Deciphering the Mechanism of Siwu Decoction Inhibiting Liver Metastasis by Integrating Network Pharmacology and In Vivo Experimental Validation

Abstract

Background:

Siwu Decoction (SWD) is a well-known classical TCM formula that has been shown to be effective as a basis for preventing and reducing liver metastases (LM). However, the active ingredients and potential molecular mechanisms remain unclear.

Objective:

This study aimed to systematically analyze the active ingredients and potential molecular mechanisms of SWD on LM and validate mechanisms involved.

Materials and methods:

The active ingredients in SWD were extracted by UHPLC-MS/MS in a latest study. Protox II was retrieved to obtain toxicological parameters to detect safety. Swiss Target Prediction database was exploited to harvest SWD targets. Five databases, Gene Cards, DisGeNET, Drugbank, OMIM, and TTD, were employed to filter pathogenic targets of LM. STRING database was utilized to construct the protein–protein interaction network for therapeutic targets, followed by Gene Ontology and the Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis. GEPIA database and the Human Protein Atlas were taken to observe the expression of core genes and proteins. ImmuCellAI algorithm was applied to analyze the immune microenvironment and survival relevant to core genes. Molecular docking was performed to verify the affinity of SWD effective ingredients to core targets. In vivo experiments were carried out to validate the anti-LM efficacy of SWD and verify the pivotal mechanisms of action.

Results:

Eighteen main bioactive phytochemicals identified were all non-hepatotoxic. PPI network acquired 118 therapeutic targets, of which VEGFA, CASP3, STAT3, etc. were identified as core targets. KEGG analysis revealed that HIF-1 pathway and others were critical. After tandem targets and pathways, HIF-1/VEGF was regarded as the greatest potential pathway. VEGFA and HIF-1 were expressed differently in various stages of cancer and normal tissues. There was a negative regulation of immunoreactive cells by VEGFA, which was influential for prognosis. Molecular docking confirmed the tight binding to VEGFA. This study revealed the exact effect of SWD against LM, and identified significant inhibition the expression of HIF-1α, VEGF, and CD31 in the liver microenvironment.

Conclusion:

This study clarified the active ingredients of SWD, the therapeutic targets of LM and potential molecular mechanisms. SWD may protect against LM through suppressing HIF-1/VEGF pathway.

Keywords

liver metastasis Siwu Decoction network pharmacology mechanism

Introduction

Liver metastasis (LM) is the major impediment in the management of malignancy, causing stifling destruction on patient survivals. Meanwhile, LM is the leading contributor to cancer-caused death in colorectal, gastric, pancreatic, melanoma, and breast cancers.¹ Especially, there is a high incidence of LM in colorectal cancer. Epidemiological data indicate that LM occurs in approximately 25% of patients with primary colorectal cancer and 40%-50% of patients with advanced colorectal cancer.² Nevertheless, direct treatment against LM is difficult to achieve gratifying results, even with hepatic metastasectomy. Fifty percent of patients still develop a recurrence in the liver after surgery.³ In contrast, robust prevention of LM carries considerable clinical value.

Hypoxia is an essential characteristic of liver pre-metastatic niche and plays a critical role in metastasis by augmenting the malignant phenotype of cancer cells and stimulating tumor angiogenesis, glycolysis, and immune escape.^4,5 Hypoxia-inducible factor-1 (HIF-1), as a prime promoter of hypoxic transcriptional response with core activity determined by HIF-1α, is identified as a decent anti-cancer target.⁶ PX-478, a developed HIF-1α inhibitor, has completed Phase I clinical trial in metastatic solid tumors with a proven favorable safety profile.⁷ There are multiple cross-linkages between HIF-1 and vasculature. HIF-1 promotes the secretion of vascular endothelial growth factor (VEGF) from tumor and stromal cells to induce neovascular growth. Furthermore, deficiency of HIF-1α in endothelial cells suppresses migration of cancer cells through the vascular endothelial barrier, thereby restraining metastasis.⁸ It is noteworthy that the liver is in an immune tolerant microenvironment in its physiological state. Hypoxic conditions further diminish the function of cytotoxic T cells. Additionally, HIF-1 enhances immune escape, thereby potentiating the immunosuppressive properties of the LM microenvironment.⁹ In particular, targeting HIF-1α boosts the efficacy of cancer immunotherapy with anti-CTLA-4 antibodies comparable to that of anti-CTLA-4 antibodies coupled with anti-PD-1 antibodies.¹⁰ It follows that HIF-1 is critical in regulating the hepatic metastatic microenvironment and deserves extensive attention.

Siwu Decoction (四物汤, SWD) is a classic formula for the treatment of diseases related to the liver meridian recorded in the Tang Dynasty and has been applied in clinical practice for more than 1000 years. Modern traditional Chinese medicine (TCM) practitioners take SWD as the basic formula plus other remedies for the treatment of LM with satisfactory results. An earlier study conducted by Ohnishi et al confirmed the anti-LM efficacy of SWD.¹¹ Subsequent studies found that Wenqing Yin (温清饮) and Shiquan Dabu Decoction (十全大补汤), which contain SWD, were determined to be capable of preventing LM, while Danggui Shaoyao San (当归芍药散), Renshen Yangrong Decoction (人参养荣汤), and Liu Junzi Decoction (六君子汤), which do not contain SWD, were not identified to exhibit anti-LM efficacy.¹² These results suggested that SWD is an effective formula against LM. Our team’s previous study, funded by the National Natural Science Foundation of China, also discovered that the herbs in SWD (Radix Angelicae Sinensis, Radix Paeoniae Alba and Rhizoma Ligustici Chuanxiong), which belong to the liver meridian, can prevent LM. Nevertheless, the mechanism of the effectiveness of SWD against LM remains to be further clarified.

We performed novel network pharmacology strategies to systematically analyze the drug-target interaction network. To raise the credibility of the phytochemicals in SWD, we referenced data derived from ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS) in a recent study.¹³ A systematic strategy including network pharmacology combined with molecular docking, human sample databases extraction, immunological and prognostic analysis, and in vivo experimental validation elucidated that the main mechanism of SWD against LM is most probably on the HIF-1/VEGF pathway. A workflow chart is shown in Figure 1.

Figure 1.

Workflow diagram of this study.

Material and Methods

Determination of the Main Phytochemicals of SWD and Evaluation of Pharmacological and Toxicological Parameters

The main bioactive phytochemicals of SWD were identified in vitro and the contents(mg/g) of 18 bioactive phytochemicals were determined by UHPLC-MS/MS from a recent study.¹³ Properties of main phytochemicals in SWD were retrieved from SwissADME web tool (https://www.swissadme.ch)¹⁴ by importing 2D chemical structures acquired from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/).¹⁵ These properties were used for subsequent screening. Compounds were screened out for further analysis in accordance with the Lipinski’s rule of 5, including molecule weight (MW) ≤500, hydrogen bond donors (Hdon) ≤5, hydrogen bond acceptors (Hacc) ≤10, rotatable bonds (Rbon) ≤ 10, and logP ≤ 5. The toxicological parameters of main phytochemicals in SWD were obtained from Protox II web (https://tox-new.charite.de/protox_II/).¹⁶ The toxicological end points were classified as active or inactive, including hepatotoxicity, carcinogenicity, immunotoxicity, mutagenicity, and cytotoxicity. Furthermore, acute oral toxicity (LD50) values were recorded.

Prediction of Therapeutic Targets for SWD Acting on LM

Firstly, Swiss Target Prediction database (http://www.swisstargetprediction.ch/)¹⁷ was employed to collect targets for bioactive ingredients of SWD. The retrieval standard was set as Probability value > 0, and the species were limited only to Homo sapiens. To search for pathogenic targets of LM, we applied the keyword of liver metastasis to retrieve the related targets in the following 5 public databases: DisGeNET database (http://www.disgenet.org),¹⁸ GeneCards database (https://www.genecards.org/),¹⁹ Drugbank (https://go.drugbank.com/),²⁰ OMIM (https://omim.org/),²¹ and TTD (https://db.idrblab.net/ttd/).²² The disease targets acquired above were organized into a collection and the repetitive data was deleted. To clarify the potential therapeutic targets, Venn diagram was drawn for overlap analysis by an online tool (https://www.bioinformatics.com.cn/). The extracted overlapping targets were considered as the therapeutic targets of SWD acting on LM.

Construction of Protein–Protein Interaction Network and Capture of Core Targets

The above genes were mapped to the STRING 11.5 database (https://string-db.org/)²³ to build a protein-protein interaction (PPI) network. The species were limited to Homo sapiens. The minimum required interaction score was set to 0.4 by default, and independent proteins were hidden. To further capture the core therapeutic targets, PPI network was visualized by Cytoscape (version 3.9.1). The network analysis ranked the targets by degree values, by virtue of Cytohubba (Cytoscape plugin), the top 15 targets were focused as core targets.

Gene Ontology and the Kyoto Encyclopedia of Genes and Genomes Pathway Enrichment Analysis

Gene Ontology (GO) functional enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of therapeutic targets were conducted by Metascape (https://metascape.org/) and visualized through an online tool (http://www.bioinformatics.com.cn/). GO terms were categorized into 3 sections, namely, biological process (BP), cellular component (CC), and molecular function (MF). Sankey dot diagram illustrated fully the linkage of KEGG pathways to core targets. Results with P-values < .05 were considered statistically significant.

Immune Cells Infiltration in Tumor Microenvironment and Prognostic Significance of Core Targets

Immunogenomic analyses were applied to calculate the abundance and proportions of immune cells in the tumor microenvironment. To reveal core gene-immunophenotype relationships, immunogenomic analysis was performed in the GSCA database (http://bioinfo.life.hust.edu.cn/GSCA/#/)²⁴ by ImmuCellAI algorithm with 24 immunes cells. Immune Cell Matrix is comprised of 18 T-cell subtypes and 6 other immune cells: B cell, NK cell, Monocyte cell, Macrophage cell, Neutrophil cell, and DC cell. Genomic associations with 4 clinical outcomes were portrayed by comparing the high and low expression groups of core genes. The expression of immune-related genes in given cell types between tumor and normal tissues was compared by CIBERSORT deconvolution analysis in the GEPIA2021 database (http://gepia2021.cancer-pku.cn/index.html).²⁵

Differences in Gene and Protein Expression Levels

The GEPIA database (http://gepia.cancer-pku.cn/detail.php)²⁶ can facilitate the release of the value of large genome data in TCGA and GTEx using a standard processing pipeline. It was utilized to perform matching normal and tumor samples, processing data from tumor samples at different pathological stages to explore gene expression differences of core targets. Immunohistochemistry data from the Human Protein Atlas (HPA) (https://www.proteinatlas.org/) were collected to analyze differences in protein expression between normal and tumor tissues.

Molecular Docking Simulations

Molecular docking between core targets and bioactive ingredients was carried out to verify the accuracy of the network pharmacology prediction. The selected 3D structures of the core targets were collected from the PDB database (http://www.rcsb.org/).²⁷ The conditions were set for the source organism to be “Homo sapiens,” the original 3D structure featuring to have at least 1 eutectic ligand and only the 3D structure with the smallest “resolution” value. Then they were processed by PYMOL (version 2.3.6). Furthermore, the 3D structures of the active ingredients were obtained from the PubChem database¹⁵ and were converted into MOL2 format by OPENBABEL. Subsequently, AutoDock Tools (version 1.5.6) and AutoDock Vina (version 1.1.2) were employed for molecular docking. According to the affinity and the number of hydrogen bonds formed, the most practical conformation of each core target was selected, and the complex of the core target and the active ingredient was output in PDB format by PYMOL. Finally, all complexes were analyzed and visualized by PLIP Web tool (https://plip-tool.biotec.tu-dresden.de/).²⁸ The binding strength was evaluated in terms of the binding affinity score with a threshold of affinity -5.0 kcal/mol.

In Vivo Validation

Drug preparation and reagents

SWD is composed of 4 Chinese medicinal herbs, specific details are described in Table 1. All herbs were purchased from Beijing Kangmei Pharmaceutical Co., Ltd (Beijing, China) and authenticated by the Department of Pharmacy, Wangjing Hospital Affiliated to China Academy of Chinese Medical Sciences. All herbs were pooled together and decocted twice in pure water. The 2 filtrates were collected and mixed. Then the filtrate was concentrated to 1.97 g/ml. The final solution was stored rather stable in a refrigerator at 4°C for 1 week. Capecitabine (State Food and Drug Administration approval number: H20133365) was purchased from Jiangsu Hengrui Pharmaceutical CO., Ltd. (Jiangsu, China). Anti-bodies (VEGF: ab2349, CD31: ab182981, and HIF-1α: ab16066) were purchased from Abcam (Cambridge, MA, United States). Antibody against β-actin was obtained from MDL (Beijing, China).

Table 1.

The Herbal Composition of Siwu Decoction (SWD).

Chinses name	Scientific name	Family	Part used	Dosage (g)
Bai Shao	Paeonia lactiflora Pall.	Paeoniaceae	Radix	12
Shu Di Huang	Rehmannia glutinosa (Gaertn.) DC.	Orobanchaceae	Radix	12
Chuan Xiong	Ligusticum chuanxiong Hort.	Apiaceae	Rhizoma	12
Dang Gui	Angelica sinensis (Oliv.) Diels	Apiaceae	Radix	12

Cells and animals

The CT26 (murine colon cancer cell line) was purchased from Cell Bank of Chinese Academy of Sciences (Shanghai, China). The CT26 cell line was cultured in complete RPMI-1640 medium. All cells were grown at 37°C in a humidified atmosphere containing 5% CO₂.

Male BALB/c mice (6 weeks old, 18-20 g) were supplied by Beijing Vital River Laboratory Animal Technology Co., Ltd. [License No.: SCXK(Jing)2021-0006]. All mice were raised in the Animal Experiment Center of China Academy of Chinese Medical Sciences under the controlled environment (temperature: 22 ± 2°C, humidity: 50 ± 10%, 12:12 h light/dark cycle, free specific pathogen) and replenished with free water and food. Before modeling, adaptive feeding was given for 7 days. All animal experiments were carried out in accordance with the procedures for the Laboratory Animal Ethics Committee of China Academy of Chinese Medical Sciences.

Mouse model and experimental design

All experimental mice were anesthetized by pentobarbital (50 mg/kg, i.p.). The model mice were established with intrasplenic injection of liver metastasis models.²⁹ CT26 cells in logarithmic growth phase were taken with a concentration of 2 × 10⁶ cells/mL, and 0.1 mL of suspension was inoculated into the lower pole of the spleen. The sham operation group mice were injected with the equivalent volume of PBS. After surgical procedures, the mice were placed in a 37°C thermostatic pad for warming and awakening.

Three days after recovery, the model mice were randomly separated into 4 groups. 5 groups (6 per group) mice commenced to receive simultaneous gavage intervention. SWDL group and SWDH group mice were respectively treated with a low-dose SWD at 9.86 g/kg/d and a high-dose SWD at 19.73 g/kg/d. The low dose is the clinical equivalent dose. Capecitabine group mice were treated by intragastric administration with capecitabine at 150 mg/kg/d.³⁰ The model group and sham operation group mice were given normal saline by gavage. All mice were given the equivalent volume of solution. Experiments were terminated on day 28.

Hepatic histopathology

To visualize the pathological changes in liver, tissues were fixed in tissue fixative, embedded in paraffin, sectioned into 4-μm-thick slices and stained with hematoxylin and eosin. The morphological changes of liver tissues were observed under a light microscope.

Immunohistochemistry staining

Paraffin-embedded liver tissues were processed in 4 μm-thick slices for immunohistochemistry staining. After antigen repair and blocking, primary antibodies against VEGF (1:100), CD31 (1:500) and HIF-1α (1:100) were separately incubated overnight at 4°C. Following incubation with corresponding secondary antibodies for 20 minutes at room temperature, immunoreactivity was visualized using diaminobenzidine (DAB). For quantification, 5 visual fields were sampled randomly in each slice (× 400) to count the number of positive cells.³¹

Western blot analysis

Western blot assay was executed as previously reported.³² Total protein of liver tissues was achieved after homogenizing with lysis buffer and collecting the supernatant after centrifugation at 4°C and 12 000 rpm for 15 minutes. Protein concentration was quantified by BCA protein assay kit. After thermal denaturing, 40 μg proteins samples were loaded and separated by 10% polyacrylamide gel and by electro-transferred to polyvinylidene fluoride (PVDF) membrane. After blocked, the membrane was separately incubated with the primary antibodies against HIF-1α (1:1000) and VEGF (1:1000) at 4°C overnight, followed by reaction with secondary antibodies for 1h at room temperature. The bands were detected with chemiluminescence imaging system.

Statistical Analysis

Statistics were undertaken using SPSS 26.0 software. Differences between individual groups were assessed using 1-way analysis of variance (ANOVA), followed by multiple comparison performed with Tukeys post-hoc analysis. All data were expressed as mean ± standard deviation. P value < .05 was considered statistically significant.

Results

Main Phytochemicals of SWD and Pharmacological and Toxicological Properties

Qualitative and quantitative analysis was carried out to identify the phytochemicals of SWD. Eighteen main bioactive phytochemicals were determined by UHPLC-MS/MS.¹³ Paeoniflorin was found to have the highest concentration (5.793 ± 0.030 mg/g SWD), followed by albiflorin (1.083 ± 0.006 mg/g SWD), Senkyunolide I (0.814 ± 0.004 mg/g SWD) and Rehmannioside D (0.581 ± 0.003mg/g SWD) (Figure 2A). The toxicological properties of main phytochemicals in SWD were accessed using Protox II against hepatotoxicity, carcinogenicity, immunotoxicity, mutagenicity, cytotoxicity, and LD50. All the phytochemicals exhibited inactive as hepatotoxic and mutagenicity (Figure 2B). Further-more, catechin manifested the highest LD50 value (10 000 mg/kg). Although senkyunolide I and senkyunolide H showed the lowest LD50 value (35 mg/kg), all the toxicological evaluations were inactive. Physicochemical properties of each phytochemical were collected by SwissADME. After data extraction and analysis, 9 ingredients satisfied the requirements the Lipinski’s rule of 5, namely, gallic acid, catechin, caffeic acid, ferulic acid, senkyunolide I, senkyunolide H, 3-butylphthalide, ligustilide, vanillin (Table 2). In addition, paeoniflorin and chlorogenic acid which held promising bioactivities but did not fulfill the above criteria, were also enrolled for further analysis.^33,34

Figure 2.

The contents and toxicological parameters of bioactive phytochemicals in Siwu Decoction (SWD). (A) The contents of chemical compositions in SWD were quantitative detected by ultra-high performance liquid chromatography. The data were derived from in a recent report 13. (B) The toxicological parameters of 18 phytochemicals in SWD, such as hepatotoxicity, carcinogenicity, immunotoxicity, mutagenicity, cytotoxicity, and acute oral toxicity (LD50, mg/kg). The red and green circles represent active or inactive toxicological end points, respectively.

Table 2.

Properties of main phytochemicals in SWD.

Compound	Formula	MW (g/mol)	Hdon	Hacc	Rbon	LogP	SAscore
Gallic acid	C₇H₆O₅	170.13	4	5	1	0.21	1.22
Oxypaeoniflorin	C₂₃H₂₈O₁₂	496.51	6	12	7	−0.70	5.48
Catechin	C₁₅H₁₄O₆	290.29	5	6	1	0.83	3.50
Chlorogenic acid	C₁₆H₁₈O₉	354.31	6	9	5	−0.39	4.16
Caffeic acid	C₉H₈O₄	180.16	3	4	2	0.93	1.81
Albiflorin	C₂₃H₂₈O₁₁	480.46	5	11	7	−0.43	5.55
Paeoniflorin	C₂₃H₂₈O₁₁	480.46	5	11	7	−0.19	5.51
Ferulic acid	C₁₀H₁₀O₄	194.18	2	4	3	1.36	1.93
Verbascoside	C₂₉H₃₆O₁₅	624.59	9	15	11	−0.60	6.37
Catalpol	C₁₅H₂₂O₁₀	362.33	6	10	4	−2.33	5.72
Senkyunolide I	C₁₂H₁₆O₄	224.25	1	3	2	1.22	4.34
Senkyunolide H	C₁₂H₁₆O₄	224.25	2	4	2	1.18	4.34
Benzoylpaeoniflorin	C₃₀H₃₂O₁₂	584.57	4	12	10	1.28	6.01
Aucubin	C₁₅H₂₂O₉	346.33	6	9	4	−2.07	5.79
3-Butylphthalide	C₁₂H₁₄O₂	190.24	0	2	3	2.81	2.51
Ligustilide	C₁₂H₁₄O₂	190.24	0	2	2	2.75	3.64
Rehmannioside D	C₂₇H₄₂O₂₀	686.61	13	20	10	−5.79	7.90
Vanillin	C₈H₈O₃	152.15	1	3	2	1.20	1.15

Therapeutic Targets of SWD for LM and Construction of PPI Network

A total of 2254 LM-related targets were retrieved through a comprehensive search of the 5 databases after deduplication (Figure 3A). About 222 protein targets of the above ingredients were identified using the Swiss Target Predication. Subsequently, the targets of the 2 sets were intersected and the 118 targets obtained were regarded as potential therapeutic targets (Figure 3B). The 118 targets were registered into the STRING 11.5 database and then visualized by Cytoscape software to construct a PPI network diagram (Figure 3C). The circles represent target nodes. Connecting lines denote the association of target nodes. The higher the degree value, the larger the circle. The color shade changes gradually from dark to light depending on the degree value. The PPI network featured 118 nodes, 1321 edges and an average node degree of 22.4. Cytohubba plug-in of Cytoscape was executed next. It was noteworthy that the 15 core targets were endorsed important in the mechanism of action of LM and were clustered out to form a new PPI network (Figure 3D). 15 targets (VEGFA, CASP3, STAT3, CCND1, ESR1, EGFR, PTGS2, AR, BCL2L1, HSP90AA1, ALB, MMP9, MAPK8, MAPK1, and IGF1R) were demonstrated by score and degree (Figure 3E).

Figure 3.

Construction of a protein-protein interaction (PPI) network with LM-related SWD targets. (A) LMrelated targets were drawn from 5 databases. (B) Venn diagram for the overlap analysis of LM and SWD targets. (C) The PPI network of the selected 118 targets. The color shade and circle size reflect the degree value. (D) 15 core targets were identified by Cytohubba plug-in of Cytoscape. The color shade is proportional to the score. (E) The core targets were ranked by score.

GO and KEGG Enrichment Analysis and the Pathway-Target Network

GO items contain 3 sections. The top 10 items of each section were listed in Figure 4A. BP mainly involved response to hormone, cellular response to nitrogen compound, and positive regulation of cell migration. CC was mainly related to side of membrane, extracellular matrix, and external encapsulating structure, etc. MF was primarily phosphotransferase activity, kinase activity, and oxidoreductase activity, etc. KEGG enrichment pathways were mapped to core therapeutic targets to build an interaction network (Figure 4B). The signaling pathways acting were mainly pathways in cancer, HIF-1 signaling pathway, TNF signaling pathway, and PI3K-Akt signaling pathway. Notably, there was a tight linkage between the highest scoring VEGFA and the HIF-1 signaling pathway (Figure 4C), and we hypothesized that both were the mechanism by which SWD exerted therapeutic effect.

Figure 4.

Enrichment analysis of therapeutic targets. (A) GO enrichment analysis. (B) KEGG enrichment analysis. (C) The critical KEGG pathway HIF-1 processed after enrichment analysis. The threshold P value < .05.

Identification of Immune Infiltration and Prognosis Signature

In consideration of the high incidence of liver metastasis in colon adenocarcinoma (COAD), we focused the cancer type on COAD in the following study. The relationship between 15 core genes and 24 immune cells was comprehensively illustrated in Figure 5A. VEGFA in focus above was inversely correlated with the immune cell infiltration score. More specifically, VEGFA was positively correlated with CD8+ naive cell, monocyte, γδT cell, nTreg, and neutrophil, and negatively correlated with cytotoxic T cell, macrophage, NK cell, CD4+ T cell, Th2 cell, Tfh cell, MAIT cell, B cell, CD8+ T cell, NKT cell, and Th17 cell. The VEGFA expression of CD8+ T cells between COAD and normal tissues was significantly different (Figure 5B). Disease Free Interval (DFI), disease Specific Survival (DSS), Overall Survival (OS), and Progression Free Survival (PFS) are designated 4 classic clinical prognosis signatures. VEGFA, STAT3, MAPK1, EGFR, and PTGS2 among 15 core genes reflected a significant relevance to prognosis (Figure 5C). Among them, there were significant differences in DFI and PFS between COAD patients with high and low VEGFA expression (Figure 5D-G). COAD patients in the VEGFA low expression group suggested superior survival characteristics.

Figure 5.

Immunity and prognostic features. (A) Correlation between 15 core genes and 24 immune cells in colon adenocarcinoma (COAD). The color of the bubble indicates the magnitude of the correlation, with redder colors indicating a positive correlation and bluer blues indicating a negative correlation. The size of the bubble indicates the magnitude of the significance. Points circled by black outlines indicate FDR < 0.05. (B) Comparison of VEGFA expression in CD8+T cells between COAD tissues and normal tissues by CIBERSORT analysis. (C) Association of 15 core genes with 4 survival types (OS, PFS, DSS, and DFI) in COAD. The bubble color from blue to red represents a low to high hazard ratio, and the bubble size is positively correlated with significance of the Cox P value. The black outline border indicates Cox P value ≤ .05. DFI (D), DSS (E), OS (F), and PFS (G) differences between VEGFA high and low gene expression groups of COAD.

Verification of Differentially Expressed Genes and Proteins in Patients

To elucidate the effect of the therapeutic target VEGFA and HIF-1 signaling pathway, we did a preliminary validation utilizing patient samples from database. The results from GEPIA database indicated that the expression of VEGFA was significantly different in different stages of COAD, while VEGFA and HIF-1 did not differ significantly between COAD and normal tissues, and HIF-1 did not differ significantly in different stages of COAD (Figure 6A-D). Immunohistochemical results from the Human Protein Atlas database showed that the protein expression of VEGFA and HIF-1A was higher in COAD than in normal tissues (Figure 6E).

Figure 6.

Expression levels of VEGFA and HIF-1. Differential expression of VEGFA between COAD and normal patients (A) and between different stages of COAD (B). Differential expression of HIF-1 between COAD and normal patients (C) and between different stages of COAD (D). (E) Differences in protein expression between normal and COAD tissues.

Molecular Docking Results of Core Phytochemicals to Therapeutic Targets

To further document the effect of bioactive phytochemicals in SWD on the core therapeutic targets, the top 3 therapeutic targets, VEGFA, STAT3 and CASP3, were picked for molecular docking analysis with 11 phytochemicals. The results demonstrated that the affinity of all conformations was less than −5 kcal/mol, implying good binding ability (Figure 7A). For the molecular docking of VEGFA, we found that paeoniflorin, catechin and chlorogenic acid were tightly bound to VEGFA with the strongest binding force. The conformations are as follows (Figure 7B).

Figure 7.

Molecular docking results. (A) Binding affinity of core phytochemicals in SWD to the top 3 therapeutic targets. The color shade represents the level of binding affinity score. (B) The conformation of VEGFA and paeoniflorin, VEGFA and chlorogenic acid, VEGFA and catechin.

Validation of the Inhibitory Effect of SWD on LM and Modulation of the HIF-1/VEGF Pathway in Mice In Vivo

To further verify the efficacy of SWD on LM, we established a LM model mouse with intrasplenic injection of CT26 and selected capecitabine as a control drug to compare the inhibition of LM (Figure 8A). The weight changes were evaluated (Figure 8B). The results showed that the weight of mice in the Sham-operated group increased steadily. The weight of mice in the model group increased rapidly after the 14th day. The weight of mice in the SWDL, SWDH and capecitabine groups increased slowly and steadily. On the 28th day, the weight of mice in the Sham-operated and model group was significantly different from that in the capecitabine and SWDH group, suggesting that the 2 treatment groups might control tumor progression. The number of metastatic nodules on the surface of the liver in the SWDL, SWDH, and capecitabine group was markedly less than that in the model group (Figure 8C). The SWDL, SWDH, capecitabine, and sham-operated groups had lighter liver weights than the model group (Figure 8D). In addition, both liver morphology (Figure 8E) and HE staining (Figure 8F) revealed inhibition of LM by SWD and capecitabine.

Figure 8.

Inhibitory effect of SWD on LM in mice. (A) Gavage dosing scheme after grouping of model mice. (B) Body weight of mice in each group. (C) Number of nodules on the surface of the liver in each group of model mice. (D) Weight of the liver was measured in different groups. (E) Liver morphology in each group of mice after 28 days. (F) The HE-stained liver sections from different groups. Data are presented as mean ± SD (**P < .01, ***P < .001, ****P < .0001).

To validate the core therapeutic targets and pathways obtained from the previous network pharmacology analysis through in vivo experiments, we examined the expression of the HIF-1/VEGF pathway. Detection of CD31 was performed to assess tumor angiogenesis. The results of VEGF, CD31, and HIF-1α assessed by immunohistochemistry were illustrated in Figure 9A and B, showing that the expression of the 3 proteins in the SWDH group was strongly weaker than that in the model group. The results of western blot of VEGF and HIF-1α demonstrated the same trend (Figure 9C and D). Both proteins were obviously dampened in the SWDH group compared with the model group.

Figure 9.

The regulating effect of SWD on HIF-1/VEGF pathway. (A) The typical images of Immunohistochemistry staining (magnification, 200×). (B) Expression levels of VEGF, CD31, and HIF-1α in each group by immunohistochemistry. (C) VEGF and HIF-1α were determined by Western blot. (D) Effects of SWD on the protein levels of VEGF and HIF-1α by Western blot. Data are presented as mean ± SD (*P < .05, **P < .01, ***P < .001).

Discussion

The liver is a prevalent site of metastasis in various malignancy. It is a characteristic tumor microenvironment which consists of a combination of tumor cells, infiltrating immune cells, endothelial cells, tumor-associated fibroblasts, and extracellular matrix.³⁵ The complex signaling crosstalk and interactions result in an acidic, hypoxic, immunosuppressive, angiogenic profile. The therapeutic practice targeting critical components, important profiles and targets in the tumor microenvironment has shown excellent clinical efficacy. For instance, the latest FRESCO-2 trial, published in Lancet, found that fruquintinib, which inhibits tumor angiogenesis, reduced the risk of death by 35% in patients with metastatic colorectal cancer.³⁶ In our study, SWD worked on critical pathways of the tumor microenvironment and showed favorable anti-liver metastatic efficacy.

After identification, all active ingredients in SWD were non-hepatotoxic and their safety characteristics suggested that SWD can be applied in the prophylactic phase. Gallic acid, catechin, caffeic acid, ferulic acid, senkyunolide I, senkyunolide H, 3-butylphthalide, ligustilide, vanillin, paeoniflorin and chlorogenic acid were the active ingredients of SWD against liver metastasis. Gallic acid, catechin, chlorogenic acid, caffeic acid and ferulic acid are all polyphenolic compounds that have been tapped for their clear anticancer effects and developed as small molecule agents for cancer prevention and treatment. Polyphenols can exert inhibitory effects on cancer cell proliferation, tumor growth, angiogenesis, metastasis, and inflammation.³⁷ The first 2 polyphenols were found to inhibit the liver metastasis process.^38,39 Catechin has been found to restrain melanoma cell migration, proliferation and to resist tumor angiogenesis.^40,41 It is worth mentioning that paeoniflorin, which is the most plentiful in SWD, has exceptional anti-inflammatory and immunomodulatory effects. Multiple studies have revealed its ability to dampen metastasis through NF-κB and STAT3 signaling pathways.³³

This study ultimately gained 15 key therapeutic targets of SWD. We focused on the top 3 scoring targets as VEGFA, CASP3, and STAT3. After correlating with the enrichment analysis, the HIF-1 signaling pathway formed an intimately affiliated crosstalk network with VEGFA and STAT3 that caught our attention, suggesting the involvement between hypoxia, angiogenesis, and immunity. HIF-1 has a direct or indirect impact on mediating significant profiles of the tumor microenvironment, and HIF-1 was discovered by Semenza and Wang back in 1991, a discovery that was awarded the Nobel Prize.⁴² Under hypoxic conditions, HIF-1α is structurally stabilized and binds to HIF-1β to form a heterodimer into the nucleus, which binds to the hypoxia response element (HRE) and stimulates transcription.⁴³ On the one hand, HIF-1 stimulates the expression of glycolytic genes and alters metabolic pathways, leading to a low PH and acidic microenvironment.⁴⁴ On the other hand, HIF-1 promotes VEGF to stimulate angiogenesis to balance hypoxia and open nutrient channels. In addition, VEGF further promotes increased HIF-1 expression, forming positive feedback.⁴⁵ However, both acidic and hypoxic microenvironments drive immune escape and tumor metastasis.^46,47 The activated STAT3/HIF-1/VEGFA axis was observed to enhance macrophage polarization and thereby facilitate gastric cancer metastasis and angiogenesis in vitro and in vivo experiments.⁴⁸ HIF-1 represents a promising therapeutic target for novel cancer treatment, focusing on the mitigation of adverse factors in the tumor microenvironment. Clinical reports revealed that HIF-1 overexpression was indeed an independent risk factor for liver metastasis from colorectal cancer, accompanied by high expression of VEGF.⁴⁹

This study, based on clinical patient data, discovered that VEGFA and HIF-1 were differentially expressed in COAD tissue and normal tissue, and that VEGFA expression was significantly higher in advanced COAD tissue than in early stages. Given the specificity of the immune status in the tumor microenvironment, we analyzed immunity and determined that CD8+ T cells are dysfunctional in tumor tissue compared to normal tissue. And there is a negative correlation between VEGFA and multiple immunologically active cells, and this effect on immunity also penetrates deeply into patient prognosis. The results of in vivo experiments in this study suggested that SWD significantly decreased the protein expression of VEGF and HIF-1α in the liver metastasis microenvironment and suppress CD31+ vascular endothelial cells, indicating that the effect mechanism of SWD against liver metastasis may be the inhibition of HIF-1/VEGF pathway. Among the active ingredients of SWD, we also retrieved approximate effects on HIF-1/VEGF axis. A study discovered that gallic acid inhibited VEGF secretion and tumor angiogenesis in a concentration-dependent manner, critically by PTEN/AKT/HIF-1α pathway.⁵⁰ Similarly, caffeic acid has been proven to restrain tumor angiogenesis by inhibiting STAT3 activity, HIF-1α and VEGF expression.⁵¹ Vanillin inhibited cell motility by suppressing STAT3-mediated HIF-1α mRNA expression in malignant melanoma cells.⁵² Chlorogenic acid depressed hypoxia-induced angiogenesis through downregulation of the HIF-1α/AKT pathway.⁵³ As such, we hypothesized that these 4 phytochemicals may be key ingredients of SWD acting on the HIF-1α/VEGF axis. SWD, as a classic formula for nourishing the liver and replenishing blood, is used to prevent liver metastasis. This follows the philosophy of TCM, which prioritize pacifying areas that have not yet been injured to minimize the total area of injuries. This study highlights the significance of therapy focused on nourishing the liver and replenishing blood in preventing and treating liver metastasis. It suggests that the mechanism involved may improve the tumor microenvironment, offering valuable guidance for clinical practice. In the end, this study has certain limitations and more in-depth studies are needed to determine the effector mechanism of SWD against liver metastasis considering the complex composition and action of herbal preparations. Moreover, due to limitations in the content of databases related to network pharmacology, the mechanisms obtained through systematic analysis may not be exhaustive.

Conclusion

In conclusion, the main ingredients of SWD against liver metastasis are gallic acid, caffeic acid, vanillin, chlorogenic acid, etc., mainly by mediating the HIF-1/VEGF pathway. This protective effect presumably inclines to inhibit angiogenesis, improve hypoxia and immune status, and ultimately safeguard prognosis.

Footnotes

Author Contributions

XC, FX, and CH contributed equally to this work. They conducted the research program and completed the paper writing. XZ, SW, and HX assisted in implementing the experiment. CA and YL performed the study data analysis. PX and SZ conceived the paper for intellectual content, carried out project design and supervised the implementation. All authors contributed to the study and approved the final manuscript.

Data Availability Statement

Data supporting the findings of this study are available from the manuscript or the corresponding author upon reasonable request.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by High level Traditional Chinese Medicine Hospital Construction Project (No. WJYY-XZKT-2023-37), the Scientific and Technological Innovation Project of China Academy of Chinese Medical Sciences (No. CI2022C002, No.CI2021B009).

ORCID iD

Xuelei Chu

References

Wang

, et al. Relaxin gene delivery mitigates liver metastasis and synergizes with check point therapy. Nat Commun. 2019;10:2993.

Tsilimigras

Brodt

Clavien

, et al. Liver metastases. Nat Rev Dis Primers. 2021;7:27.

Petrelli

NJ.

Perioperative or adjuvant therapy for resectable colorectal hepatic metastases. J Clin Oncol. 2008;26:4862-4863.

Bosco

D'Orazi

Del Bufalo

Targeting hypoxia in tumor: a new promising therapeutic strategy. J Exp Clin Cancer Res. 2020;39:8.

Petrova

Annicchiarico-Petruzzelli

Melino

Amelio

The hypoxic tumour microenvironment. Oncogenesis. 2018;7:10.

Choudhry

Harris

AL.

Advances in hypoxia-inducible factor biology. Cell Metab. 2018;27:281-298.

Lee

Kim

HM.

A novel approach to cancer therapy using PX-478 as a HIF-1α inhibitor. Arch Pharm Res. 2011;34:1583-1585.

Branco-Price

Zhang

Schnelle

, et al. Endothelial cell HIF-1α and HIF-2α differentially regulate metastatic success. Cancer Cell. 2012;21:52-65.

You

Nepovimova

, et al. Hypoxia-inducible factors: master regulators of hypoxic tumor immune escape. J Hematol Oncol. 2022;15:77.

10.

Bailey

Liu

, et al. Targeting HIF-1α abrogates PD-L1-mediated immune evasion in tumor microenvironment but promotes tolerance in normal tissues. J Clin Investig. 2022;132:e150846.

11.

Ohnishi

Fujii

Hayakawa

, et al. Oral administration of a Kampo (Japanese herbal) medicine Juzen-taiho-to inhibits liver metastasis of colon 26-L5 carcinoma cells. Jpn J Cancer Res. 1998;89:206-213.

12.

Onishi

Yamaura

Tauchi

, et al. Expression of the anti-metastatic effect induced by Juzen-taiho-to is based on the content of Shimotsu-to constituents. Biol Pharm Bull. 1998;21:761-765.

13.

Qiao

Wang

, et al. Dissolution rule of Siwu decoction by different decocting methods based on chemistry holography. Chin Tradit Herb Drugs. 2020;51:4960-4971.

14.

Daina

Michielin

Zoete

SwissADME: a free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules. Sci Rep. 2017;7:42717.

15.

Kim

Chen

Cheng

, et al. PubChem in 2021: new data content and improved web interfaces. Nucleic Acids Res. 2021;49:D1388-D1395.

16.

Banerjee

Eckert

Schrey

Preissner

ProTox-II: a webserver for the prediction of toxicity of chemicals. Nucleic Acids Res. 2018;46:W257-W263.

17.

Daina

Michielin

Zoete

SwissTargetPrediction: updated data and new features for efficient prediction of protein targets of small molecules. Nucleic Acids Res. 2019;47:W357-W364.

18.

Piñero

Saüch

Sanz

Furlong

LI.

The DisGeNET cytoscape app: Exploring and visualizing disease genomics data. Comput Struct Biotechnol J. 2021;19:2960-2967.

19.

Stelzer

Rosen

Plaschkes

, et al. The GeneCards suite: from gene data mining to disease genome sequence analyses. Curr Protoc Bioinformatics. 2016;54:1.30.31-31.30.33.

20.

Wishart

Feunang

Guo

, et al. DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Res. 2018;46:D1074-d1082.

21.

Amberger

Hamosh

Searching online Mendelian inheritance in Man (OMIM): a knowledgebase of human genes and genetic phenotypes. Curr Protoc Bioinformatics. 2017;58:1.2.1-1.2.12.

22.

Zhou

Zhang

Lian

, et al. Therapeutic target database update 2022: facilitating drug discovery with enriched comparative data of targeted agents. Nucleic Acids Res. 2022;50:D1398-d1407.

23.

Szklarczyk

Gable

Nastou

, et al. The STRING database in 2021: customizable protein-protein networks, and functional characterization of user-uploaded gene/measurement sets. Nucleic Acids Res. 2021;49:D605-D612.

24.

Liu

Xie

, et al. GSCA: an integrated platform for gene set cancer analysis at genomic, pharmacogenomic and immunogenomic levels. Brief Bioinform. 2023;24:bbac558.

25.

Tang

Zhang

Liu

GEPIA2021: integrating multiple deconvolution-based analysis into GEPIA. Nucleic Acids Res. 2021;49:W242-W246.

26.

Tang

Kang

, et al. GEPIA: a web server for cancer and normal gene expression profiling and interactive analyses. Nucleic Acids Res. 2017;45:W98-W102.

27.

Waman

Orengo

Kleywegt

Lesk

AM.

Three-dimensional structure databases of biological macromolecules. Methods Mol Biol. 2022;2449:43-91.

28.

Adasme

Linnemann

Bolz

, et al. PLIP 2021: expanding the scope of the protein-ligand interaction profiler to DNA and RNA. Nucleic Acids Res. 2021;49:W530-W534.

29.

Zhang

Zhu

Wang

, et al. Organ-specific cholesterol metabolic aberration fuels liver metastasis of colorectal cancer. Theranostics. 2021;11:6560-6572.

30.

Jin

Lan

Xie

, et al. Antitumor effects of FP3 in combination with capecitabine on PDTT xenograft models of primary colon carcinoma and related lymphatic and hepatic metastases. Cancer Biol Ther. 2012;13:737-744.

31.

Buckup

Rice

Hsu

, et al. Plectin is a regulator of prostate cancer growth and metastasis. Oncogene. 2021;40:663-676.

32.

Shi

, et al. Neuroprotective effect of Taohong Siwu decoction on cerebral ischemia/reperfusion injury via mitophagy-NLRP3 inflammasome pathway. Front Pharmacol. 2022;13:910217.

33.

Wang

Xia

Zhang

, et al. The multifaceted mechanisms of paeoniflorin in the treatment of tumors: State-of-the-Art. Biomed Pharmacother. 2022;149:112800.

34.

Gupta

Atanasov

Kumar

Bishayee

Chlorogenic acid for cancer prevention and therapy: current status on efficacy and mechanisms of action. Pharmacol Res. 2022;186:106505.

35.

Arneth

Tumor microenvironment. Medicina. 2019;56:15.

36.

Dasari

Lonardi

Garcia-Carbonero

, et al. Fruquintinib versus placebo in patients with refractory metastatic colorectal cancer (FRESCO-2): an international, multicentre, randomised, double-blind, phase 3 study. Lancet. 2023;402:41-53.

37.

Niedzwiecki

Roomi

Kalinovsky

Rath

Anticancer efficacy of polyphenols and their combinations. Nutrients. 2016;8:552.

38.

Ohno

Inoue

Ogihara

Cytotoxic activity of gallic acid against liver metastasis of mastocytoma cells P-815. Anticancer Res. 2001;21:3875-3880.

39.

Maruyama

Murata

Nakayama

, et al. Epigallocatechin-3-gallate suppresses liver metastasis of human colorectal cancer. Oncol Rep. 2014;31:625-633.

40.

di Leo

Battaglini

Berger

, et al. A catechin nanoformulation inhibits WM266 melanoma cell proliferation, migration and associated neo-angiogenesis. Eur J Pharm Biopharm. 2017;114:1-10.

41.

Zhao

Wang

, et al. Pull the plug: anti-angiogenesis potential of natural products in gastrointestinal cancer therapy. Phytother Res. 2022;36:3371-3393.

42.

Semenza

Wang

GL.

A nuclear factor induced by hypoxia via de novo protein synthesis binds to the human erythropoietin gene enhancer at a site required for transcriptional activation. Mol Cell Biol. 1992;12:5447-5454.

43.

Lee

Golinska

Griffiths

.HIF-1-independent mechanisms regulating metabolic adaptation in hypoxic cancer cells. Cells. 2021;10:2371.

44.

Kierans

Taylor

CT.

Regulation of glycolysis by the hypoxia-inducible factor (HIF): implications for cellular physiology. J Physiol. 2021;599:23-37.

45.

Manuelli

Pecorari

Filomeni

Zito

Regulation of redox signaling in HIF-1-dependent tumor angiogenesis. FEBS J. 2022;289:5413-5425.

46.

Huber

Camisaschi

Berzi

, et al. Cancer acidity: an ultimate frontier of tumor immune escape and a novel target of immunomodulation. Semin Cancer Biol. 2017;43:74-89.

47.

Damgaci

Ibrahim-Hashim

Enriquez-Navas

, et al. Hypoxia and acidosis: immune suppressors and therapeutic targets. Immunology. 2018;154:354-362.

48.

Zhu

Dong

, et al. Calmodulin 2 facilitates angiogenesis and metastasis of gastric cancer via STAT3/HIF-1A/VEGF-A mediated macrophage polarization. Front Oncol. 2021;11:727306.

49.

Shimomura

Hinoi

Kuroda

, et al. Overexpression of hypoxia inducible factor-1 alpha is an independent risk factor for recurrence after curative resection of colorectal liver metastases. Ann Surg Oncol. 2013;20:S527-S536.

50.

Chen

Rojanasakul

Rankin

Chen

YC.

Gallic acid, a phenolic compound, exerts anti-angiogenic effects via the PTEN/AKT/HIF-1α/VEGF signaling pathway in ovarian cancer cells. Oncol Rep. 2016;35:291-297.

51.

Jung

Kim

Lee

, et al. Caffeic acid and its synthetic derivative CADPE suppress tumor angiogenesis by blocking STAT3-mediated VEGF expression in human renal carcinoma cells. Carcinogenesis. 2007;28:1780-1787.

52.

Park

Lee

, et al. Vanillin suppresses cell motility by inhibiting STAT3-Mediated HIF-1α mRNA expression in malignant melanoma cells. Int J Mol Sci. 2017;18:532.

53.

Park

Hwang

Park

Lee

HJ.

Chlorogenic acid inhibits hypoxia-induced angiogenesis via down-regulation of the HIF-1α/AKT pathway. Cell Oncol. 2015;38:111-118.