The human liver cancer HepG2 (RRID: CVCL_0027), human HCC PLC/PRF/5 cell line (RRID: CVCL_0485) and murine HCC Hepa1-6 cell line (RRID: CVCL_0327) were authenticated by short tandem repeat analysis and were provided by Procell Life Science and Technology Co., Ltd. All cell lines were authenticated within the last 3 years by Procell Life Science and Technology Co., Ltd. Mycoplasma-free cells were used for all experiments. ACT extracted from Cistanche tubulosa (Schenk) Wight (cat. no. 103200) was purchased from Yongjian Pharmaceutical Technology, with a purity >98%. OXA (cat. no. 100584) was from National Institutes for Food and Drug Control.

HepG2, PLC/PRF/5 and Hepa1-6 cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM), minimum essential medium (MEM) (both from HyClone; Cytiva) and DMEM containing 10% fetal bovine serum (FBS; Corning Life Sciences), 100 U/ml penicillin and 100 µg/ml streptomycin, respectively. Cells were cultured at 37°C in 100% humidity and 5% CO₂. Cells were dissociated to single cells using trypsin-EDTA upon reaching 80-90% confluence, followed by termination of enzymatic digestion with complete medium.

A total of 5×10³ cells were cultured in a 96-well plate filled with complete medium for 48 h. OXA (1.56, 3.13, 6.25, 12.50, 25.00, 50.00, 75.00, 100.00 µM) and ACT (3.13, 6.25, 12.50, 25.00, 50.00, 100.00, 200.00, 400.00 µM) intervened three HCC cells (HepG2, PLC/PRF/5, Hepa1-6) for 48 h at 37°C. Detection of the IC₅₀ of HCC cells by Cell Counting Kit-8 (CCK-8) assay. HepG2 cells were grouped as follows: i) Control, untreated; ii) OXA (1.56, 3.13, 6.25, 12.50 and 25.00 µM); iii) ACT (6.25, 12.50, 25.00, 50.00 and 100.00 µM) and iv) OXA + ACT (1.56+6.25, 3.13+12.50, 6.25+25.00, 12.50+50.00 and 25.00+100.00 µM) at the same time. Hepa1-6 cells were grouped as follows: i) Control, untreated; ii) OXA, (1.56, 3.13, 6.25, 12.50 and 25.00 µM); iii) ACT (25.00, 50.00, 100.00, 200.00 and 400.00 µM) and iv) OXA + ACT (1.56+25.00, 3.13+50.00, 6.25+100.00, 12.50+200.00 and 25.00+400.00 µM) at the same time. The OXA/ACT concentration ratios were determined based on half-maximal inhibitory concentration (IC₅₀) values.

HepG2 and Hepa1-6 cells were incubated for 4 h in a 96-well plate with 10 µl/well CCK-8 reagent (Beijing Solarbio Science & Technology Co., Ltd.). Cell viability was measured using a multi-detection microplate reader (Multiskan™ FC; Thermo Fisher Scientific, Inc.) at 450 nm. CI of OXA combined with ACT was calculated using CompuSyn 1.0 (Biosoft) and the CI curve was drawn as previously described (32). CI=0.9-1.1, <0.9 and >1.1 indicated an additive effect, synergism and antagonism, respectively.

Cell invasion was evaluated using a Transwell assay as previously described (33). The Matrigel^® matrix (cat. no. 356234; lot no. 8155016; Corning, Inc.) was thawed at 4°C and then diluted with serum-free DMEM at a 1:8 volume ratio for coating overnight at 37°C. Transwell chambers. Cells (1×10⁵) suspended in 200 µl high-glucose FBS-free DMEM were seeded in the upper chamber. High-glucose DMEM (600 µl) supplemented with 10% FBS was placed in the lower chamber. Following incubation for 48 h at 37°C, cells on the upper surface were removed. Next, cells fixed for 30 min at room temperature using 95% ethanol, then stained for 30 min at room temperature with 0.1% crystal violet and observed under a light microscope (magnification, ×100).

Cell migration was assessed using a wound healing assay as previously described (33). Cells were seeded in 6-well plates until they reached 100% confluence. After wounding with a 200-µl pipette tip and washing with PBS, cells were incubated in serum-free DMEM (HyClone; Cytiva) at 37°C. Wound images at 0, 12, 24 and 48 h were captured under a light microscope (magnification, ×200). Wound distance was expressed as ratio to the initial distance at 0 h.

Cells (5×10⁵) were incubated in 6-well plates for 24 h at 37°C in 5% CO₂ and were cultured for 4 h in serum-free DMEM. Cells were then digested with 0.25% pancreatin. Following centrifugation at 1,000 × g for 5 min at room temperature, cells were suspended with PBS buffer to a concentration of 1-5×10⁵ cells/ml. After PI and RNase A (cat. no. C1052, Beyotime biotechnology) staining at 37°C away from light for 30 min, cells were subjected to flow cytometry using a flow cytometer (DXFLEX, Beckman, USA). FlowJo (version 10.6.2; Tree Star, Inc.) was used to analyze data.

Total RNA of HepG2 cells was isolated using TRIzol^® (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. Purity was determined using the NanoDrop ND-1000 (NanoDrop). RNA were reverse-transcribed by SuperScript™ II Reverse Transcriptase (Invitrogen, cat. 1896649, USA), which were used to synthesize U-labeled second-stranded DNAs with E. coli DNA polymerase I (NEB, cat.m0209, USA), RNase H (NEB, cat. no. m0297, USA) and dUTP Solution (Thermo Fisher, cat.R0133, USA). An A-base is then added to the blunt ends of each strand, preparing them for ligation to the indexed adapters. Each adapter contains a T-base overhang for ligating the adapter to the A-tailed fragmented DNA. Single- or dual-index adapters are ligated to the fragments, and size selection was performed with AMPureXP beads. After the heat-labile UDG enzyme (NEB, cat.m0280, USA) treatment of the U-labeled second-stranded DNAs, the ligated products are amplified with PCR. For each sample, 10 pM RNA was selected for sequencing library construction. The average insert size for the final cDNA library was 300±50 bp. At last, we performed the 2×150 bp paired-end sequencing (PE150) on an Illumina Novaseq™ 6000 (LC-Bio Technology CO., Ltd.) with a KAPA Stranded RNA-seq Library Prep kit (cat. no. KK8401; Illumina, Inc.) following the vendor's recommended protocol. The differentially expressed genes (DEGs) between control group and ACT group were selected by R package (version 4.0.2, cran.r-project.org/bin/windows/) (34). The sequencing data can be accessed at https://www.ncbi.nlm.nih.gov/sra/PRJNA1219772.

Metabolomics analysis was performed by LC-Biotechnology. Briefly, six cell samples per group were randomly used for metabolomics analysis. Cell pre-processing was performed as previously described (35). HepG2 cell extracts were analyzed on a Triple TOF 5600 plus mass spectrometer (SCIEX). Quadrupole time-of-flight mass spectrometry was operated in negative and positive ion modes. Data were acquired in data independence initiative mode, with 60-1, 200 Da TOF mass and 0.56 sec total cycle. For each scan, 4-time bins were aggregated at an 11-kHz pulse frequency by monitoring a 40-GHz multi-channel thermal conductivity detector with four-anode/channel detection, dynamically excluded for 4 sec. The mass accuracy was calculated per 20 samples. A quality control sample (aliquots of all samples were mixed) was injected at the beginning to equilibrate the system and then every 10 samples to monitor the stability of the system. Kyoto Encyclopedia of Genes and Genomes (URL: http://www.genome.jp/kegg) enrichment analysis on transcript and metabolite profiles was conducted to explore the potential link between mRNA-metabolomics pathways (impact value >0.1) and DEGs. Data processing and analysis were performed as previously described (36).

A total of 50 male BALB/c nude mice (SPF grade; age, 5-6 weeks; weight, 18-22 g) were purchased from Beijing Huafukang Biotechnology Co., Ltd. (license no. SCXK 2019-0008), reared under standard environmental conditions (26-28°C; relative humidity, 40-60%; 12/12-h light/dark cycle, with food and water provided ad libitum. All animal experiments were conducted following the ethical guidelines approved by the Experimental Animal Ethics Committee of the First Affiliated Hospital of Xinjiang Medical University (approval no. IACUC-20210301-179; Urumqi, China).

A xenograft mouse model was established by subcutaneously injecting HepG2 cells (1×10⁷ cells/200 µl of 0.9% physiological saline) into the axillary region as previously described (33). A total of 2 weeks post-injection, mice with tumor volumes >100 mm³ were randomly assigned to the following groups (n=11/group): i) Model group, receiving an equivalent volume of 5% glucose injection; ii) OXA group, administered 5 mg/kg OXA weekly; iii) ACT group, treated with 80 mg/kg ACT every 3 days; and iv) OXA + ACT group, receiving 5 mg/kg OXA weekly in combination with 80 mg/kg ACT every 3 days (Fig. 2). In addition, the control group was BALB/c nude mice not injected with HepG2 cells (n=6). OXA was dissolved in 5% glucose according to the manufacturer's instructions, while ACT was dissolved in sterilized water. All treatments were administered via intraperitoneal injection. All mice were weighed, and the tumor volume was calculated (0.5 a x b²; a, length; b, width) every 3 days. After 2 weeks, six randomly selected mice in each group were anesthetized with sodium pentobarbital (50 mg/kg). Blood samples were collected from the eyeballs, followed by euthanasia via intraperitoneal injection of sodium pentobarbital (150 mg/kg). Tumor mass and small intestine tissues were isolated for further analysis. The tumor inhibition efficiency (%) was calculated as (1-W_i/W_m) ×100, where W_i is the mean tumor weight in each treatment group, and W_m is the mean tumor weight in the model group. Coefficient of drug interaction (CDI) was used to determine the interaction of OXA + ACT, calculated as AB/(A x B), where AB was the ratio of the tumor growth efficiency (W_i/W_m) in the two-drug combination group to the control group, and A or B was the ratio of the tumor growth efficiency in a single drug group to the control group (37). CDI values of <1, =1 and >1 indicated synergism, additive effects and antagonism, respectively. CDI <0.7 indicated significant synergistic effects. To evaluate the roles of OXA + ACT on the survival of HepG2 tumor-bearing mice using Kaplan-Meier curve, the remaining five mice in each group received the same regimen for another 60 days. Humane endpoints were as follows: i) Tumor volume exceeded 2,000 mm³; b. Tumor diameter exceeded 20 mm; c. Rapid or continuous weight loss of more than 20% within 72 h; d. Tumor ulceration or necrosis resulted in skin rupture or persistent exudation for more than 48 h; e. Significant abdominal distension or ascites burden exceeding 10% of body weight, with an appearance similar to that of a pregnant mouse compared to age-matched controls; f. The survival observation period was set at 60 days, and all mice were euthanized at the end of this phase. Sodium pentobarbital (150 mg/kg) was administered via intraperitoneal injection to induce respiratory arrest. The survival time of mice in each group was recorded to construct survival curves using Graphpad prism 8.0.

Tumor tissue sections were stained with H&E for histological evaluation, as previously described (38). Images were captured using a light microscope and percent of tumor necrotic area of total tumor from mice was semi-quantified by ImageJ (version1.54g) software (National Institutes of Health). Ki67 and proliferating cell nuclear antigen (PCNA) were evaluated as tumor cell proliferation markers (39) detected in tumor tissue sections to assess cell viability. Incubate sections (thickness, 0.2-0.3 cm) were fixed in methanol for 30 min at room temperature, rinsed with xylene and dehydrated by gradient ethanol. Immerse the slides in EDTA antigen retrieval buffer (pH 8.0) and maintain at a sub-boiling temperature for 8 min, standing for 8 min and then followed by another sub-boiling temperature for 7 min. Wash three times with PBS (pH 7.4) in a Rocker device, 5 min each. Add 3% BSA (Wuhan Servicebio Technology Co., Ltd, China, https://www.servicebio.cn/) to block non-specific binding for 30 min at room temperature. The primary anti-PCNA (1:400; cat. no. GB11010-100) and anti-Ki67 (1:500; cat. no. GB121141-100, both Wuhan Servicebio Technology Co., Ltd, China) was added at 4°C overnight. Following three washes with PBS, the Alexa Fluor 488-labeled goat anti-mouse IgG secondary antibody (1:500; cat. no. GB25301, Wuhan Servicebio Technology Co., Ltd.) and Cy3-labeled goat anti-rabbit IgG secondary antibody (1:500; cat. no. GB21303, Wuhan Servicebio Technology Co., Ltd, China) was added at room temperature for 50 min. Sections were stained with DAPI at room temperature for 10 min. Then, the tumor tissue sections are observed using a fluorescence microscope (Nikon Eclipse C1, Nikon) at a magnification of ×40. A total of three random fields of view were selected for each section. The fluorescence intensities of Ki67 and PCNA were calculated using ImageJ 1.45 (National Institutes of Health) as follows: Fluorescence intensity (%)=fluorescence-stained area/DAPI-stained area ×100.

Fresh tumor tissue was fixed with 2.5% glutaraldehyde (cat. no. G1102, Wuhan Servicebio Technology Co., Ltd, China) for 24 h at 4°C. Tissue was embedded in EMBed 812, and then keep in 37°C oven overnight. The resin blocks were cut to 60-80 nm thin on the ultra-microtome, and the tissues were fished out onto the 150 meshes cuprum grids with formvar film (cat. no. HT7700, Hitachi Ltd., Japan). Sections were stained with 2% uranium acetate saturated alcohol solution for 8 min at room temperature, rinsed in 70% ethanol (cat. no. 100092183, Sinaopharm Group Chemical Reagent Co. Ltd., China) and ultra-pure water three times. 2.6% Lead citrate avoid CO₂ staining for 8 min, and then rinsed with ultra-pure water for 3 times. Then, the tumor tissue sections are sent to the electron microscopy room at Xinjiang Medical University for observation using a transmission electron microscope (JEM-100CX II TEM, JEOL Ltd.).

Whole blood was allowed to set at room temperature for 1 h, centrifuged at 1,509.3 g at 4°C for 10 min, and serum was collected. ELISA was performed to detect serum levels of IL-6, hypersensitive C-reactive protein (hs-CRP), α fetoprotein (AFP) and TNF-α using Human hs-CRP ELISA Kit (cat. no. JL10346), Human aFP ELISA Ki (cat. no. JL40021) and Human TNF-α ELISA Kit (cat. no. JL10484), according to the manufacturer's instructions (Shanghai Jianglai Biotechnology Co., Ltd.). All the tests were performed at least in triplicate.

H&E staining was performed on the intestinal tissue sections to evaluate the effects of ACT on OXA-induced digestive toxicity, as aforementioned. White blood cells (WBCs), red blood cells (RBCs) and platelets (PLTs) were counted to assess the effects of ACT on OXA-induced hematological toxicity. In addition, the effects of ACT on OXA-induced OXIN were evaluated by measuring the paw withdrawal mechanical threshold (PWMT) using Von Frey hairs. Levels of serum alanine aminotransferase (ALT; cat. no. C009-2-1), aspartate aminotransferase (AST; cat. no. C010-2-1), alkaline phosphatase (AKP; cat. no. A059-2-2), lactate dehydrogenase (LDH; cat. no. A020-2-2), creatinine (CRE; cat. no. C011-2-1) and blood urea nitrogen (BUN; cat. no. C013-2-1) were detected according to the manufacturer's instructions (Nanjing Jiancheng Bioengineering Institute, Ltd.) to assess the OXA-induced hepatotoxicity and nephrotoxicity in the presence of ACT.

Protein of tumor tissue was extracted with RIPA lysis buffer (Thermo Fisher Scientific, Inc.) and protein concentration was detected using the bicinchoninic acid assay method. A total of 10 µg/lane protein was separated by 10% SDS-PAGE, blocked in 5% non-fat milk for 1h at 4°C, then transferred to PVDF membrane (Merk Millipore, cat. no. R1CB35759) and incubated with appropriate primary antibodies at 4°C overnight and secondary antibodies at room temperature for 4 h. The PVDF membrane was rinsed with TBST (containing 0.1% Tween-20). The primary antibodies β-actin (cat. no. bs-0061R; 1:5,000), AKT1 (cat. no. bsm-52010R; 1:500), phosphorylated (p)-AKT1 (cat. no. bs-5194R; 1:500), PIK3R1 (cat. no. bs-0128R; 1:500), p-PIK3R1 (cat. no. bs-6417R; 1:500), PRKCA (cat. no. bsm-54393R; 1:500), INPP5D (cat. no. bs-3567R; 1:500), PTEN (cat. no. bsm-33319M; 1:500). and the secondary antibodies goat anti-rabbit IgG H&L (cat. no. bs-0295G; 1:5,000) and anti-mouse IgM, HRP conjugated (cat. no. bs-0368G-HRP; 1:5,000) purchased from Beijing Bioss Biotechnology Co., Ltd (http://bioss.com.cn/). The primary antibodies PLCB4 (cat. no. DF2564; 1:500), INPP4B (cat. no. DF7975; 1:1,000) and PIKFYVE (cat. no. DF8707; 1:1,000) purchased from Shanghai Fushen Biotechnology Co., Ltd (https://www.fsbio-mall.com/). Blots were visualized using an enhanced chemiluminescence (Hefei White Shark Biotechnology Co., Ltd. cat. no. 1027a01), and relative intensity was semi-quantified by ImageJ (version1.54g) software (National Institutes of Health) and normalized to the intensity of β-actin. Antibodies against PLCB4, INPP4B and PIKFYVE were from Affinity Biosciences. All other antibodies were from Boaosen Biotechnology.

All experiments were performed in triplicate. GraphPad Prism 5.0 and CompuSyn software were used to calculate IC₅₀ and CI values, respectively. SPSS 22.0 (IBM Corp.) was used for statistical analyses. Data are presented as the mean ± standard deviation. Comparisons were performed using one-way ANOVA followed by Bonferroni's post hoc test. P<0.05 was considered to indicate a statistically significant difference.

OXA and ACT were used to treat liver cancer cell lines (PLC/PRF/5, HepG2 and Hepa1-6) for 48 h. For OXA treatment (Fig. S1A), the IC₅₀ values in HepG2, PLC/PRF/5 and Hepa1-6 cells were 15.81, 22.56 and 16.79 µM, respectively. For ACT (Fig. S1B), the IC₅₀ values in HepG2, PLC/PRF/5 and Hepa1-6 cells were 71.72, 74.74 and 241.40 µM, respectively. HepG2 cells showed the greatest sensitivity (lowest IC₅₀ value) and were selected as the most suitable liver cell line for animal studies.

The CI value for ACT + OXA was <0.9, suggesting this combination synergistically inhibited the viability of HepG2 and Hepa1-6 cells (Figs. S2 and S3). In HepG2 cells, the CI value for the 50 µM ACT + 12.5 µM OXA treatment group was 0.25±0.41, which was lower than that of other dose combinations (Table SI). In Hepa1-6 cells, 200 µM ACT + 12.5 µM OXA yielded the lowest CI value (0.07±0.01; Table SII). Regarding safety, ACT/OXA concentration ratios of 50:12.5 and 200:12.5 µM were selected for subsequent experiments on HepG2 and Hepa1-6 cells, respectively.

Cell viability in OXA and/or ACT-treated groups was significantly decreased compared with that in the control group (Figs. 3A and 4A). In addition, cell viability in the OXA + ACT group showed a significant decrease compared with the OXA and ACT groups. The Transwell assay showed cell invasion was decreased in the OXA compared with that in the control group, and in the OXA + ACT group compared with that in OXA group (Figs. 3B and 4B).

The wound healing assay showed that the wound size in the OXA + ACT group increased, whereas in the control, OXA, and ACT groups, get smaller. (Figs. 3C and 4C). After 48 h, the wound size in the OXA + ACT group was significantly increased compared with that in the OXA and ACT groups. This finding suggested that cell migration was reduced in the OXA + ACT group compared to the other groups (Figs. 3D and 4D).

The rate of early cell apoptosis was increased in the OXA and/or ACT groups compared with that in the control group. The early apoptotic rate of the OXA + ACT group was higher than that in the OXA and ACT groups (Figs. 3E and G, and 4E and G).

In the OXA + ACT group, the proportion of G₀/G₁ phase cells showed a significant increase compared with that in the OXA group, together with a significant decrease in the proportion of S phase cells (Figs. 3F and H and 4F and H). These results indicated that OXA + ACT could arrest cells in G₀/G₁ phase.

All six mice in each group survived without any fatalities. The body weight of mice in the OXA + ACT group showed a decrease within 9 days of the intervention and increased after day 9. Notably, the body weight of mice in the OXA + ACT group was significantly increased compared with that of mice in the OXA group on day 12 (Fig. 5A). The maximum diameter and volume of the tumors detected during the 2-week treatment phase were 17.48 mm and 1,997.32 mm³, respectively. Tumor volume in the OXA and OXA + ACT groups was decreased compared with that in the model group after 12 days, although the difference was not significant (Fig. 5D). In the model group, the first nude mouse death occurred on day 25, and all nude mice died within 50 days. By contrast, in the OXA, ACT and OXA + ACT groups, the first mouse death was observed on days 35, 45 and 40, respectively. There was no significant difference in median survival time between the OXA + ACT group, and the model or OXA groups (Fig. 5B). Mice used for 60-day survival analysis in each group euthanized due to tumor growth, tumor ulceration and nutritional depletion (Table SIII).

The tumor volume in the OXA + ACT group was lower than that in the model group (Fig. 5C, D and F). Furthermore, compared with in the OXA group, tumor weight was significantly decreased in the OXA + ACT group (0.319±0.147 vs. 0.487±0.099 g; Fig. 5E; Table SIV). Moreover, the CDI of the OXA + ACT group was 0.93, indicating a synergistic effect of ACT and OXA in HCC.

IL-6, hs-CRP, TNF-α and AFP levels were significantly decreased in the OXA and OXA + ACT groups compared with those in the model group (Fig. 5G). In addition, serum hs-CRP levels were significantly decreased in the OXA + ACT group compared with those in the OXA group and serum IL-6, hs-CRP, TNF-α and AFP levels were significantly decreased in the OXA + ACT group compared with those in the ACT group.

In each group, the nuclei of the tumor necrosis area were fragmented, exhibited pyknosis or were absent while the viable tumor survival area showed intact nuclei with clear boundaries (Fig. 6A). Compared with in the model group, tumor necrosis in the OXA and/or ACT groups was significantly increased (Fig. 6B). Furthermore, tumor necrosis in the OXA + ACT group showed a significant increase compared with that in the OXA and ACT groups.

The Ki-67 levels in tumor tissues in the OXA and/or ACT groups were significantly increased compared with those in the model group, whereas those in the OXA + ACT group were significantly decreased compared with those in the OXA group. By contrast, PCNA levels in the OXA and/or ACT groups showed a significant decrease compared with those in the model group, and the OXA + ACT group showed a further significant decrease in PCNA levels compared with in the OXA and ACT groups (Fig. 6C and D).

Regarding the results of transmission electron microscopy, heteromorphic nuclei, partially depressed nuclear membranes, nuclear notches, binucleated cells and depleted organelles were observed in the model group (Fig. 6E). The mitochondrial ridges were dissolved, swollen and vacuolized in the tumor cells in the OXA group. The tumor cells in the ACT group showed lysed nuclear membranes, heteromorphic flocculent substances and lysed mitochondrial ridges. The combination of OXA + ACT resulted in disintegration of cell membranes, widening of the perinuclear spaces, cavitation of mitochondria, complete dissolution of the outer mitochondrial membranes and rough endoplasmic reticula with degranulation and marked expansion.

The small intestinal tissues in both the control and model groups exhibited regular pathological morphology, with intact intestinal villi and clear boundaries (Fig. 7A). By contrast, the small intestinal villi in the OXA group were damaged, with some villi ruptured and unclear boundaries. However, the OXA + ACT group showed alleviated damage, with better preserved intestinal villi compared with model group. The PWMT value in the OXA + ACT group was significantly higher than that in the model group, and significantly lower than that in the OXA group (Fig. 7B). Furthermore, compared with model group, WBC, RBC and PLT counts in the OXA group were significantly reduced (Fig. 7C), whereas these values were increased in the OXA + ACT group, particularly WBC and RBC counts. Compared with in the model group, the OXA group showed elevated levels of AST and decreased levels of AKP. The OXA + ACT group exhibited significantly lower AST and LDH levels compared with those in the OXA group. There was no significant change in ALT levels between groups (Fig. 7D). In addition, CRE levels in the model group were significantly higher than those in the control group, whereas CRE levels in the OXA + ACT group lower than those in the model group. BUN levels in the OXA group were significantly higher than those in the model group (Fig. 7E), whereas BUN levels in the OXA + ACT group were significantly lower than those in the OXA group (Fig. 7E).

Principal component analysis showed that the mRNA expression data of the HepG2 cells in the control and the ACT group were distinguishable (Fig. 8A and B). Compared with in the control group, 7,205 DEGs, including 3,010 upregulated and 4,195 downregulated genes, were screened in the ACT group (Fig. 8C and D). These data indicated that ACT may inhibit HCC progression by downregulating multiple signaling pathways. The volcano plots demonstrated 235 upregulated and 16 downregulated metabolites in the ACT group (Fig. 8E). Cluster analysis indicated 16 significant differentially expressed secondary metabolites between the control group and the ACT group (Fig. 8F and G). The top 20 enriched pathways in positive ion mode including 'phosphatidylinositol signaling system', 'inflammatory mediator regulation of TRP channels', 'inositol phosphate metabolism' and 'amino sugar and nucleotide sugar metabolism', as well as 'aldosterone synthesis and secretion'. Additionally, there were 20 enriched pathways in negative ion mode, including 'neuroactive ligand-receptor interaction', 'serotonergic synapse', 'protein digestion and absorption', 'amino sugar and nucleotide sugar metabolism' and 'glycerolipid metabolism' (Fig. 8H).

Following mRNA-metabolomics integration analysis through signal pathways, eight target genes/proteins involved in 'phosphatidylinositol signaling system' were screened, including AKT1, PIK3R1, PLCB4, PRKCA, PTEN, INPP4B, INPP5D and PIKFYVE (Table I and Fig. 9). As determined by western blotting, the protein expression levels of p-AKT1, AKT1, p-PIK3R1 p-PIK3R1/PIK3R1 and PRKCA showed a significant decrease in the OXA compared with those in the model group, whereas p-AKT1/AKT1, PLCB4, PTEN, INPP4B and INPP5D protein expression levels showed a significant increase. AKT1, p-PIK3R1 and PIK3R1 protein expression levels were significantly decreased in the ACT group compared with those in the model group, whereas PTEN, INPP4B and INPP5D protein expression levels showed a significant increase. In addition, compared with those in the OXA group, a significant increase was observed in p-AKT1, AKT1, p-AKT1/AKT1, p-PIK3R1, INPP4B and INPP5D protein expression levels in the OXA + ACT group. Moreover, compared with those in the ACT group, a decrease in AKT1, p-PIK3R1, and an increase in INPP4B and INPP5D protein expression levels were observed in the OXA + ACT group. Changes in both p-AKT1 and AKT1, p-PIK3R1 and PIK3R1 may be the result of the intervention of OXA and ACT, which led to an increase in the necrotic areas of the tumor tissues, resulting in changes in their protein content. The above results show that OXA+ACT co-administration up-regulated the expression of oncogene INPP4B/PTEN through the multi-signaling pathway in the phosphatidylinositol signaling system inhibiting the PI3K(PIK3R1)/AKT signaling pathway to play a synergistic role in OXA to exert an anti-HCC effect while alleviating oxaliplatin-induced toxicity.