Liver cancer tissues and corresponding non-tumor tissues were obtained from the Beijing 302 Hospital. Sixteen HCC samples (14 males and 2 females; range, 49–65 years) were obtained from consecutive patients undergoing initial hepatectomy from January 2015 to December 2017. All human specimens were collected in accordance with the Declaration of Helsinki, and the protocols involving clinical samples were approved by the Research Ethics Committees of the Academy of Military Medical Sciences, Beijing, China. Written informed consent was obtained from all patients.

Liver cancer cells including Huh7, Hep3B, Li-7 and PLC/PRF/5, (CRL-8024™; also referred to as PLC) from the American Type Culture Collection (ATCC; Manassas, VA, USA), the MHCC97H (briefly, 97H), MHCC97L (briefly, 97L), HCCLM3 (briefly, LM3) cells were obtained from the Liver Cancer Institute of Fudan University (Shanghai, China) (25). The HepG2 cells were obtained from the National Platform for Experimental Cell Resources (Beijing, China). These cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin and 100 ng/ml streptomycin at 37°C under 5% CO₂.

The TCGA database (https://tcga-data.nci.nih.gov/tcga/) was used to analyze differentially expressed miRNAs between liver cancer and adjacent non-tumor tissues.

Total RNAs from liver cancer specimens and cells were extracted using TRIzol (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA). Reverse transcription of total RNA into cDNA was performed with the miScript Reverse Transcription kit (Qiagen Sciences, Inc., Gaithersburg, MD, USA), and qPCR was performed using the miScript SYBR-Green PCR Master Mix on the ABI Prism 7900 system (Applied Biosystems; Thermo Fisher Scientific, Inc.). Briefly, a 20 µl RTqPCR system was performed for 40 cycles according to the following conditions: An initial denaturation was performed at 95°C for 3 min, followed by denaturation at 95°C for 15 sec, annealing at 58°C for 30 sec, and extension at 72°C for 7 min. Relative quantification of miRNA or mRNA expression was calculated using the 2^−∆∆Cq method (26). 18S was used as an mRNA internal control. U6 RNA was used as a miRNA internal control. Each experiment was conducted with at least three independent replicates. Primer sequences are listed in Table I.

Fluorescence-activated cell sorting was performed to enrich EpCAM⁺ and CD13⁺ liver cancer cells. Briefly, Huh7 cells were trypsinized and washed twice with phosphate-buffered saline (PBS), then incubated with APC-anti-human EpCAM (dilution 1:200; cat. no. 324208; BioLegend, San Diego, CA, USA) or APC-anti-human CD13 antibody (dilution 1:100; cat. no. 301706; BioLegend) for 30 min at 4°C. After incubation, Huh7 cells were washed 3 times and then sorted on the FACSAria II instrument (BD Biosciences, San Jose, CA, USA). Huh7-EpCAM⁺ (or CD13⁺) and Huh7-EpCAM⁻ (or CD13⁻) cells were collected for qPCR analysis.

A total of 1,000 tumor cells were seeded in the 6-well ultra-low attachment plates (Corning Inc., Corning, NY, USA) and cultured in DMEM/F-12 supplemented with B27, N2 (both from Invitrogen; Thermo Fisher Scientific, Inc.), 10 ng/ml of epidermal growth factor (EGF), 5 ng/ml of basic fibroblast growth factor (bFGF), 100 U/ml penicillin and 100 ng/ml streptomycin at 37°C under 5% CO₂ for 7–10 days as we previously described (27,28).

Transient expression of the miR-486 mimics or miR-486 inhibitors have been previously described (29). Scramble miRNA, lenti-miR-486 and lenti-miR-486 inhibitor were purchased from Suzhou GenePharma Co., Ltd. (Suzhou, China). The sequence of hsa-miR-486 mimics was: TCCTGTACTGAGCTGCCCCGAG, that of hsa-miR-486 inhibitor was: ACCCCTATCACGATTAGCATTAA, and that of the miRNA inhibitor negative control (NC) was: CAGTACTTTTGTGTAGTACAA.

To generate lentivirus plasmids for stable RNA interference, short hairpins were designed using online software (http://rnaidesigner.lifetechnologies.com/rnaiexpress/design.do) as previously described (30). Based on the Sirt1 sequence, the small interfering RNA (siRNAs) Sirt1 sequence was 5′-GGUGCCGUGCUACUCAUAUTT-3′. The Sirt1-specific short hairpin (shRNA) expression vector and the scrambled ineffective shRNA cassette were cloned into the pSicoR-GFP plasmid. Virus packaging was performed in 293T cells after co-transfection of lentiviral expression plasmids with the packaging plasmids (pRRE, pCMV-VSVG and pRSV-REV; Addgene, Inc., Cambridge, MA, USA) using Lipofectamine 2000. The viral supernatant was used to infect liver cancer cells, and stable GFP-expressing clones were selected by fluorescence-activated cell sorting (FACS).

The 3′UTR sequences of Sirt1 that contained the predicted complimentary sites of miR-486 were cloned into the pGL3 basic reporter vector (Promega Corp., Madison, WI, USA). The pGL3-Sirt1-3′UTR vector or pGL3 basic vector was co-transfected with miR-486 mimics or NC into 293T cells, with the Renilla luciferase vector used as an internal control. After 48 h, cells were harvested, and the luciferase activity was detected using the Dual-Luciferase assay kit (Promega). The ratio of firefly/Renilla luciferase activities was calculated and designated as the relative promoter activity.

Cell invasion assays were performed in Transwell chambers coated with Matrigel (8 µm Transwell inserts; BD Biosciences). A total of 1×10⁶ cells in serum-free DMEM were seeded in the upper chamber, and DMEM with 5% FBS was added to the bottom chamber as an attractant. After 48 h of incubation, the penetrated cells on the filters were fixed in 20% methanol and stained with crystal violet. Ten randomly selected fields (magnification, ×100) in each well were counted under a light microscope.

The effects of miR-486 on the migration of tumor cells were determined by wound healing assays. A total of 5×10⁵ cells/well were seeded and cultured in 6-well plates to create a confluent monolayer. After the cells attached overnight, a straight line was scraped with a 200-µl pipette tip across the cell monolayer to create a ‘scratch’. After 48 h, microscopic images of the ‘scratch closure’ were captured under a light microscope at an original magnification of ×4.

Female nude mice (6–8 weeks old) were purchased from Weitonglihua Company (Beijing WeitongLihua Experimental Animal Technology Co., Ltd., Beijing, China) and raised under specific pathogen-free conditions. Mice were allowed ad libitum access to food and water and were maintained on a constant light dark cycle with constant temperature and humidity. All animal experiments in the present study were approved by the Ethics Committee of the Academy of Military Medical Sciences. Huh7-miR-486 or HepG2-miR-486 and corresponding control cells (2×10⁶) were subcutaneously injected into the right and left side of mice, respectively. Similarly, Hep3B-shSirt1 or HepG2-shSirt1 and control cells were subcutaneously injected into nude mice. Approximately 5 weeks after injection, mice were anesthetized with 1% pentobarbital sodium (75 mg/kg) through the intraperitoneal (IP) administration, and sacrificed by cervical dislocation, and the tumors were weighed as previously described (23). Non-retrospective ethical approval was obtained for the animal experiments conducted in the study. Tumor burden did not exceed the recommended dimensions.

The data are presented as the means ± standard deviation (SD) from at least 3 independent experiments. Quantitative results were compared using GraphPad Prism version 5.0 software (GraphPad Software, Inc., La Jolla, CA, USA). A two-tailed Student's paired t-test was used to test for significance between two groups. Multiple groups were compared by a one- or two-way analysis of variance with Tukey's post hoc correction. Statistical significance was regarded as P<0.05 or P<0.01.

To identify differentially expressed miRNAs between CSC-like cells and adherent cells, we first used the TCGA miRNA sequence database to analyze the altered miRNAs between liver cancer specimens (n=369) and adjacent non-tumor tissues (n=50). We mainly focused on the downregulated miRNAs in liver cancer tissues, and identified a set of 59 miRNAs that were significantly downregulated in liver cancer specimens compared with adjacent non-tumor tissues, including miR-486, miR-99a, miR-195 and miR-154 (Fig. 1A). Then, we enriched liver CSC-like cells using tumor sphere models (Fig. 1B), and qPCR assays were further utilized to compare the differentially expressed miRNAs between tumor spheres and adherent cells. As shown in Fig. 1C, the expression levels of miR-99a, miR-10a, miR-486, miR-195, miR-154 and miR-138 in liver tumor spheres were significantly decreased compared with adherent cells. Moreover, liver CSCs were enriched with CSC surface markers including EpCAM and CD13. Thus, we performed fluorescence-activated cell sorting (FACS) to separate Huh7 cells into CD13⁻ and CD13⁺, and EpCAM⁻ and EpCAM⁺ subsets, respectively (Fig. 1D). According to the qPCR results, miR-486 was significantly downregulated in EpCAM⁺ and CD13⁺ subpopulations (Fig. 1E). Moreover, miR-486 levels were significantly decreased in tumor samples compared with corresponding non-tumor tissues (Fig. 2A) and most tumor cell lines (Fig. 2B). Among the 10 most frequently used liver cancer cell lines, miR-486 was lowly expressed in Huh7 (highly proliferative) (23,29) and 97H cells (highly invasive) (25,31). In our laboratory, only 3 liver cancer cell lines including Huh7, PLC/PRF/5 and Li-7 cells could form tumor spheres easily in vitro (data not shown). The tumor sphere-forming cells from tumor patients and cell lines possessed more CSC-like features and tumorigenic potential. In addition, the Huh7 cells could easily form tumors in vivo than PLC/PRF/5 and Li-7 cells (data not shown). These results indicated that miR-486 was significantly downregulated in liver CSCs and liver cancer tissues.

First, we confirmed the relative expression of miR-486 in miR-486 overexpression and inhibitor expression in liver cancer cells (Fig. 3A). Compared to the NC group, the number of tumor spheres was decreased in Huh7 cells transfected with lentivirus miR-486 (LV-miR-486) (Fig. 3B). miR-486 overexpression suppressed the expression of stemness-related gene CD13. Conversely, miR-486 inhibitor promoted the expression of CD13 in liver cancer cells (Fig. 3C). The Cell Counting Kit-8 (CCK-8) assays indicated that miR-486 overexpression significantly suppressed the proliferation of Huh7 and LM3 cells in vitro (Fig. 3D). Moreover, the clone-forming capacity in miR-486 overexpression Huh7 cells was significantly decreased compared with NC group (Fig. 3E). Then, Huh7-miR-486 and Huh7-NC cells were subcutaneously injected into nude mice. Consequently, the tumor sizes and weights in the Huh7-miR-486 group were significantly decreased compared to the Huh7-NC group (P<0.01; Fig. 3F, left panel). Similar results were observed in mice injected with HepG2-miR-486 cells (P<0.01) (Fig. 3F, right panel). The cell invasion and scratch healing assays revealed that miR-486 mimics significantly decreased the number of invading cells (Fig. 4A) and the migration distance of liver cancer cells (Fig. 4B). Collectively, these results demonstrated that miR-486 has tumor suppressive functions in CSC traits.

miRNAs usually bind to the complementary sites in the 3′UTRs of target mRNAs and trigger RNA degradation or translation repression (32). Both TargetScan and miRanda database predicted that Annexin A13 (ANXA13), Sirt1 and tenascin-C (TNC) could serve as potential target genes of miR-486 (Fig. 5A). The results of the qPCR analysis confirmed that miR-486 suppressed the expression of Sirt1 and TNC in HepG2 and PLC cells. Conversely, Sirt1 and TNC mRNA levels were increased with transfection of a miR-486 inhibitor (Fig. 5B). Our preliminary results indicated that knockdown of TNC did not inhibit the growth of tumor spheres in vitro (data not shown). Thus, in the present study, we mainly focused on Sirt1 as the target gene of miR-486. To further elucidate the interaction between miR-486 and Sirt1 3′UTR, a luciferase assay was subsequently performed in 293T cells. The pGL3-Sirt1-3′UTR vector or pGL3 basic vector was co-transfected with miR-486 mimics or NC into cells, and the luciferase activity was detected using the Dual-Luciferase Assay System. Co-transfection with miR-486 mimics significantly decreased the firefly luciferase activity of Sirt1-3′UTR reporter but not that of the pGL3 reporter (P<0.01; Fig. 5C). Next, we also observed downregulation of Sirt1 in the miR-486 overexpression group by western blot analysis (Fig. 5D). Moreover, the IHC staining indicated that Sirt1 expression was downregulated in miR-486 overexpression cell-derived tumor tissues (Fig. 5E). In summary, these findings indicated that Sirt1 is a direct target of miR-486.

Sirt1 mRNA levels were assessed in tumor spheres and adherent cells. The expression levels of Sirt1 were significantly higher in tumor spheres compared with adherent cells (P<0.01; Fig. 6A). This was consistent with our results revealing that miR-486 expression was decreased in tumor spheres and that Sirt1 expression was inversely associated with miR-486 levels. Additionally, increased Sirt1 expression was observed in liver cancer tissues compared with adjacent non-tumor tissues (Fig. 6B). Then, the expression of Sirt1 in liver cancer cell lines (Fig. 6C) was analyzed. To understand the role of Sirt1 in the maintenance of liver CSC characteristics, we transfected liver cancer cells with Sirt1-shRNA lentivirus, and confirmed that its expression was suppressed by qPCR analyses (Fig. 6D). To determine the biological function of Sirt1 in maintaining CSC-like characteristics, tumor sphere assays were performed. As revealed in Fig. 6E, knockdown of Sirt1 markedly inhibited tumor sphere formation. Conversely, overexpression of Sirt1 (oeSirt1) was capable of promoting tumor sphere formation (Fig. 6F). To further investigate the biological functions of Sirt1 in liver cancer progression in vivo, HepG2-shSirt1 and HepG2 control cells were subcutaneously injected into nude mice to monitor tumor growth. Tumor weights were reduced in the HepG2-shSirt1 cells as compared with the control group (Fig. 6G). Similar results were obtained with Hep3B-shSirt1 cells (Fig. 6G). Moreover, Sirt1 knockdown reduced the invading capacity of the miR-486 inhibitor-transfected LM3 cells (Fig. 6H). Collectively, these data indicated that Sirt1 was responsible for the maintenance of self-renewal of liver CSCs and tumorigenesis in vivo.