HCCLM3 and HepG2 cells were obtained from Shanghai TongBai Biological Technology Co., Ltd., and were authenticated by STR profiling. Dimethyl sulfoxide (DMSO; product no. 276855), SC-79 (product no. 123871) and LY294002 (product no. 528108), antibodies to Cx32 (product no. MAB3069), and secondary antibodies including HRP-conjugated goat anti-rabbit IgG (H + L) (cat. AP307P) and HRP-conjugated goat anti-mouse IgG (H + L) (cat. AP308P) were purchased from Sigma-Aldrich; Merck KGaA. Antibody to Sox-9 (cat. no. PA5-81966) was obtained from Invitrogen; Thermo Fisher Scientific, Inc. Antibodies to phosphorylated (p)-Akt (product code ab81283), total Akt (product code ab8805), EpCAM (product code ab223582), CD133 (product code ab222782), Nanog (product code ab14959), Oct4 (product code ab181557), c-Myc (product code ab32072), β-actin (product code ab8226) were obtained from Abcam. Short hairpin RNA (shRNA)-Cx32 and overexpression (OE)-Cx32 plasmid were obtained from Shanghai GenePharma Co., Ltd.

The specimens of 85 patients (63 males and 22 females; aged 25–78 years) with liver cancer were collected from the Department of Hepatobiliary Surgery, the First Affiliated Hospital of Bengbu Medical College (Bengbu, China) from January 2014 to December 2015. A total of 124 patients were followed-up to December 2020, but 39 patients were lost to follow-up in the present study. Inclusion criteria were as follows: A pathological diagnosis of liver cancer, complete clinical data records, no history of chemotherapy, radiation, immunotherapy, or other related treatment, no history of other cancers, and no metastases from other organs. The collected specimens included HCC tissues and the corresponding paracancerous tissues. The paracancerous tissues were non-cancerous liver tissues located at least 5 cm from the lesions. This study was approved by the Ethics Committee of Bengbu Medical College (approval no. 2020047).

Proteins of tissues and cells were extracted using a lysis buffer containing a protease inhibitor (product no. P0013B; Beyotime Institute of Biotechnology) and quantified using a BCA protein assay kit (product no. P0012; Beyotime Institute of Biotechnology). Total proteins (20 µg per lane) were separated using 10% SDS-PAGE. The resolved proteins were transferred to a PVDF membrane. The PVDF membrane was then incubated at 4°C for 1 h with 1X Protein Free Rapid Blocking Buffer (cat. no. PS108P; Shanghai Epizyme Biomedical Technology Co., Ltd.). Following sealing, the membrane was incubated at 4°C overnight in the corresponding primary antibody solution containing EpCAM (1:1,000), CD133 (1:2,000), Nanog (1:2,000), Oct4 (1:2,000), Sox9 (1:1,000), c-Myc (1:1,000), Cx32 (1:1,000), p-Akt (1:2,000), total Akt (1:1,000), or β-actin (1:1,000). The membrane was washed the following day and incubated with the secondary antibodies, HRP-conjugated goat anti-rabbit IgG or HRP-conjugated goat anti-mouse IgG (1:2,000 or 1:4,000), incubated at 4°C for 1 h. Finally, the membrane was developed in the dark after incubating with enhanced chemiluminescence reagent (product no. RPN2235; Cytiva). The gray value of the protein bands was analyzed by ImageJ software [version 1.46r; National Institutes of Health (NIH)].

The expression of Cx32 and p-Akt was detected by IHC. Paraffin-embedded specimens were cut into 5 µm-thick slices using a microtome. The tissue sections were dewaxed and hydrated, endogenous peroxidase was inactivated with 3% hydrogen peroxide, antigen repair was performed by microwave irradiation. The tissue sections were blocked at room temperature in 5% bovine serum albumin (BSA)/50 mM PBS (pH 7.4) and incubated with the Cx32 antibody (1:100; product no. MAB3069; Sigma-Aldrich; Merck KGaA) and p-Akt antibody (1:100; product code ab81283; Abcam) at 4°C overnight. The sections were washed with phosphate-buffered solution and incubated with secondary antibodies, goat anti-mouse IgG HRP conjugate (1:200; product no. 12-349) and goat anti-rabbit IgG HRP conjugate (1:200; product no. 12-348; both from Sigma-Aldrich; Merck KGaA), followed by color development with 3,3′ diaminobenzidine (DAB) kit (product no. D3939; Sigma-Aldrich, Merck KGaA), redyeing, dehydration, and sealing. Each section was observed and images were captured using model IX71 optical microscope (Olympus Corporation). All imaged sections stained by immunohistochemistry were analyzed using ImageJ software version 1.46r (NIH). The criteria for staining cells were the same as those previously reported (15).

Agarose (1%) was spread on a 10-cm cell culture dish and allowed to coagulate. A total of 3×10⁵ cells/ml were plated per well of a gel-coated petri dish and maintained with B27 serum-free microsphere medium (Gibco; Thermo Fisher Scientific, Inc.). N2 (Gibco; Thermo Fisher Scientific, Inc.), 20 ng/ml epidermal growth factor, and 20 ng/ml basic fibroblast growth factor (Invitrogen; Thermo Fisher Scientific, Inc.) were added to DMEM/F12 basal culture medium (Gibco; Thermo Fisher Scientific, Inc.). After 3–4 days, the cells were supplemented with microsphere medium and cultured for another 4–6 days. Images of the spheroids were captured and counted by fluorescence microscope (Olympus Corporation).

TCGA Liver Hepatocellular Carcinoma (LIHC) data were obtained from UCSC Xena (URL: http://xena.ucsc.edu). Screened samples expressed GJB1 (Cx32) and had sufficient survival information. The R language Survival package (https://CRAN.R-project.org/package=survival) was used to evaluate the association between Cx32 expression level and survival outcomes, such as overall survival (OS), disease-free interval (DFI), progression-free interval (PFI), and disease-specific survival (DSS).

All animal experiments were conducted in accordance with the regulations approved by the Laboratory Animal Management and Ethics Committee of Bengbu Medical College. A total of 64 male BALB/c mice (4 weeks of age, 20–22 g) were purchased from Beijing Weitong Lihua Experimental Animal Technology Co., Ltd. They were bred in individually ventilated cages (IVC) with specific pathogen-free (SPF) conditions in a 12-h light/dark cycle, 40–70% relative humidity, and controlled temperature (24±2°C). During the experiment, all mice had free access to standard mice chow and water. Following one-week acclimation, the mice were divided randomly into two groups: HCCLM3 overexpression empty vector (HCCLM3 EV) group and HCCLM3 overexpression (HCCLM3 OE) group. Different numbers of HCC cells (1×10³, 5×10³, 1×10⁴, and 5×10⁴) were suspended in Matrigel (50%; BD Biosciences) and injected subcutaneously into the nude mice to observe tumor growth. Finally, the mice were divided into eight groups, with eight mice in each group. Two months later, the mice were euthanized by cervical dislocation under anesthesia (1% pentobarbital sodium, 80 mg/kg, ip) and tumorigenicity of the cells in vivo was evaluated. The evaluation criterion was negative if no obvious tumor nodules were found at the injection site of the nude mice. The animal experiments were approved by the Ethics Committee of Bengbu Medical College (approval no. 2020090).

Total RNA of cells was extracted using TRIzol^® (cat. no. 15596026; Invitrogen; Thermo Fisher Scientific, Inc.). RNA purity was evaluated by Agilent Bioanalyzer 2100 (Agilent Technologies, Inc.) before RT-qPCR analysis. Subsequently, the cDNA was reverse-transcribed using TaqMan Reverse Transcription reagent (cat. no. N8080234; Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol, and was detected by fluorescence qPCR using SYBR^™ Green PCR Master Mix (cat. no. 4309155; Invitrogen; Thermo Fisher Scientific, Inc.). PCR amplification conditions were as follows: Pre-denaturation at 95°C for 15 sec, denaturation at 95°C for 5 sec, and annealing and extension at 62°C for 30 sec. These steps were repeated for 45 cycles. The absorbance value was read at each extension stage. β-actin was used as the reference gene. The relative gene expression levels of each group were calculated according to quantification cycle (Cq) value (23). Primers of β-actin, Sox-9, CD133, EpCAM, Nanog, Oct4 and c-Myc were obtained from Shanghai Shenggong Biological Technology Co., Ltd. Primers for each gene are listed in Table I.

Negative control (NC)-shRNA lentivirus and Cx32-shRNA lentivirus were prepared by inserting NC-shRNA and Cx32-shRNA into the respective lentivirus vectors. Lentivirus were produced in 293T cells (Shanghai TongBai Biological Technology Co., Ltd.) using a second generation lentiviral system (Invitrogen; Thermo Fisher Scientific, Inc.). Briefly, 4×10⁶ 293T cells were seeded in a 100-cm culture dish at 24 h before transfection. Subsequently, 6 µg lentiviral construct, 3 µg psPAX2 (packaging plasmid), and 1.5 µg pDM2.G (envoloping plasmid; both from Sangon Biotech Co., Ltd.) were co-transfected into 293T cells (the mixed ratio was pDM2.G: psPAX2: lentivirus, 1:2:4) using Lipofectamine^™ 2000 (Invitrogen; Thermo Fisher Scientific, Inc.). Following transfection for 6 h at 37°C in a CO₂ incubator, the medium was replaced with normal culture medium. After 48 h, the lentivirus-containing supernatants were harvested, centrifuged at 800 × g for 4 min to pellet cell debris. The HepG2 cells were plated at 30–50% confluence, transfected with lentivirus supernatants at a multiplicity of infection (MOI) of 20 TU/ml. A total of 24 h after transfection, the transformed cells were screened by doxycycline (final concentration, 2 µg/ml) for 7–10 days. The transfection results were identified by western blotting. The sequences for the shRNA targeting Cx32 (shCx32) were: shRNA1, 5′-CCGGGCCGTCTTCATGTATGTCTTTCTCGAGCCCTCACTACATGAAGACGGCTTTTTG-3′; and shRNA2, 5′-CCGGCGTTTGCTATGACCAATTCTTCTCGAGAAGAATTGGTCATAGCAAACGTTTTTG-3′; and the sequence for the NC was: 5′-CCTAAGGTTAAGTCGCCCTCGCTCGAGCGAGGGCGACTTAACCTTAGG-3′.

SPSS 23.0 (IBM Corp.) and SigmaPlot 10.0 software (Jandel Corporation; Systat Software, Inc.) were used for statistical analysis. Data represented the mean ± SEM from three independent experiments. Parametric data were analyzed using unpaired t-test or one-way ANOVA with Tukey's multiple comparison test for post hoc comparison. The association between Cx32 expression and clinicopathological characteristics was determined by Pearson χ² test. The chi-square test and Fisher's exact probability tests were used as appropriate to evaluate the significance of differences in data between groups. Survival curves were analyzed using the Kaplan-Meier method, and significance was assessed by the Gehan-Breslow-Wilcoxon test or log-rank test. P<0.05 was considered to indicate a statistically significant difference.

TCGA data analysis showed that the OS of patients with high expression of Cx32 was significantly higher than those with low expression of Cx32 (Fig. 1A). The same results were observed for the DSS, DFI and PFI (Fig. 1B-D). The data of 85 patients with liver cancer was collected and the expression of Cx32 was assessed in cancer tissues and corresponding paracancerous tissues by IHC. The positive expression rate of Cx32 was 41.2% (35/85) in cancer tissues and 82.4% (70/85) in paracancerous tissues (Fig. 1F). The rate in paracancerous tissue was significantly higher than the rate in cancer tissues. The 85 patients were followed-up for 5 years (Table II). Survival analysis revealed an overall survival rate of 46.39% in the Cx32-positive group and 10.31% in the Cx32-negative group (Fig. 1G; P=0.007), indicating a significant association of Cx32 with prognosis of HCC.

CD133⁺ and EpCAM⁺ liver CSCs, and CD133⁻ and EpCAM⁻ non-liver CSCs were sorted from primary HCC tissues. RT-qPCR was used to determine the level of Cx32. Cx32 was significantly reduced in CD133⁺ and EpCAM⁺ liver CSCs (Fig. 2A and B). Cx32 mRNA expression was also decreased in HCC spheres derived from human primary HCC cells, compared with adherent cells (Fig. 2C). Consistent with these results, Cx32 was reduced in CD133⁺ and EpCAM⁺ liver CSCs sorted from spheres of HepG2 and HCCLM3 cells (Fig. 2D and E). Moreover, compared with the attached cells, Cx32 was obviously decreased in the self-renewing spheroids (Fig. 2F). These results indicated that Cx32 regulated the expansion of LCSCs.

To explore the effect of Cx32 on the expansion of LCSCs, HepG2 cells (that express a high level of Cx32) and HCCLM3 (that express a low level of Cx32) were used (Fig. 3A and B), and Cx32 was silenced in HepG2 cells and overexpressed in HCCLM3 cells (Fig. 3C-F). RT-qPCR and western blotting were used to observe the expression levels of stemness-associated genes, including EpCAM, CD133, Sox9, Nanog, Oct4, and c-Myc. Self-renewal ability was detected by the spheroid formation assay. Cx32 knockdown in HepG2 cells significantly enhanced the mRNA expression of stemness-associated genes (Fig. 4A), and western blotting results were consistent with those of RT-qPCR (Fig. 4C and D). Moreover, after Cx32 was silenced in HepG2 cells, the numbers of spheres were significantly increased (Fig. 4G and H). By contrast, overexpression of Cx32 in HCCLM3 cells significantly decreased the expression of stemness-associated genes (Fig. 4B, E, and F) and cells formed smaller and fewer spheroids than the control cells (Fig. 4I and J). In vivo limiting dilution assay results showed that overexpression of Cx32 markedly downregulated the tumorigenicity capacity of HCCLM3 cells (Fig. 4K-N). Collectively, the results indicated that Cx32 regulated expansion of LCSCs.

It was previously confirmed by the authors that Cx32 regulates the PI3K/Akt pathway in HCC (16). In the present study, IHC was used to detect the expression of p-Akt in HCC tissues and corresponding paracancerous tissues. Only 9.4% (8/85) of patients with HCC had positive expression of p-Akt in the corresponding paracancerous tissues, while the positive expression rate of p-Akt in HCC tissues was 71.8% (61/85) (Fig. 5A and B). The results indicated that p-Akt was activated in HCC tissues. The PI3K/Akt signaling pathway was also activated in HCCLM3 cells. Cx32 was silenced in HepG2 cells and overexpressed in HCCLM3 cells. Western blotting was used to examine PI3K/Akt pathway activity. Cx32 knockdown in HepG2 cells significantly increased the expression level of p-Akt, whereas Cx32 overexpression significantly decreased the expression level of p-Akt (Fig. 5C and D). These findings suggested that Cx32 regulated the activity of the PI3K/Akt signaling pathway in HCC cells.

It was next investigated whether Cx32 affected expansion of LCSCs by regulating the PI3K/Akt signaling pathway. HepG2 cells were exposed to the AKT agonist SC-79 to activate the PI3K/Akt signaling pathway. HCCLM3 cells were exposed to the Akt antagonist LY294002 to inhibit the PI3K/Akt pathway. SC-79 significantly enhanced the expression of stemness-associated genes and the numbers of spheres were significantly upregulated (Fig. 6). However, LY294002 obviously decreased the expression of stemness-associated genes, and hepatoma cells formed small and fewer spheroids. The results suggested that PI3K/Akt regulated expansion of LCSCs. In addition, OE-Cx32-HCCLM cells that stably expressed Cx32 were stimulated by the AKT agonist SC-79 and the effect of SC-79 on expansion of LCSCs in these cells was observed. Overexpression of Cx32 in HCCLM3 cells obviously decreased the expression of stemness-associated genes, and cells formed smaller and fewer spheroids, but SC-79 reversed these effects (Fig. 7). These results indicated that Cx32 regulated expansion of LCSCs by the PI3K/Akt signaling pathway.