HepG2 cells were cultured in DMEM, supplemented with 10% FBS, penicillin (100U/mL) and streptomycin (100 μg/mg). Cells were incubated at 37°C in a humidified atmosphere of 95% air and 5% CO₂.

α-tomatine (purity>97%), XAV939 and RS47 were purchased from Selleck (Shanghai, China). Wnt3a was purchased from Peprotech (Shanghai, China). Dimethyl sulfoxide (DMSO) was purchased from Sigma Aldrich (Shanghai, China). Penicillin/streptomycin, DMEM, RPMI-1640, fetal bovine serum (FBS), Lipofectamine 3,000 transfection reagent and MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenytetrazodium bromide) were purchased from Invitrogen (Shanghai, China).

The cells were seeded at a density of 6 × 10⁵ cells per well on 12-well plate and grown to about 90% confluence before assay. After removing the medium, the cells were wounded by manually scraping with plastic tip. Then the cells were pre-treated with α-tomatine (2 μM) or DMSO for 12 h before scaraping. The images were captured every 12 h post wounding with microscope.

Cells were treated with α-tomatine (2 μM) or DMSO for 12 h before resuspending. After treatment with α-tomatine or DMSO, the cells were plated at 100 μL cell suspension in chambers with pores (Corning). These chambers were then put in wells with 10% FBS fresh medium. After incubation at 37°C in 5% CO2 for 24°h, the cells invaded the surface of filter were fixed by 4% paraformaldehyde and stained with crystal violet. The images were captured by microscope.

Cell cycle was examined by the staining method using PI (BD bioscience, Shanghai, China) according to the manufacturer’s instructions. Briefly, control (DMSO) and α-tomatine (2 μM) treated cells treated cells were collected after 12 h of each treatment, washed in cold PBS twice before fixation with 70% ethanol for 15 min and stained with PI dyes for 15 min under room temperature. Then we used Flow Cytometry 500 to detect the PI stained cells.

Proteins were extracted with RIPA buffer (10 mM Tris, 150 mM NaCl, 0.5% NP-40, 0.1% SDS, 0.1% deoxycholate, 1 mM PMSF, 2 mM sodium fluoride, and 1 mM sodium orthovanadate), then centrifuged at 12,000 rpm for 30 min at 4°C. Briefly, equivalent amounts of proteins were analyzed by 12% SDS-PAGE, then transferred to PVDF membranes (Millipore, Shanghai, China), which were then incubated with specific primary antibodies anti-β-actin (1:5,000, Cat a1978, Sigma), anti-tubulin (1:5000, Cat T8203, Sigma), anti-RelB (1:1000, Cat 4922, CST), anti-p100 (1:1000, Cat 05–361, Sigma), anti-IKK1 (1:1000, Cat 2682, CST), β-catenin (1:2000, Cat 9582, CST), anti-Axin2 (1:1000, Cat 20540-1-AP, proteintech), anti-cyclin A (1:1000, Cat sc-271682, Santa Cruz), anti-cyclin B1 (1:1000, Cat MA1-155, ThermoFisher), anti-p27 (1:1000, Cat sc-1641, Santa Cruz). Finally, proteins were visualized with peroxidase-coupled secondary antibody, using the ECL system for detection.

RNA was used for RNA-seq according to the manual of Illumina bead arrays. Briefly, libraries were prepared by using 1 μg of RNA per samples measured with Qubit 2.0 fluorometer (Thermo Fisher Scientific) by the KAPA stranded mRNA-seq Kit Illumina platform KR0960. Final libraries were tested via agarose gel and multiplexed with a maximum of 24 samples per sequencing reaction. Libraries were sequenced using an Illumina HiSeq 3,000 with single-end 50 bp reads.

Reads were trimmed using cutadapt (cutoff q = 20) and mapped to the mm10 genome. Processed reads showed high quality reads and alignment scores. Count per minute (cpm) values were generated using edgeR to normalize the raw counts data based on sequencing depth. Induced genes were selected by using a cutoff of fold change greater than 1.5 for any treatment time point relative to the control. Transcripts with empty gene names were removed. Data were z-scored and plotted using the heatmap R package. Gene ontology and IPA analysis were performed using Enrichr and Qiagen Ingenuity Pathway Analysis. Anti-tumor drug treated cancer related dataset, GSE235401, GSE272162, and GSE164561 were selected from GEO database (http://www.ncbi.nlm.nih.gov/geo), and the data characteristics were assessed using the GEO2R tool, which generated volcano diagrams.

The gRNA sequence targeting RelB 5′-TCG​CCG​CGT​CGC​CAG​ACC​GC-3′ was designed using CRISPOR (www.crispor.tefor.net). The gRNA was subcloned into the LentiCRISPR Puro v2 (Cat 98290, Addgene, China). HepG2 cells were pre-cultured to 60%–80% confluence in 6-well plate and transfected using Lipofectamine 2000 (Cat 12566014, Thermofisher, China) for 6 h. To obtain stably transfected clones, cells were selected in the medium containing Puromycin (10 μg/mL, Cat ant-pr-1, Invivogen, China) for 7 days. The cDNA sequence of human full length RelB was subcloned into pBabe plasmid (Cat 17383, Addgene, China) for overexrepsssion of RelB.

To measure Wnt/β-catenin signaling activity, the 0.3 μg of Super8XTOPFlash and 0.01 μg of Renilla luciferase plasmids were transfected in HepG2 cells followed by Wnt3a (250 ng/mL) and α-tomatine (2 μM) treatment for 12 h or in HepG2 cells with overexpression of RelB or RelB knockout, and the luciferase activities were measured. To measure NFkB signaling activity, the 0.2 μg of 5XNFkB reporter and 0.01 μg of Renilla luciferase plasmids were transfected in HepG2 cells followed by α-tomatine (2 μM) or Wnt3a (250 ng/mL) treatment for 12 h, and the luciferase activities were measured.

MTT assay was used to measure cell viability according to manufacturer’s instruction. Cells were plated at a density of 4 × 10⁴/well into 96-well plates and treated with indicated regents. Before measurement, 10 μL MTT solution was added to each well and cells were incubated at 37°C for 4–6 h. Formazan crystals dissolved in 150 μL DMSO were added for 10 min with agitation. Absorbance was measured at 490 nm in a plate reader with SPECTROstarNano (BMG Labtech).

Six-week old female BALB/c nude mice were obtained from the Jackson Laboratory. 5 × 10⁶ HepG2 cells in PBS were implanted subcutaneously into the right flank of nude mice. The mice were randomly divided into three groups and treated with vehicle (corn oil, Cat C8267, Sigma), α-tomatine (25 mg/kg) or α-tomatine (25 mg/kg) together with Wnt3a (2 ug/kg) everyday by i.p. injection. For the injection of WT and RelB KO HepG2 cells, 5 × 10⁶ WT or RelB KO HepG2 cells in PBS were implanted subcutaneously into the right flank of nude mice. After injection of WT or RelB KO HepG2 cells, the mice receiving each cells were randomly divided into two groups and treated with vehicle or α-tomatine (25 mg/kg) everyday by i.p. injection (n = 5). The mice were sacrifieced 21 days later after injection, and the tumors were weighed and photographed. All the animals were treated according to protocols approved by Animal Care and Use Committee of the East China Normal University and ARRIVE guidlines.

Prism software (GraphPad Software) was used for statistical analyses. Values are shown as mean ± s.e.m. Statistical significance between two samples was determined with two-tailed Student’s t-test. Statistical differences of time courses were assessed by a one-way ANOVA (Kruskal-Wallic) followed by Dunn’s post hoc test. *P < 0.05; **P < 0.01; ***P < 0.001.

α-Tomatine is a type of tomatine alkaloid extracted from tomatoes, composed of a sugar group (tomatine) and a tetrasaccharide group (β-tetra-glucose). The tetrasaccharide group consists of two glucose molecules, one galactose, and one xylose; the four monosaccharides form a branched structure attached to the C-3 position of the sugar (Figure 1A). To evaluate the effect of α-tomatine on cell viability of liver cancer cells, MTT assay was performed. Treatment of α-tomatine to HepG2 cells resulted in a significant dose-dependent (from 0.2 to 5.0 μM) inhibition of cell growth (Figure 1B). The EC₅₀ value at 24 h post-treatment with α-tomatine for HepG2 cells was estimated at 0.5418 ± 0.077 μM. Results of HepG2 cells treated with α-tomatine were used for wound healing assays (Figure 1C) and transwell migration assays (Figure 1D) showed that α-tomatine significantly inhibited the ability of cell migration of HepG2 cells. Cell cycle distribution of the treated cells was assessed by labelling the cell’s DNA with Propidium (PI). The G2/M population after α-tomatine treatment were markedly increased compared to control cells after α-tomatine treatment for 12 h (Figure 1E). To further investigate the mechanism of α-tomatine induced cell cycle arrest in HepG2 cells, the levels of G2/M related genes were analyzed. It was shown that α-tomatine decreased the expression of cyclin A and cyclin B1, while the p27 expression remained the same after treatment of α-tomatine (Figure 1F), reflecting the G2/M phase arrest after α-tomatine treatment in the way independent of p27. These data suggested that the proliferation, capacity of migration and invasion are significantly hindered by α-tomatine treatment in liver cancer HepG2 cells followed by G2/M phase arrest.

To further study the outcomes and mechanisms of α-tomatine treatment in liver cancer cell, we performed RNA sequencing of HepG2 cell lines after treatment of α-tomatine for 12h and 24 h with four biological replicates in each group under 0.05 cut-off of False Discovery Rate (FDR). Heatmaps and volcano plot showed that α-tomatine treatment had a large effect on the expression of various genes, including 375 Differential Expressed Genes (DEGs) with fold change larger than 1.5 and adjusted p-value less than 0.05 (Figures 2A,B). Among 247 genes whose expression was reduced after α-tomatine treatment, the expression of 194 genes was attenuated in both 12h and 24 h of α-tomatine treatment (Figure 2C). The downregulated genes were associated with terms invoking RNA metabolic process, extracellular matrix binding, positive regulation of signaling and transcription by RNA polymerase II in gene ontology (GO) analysis (Figure 2C), which is consistent with the attenuated proliferation, migration and invasion in HepG2 cell lines after α-tomatine treatment. Moreover, α-tomatine resulted in increased expression of 128 genes, and the expression of 68 genes was elevated in both 12h and 24 h of α-tomatine treatment (Figure 2D). The GO analysis revealed the response to cytokine, inflammatory and defense response, apoptotic process among the top terms (Figure 2D). The Ingenuity Pathway Analysis (IPA) of differentially expressed genes further revealed a compromised activation of WNT response, extracellular matrix organization, NOTCH expression and processing, serotonin receptor signaling and KEAP1-NFE2L2 pathway in the downregulated genes, an enhanced activation of activin inhibin signaling pathway, inflammatory response pathway, IL17 signaling and HMGB1 signaling in the upregulated genes (Figure 2E). These data suggested that α-tomatine can induce or inhibit the expression of various genes which is involved in multiple processes related to proliferation, apoptosis and immune response.

Building upon RNA-seq findings demonstrating α-tomatine-induced impairment of WNT pathway activation (Figure 3A), we thought to characterize the functional involvement of canonical Wnt/β-catenin signaling in mediating α-tomatine’s therapeutic effects in HepG2 cells. Western blot quantification demonstrated that α-tomatine treatment significantly reduced Axin2 and β-catenin protein levels in a time-dependent manner (Figure 3B). Luciferase reporter assay by using Super8XTOPFlash reporter plasmid showed that WNT signaling activity reduced after α-tomatine treatment after 12 and 24 h in HepG2 cells (Figure 3C). Furthermore, activation of Wnt/β-catenin signaling by adding of Wnt3a significantly rescued cell viability induced by α-tomatine, whereas inhibition of Wnt/β-catenin signaling by adding of XAV939 markedly reduced cell viability (Figure 3D). To further validate the roles of Wnt signaling in the anti-tumor effects of α-tomatine, the HepG2 xenograft model was applied. The mice were randomized into Control group (s.c. injection of HepG2 cells; i.p. injection of corn oil; n = 5), α-tomatine group (s.c. injection of HepG2 cells; i.p. injection of α-tomatine; n = 5) and α-tomatine + Wnt3a group (s.c. injection of HepG2 cells; i.p. injection of α-tomatine combined with Wnt3a; n = 5). The in vivo experiment showed that the addition of Wnt3a during α-tomatine treatment result in larger tumor sizes and weights (Figures 3E–H). The above results demonstrated that the attenuated Wnt/β-catenin signaling is one of the mechanisms of α-tomatine that result in anti-proliferation in HepG2 cells.

RNA-seq profiling revealed significant activation of cytokine storm-associated signaling in α-tomatine-treated HepG2 cells. Notably, the non-canonical NF-κB transcription factor RelB exhibited sustained upregulation at both 12h and 24 h post-treatment (Figure 4A). To explore if anti-tumor drugs also induce expression of RelB in other cancer cells, we searched GEO data sets. Upregulated DEGs in gastric cancer cell SNU1 treated with anti-tumor drug Palbociclib from GSE235401 and non-small cell lung cancer cell H1975 with anti-tumor drug Osimertinib resistance from GSE272162 revealed that RelB is one of the genes increased expression (Figure 4B). RelB was also shown in the upregulated DEGs in HepG2 cells treated with anti-tumor drug Berberine from GSE164561 (Figure 4C). Clinical correlation analysis demonstrated that hepatocellular carcinoma patients with high RelB expression exhibited significantly reduced overall survival and disease-free survival compared to low-expressing counterparts (Figure 4D). Western blot analysis confirmed α-tomatine-induced elevation of RelB and its upstream regulators NIK and phosphorylated p100, though p52 subunit expression remained unaltered (Figure 4E). Luciferase reporter assay by using 5xNFkB reporter plasmid showed that treatment of α-tomatine result in upregulation of NFkB luciferase activity, but Wnt3a stimulation did not alter NFkB luciferase activity in HepG2 cells (Figure 4F). To further elucidate the functional impact of RelB in the context of α-tomatine treatment, we genetically modulated its expression in HepG2 cells by either overexpressing or knocking it out using CRISPR/Cas9 technology (Figures 4G,H). Luciferase reporter assay by using Super8XTOPFlash reporter plasmid in RelB overexpressed or RelB knockout HepG2 cells did not show altered WNT signaling activity, indicating that α-tomatine might affect WNT and NFkB signaling pathways in independent manners. Overexpression of RelB led to increased proliferation of HepG2 cells in the absence of α-tomatine, while concurrently diminishing the antiproliferative efficacy of α-tomatine (Figure 4J). Conversely, RelB knockout in HepG2 cells curtailed their proliferative capacity without α-tomatine exposure and augmented the antiproliferative response to α-tomatine (Figure 4K). Interestingly, treatment of 2 μM RelB specific inhibitor RS47 (Li et al., 2025) alone lead to reduced proliferation of HepG2 cells, and combined treatment of RS47 and α-tomatine further augment the inhibitory effects of proliferation (Figure 4L). Collectively, these findings underscore a correlation between elevated RelB expression and poorer survival outcomes in liver hepatocellular carcinoma patients, highlighting that α-tomatine-induced RelB upregulation may compromise its therapeutic potential against HepG2 cell proliferation.

To further probe the impact of RelB on α-tomatine’s antitumor efficacy, we employed a HepG2 xenograft model for in vivo validation. Mice were stratified into five cohorts: WT control (s.c. injected with WT HepG2 cells; i.p. administered corn oil; n = 5), KO control (s.c. injected with RelB KO HepG2 cells; i.p. administered corn oil; n = 5), WT tomatine (s.c. injected with WT HepG2 cells; i.p. treated with α-tomatine; n = 5), and KO tomatine (s.c. injected with RelB KO HepG2 cells; i.p. treated with α-tomatine; n = 5). Treatments—corn oil as vehicle or α-tomatine—were administered daily for 21 consecutive days. The data revealed that while α-tomatine treatment alone or RelB deficiency suppressed tumor volume in WT HepG2 models, combining α-tomatine with RelB KO HepG2 cells led to a significant reduction in tumor volume (Figures 5A–C). Additionally, α-tomatine treatment markedly decreased tumor weights in RelB KO HepG2 mice (Figure 5D). Notably, RelB expression within tumors was upregulated following 21 days of α-tomatine exposure (Figure 5E). In summary, our findings illustrate that α-tomatine-induced elevation of RelB diminishes its in vivo antitumor effectiveness.