Human liver cancer cell lines Huh7 and Hep3B and normal human liver cell line L02 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Huh7 and Hep3B CSCs were enriched and maintained on poly-HEMA coated plates in serum-free DMEM/Nutrient Mixture F-12 (F-12) medium containing 20 ng/ml epidermal growth factor (EGF) (236-EG-200, R&D Systems), 10 ng/ml fibroblast growth factor (FGF) (233-FB-025, R&D Systems), and 1% penicillin/streptomycin (Pang et al., 2010; Li et al., 2015). For preparing poly-HEMA coated plates, 6-well plates were pre-coated with 1.2% (w/v) poly-HEMA (Re et al., 1994).

Cell viability was measured by Cell Counting Kit-8 (CCK-8) (Dojindo Laboratories) according to the user’s manual. The cell viability in each group is expressed as the percentage of untreated control cell viability (Wu et al., 2017).

To examine the expression of CD133 and CD44, Huh7 and Hep3B stem cells were digested with 0.05% trypsin. Next, 10⁶ cells/100 μl of single cells were resuspended and incubated with PE-labeled CD133 (1:50, Miltenyi Biotec) or CD44 (1:50, Miltenyi Biotec) in the dark for 15 min, washed twice with cold phosphate-buffered saline (PBS), resuspended in 400 μl PBS, and analyzed using flow cytometry (Becton Dickinson FACS Vantage SE, San Jose, CA, United States).

To analyze cell apoptosis, Huh7 stem cells were digested with 0.05% trypsin. Then, 1 × 10⁶ single cells were resuspended and mixed with 10 μl Annexin V-fluorescein isothiocyanate (FITC, 130-097-928, Miltenyi Biotec), incubated in darkness for 15 min, washed with 1 ml 1× Annexin V Binding Buffer and resuspended in 500 μl 1× Annexin V Binding Buffer, mixed with propidium iodide (PI) solution, and then analyzed by flow cytometry (Cheng et al., 2017).

Total RNA was isolated using a Tissue RNA Kit (R6311-01, Biomiga). RNA (1 μg) was reverse-transcribed into cDNA using GoScript Reverse Transcriptase (A5001, Promega). Quantitative real-time PCR was completed using the Power Up SYBR Green Master Mixture (Thermo Fisher) with the StepOne Plus Real-Time PCR System (Thermo Fisher), according to a protocol from a previous study (Wu et al., 2017). Specific primers for CD90 and EPCAM were created according to Luo et al. (2015). Specific primers for CD133, OCT4 (POU5F1), NANOG, SOX2, and beta-actin were created according to Ma et al. (2010).

All of the mice were maintained in a pathogen-free facility, and all of the animal experiments were approved by the Committee on the Ethics of Animal Experiments of the Naval Medical University, China. For the animal experiments, 6-week-old female nude BABL/c mice were used, and 2 × 10⁶ Huh7 or Hep3B cells were subcutaneously inoculated into the nude mice (Ma et al., 2018; Yuan et al., 2015). Three weeks later, PBS (control group) or 4 g/kg AA was injected intraperitoneally twice daily for 26 days. The tumor volume was calculated as: total volume = (length × width²)/2 (Naito et al., 1986). Lung and liver tissues were fixed with 4% polyformaldehyde, and serial sections (four sections per tissue with a 30-μm step) were created and stained with hematoxylin and eosin (HE) (Cheng et al., 2017).

Western blot was completed according to a protocol from a previous study (Wu et al., 2017). Briefly, cells or tissues were lysed with Radioimmunoprecipitation Assay (RIPA) Lysis Buffer (P0013C, Beyotime Biotechnology, China) and centrifuged at 13,000 rpm for 15 min. The supernatant was separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel and transferred to a polyvinylidene difluoride (PVDF) membrane. The membrane was incubated overnight with anti-NANOG (1:500, ab109250, Abcam), anti-SOX2 (1:500, ab92494, Abcam, UK), anti-ALDH1A1 (1:1,000, ab52492, Abcam), or anti-β-actin (1:1,000, 3700S, Cell Signaling Technology) primary antibodies, washed with Tris-buffered saline (TBS) containing 0.1% Tween-20 (TBST) three times, incubated with secondary antibody (926-32210, 1:20,000 for β-actin and 926-32211, 1:5,000 for others, LI-COR, Biosciences), and analyzed with the Odyssey Infrared Imaging System (LI-COR, Biosciences).

The H₂O₂ concentration was measured using a H₂O₂ Assay Kit (S0038, Beyotime Biotechnology, China) according to the user’s manual. Simply, 1 × 10⁶ cells were lysed in 200 μl lysis buffer and centrifuged for 5 min at 12,000 rpm. Every 50 μl of the supernatant was mixed with 100 μl of H₂O₂ detection reagent and incubated for 30 min at room temperature. Absorbance was determined at 560 nm using an Epoch Microplate Spectrophotometer (BioTek). For catalase experiments, catalase was added prior to AA treatment.

For the sphere formation experiment, cells were digested into single cells with trypsin. Then, 100 cells/well were plated into a 96-well ultra-low attachment plate and cultured for 2 weeks in serum-free DMEM/F-12 medium containing 20 ng/ml EGF, 10 ng/ml FGF, and AA (0, 0.5, or 1 mM). The number of spheres was counted and photographed.

For the colony formation experiment, 1,000 cells/well were plated into 6-well plates. The colonies were cultured in DMEM containing 10% fetal bovine serum, 1% penicillin/streptomycin, and AA (0, 0.5, or 1 mM). The colonies were then stained with 1% crystal violet.

Statistical analysis was performed using unpaired t tests when comparing two different groups or one-way ANOVA with Tukey’s multiple comparison tests. IC50 values were calculated using Prism software (GraphPad, San Diego, CA, USA) by nonlinear regression to dose-response curves, and expressed as mean and 95% confidence intervals (CI). The data are expressed as the mean ± SEM. p < 0.05 was considered statistically significant.

Two human liver cancer cell lines (Huh7 and Hep3B), the respective CSCs, and a normal human liver cell line L02 were treated with AA at the concentrations of 0, 0.5, or 1 mM, which are easily achievable clinically by intravenous infusion (Chen et al., 2008) (Hoffer et al., 2008). The results showed that AA inhibited the viabilities of liver cancer cells and liver CSCs in a concentration-dependent manner (Figures 1A–D). AA at the concentration of 1 mM decreased the viabilities of Huh7 and Hep3B cells to 12.15 and 5.77%, respectively (Figures 1A,B). For Huh7 and Hep3B CSCs, the viabilities were decreased to 52.37 and 33.04%, respectively, at 1 mM concentration of AA (Figures 1C,D). The IC50 values of AA for Huh7, Hep3B, and Huh7 CSCs and Hep3B CSCs were 0.67, 0.32, 1.21, and 0.52 mM, respectively (Figure 1E). However, AA did not display significant inhibitory effects on the viability of L02 cells at 0.5 mM or 1 mM concentrations (Figure 1F). Together, these data indicated that AA was responsible for selective inhibitory effects on the viabilities of liver cancer cells and liver CSCs.

We further examined the effects of AA on sphere formation and colony formation. As shown in Figure 2A, AA treatment reduced the volume of spheres formed by Huh7 cells. The number of spheres larger than 50 μm in diameter was markedly decreased in a concentration-dependent manner in AA-treated Huh7 cells (Figure 2B). Twenty-two spheres were formed for every 100 cells in the control group, whereas only two spheres were formed for every 100 cells in the group treated with 1 mM AA. Similar results were obtained for Hep3B cells (Figures 2C,D). As shown in Figure 2E, AA treatment also markedly decreased colony formation in a concentration-dependent manner in Huh7 and Hep3B cell lines. Collectively, our data showed that AA reduced sphere formation and colony formation by liver cancer cells, indicating the inhibitory effects of AA on self-renewal and tumorigenicity of liver cancer cells.

We determined the effects of AA on tumor growth in mice bearing Huh7 and Hep3B xenografts. As mentioned above, AA concentrations in human plasma and cells were tightly controlled. With the oral ingestion of high doses of vitamin C, even at 100 times the recommended dietary allowance, the plasma concentration rarely exceeds 200 µM. Both i.v. and i.p. administration of ascorbate induced pharmacologic serum ascorbate concentrations up to 20 mmol/L. To obtain a pharmacologic serum ascorbate concentration, the i.p. administration method was selected. Compared with the PBS control group, AA treatment significantly suppressed the growth of Huh7 and Hep3B xenograft tumors in vivo (Figures 3A,B) without significantly changing the animal’s body weight (Figures 3C,D).

As shown in Figures 4A,B, AA-treated mice developed fewer metastatic lung tumors as compared to the control group. The number of metastatic lung tumors in AA-treated mice was 0.90 ± 0.40 (n = 5), and that in the control mice was 6.25 ± 2.27 (n = 5) (Figure 4C). The area ratio of metastatic lung tumors in AA-treated mice was 0.29 ± 0.17 (n = 5), and that in control mice was 14.61 ± 6.91 (n = 5) (Figure 4D). The metastatic tumors in the livers of either the control or AA groups were small (Figures 4E,F). In the control group, 5 of 5 mice developed metastatic lung tumors, whereas 3 of 5 mice exhibited metastatic lung tumors in the AA-treated group (Figure 4G). Additionally, in the control group, 4 of 5 mice developed metastatic liver tumors, while in the AA-treated group, 1 of 5 mice developed metastatic liver tumors (Figure 4H). In summary, our data demonstrated that AA treatment reduced liver and lung metastasis of liver cancer cells inoculated subcutaneously into nude mice.

We investigated the effects of AA on the expression of stemness genes. Flow cytometric analysis showed that AA treatment increased CD133⁺ cells and CD44⁺ cells in both Huh7-and Hep3B-derived stem cells (Figures 5A,B). CD133 antigen was identified as a CSC marker in various cancer types, including liver cancer. CD44, a transmembrane glycoprotein, is also considered as an important liver CSC marker (Zhu et al., 2010; Yang et al., 2008). For Huh7 CSCs, AA at 1 mM increased CD133⁺ cells and CD44⁺ cells from 2.90 to 14.70%–4.29 and 24.19%, respectively (Figures 5A,B). For Hep3B CSCs, CD133⁺ cells and CD44⁺ cells were increased by AA from 20.40 to 0.75%–24.22 and 4.51%, respectively (Figures 5A,B). Western blot analysis showed that the protein levels of embryonic stem cell markers NANOG and SOX2 as well as liver CSC marker ALDH1A1 were increased after treatment with AA in Huh7-and Hep3B-derived stem cells (Figures 5C,D).

We also examined the effects of AA on the expression of stemness genes in liver tumors in vivo. Consistent with the in vitro results, the mRNA expression levels of NANOG, OCT4, SOX2, EPCAM, CD133, and CD90 were upregulated in the AA-treated tumors (Figure 6A). Also, the protein level of NANOG was increased in the AA-treated group as compared with that of the control group (Figures 6B,C). Collectively, our data showed that AA upregulated the expression of stemness genes in liver cancer cells in vitro and in vivo.

It was reported that H₂O₂ plays an important role in AA’s anticancer activity (Lennicke et al., 2015; Chaiswing et al., 2018). To determine the role of H₂O₂ in the inhibitory effect of AA on liver CSCs, we first evaluated the concentrations of H₂O₂ in Huh7-derived CSCs with or without AA treatment. As shown in Figure 7A, AA treatment increased the concentration of H₂O₂ in Huh7-derived CSCs. Furthermore, AA increased the protein levels of cleaved poly (ADP-ribose) polymerase (PARP) and cleaved caspase-7 (Figure 7B) and promoted cell apoptosis (Figures 7C,D).

Catalase, as a specific H₂O₂ scavenger, converts the ROS H₂O₂ to water and oxygen and thereby mitigates the cytotoxic effects of H₂O₂. We also found that the addition of catalase reversed the effects of AA on the production of H₂O₂ and the cleavage of PARP and caspase-7 (Figures 7E,F). More importantly, the addition of catalase reduced the inhibitory effects of AA on liver CSC viability (Figure 7G), which was consistent with previous reports describing the dependence of AA’s cytotoxicity on the generation of H₂O₂ (Du et al., 2010; Chen et al., 2015; Chen et al., 2005). In conclusion, our data indicate that AA exerts its inhibitory effects on liver CSCs through the production of H₂O₂ and the promotion of cell apoptosis.