CCT2 promoter was cloned into pGL3-basic, generating the vector pGL3-basic-CCT2 promoter. HEK-293 T cells were transfected with pGL3-basic-CCT2 promoter combined with si-E2F6/si-NC. For detection of the interaction relationship among hsa_circ_001726, miR-671-5p and PRMT9, the wild type or mutant hsa_circ_001726/PRMT9 containing the predictive binding site of miR-671-5p was cloned into the pmirGLO Dual-Luciferase miRNA Target Expression Vector (Promega, Madison, WI, USA), generating the vectors pmirGLO-hsa_circ_001726-WT (5’-UCC UAU UUA GAU GAU GGC UUC CU-3’), pmirGLO-hsa_circ_001726-MUT (5’-UCC UAU UUA GAU GAU CCG AAG GU-3’), pmirGLO-PRMT9-WT (5’-UGG GAA ACA CUG AAA GGC UUC CA-3’) and pmirGLO-PRMT9-MUT (5’-UGG GAA ACA CUG AAA CCG AAG GA-3’). HEK-293 T cells were transfected with WT/MUT of pmirGLO vectors and miR-671-5p mimic or mimic NC. At 48 h post-transfection, the luciferase assay was performed using the Dual Luciferase Reporter Assay System (Promega).

Applying Magna RIP kit (Millipore, Billerica, MA, USA), RIP assay was performed to verify the relationship between hsa_circ_001726 and miR-671-5p. Huh7 cells were treated with RIPA Lysis Buffer, cell lysates were collected for further use. Protein A/G magnetic beads were incubated with anti-AGO2 (Abcam) at room temperature for 1 h. Then, cell lysates were incubated with antibody-conjugated magnetic beads on a rotator at 4 °C for 12 h. IgG antibody (Abcam) served as control. The co-precipitated RNAs were purified from the bead-antibody-RNA complex, and then used for qRT-PCR analysis of hsa_circ_001726 and miR-671-5p.

Immunodeficient BABL/c male nude mice (4–5 weeks old) were purchased from Shanghai SLAC Laboratory Animal Co. Ltd. Mice were maintained specific pathogen-free conditions with constant temperature (22–24 °C) and humidity (40–60%). Mice take food and water freely. The position of the mouse cage was rotated once a week, and the measurements of the mice were carried out in a fixed order. During the study, all protocols were carried out by experienced researcher applying double-blind method. Animal experiments were performed according to the Laboratory Animals-Guidelines for Ethical Review of Animal Welfare (GB/T 35892–2018) and the ARRIVE guidelines.

According to statistical requirements and the feasibility of experimental operations, mice were randomly divided into two groups (n = 6): LV-sh-NC group and LV-sh-hsa_circ_001726 group. Mice were subcutaneously injected with 200 μL of Huh7 cells following transfection of LV-sh-hsa_circ_001726 or LV-sh-NC. Tumor volume was monitored every week for 5 weeks. On the 25^th day of modeling, all mice were anesthetized with intraperitoneal injection of ketamine (100 mg/kg body weight) and then euthanized with cervical dislocation. Tumor tissues were separated, and tumor weight was detected. After one week of inoculation, mice with subcutaneous tumor formation observed with the naked eye were included in the experiment. The weight of tumor tissues should not exceed 10% of the mouse's body weight, and the average tumor diameter should not exceed 20 mm. If ulceration occurs and severe metastasis leads to infection or necrosis, the experiment should be stopped and the animal should be euthanized.

Orthotopic transplantation tumor model was constructed as previous study described [28]. When the xenograft tumors grew to about 1 cm³ in volume, the tumor tissues were removed from the mice in each group (LV-sh-hsa_circ_001726 and LV-sh-NC), and then cut into small pieces (1 mm³). Mice were randomly divided into two groups (n = 6): LV-sh-NC group and LV-sh-hsa_circ_001726 group. Following anesthetization with 1% ketamine (100 mg/kg), the clipped tumor tissues were transplanted into the subcapsular region of the left medial lobe of the liver of nude mice. The 8–0 nylon surgical sutures were used to suture the liver incision. When the liver tumors reached about 20 mm in length, mice were euthanized with ketamine combined with cervical dislocation. Lung tissues were collected from mice. The number of metastatic nodules in the lung tissues were counted under a dissecting microscope. If ulceration occurs and severe metastasis leads to infection or necrosis, the experiment should be stopped and the animal should be euthanized.

Lung tissues were fixed with 4% paraformaldehyde and embedded with paraffin. The paraffin sections with 4 μm thick were subjected to deparaffinization and hydration. After that, the sections were stained with hematoxylin for 5 min and incubated with eosin for 30 s applying HE staining kit (Beyotime). Finally, the pathological changes of lung tissues were observed under an optical microscope.

To investigate the key circRNAs involved in metastasis of HCC, we analyzed the differentially expressed circRNAs among adjacent nontumor tissues and tumor tissues without metastasis based on GSE121714 dataset. Compared with adjacent nontumor tissues, 21 down-regulated circRNAs and 75 up-regulated circRNAs were found in HCC tumor tissues without metastasis (Fig. 1A). There were 62 down-regulated circRNAs and 190 up-regulated circRNAs in HCC tumor tissues with metastasis as compared to HCC tumor tissues without metastasis (Fig. 1B). Venn diagram and heatmap showed there were 11 overlapping differentially expressed circRNAs (Fig. 1C and D). Hsa_circ_001726 is a novel discovered differentially expressed RNA with the highest expression in HCC. Hsa_circ_001726 is located on chr12 (69,983,264–69,985,939) with a length of 304 bp. Like all circRNAs, hsa_circ_001726 exhibits a closed ring structure. It is not easily degraded by nucleic acid exonucleases and is more stable than linear RNA. In light of the fact that hsa_circ_001726 is transcribed from CCT2 (circBase database). Then, we examined the expression of CCT2 in HCC applying TCGA-LIHC database. As shown in Fig. 1E and F, CCT2 was highly expressed in tumor tissues, which indicated a poor prognosis. Then, the structure of hsa_circ_001726 and CCT2 was examined by qRT-PCR. Following RNase R digestion, hsa_circ_001726 could be amplified, but not linear CCT2 (Fig. 1G). Actinomycin D assay evaluated the stability of hsa_circ_001726 and CCT2, showing that hsa_circ_001726 was much more stable than CCT2 (Fig. 1H). Moreover, the distribution of hsa_circ_001726 and CCT2 in the nuclear and cytoplasm of Huh7 cells was detected by qRT-PCR. The expression of hsa_circ_001726 and CCT2 was abundant in the cytoplasm, but rarely in the nucleus (Fig. 1I). Consistently, RNA-FISH assay uncovered that hsa_circ_001726 was localized in the cytoplasm of Huh7 cells (Fig. 1J). Thus, these data showed that hsa_circ_001726 was transcribed from CCT2, and served as an oncogene in HCC.

To understand the involvement of hsa_circ_001726 in HCC, the expression of hsa_circ_001726 in tumor tissues of HCC patients was examined by qRT-PCR. Compared with adjacent nontumor tissues, up-regulation of hsa_circ_001726 was observed in tumor tissues of primary HCC patients. The expression of hsa_circ_001726 was even higher in tumor tissues of metastatic HCC patients compared to those of tumor tissues of primary HCC patients (Fig. 2A). In addition, the precise mechanism of hsa_circ_001726 in HCC was determined in vitro. The results of qRT-PCR revealed that hsa_circ_001726 was up-regulated in HCC cell lines with respect to normal human hepatic cell line LO2 (Fig. 2B). Subsequently, we employed siRNA to silence hsa_circ_001726 and examined its impact on the malignant characteristics of HCC cells. Following transfection of si-hsa_circ_001726, the expression of hsa_circ_001726 was significantly decreased in MHCC97H and Huh7 cells (Supplementary Fig. 1A). Results obtained from wound healing and transwell assays showed that hsa_circ_001726 deficiency repressed migration and invasion of MHCC97H and Huh7 cells (Fig. 2C and D). Furthermore, a xenograft mouse model was constructed to verify the role of hsa_circ_001726 silencing in HCC in vivo. The expression of hsa_circ_001726 was severely decreased in Huh7 cells in the presence of LV-sh-hsa_circ_001726, as determined by qRT-PCR (Supplementary Fig. 1B). Hsa_circ_001726 knockdown led to a decrease in the volume and weight of tumor tissues in xenograft mice (Fig. 2E-G). Additionally, orthotopic transplantation tumor model was constructed to explore the influence of hsa_circ_001726 on tumor metastasis. Lung tissues are the most common site of extrahepatic metastasis in HCC [29]. The occurrence of lung metastasis is to some extent a manifestation of accelerated disease progression in patients, which increases their risk of death [6]. We observed the lung metastasis of mice. Hsa_circ_001726 deficiency obviously reduced the number of lung metastatic foci. Compared with LV-shNC group, orthotopic transplantation tumor mice with hsa_circ_001726 knockdown exhibited more serious lesions in lung tissues (Fig. 2H-J and Supplementary Fig. 2). Thus, these results collectively indicate that hsa_circ_001726 silencing impeded malignant progression of HCC in vivo and in vitro.

Here, we analyzed the miRNA that targeted by hsa_circ_001726 based on circular RNA interactome database. Applying TargetScan and miRDB databases, the miRNA that interacted with PRMT9 was predicted. This analysis revealed an overlapping miRNA, miR-671-5p, which possesses binding sites in both hsa_circ_001726 and PRMT9 (Fig. 3A). Notably, miR-671-5p displayed significant downregulation in the tumor tissues of primary HCC patients compared to adjacent non-tumor tissues. The expression of miR-671-5p was lower in tumor tissues of metastatic HCC patients than that in primary HCC patients (Fig. 3B). The results of qRT-PCR exhibited a negative correlation between miR-671-5p expression and hsa_circ_001726 levels (Fig. 3C). Compared with normal human hepatic cell line LO2, miR-671-5p expression was also decreased in HCC cell lines (Fig. 3D). To verify the interaction between hsa_circ_001726 and miR-671-5p, luciferase reporter assay and RNA immunoprecipitation were carried out. The luciferase activity was notably decreased in HEK-293 T cells upon transfection with both miR-671-5p mimic and pmirGLO-hsa_circ_001726-WT, demonstrating a direct interaction between miR-671-5p and hsa_circ_001726 (Fig. 3E and F). AGO2 is an indicator protein for circRNA to exert sponge effects. MiRNA interacts with AGO2 protein to form RNA-induced silencing complex [30]. Anti-AGO2 increased the expression of hsa_circ_001726 and miR-671-5p as compared with anti-IgG (Fig. 3G), indicating that hsa_circ_001726 acted as sponge for miR-671-5p. Results of qRT-PCR showed that miR-671-5p was highly expressed in MHCC97H and Huh7 cells in the presence of si-hsa_circ_001726 (Fig. 3H). Similarly, miR-671-5p was up-regulated in lung tissues of orthotopic transplantation tumor mice with hsa_circ_001726 knockdown, as determined by qRT-PCR (Fig. 3I). Moreover, results of luciferase reporter assay showed that the luciferase activity was reduced in HEK-293 T following transfection of miR-671-5p mimic and pmirGLO-hsa_circ_001726-WT, indicating that miR-671-5p interacted with 3’ UTR of PRMT9 (Fig. 3J and K). Additionally, qRT-PCR and western blotting examined the impact of hsa_circ_001726 on the expression of PRMT9 in HCC cells. Hsa_circ_001726 silencing caused a down-regulation of PRMT9 in MHCC97H and Huh7 cells (Fig. 3L-N). In vivo, hsa_circ_001726 knockdown also reduced PRMT9 expression in lung tissues of orthotopic transplantation tumor mice (Fig. 3O). Then, miR-671-5p was overexpressed in MHCC97H and Huh7 cells applying transfection of miR-671-5p mimic (Supplementary Fig. 1C). MiR-671-5p mimic repressed PRMT9 expression in MHCC97H and Huh7 cells (Fig. 3P, Q). Furthermore, miR-671-5p was silenced in MHCC97H and Huh7 cells following transfection of miR-671-5p inhibitor (Supplementary Fig. 1D). QRT-PCR and western blotting results showed that hsa_circ_001726 knockdown repressed PRMT9 expression in MHCC97H and Huh7 cells, which was reversed by miR-671-5p inhibitor (Fig. 3R and S). All these data indicated that hsa_circ_001726 sponged miR-671-5p to elevate PRMT9 expression.

The mechanism of hsa_circ_001726/miR-671-5p/PRMT9 axis in HCC development was investigated. PRMT9 overexpression enhanced PRMT9 expression in MHCC97H and Huh7 cells, as determined by qRT-PCR and western blotting (Supplementary Fig. 1E, F). Then, the influence of hsa_circ_001726 on proliferation of HCC cells was detected by EdU staining. Hsa_circ_001726 silencing reduced proliferation of MHCC97H and Huh7 cells. Proliferation ability was elevated in MHCC97H and Huh7 cells in the presence of si-hsa_circ_001726 and PRMT9-OE (Fig. 4A and B). Wound healing and transwell invasion assays showed that hsa_circ_001726 silencing-mediated inhibition in migration and invasion of MHCC97H and Huh7 cells was rescued by PRMT9 overexpression (Fig. 4C-F). PRMT9 up-regulation reversed si-hsa_circ_001726-mediated promotion in apoptosis of MHCC97H and Huh7 cells, as determined by TUNEL staining (Fig. 4G and H). Moreover, miR-671-5p mimic severely repressed migration and invasion of MHCC97H and Huh7 cells (Supplementary Fig. 3A, B). Thus, hsa_circ_001726 silencing repressed malignant phenotypes of HCC cells, which may attribute to target regulate miR-671-5p/PRMT9 axis.

EMT is an important mechanism of tumor metastasis [31]. Our previous study has confirmed that PRMT9 accelerates invasion and metastasis of HCC cells through EMT process [23]. We tested if hsa_circ_001726/miR-671-5p/PRMT9 affect HCC progression through EMT process. The expression of EMT markers in tumor tissues with high expression of hsa_circ_001726 was examined by immunohistochemistry (IHC) assay. Up-regulation of N-cadherin and down-regulation of E-cadherin were observed in primary and metastatic tumor tissues (Fig. 5A). The correlation between hsa_circ_001726 and N-cadherin/E-cadherin in HCC tumor tissues was analyzed. The results of qRT-PCR showed that hsa_circ_001726 expression was positively correlated with N-cadherin expression, and negatively correlated with E-cadherin expression in HCC tumor tissues (Fig. 5B). In an orthotopic transplantation tumor model, hsa_circ_001726 silencing led to an increase in E-cadherin expression and a decrease in N-cadherin expression in lung tissues, as determined by IHC assays (Fig. 5C). Moreover, the expression of EMT-related proteins in HCC cells was examined by western blotting analysis. Silencing of hsa_circ_001726 led to an up-regulation of E-cadherin and ZO-1, and a down-regulation of N-cadherin and Vimentin in MHCC97H and Huh7 cells, which was abolished by PRMT9 overexpression (Fig. 5D). Additionally, Notch1/Hes1 signaling pathway participates in EMT of cancer cells [32]. We also examined the Notch1/Hes1 signaling pathway, which is known to participate in the EMT of cancer cells. The expression of Notch1 and Hes1 was decreased in MHCC97H and Huh7 cells in the presence of si-hsa_circ_001726, while increased in MHCC97H and Huh7 cells following transfection of si-hsa_circ_001726 and PRMT9-OE (Fig. 5E). These results collectively indicated that hsa_circ_001726 silencing repressed HCC development through Notch1/Hes 1-mediated EMT process.

The carcinogenic role of hsa_circ_001726 in HCC propelled us to further examine the underlying mechanism of its high expression. Hsa_circ_001726 is transcribed from CCT2. We speculated that the transcription factors involved in the transcription of CCT2 may have a similar effect on hsa_circ_001726. Applying PROMO database, we found there were 85 transcription factors may bind to the CCT2 promoter region. Among them, transcription factors with a correlation coefficient greater than 0.7 with CCT2, including E2F4, E2F5, and E2F6, especially E2F6 (Fig. 6A). Analysis of TCGA-LIHC database showed that E2F4, E2F5, and E2F6 were upregulated in tumor tissues of HCC compared to normal tissues (Fig. 6B). Furthermore, the expression of E2F6 in HCC cell lines was detected by qRT-PCR and western blotting, showing significant upregulation in Huh7, MHCC97H, Sk-Hep1, and Bel7402 cells compared to the normal hepatic cell line LO2 (Fig. 6C and D). These data indicated that E2F6 has a role in HCC development. To test if E2F6 can interact with CCT2 promoter, ChIP assay was carried out. Anti-E2F6 notably elevated the level of CCT2 in Huh7 and MHCC97H cells, indicating that E2F6 binds to the CCT2 promoter (Fig. 6E). Luciferase reporter assay revealed that the activity of luciferase was inhibited by E2F6 knockdown, showing that E2F6 knockdown repressed the activity of CCT2 promoter (Fig. 6F). Thus, we speculated that E2F6, as a transcription factor of CCT2, regulated the transcription of CCT2 and also affected the expression of hsa_circ_001726. E2F6 was silenced in Huh7 and MHCC97H cells, and the influence of E2F6 on hsa_circ_001726 expression was evaluated by qRT-PCR. E2F6 was severely down-regulated in Huh7 and MHCC97H cells in the presence of si-E2F6-2 and si-E2F6-3, especially si-E2F6-3 (Supplementary Fig. 1G, H). The expression of hsa_circ_001726 was notably decreased in Huh7 and MHCC97H cells upon transfection of si-E2F6 (Fig. 6G). Taken together, these findings suggested that hsa_circ_001726 was regulated by E2F6 in HCC.