The research first retrieved the compounds of Bos taurus domesticus Gmelin, Boswellia carterii Birdw, and Commiphora myrrha Engl through the TCMSP database [20], and the compounds in Moschus were obtained from the herb [21] and batman databases [22]. Oral bioavailability (OB) of $\ge$30% and drug-likeness (DL) $\ge$0.18 were used as component screening thresholds. The TCMSP database, the UniProt database [23], GeneCards database [24], NCBI gene database [25], and DisGeNET database [26] were utilized for both forecasting and screening. To construct a Protein-Protein Interaction (PPI) network, the common targets of drugs and diseases obtained from screening were inputted into the String database [26]. The PPI network was imported into Cytoscape 3.8.0 (GraphPad Software, Inc., San Diego, California, USA) [27, 28]. Topological analysis was performed by the Network Analyzer tool, and genes with scores greater than average were selected as key targets by degree sorting. Further, Cytoscape 3.8.0 was adopted to construct a component/disease/target network diagram. Additionally, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed.

We evaluated the core components of Xihuang Pills via UPLC-MS using the Shimadzu Prominence UPLC system (SCIEX, ExionLC AD) coupled with the QTRAP MS (SCIEX, QTRAP 5500). The flow rate, the injection tray temperature, and the column temperature was 0.3 mL/min, 4 ℃, and 40 ℃, respectively. The column used was a Waters HSS T3 column (100 $\times$ 2.1 mm, 1.7 $\mu$m) with (A) H${}_2$O (0.1% FA) and (B) Acetonitrile. The liquid gradient conditions are shown in Supplementary Table 1. Retention time was compared with that of a quercetin standard.

H22 and HepG2 cell lines were purchased from Procell Life Science & Technology Co., Ltd (Wuhan, China). The H22 and HepG2 cell lines were cultured in 1640 medium (Sigma, Saint Louis, R8758, USA) and Dulbecco’s minimum essential medium (DMEM) (Sigma, D5796, USA) containing 10% fetal bovine serum (FBS, Gibco, Grand Island, 10099141, USA), respectively. The cells were divided into four groups: a control group, a low-dose group, a medium-dose group, and a high-dose group [29], which were treated with quercetin (MedChemExpress, New Jersey, HY-18085, China) at concentrations of 0, 25 $\mu$M, 50 $\mu$M and 100 $\mu$M, respectively.

After digestion, H22 and HepG2 were counted and divided into four groups. Each well was seeded with 500 $\mu$L of cell suspension, the number of cells/well was adjusted to 5 $\times$ 10${}^4$, and each group was seeded in triplicate. After cell adherence, cells were treated with quercetin for 24, 48, and 72 h. After removing the drug-containing medium, 10% CCK8 (DOJINDO, Kumamoto, NU679, Japan) was added to each well and after incubating for four hours, the optical density (OD) value was measured at 450 nm.

Matrigel (354262, Becton Dickenson, Franklin Lakes) was diluted and placed in the upper chamber and medium containing 10% FBS was placed in the lower chamber. HepG2 cells were treated with different concentrations of quercetin for 72 h, digested into single cells, and then resuspended at 2 $\times$ 10${}^6$/mL in serum-free medium. 100 $\mu$L/well of cell suspension was added to the upper chamber and the plate incubated at 37 ℃ for 48 h. After this, the upper chamber was removed and the cells in the lower chamber were fixed with 4% paraformaldehyde for 30 minutes, stained with 0.1% crystal violet for 5 minutes, placed on a glass slide, and photographed under a microscope (DSZ2000X, Beijing Zhongxian Hengye Instrument Co., Ltd., China). After destaining, the OD value was obtained at 550 nm.

The cells (H22 and HepG2) were collected and stained with 5 $\mu$L Annexin V-APC and 5 $\mu$L Propidium Iodide (Keygen Biotech, Jiangsu, KGA1030, China). Binding buffer (500 $\mu$L) was used to resuspend cells and apoptotic cells were detected by flow cytometry (Beckman, Fullerton, USA).

Approximately 5 $\times$ 10${}^5$ HepG2 cells were added to each well of a 6-well plate. After the cells reached confluence, a pipette tip was used to scratch the cell monolayer. Scratches at 0 h were photographed, and three fields of view were photographed from each group. The scratches were subsequently re-photographed after culturing cells at 37 ℃ for 24 h and 48 h.

The cells (H22 and HepG2) and tissue were washed twice with PBS, and the protein supernatant was obtained by lysing with RIPA buffer and clearing the lysate by centrifugation. Proteins were separated by SDS-PAGE and transferred to the polyvinylidene fluoride (PVDF) membrane. Then, the membranes were blocked with 5% nonfat milk for 90 min. The membrane was incubated with primary antibodies to MMP-2, MMP-9, CD86, CD206, PD-L1, LC3, P62, I$\kappa$B$\alpha$, p-I$\kappa$B$\alpha$, P65, p-P65, and $\beta$-actin overnight at 4 ℃. After washing in triplicate, membranes were incubated with diluted secondary antibodies for 90 min at room temperature. Finally, membranes were incubated for 1 min using Enhanced Chemiluminescence (ECL) solution (Abiowell, Changsha, AWB0005, China) and then imaged by a gel imaging system (CLINX, Shanghai, China). The antibodies used are provided in Supplementary Table 2.

A total of 32 four weeks old male BALB/c mice were purchased from Hunan SJA Laboratory Animal Co., Ltd. (Changsha, China). Animals were given free access to water and food under standard animal rearing conditions (temperature 25 $\pm$ 2 ℃, relative humidity 50–60%). Animals were maintained on a 12 h/12 h dark/light cycle throughout the experiments. After a week of adaptive feeding, the mice were subcutaneously injected with H22 cells (1 $\times$ 10${}^6$) in the left armpit and the tumor was measured twice a week. The mice were divided into four groups, the control group, low-dose group (25 mg/kg quercetin), medium-dose group (50 mg/kg quercetin), and high-dose group (100 mg/kg quercetin). Mice were given drug intervention by gavage, and the control group was given the same volume of normal saline once a day for 21 days [30, 31]. At the end of the experiment, peripheral blood was obtained from mice after anesthesia, mice were subsequently sacrificed and the tumor tissue was dissected and weighed.

Tumor tissues were fixed in 4% paraformaldehyde. The sections were deparaffinized and stained according to instructions provided with the Hematoxylin-Eosin (H&E) staining kit (Abiowell, Changsha, China). The sections were observed and photographed under a light microscope.

First, peripheral blood samples and tumor tissue homogenates were prepared. Subsequently, the concentration of granulocyte-macrophage (GM-CSF, KE10015) and granulocyte colony-stimulating factor (G-CSF, KE10025) in cancer tissue homogenates, and TNF-$\alpha$ (KE10002), IL-6 (KE10007), and IL-17A (KE10020) in peripheral blood were determined according to the instructions provided with the ELISA kit (Proteintech, Chicago, USA).

After resuspending the cells, CD86 antibody (eBioscience, San Diego, 12-0862-82, USA) and CD206 antibody (eBioscience, 12-0114-82, USA) were added and incubated with the cells in the dark for 30 min. The cells were subsequently resuspended with PBS, filtered through a nylon mesh, and analyzed by a flow cytometry (Beckman, USA).

GraphPad 8.0 software (GraphPad Software, Inc., San Diego, California, USA) was used for statistical analysis, and three independent experimental data points were expressed as mean $\pm$ standard deviation (SD). Student’s t-test or one-way ANOVA was used to analyze differences between two or more groups. p 0.05 was considered statistically significant.

189 potential drug targets and 2998 liver cancer-related genes were acquired through database screening and 135 common targets were obtained (Fig. 1A). JUN, MAPK1, HSP90AA1, and AKT1 could be found to be highly correlated targets in the PPI network (Fig. 1B). Through the network diagram of ingredients/diseases/targets (Fig. 1C), we observed a complex interaction relationship among components, such as Bos taurus domesticus Gmelin, Moschus, Commiphora myrrha Engl and Boswellia carterii Birdw, liver cancer, and corresponding targets. These common targets were principally enriched in responses to lipopolysaccharide, responses to molecules of bacterial origin, cellular responses to chemical stress, epithelial cell proliferation, responses to oxidative stress, lipid and atherosclerosis, chemical carcinogenesis-receptor activation, hepatitis B, human cytogenetic infection, Kaposi sarcoma-associated, herpesvirus infection and other processes (Fig. 1D,E).

Fig. 1.

Network pharmacological analysis of Xihuang Pills. (A) Venn diagram shows the targets of Xihuang Pills and liver cancer. (B) PPI network shows the interaction of common targets of drug diseases. (C) The component-disease-target network diagram in Xihuang Pills. The main components, the ID of active ingredients, liver cancer-related targets, related pathways and liver cancer are marked with red, purple, yellow and green, respectively. (D, E) Bubble chart of GO and KEGG analysis.

From the ingredients/diseases/pathways/targets network diagram of Xihuang Pills, we found that the principal active ingredients of Xihuang Pills were quercetin, beta-sitosterol, pelargonidin, ellagic acid, morin, stigmasterol, and 3-methoxyfuranoguaia-9-en-8-one. Among these, quercetin had the highest degree score (Fig. 2A and Table 1). The core components of Xihuang Pills were identified via UPLC-MS and it was confirmed that quercetin was one of the key components of the Xihuang Pills (Fig. 2B) and thus we focused our investigation on quercetin action in HCC.

Fig. 2.

The identification of drug components in Xihuang Pills and the network analysis of liver cancer targets and pathways. (A) The network diagram of ingredients-diseases-pathways-targets in Xihuang Pills. The key active ingredients, liver cancer-related targets, related pathways and liver cancer are marked with blue, pink, green and orange, respectively. (B) HPLC was used to identify the core active ingredients of Xihuang Pills (quercetin).

To verify whether quercetin has an anti-cancer effect, we first explored the effect of quercetin on HCC in vitro. CCK8 was used to detect cell proliferation and we observed that the proliferation ability of cells treated with quercetin decreased compared to the control group, and that this was proportional to the quercetin concentration (Fig. 3A). After 72 h of quercetin treatment, the average inhibition rate was the highest, thus this time point was adopted for all subsequent cell experiments. Invasion characteristics of HepG2 were detected by transwell migration assays. Data gathered (Fig. 3B) indicates that with increasing quercetin concentrations, reduced cell migration was observed, suggesting that quercetin inhibits the cell invasion ability. We found that quercetin could promote the apoptosis of HCC cells while inhibiting the migration in a concentration-dependent manner, as shown in Fig. 3C,D. The expression of MMP-2 and MMP-9 in HCC cells gradually decreased with increased quercetin concentration (Fig. 3E). These results suggest that quercetin could significantly inhibit HCC invasiveness.

Fig. 3.

The effects of quercetin on the proliferation, apoptosis, invasion and migration of HCC. (A–D) The proliferation, invasion, apoptosis, and migration of HCC cells after quercetin treatment were measured by CCK8, transwell, FCM, and scratch test, respectively. (E) The MMP-2 and MMP-9 expression were detected via WB. *p 0.05, vs. the control.

In order to evaluate whether quercetin also plays a role in inhibiting HCC in vivo, we employed animal modeling. The experimental results obtained were consistent with the results obtained from cultured cells. The tumor volume and weight in the quercetin treatment group were significantly reduced (Fig. 4A–C). With increased quercetin concentration, its inhibitory effects on tumor development were measurably stronger. Similarly, after quercetin treatment, the density of cells decreased, and the nuclei appeared pyknotic and fission. The highest quercetin concentration group displayed the most significant results on tumor development (Fig. 4D).

Fig. 4.

Quercetin inhibits the development of HCC by inducing M1 polarization of macrophages. (A) Representative pictures of tumor morphology. (B,C) Quantitative statistics of changes in tumor volume and weight. (D) Representative images in the pathological changes. (E) The concentration of GM-CSF and G-CSF in cancer tissue homogenate was detected via ELISA. (F) The percentage of CD86 and CD206+ macrophages in tumor tissue were analyzed by FCM. (G) The CD86, CD206, and PD-L1 levels were measured by WB. n = 8 mice/group. *p 0.05, vs. the control.

GM-CSF and G-CSF are important factors regulating the immune microenvironment [32, 33]. Therefore, the GM-CSF and G-CSF were measured by ELISA in HCC tissue homogenates. The concentrations of GM-CSF and G-CSF after quercetin treatment were lower than those in the control group (Fig. 4E). In addition, flow cytometry was used to analyze the percentages of M1 macrophages (CD86+) and M2 macrophages (CD206+). After quercetin treatment, the abundance of M1 macrophages increased, while the abundance of M2 macrophages decreased (Fig. 4F). At the protein level, CD86 and CD206 levels were consistent with flow cytometry results (Fig. 4G), suggesting that macrophages in the HCC tissue were induced to polarize towards the M1 phenotype after quercetin treatment. PD-L1 is significantly locally upregulated in the tumor microenvironment of HCC [34] and it was observed that PD-L1 was obviously down-regulated in the low, medium, and high-quercetin groups (Fig. 4G). We speculate that these results indicate that quercetin may inhibit tumor growth by regulating macrophage polarization.

Autophagy is critical for antigen presentation and homeostasis in immune cells and the tumor microenvironment [17]. To explore the effects of quercetin on autophagy, we detected the expression of autophagy-related proteins LC3 and p62. The results showed that the LC3II/LC3I ratio was increased with the increase of quercetin concentration, while the p62 expression was decreased with the increase of quercetin concentration. Moreover, both findings were statistically significant (Fig. 5A). The levels of TNF-$\alpha$, IL-6, and IL-17A in peripheral blood were also assayed and these pro-inflammatory factors were significantly reduced by quercetin (Fig. 5B).

Fig. 5.

Quercetin affects autophagy through the NF-$\kappa$B pathway. (A) The expression levels of autophagy-related proteins (LC3 and p62) were analyzed by WB in tumor tissue. (B) ELISA was used to detect the concentration of TNF-$\alpha$, IL-6, and IL-17A in peripheral blood. (C) The expression levels of NF-$\kappa$B pathway proteins (I$\kappa$B$\alpha$, p-I$\kappa$B$\alpha$, p65 and p-p65) were elucidated by WB. n = 8 mice/group. *p 0.05, vs. the control.

NF-$\kappa$B can influence HCC progression by regulating inflammation and downstream metabolic pathways [35]. In addition, the NF-$\kappa$B pathway is a classical pathway that regulates autophagy [36]. We speculated that the effects of quercetin on autophagy might be related to the NF-$\kappa$B pathway and to test this expression of I$\kappa$B$\alpha$, p-I$\kappa$B$\alpha$, p65, and p-p65 in HCC tissue were analyzed. Results showed that the phosphorylation levels of I$\kappa$B$\alpha$ and p65 were significantly reduced (Fig. 5C). Further, these results indicate that quercetin might promote autophagy in HCC by regulating the NF-$\kappa$B pathway.

In sum, our findings indicate that quercetin inhibited HCC progression by inhibiting migration and invasion, promoting apoptosis and M1-type polarization (Fig. 6).

Fig. 6.

Schematic diagram of this study. Quercetin, the key active ingredient of Xihuang Pills, was screened out by network pharmacology. Next, quercetin was used to treat mice and cells in vivo and in vitro. In vitro experiments showed that quercetin might inhibit cell invasion and migration via the MMP-2/MMP-9 pathway. In vivo experiments clarified that quercetin might blunt HCC development by inhibiting the NF-$\kappa$B pathway, promoting autophagy, and regulating macrophage polarization.