Yi-qi-hua-yu-jie-du Decoction Inhibits Gastric Cancer Progression by Inducing Tumor-Associated Macrophage Polarization via Modulating PP2A Enzyme Activity to Mediate the PI3K/AKT/NF-κB Axis

Introduction

With an estimated incidence ranking fifth and mortality ranking fourth among all cancers worldwide,¹ gastric cancer (GC) remains a major global health challenge as documented by the International Agency for Research on Cancer. Although the application of novel therapeutics, including targeted therapies and immunotherapies, has enabled current comprehensive treatments to extend the survival of some GC patients beyond one year, recurrence and metastasis rates remain persistently high.² Concurrently, systemic adverse effects and sequelae induced by radiotherapy, chemotherapy, and surgical interventions severely compromise patients’ quality of life.³ Overcoming this therapeutic impasse necessitates the exploration of optimized treatment strategies for GC.

The tumor microenvironment (TME) functions as a complex biological niche where transformed cells establish dynamic reciprocal relationships with resident stromal elements, matrix constituents, and metabolic waste products, synergistically promoting oncogenic development, metastatic spread, and therapeutic resistance.⁴ Tumor-associated macrophages (TAMs) represent the predominant immune population within the TME, demonstrating remarkable adaptability in their functional states. Macrophages exist along a functional continuum, with classically activated (M1) subtypes demonstrating antitumor and proinflammatory activities, while alternatively activated (M2) counterparts display protumorigenic and immunosuppressive features. TAMs promote GC cell proliferation, invasion, and migration^5,6 and play pivotal regulatory roles in GC cell immunomodulation,^7,8 metabolism,^9,10 and drug resistance.¹¹ Of particular significance, the phenotypic transition from tumor-suppressive M1 to tumor-promoting M2 macrophages represents a defining characteristic of advancing malignancy.^12,13 Therefore, identifying effective strategies that can reprogram TAMs and reverse their pro‑tumor phenotype has become a crucial direction in GC treatment. Achieving this goal hinges on discovering molecular hubs capable of coordinately regulating multiple signaling pathways in macrophages.

Protein phosphatase 2A (PP2A), a serine/threonine-specific phosphatase, consists of structural (A), regulatory (B), and catalytic (C) subunits. This enzyme modulates key cellular functions including glycolytic metabolism, DNA synthesis, protein biosynthesis, and immunoregulation,^14–16 and is established as a major tumor suppressor.¹⁷ Although emerging evidence implicates PP2A in macrophage polarization dynamics,^18–20 its specific role in regulating TAM polarization within the GC context has yet to be investigated.

The unique advantages of Traditional Chinese Medicine (TCM) in achieving “holistic, balanced, multi-system, and multi-target” regulation of the TME are gaining widespread attention among researchers.^21,22 Its capacity to synergistically modulate multiple signaling pathways offers novel avenues for the comprehensive intervention of TAMs polarization. Yi-qi-hua-yu-jie-du decoction (YJD) was developed by the renowned TCM physician Professor Liu Shenlin through clinical optimization of the classical formula “Gui Shao Liu Jun Zi”—a time-honored prescription historically used to treat spleen-stomach deficiency syndromes presenting with gastrointestinal dysfunction, including GC-related symptoms. To enhance its therapeutic efficacy against GC, YJD incorporates supplementary herbs with blood-activating and stasis-resolving properties. Our prospective multicenter study involving 489 stage II/III GC patients (2009–2016) revealed YJD’s dual benefits: a significant decrease in postoperative recurrence/metastasis rates and measurable quality-of-life improvements.²³ Preliminary studies from our group indicate that YJD exerts anti-GC effects through mechanisms including reversing drug resistance and inducing ferroptosis.^24–26 However, whether YJD can inhibit GC progression by reprogramming TAMs polarization remains a critical unanswered question. Notably, as a master regulator of diverse cellular functions, PP2A exhibits characteristics that align well with the therapeutic philosophy of TCM formulations, which often exert effects through multi‑component, multi‑pathway synergism. This leads us to hypothesize whether PP2A could serve as a core molecular target through which YJD modulates TAMs and exerts its anti‑GC effects.

In this study, we reveal, for the first time, that YJD enhances PP2A enzyme activity, thereby mediating the PI3K/AKT/NF-κB axis to induce TAMs polarization towards the M1 phenotype. This cascade consequently suppresses proliferative and migratory capacities in GC cells, manifesting tumor-suppressive activity. We identified PP2A as the critical target for YJD-mediated TAMs polarization. These findings elucidate the scientific basis of YJD’s multi-target anti-GC effects from the perspective of the TME, providing novel perspectives and evidence for GC treatment.

Materials and Methods

Network Pharmacology Analysis

Screening of YJD Active Components and Prediction of Potential Targets

Based on our prior research,²⁵ 41 active components of YJD exhibiting favorable ADME (Absorption, Distribution, Metabolism, Excretion) properties were identified. This was achieved using HPLC-Q-TOF-MS/MS combined with screening via the TCMSP Platform (https://tcmsp-e.com/index.php).²⁷(Note: As YJD is a modified formula derived from “Gui Shao Liu Jun Zi”, it was previously designated as mGSLJZ in our studies). The SMILES structures of these active components were retrieved from the PubChem database (https://pubchem.ncbi.nlm.nih.gov)²⁸ and subsequently imported into the SwissTargetPrediction database (http://www.swisstargetprediction.ch/). Potential biological targets were predicted for each component, retaining their top 50 predicted targets for further analysis.

Screening of Macrophage Polarization and GC-Related Targets

The keywords “Gastric cancer” and “Macrophage polarization” were searched within the GeneCards database (https://www.genecards.org/). Potential therapeutic targets were identified by applying a median-based cutoff for relevance scores, with values meeting or exceeding this threshold designated as GC-associated and macrophage polarization-related targets, respectively. Common targets shared among the YJD active components, macrophage polarization-related targets, and GC-related targets were identified using bioinformatics tools (http://www.bioinformatics.com.cn/) and visualized using a Venn diagram.

Protein-Protein Interaction (PPI) Network

Common targets were analyzed using STRING 11.5 database (https://cn.string-db.org/) (Szklarczyk et al, 2023) under Homo sapiens designation. A confidence threshold ≥0.700 was set for protein interactions. The resultant protein interactome was downloaded and visualized in Cytoscape software (version 3.7.1). The CytoNCA plugin was then employed to calculate key network centrality measures: Degree Centrality (DC), Betweenness Centrality (BC), and Closeness Centrality (CC). Core targets were defined as nodes exceeding median values for all three centrality measures. The top 10 targets based on DC values were selected for subsequent analysis.

GO and KEGG Enrichment Analysis

The common targets were submitted to the DAVID database (https://david.ncifcrf.gov/) for Gene Ontology (GO) functional enrichment analysis (covering Biological Process (BP), Cellular Component (CC), and Molecular Function (MF)) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. Terms/pathways significantly enriched (P < 0.01) were considered statistically significant. Bioinformatics tools (http://www.bioinformatics.com.cn/) were used to generate visualizations of the top 20 enriched GO terms, the top 20 enriched KEGG pathways.

Construction of the “Active Component-Target-Pathway” Network

The 41 YJD active components, the core targets identified from the PPI network, and the top 20 enriched KEGG pathways (P < 0.01) were imported into Cytoscape 3.7.1 to construct an “Active Component-Target-Pathway” network.

Molecular Docking Validation

To further validate potential binding interactions between the core active components of YJD and the top 10 key targets as well as the PP2A catalytic subunit alpha isoform (PPP2CA), molecular docking was performed. The three-dimensional (3D) crystal structures of the target proteins were retrieved from the RCSB Protein Data Bank (PDB, https://www.rcsb.org).²⁹ Using PyMOL software (version 2.5.0), proteins were prepared. The structures of the key active components were converted to the mol2 format using Open Babel (version 2.4.0).³⁰ Molecular docking simulations were executed using AutoDock Vina (version 1.2.3).³¹ The resulting docking poses were visualized and analyzed using PyMOL 2.5.0.

Bioinformatics Analysis

The mRNA and protein expression information of TOP10 key targets and PPP2CA was obtained from the GEPIA2 platform (http://gepia.cancer-pku.cn/index.html)³² and the HPA database (http://www.proteinatlas.org). The differences in gene expression levels between GC tissues and normal gastric mucosa tissues were evaluated. Using the Correlation module in the single-gene analysis tool of GEPIA2, the correlations between key targets, PPP2CA, and macrophage polarization biomarkers in gastric cancer were analyzed.

Experimental Validation

Reagents and Antibodies

5-Fluorouracil (5-Fu) was sourced from Tianjin Jinyao Amino Acid Co., Ltd. (China). The PP2A inhibitor LB-100 (Purity: 99.91%; #S7537) was purchased from Selleck Inc. (China). Primary antibodies against iNOS (#AF0199), TNF-α (#AF7014), Arg-1 (#DF6657), IL-10 (#DF6894), Bax (#AF0120), Bcl-2 (#AF6139), NF-κB (#AF5006), and p-NF-κB (#AF2006) were sourced from Affinity Biosciences, Inc. (China). Primary antibodies against AKT (#10176-2-AP), p-AKT (#66444-1-Ig), E-cadherin (#60335-1-Ig), N-cadherin (#66219-1-Ig) and β-actin (#66009-1-Ig) were acquired from Proteintech Group, Inc (China). Primary antibodies against PP2A catalytic subunit (PP2Ac) (#T55564), PI3K(#T40115), and p-PI3K (#T40116) were purchased from Abmart, Inc (China). Secondary antibodies goat anti-mouse IgG (#ZB-2305) and goat anti-rabbit IgG (#ZB-2301) were sourced from ZSGB-BIO (China).

Preparation and Quality Control of YJD

The composition of YJD used in this study is detailed in Table S1. All herbal materials were procured from the institutional pharmacy at Nanjing University of Chinese Medicine Affiliated Hospital. The herbs were decocted following soaking, and the resulting decoction was lyophilized to produce a powder (extraction yield: 16%). The lyophilized powder was dissolved in complete cell culture medium and sterilized by filtration. The quality of the YJD preparation was confirmed using Liquid Chromatography-Mass Spectrometry (LC/MS) as described in our previous study.²⁴ The concentrations of the marker compounds in YJD are provided in Table S2. Detailed information regarding the standard reagents, the YJD preparation method, and the corresponding mass spectra are available in Method S1 and Figure S1, respectively.

Cell Culture Procedures

The human gastric cancer cell lines (MKN-45 and AGS) and the human monocytic cell line THP-1 were obtained from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China) or Sunncell Biotechnology Co., Ltd. (Wuhan, China). Cultures were maintained in RPMI 1640 (Gibco, USA) containing 10% fetal bovine serum (Newzerum, New Zealand) under standard conditions (37°C, 5% CO₂).

To establish a tumor cell-macrophage co-culture system, gastric cancer cells were seeded into the upper chamber of a 0.4-μm pore size Transwell insert (Corning, USA) at a density of 5 × 10⁵ cells per insert and allowed to adhere for 24 h. THP-1 cells were seeded into the bottom of a 6-well plate at 1 × 10⁶ cells per well and differentiated into M0 macrophages by treatment with 50 ng/mL phorbol 12-myristate 13-acetate (PMA; MedChemExpress, USA) for 24 h.³³ The Transwell insert containing gastric cancer cells was then placed directly above the well containing the M0 macrophages. Co-culture was performed in serum-free RPMI 1640 medium for 24 h. For specific experiments, the positioning of the cell types within the co-culture system was reversed. Conditioned medium containing secreted factors from the co-culture was collected for subsequent cell viability assay and wound healing assay.

Cell Viability Assay

Cell viability was assessed with the Cell Counting Kit-8 (CCK-8; GlpBio, USA). THP-1-derived macrophages were treated with various concentrations of YJD. After 24 h or 48 h, the supernatant was collected. About 10 μL CCK-8 was added per well, followed by 1 h incubation at 37°C (dark). Absorbance was read at 450 nm. Separately, gastric cancer cells were seeded into new 96-well plates. After adhesion, cells were treated with the conditioned medium collected from YJD-treated macrophages as described above. The procedure was repeated to obtain OD values for all wells.

Western Blotting

SDS-PAGE-separated proteins were electroblotted onto methanol-activated PVDF membranes (Millipore, USA). After blocking with 5% BSA (1.5 h), membranes were incubated overnight at 4°C with primary antibodies targeting: iNOS, Arg-1, TNF-α, IL-10, β-actin, PI3K, p-PI3K, AKT, p-AKT, NF-κB, p-NF-κB, PP2Ac, Bax, Bcl-2, E-cadherin, and N-cadherin. Following secondary antibody incubation (1:10,000, 1 h, RT), proteins were detected by enhanced chemiluminescence (Biosharp, China) and quantified with ImageJ software.

Flow Cytometry

For in vitro macrophage polarization: Cells were stained with FITC anti-human CD86 antibody and PE anti-human CD206 antibody (BioLegend, USA) for 30 minutes at room temperature. For in vivo macrophage polarization: Single-cell suspensions were stained with PerCP anti-mouse CD45, APC- anti-mouse/human CD11b, FITC anti-mouse F4/80, and PE anti-mouse CD206 antibodies (BioLegend, USA) for 30 minutes at room temperature.³⁴ Stained cells were analyzed using the FACSCelesta flow cytometer (BD Biosciences, USA). Data analysis for all flow cytometry experiments was performed using FlowJo software.

Assay of PP2A Activity

PP2A enzyme activity in cells and tissues was measured using a commercial Serine/Threonine Phosphatase Assay System (Promega, USA) based on a molybdate dye-based colorimetric method. Cell or tissue lysates were prepared, and endogenous free phosphate was removed from the supernatant using a desalting column prior to the assay. A phosphate standard curve was generated for quantification. Absorbance was measured at 630 nm after stopping the reaction.^35–37

Wound Healing Assay

MKN-45 cells (5 × 10⁵ cells/well) were seeded into 6-well plates. Upon reaching ~80% confluence, a straight scratch was created in the monolayer. The collected conditioned medium from co-cultures under different treatments was then added to the wells. Images were captured again at 0 h, 24 h and 48 h using an inverted phase-contrast microscope. The remaining wound area was quantified using ImageJ software.

ELISA Analysis

The concentrations of TNF-α and IL-10 in the cell culture supernatants were measured using Enzyme-Linked Immunosorbent Assay (ELISA) kits (Jiyinmei, China), according to the manufacturer’s instructions.

Animal Experiments

Male BALB/c nude mice (SPF grade, 6–8 weeks old, 18–22 g) were purchased from SPF Biotechnology Co., Ltd. (China). GC xenograft models were established by subcutaneous injection of MKN-45 cells (5 × 10⁶ cells/mice) into the right axilla. Seven days post-inoculation, when tumor volumes reached approximately 100 mm³, were randomized into four experimental cohorts (n = 6/group): Control group (normal saline, 0.2 mL/10 g, via gavage, every day),5-Fu group (5-Fu, 25 mg/kg, via intraperitoneal injection, alternate days), YJD-L group (YJD 11.7 g/kg, 0.2 mL/10 g, via gavage, every day), and YJD-H group (YJD 23.4 g/kg, 0.2 mL/10 g, via gavage, every day). The dosing of YJD was based on its standard clinical use in human patients. The clinical daily dose for adults is 180 g (per 70 kg body weight). Using a human-to-mouse equivalent dose conversion factor of 9.1 (based on body surface area normalization), the mouse equivalent dose was calculated as 23.4 g crude herb equivalents/kg, which was set as the high-dose group (YJD‑H). The low-dose group (YJD‑L) received half of this amount, ie., 11.7 g crude herb equivalents/kg. In this study, the extraction yield of YJD was 16%; therefore, the actual administered doses of the YJD water extract were 1.87 g/kg for YJD‑L and 3.74 g/kg for YJD‑H.

Mice received continuous treatment for 14 days. Tumor dimensions and animal body mass were recorded at 3-day intervals. At the end of the experiment, deep anesthesia was performed in all mice via isoflurane inhalation to ensure the absence of response to a painful stimulus. Euthanasia was then confirmed by cervical dislocation. All animal euthanasia procedures were carried out in accordance with the American Veterinary Medical Association (AVMA) Guidelines for the Euthanasia of Animals. Subsequently, a necropsy was performed on the euthanized specimens, and tissues from the spleen, liver, kidney, and tumor were collected for further evaluation.

Immunofluorescence (IF) Staining

Tumor tissues were fixed in 4% paraformaldehyde, dehydrated, paraffin-embedded, and sectioned. After deparaffinization and antigen retrieval, sections were blocked and incubated overnight (4°C) with anti-F4/80, CD86 and CD206 antibodies, followed by secondary antibodies and DAPI counterstaining. Imaging used a fluorescence microscope (Nikon, Japan), with co-localization analyzed in SlideViewer.

Hematoxylin and Eosin (H&E) Staining

Tissues were fixed in 4% paraformaldehyde, paraffin-embedded, and sectioned. Sections were deparaffinized, rehydrated, stained with hematoxylin and eosin, dehydrated, and mounted. Images were acquired using a virtual slide scanner (Olympus VS200, Japan).

Immunohistochemistry (IHC) Staining

Following paraffin embedding (per H&E protocol), sections were immunostained with Ki-67, F4-80, and CD206 antibodies (4°C overnight), incubated with HRP-secondaries (30 min, RT), developed with 3,3′-diaminobenzidine (DAB), and counterstained with hematoxylin. Whole-slide imaging used the virtual slide scanner (Olympus VS200, Japan).

Statistical Analysis

Experiments were performed in triplicate. Data are presented as mean ± standard deviation (SD). Statistical analysis was performed using GraphPad Prism (version 7.0). The sample size for animal studies (n = 6 per group) was determined based on preliminary experimental results and the effect size observed in similar animal models. The normality of all quantitative data in this study was assessed using the Shapiro–Wilk test. For comparisons among multiple groups that met the normality assumption, One-way ANOVA followed by Tukey’s post hoc test was employed. * P < 0.05, ** P < 0.01 and *** P < 0.001 were considered statistically significant.

Results

Network Pharmacology Analysis

Screening of Potential Targets for YJD’s Regulation of Macrophage Polarization Against GC

Building upon our prior research, we identified 41 active components within YJD possessing favorable ADME properties. Structural representations in SMILES format appear in Table S3. Importing these 41 components into the SwissTargetPrediction database and retaining the top 50 targets based on prediction probability yielded 616 potential targets. Searching the GeneCards database with the keywords identified 6865 GC-related targets and 5,241 macrophage polarization-related targets, respectively. Intersecting these target sets with the YJD-derived targets and removing duplicates resulted in 320 potential targets implicated in YJD’s regulation of macrophage polarization against GC (Figure 1A).

Figure 1 Key targets and pathways of Yi-qi-hua-yu-jie-du decoction (YJD) in regulating macrophage polarization against gastric cancer (GC). (A) Venn diagram of potential targets for YJD-regulated macrophage polarization in GC treatment (320 common targets). (B) Protein-protein interaction (PPI) network of YJD-regulated macrophage polarization against GC. (C–F) Bubble plots of the top 20 enriched biological processes (BPs, C), cellular components (CCs, D), molecular functions (MFs, E), and KEGG pathways (F). Bubble area scales with enriched gene count, while color intensity corresponds to enrichment significance.

PPI Network Construction

The 320 common targets were subjected to PPI network construction. The resulting interactome was imported into Cytoscape 3.7.1 for visualization and analysis (Figure 1B), where node color intensity corresponds to the DC value. Using the CytoNCA plugin, median values for DC, BC, and CC were calculated as 8.0, 111.57857, and 0.34822452, respectively. Nodes exceeding all three median values (DC ≥ 8, BC ≥ 111.57857, CC ≥ 0.34822452) were designated as core targets, yielding a set of 100 hub genes (Figure 1B). Based on DC ranking, the top decile of network hubs was selected: EGFR (DC = 78), AKT1 (DC = 71),STAT3 (DC = 71),HSP90AA1 (DC = 68),SRC (DC = 65),TNF (DC = 58),BCL2 (DC = 57),IL1B (DC = 55),HSP90AB1 (DC = 52),JUN (DC = 48). Notably, the PP2A catalytic subunit alpha isoform (PPP2CA) ranked 87th within the PPI network (DC = 14.0, BC = 2056.9915, CC = 0.4010554), indicating significant connectivity to other network targets and suggesting its potential role as a network hub. More importantly, given its established function as a master regulator of the PI3K/AKT and NF-κB signaling pathways,^38–43 along with the unique, activity-dependent nature of phosphatase function, we selected PPP2CA for further experimental validation.

GO and KEGG Enrichment Analysis

Significant enrichment associations (P < 0.01) were identified across 541 BPs, 84 CCs, 121 MFs, and 148 pathways. The top 20 enriched GO terms and KEGG pathways are visualized in Figure 1C–F. YJD’s effects primarily involved BPs such as signal transduction, protein phosphorylation (Figure 1C). These processes were associated with CCs including the cytosol, plasma membrane, cytoplasm, nucleus, and nucleoplasm (Figure 1D). Key MFs influenced by YJD included protein binding, ATP binding, identical protein binding, protein serine/threonine/tyrosine kinase activity, and enzyme binding (Figure 1E). The top 5 statistically significant pathways (P < 0.01) were: Pathways in cancer, PI3K-Akt signaling pathway, Lipid and atherosclerosis, Chemical carcinogenesis - receptor activation, and Proteoglycans in cancer (Figure 1F). Significantly, the NF-κB signaling pathway, a classical inflammatory pathway crucial for M1 macrophage polarization regulation,⁴⁴ was enriched with 15 genes (P < 0.01). The PI3K-Akt signaling pathway is ranked as the second most significant pathway. Previous studies indicate crosstalk between the PI3K/Akt and NF-κB pathways, with both being modulated by PP2A enzyme activity.³⁸ This finding further suggests that YJD may coordinately regulate macrophage polarization against GC by modulating PP2A activity and impacting the PI3K/AKT/NF-κB axis.

“Active Component-Target-Pathway” Network

Integrating the 41 YJD active components, the 100 core targets, and the top 20 enriched KEGG pathways (P < 0.01) into Cytoscape 3.7.1 generated an “Active Component-Target-Pathway” network comprising 160 nodes and 844 edges (Figure 2A). This network included 40 compounds (excluding perlolyrine), 100 targets, and 20 pathways. Active components demonstrated broad target engagement, with each component interacting with an average of 9.45 targets. Notably, 45% of components interacted with more than 9 core targets, and seven components exhibited particularly extensive connectivity (≥14 targets): Nobiletin (YJD39), Zedoalactone A (YJD19), trans-Caffeic acid (YJD4), Germacrone-4,5-epoxide (YJD17), Methylnissolin-3-O-glucoside (YJD28), Alizarin (YJD35), and Sinensetin (YJD34). Meanwhile, core targets showed an average interaction with 3.86 active components, with 18% of targets binding to over 5 components. Key highly connected targets included ESR2 (24 components), CYP19A1 (23 components), ESR1 (18 components), and EGFR (18 components). Regarding signaling pathways, each pathway involved an average of 23.35 core targets, with 75% of pathways engaging more than 20 targets. As illustrated in Figure 2A, the network demonstrates the characteristic TCM principle: a single component can influence multiple targets and pathways, while different components can converge on the same targets and pathways.

Figure 2 Active component-target-pathway network and molecular docking analysis of YJD. (A) Integrated “Active Component-Target-Pathway” network. Blue polygons: active components; purple arrows: pathways; Orange circles: core targets. Node connectivity reflects functional importance. (B) Heatmap of molecular docking scores between YJD active components and targets. Color intensity corresponds to binding affinity strength (absolute binding energy). (C) Representative 3D binding conformations of active components with target proteins.

Molecular Docking Validation

Molecular docking assesses binding affinity, where lower binding energy scores indicate stronger binding and greater conformational stability. We performed docking between the top 10 key targets (from 3.1.2), PPP2CA, and the 7 highly connected active components identified in the network (3.1.4). As shown in the heatmap (Figure 2B), all calculated binding energies were below −5 kcal/mol, indicating robust binding potential. The optimal binding conformations between representative targets and active components are visualized in Figure 2C.

Bioinformatics Analysis

As shown in Figure 3A, compared to normal gastric mucosa tissues, the mRNA expression levels of HSP90AA1 and PPP2CA were significantly upregulated in GC tissues, with statistically significant differences (P < 0.05). In GC tissues, the mRNA expression of EGFR, AKT1, STAT3, SRC, TNF, IL1B, and HSP90AB1 increased, while the mRNA expression of BCL2 and JUN decreased; however, these differences were not statistically significant (P > 0.05). This suggests that HSP90AA1 and PPP2CA may be associated with the development and progression of GC. The results (Figure 3B) revealed that AKT1, STAT3, HSP90AA1, SRC, TNF, BCL2, IL1B, and PPP2CA were significantly positively correlated with CD86, while JUN was significantly negatively correlated with CD86. Additionally, AKT1, STAT3, BCL2, and JUN were significantly positively correlated with CD163, whereas SRC and PPP2CA were significantly negatively correlated with CD163. All these correlations were statistically significant (P < 0.05) (Figure 3B).

Figure 3 Differential mRNA expression of targets in GC versus normal mucosa and their correlation with macrophage markers. (A) Differential mRNA levels of key targets and PPP2CA between GC and normal gastric mucosa tissues. Red boxes: tumor tissues; blue boxes: normal mucosa. Red asterisks indicate statistically significant differences compared to normal mucosa (*P < 0.05). (B) Correlation analysis between key targets/PPP2CA and macrophage markers.

Experimental Validation

In vitro TAMs Model Exhibits Pro-Tumor M2 Phenotype Associated with PP2A Enzyme Activity

Morphological assessment revealed untreated THP-1 cells displayed a round, translucent, suspension-growth morphology. Following stimulation with PMA, cells differentiated into M0 macrophages, exhibiting a characteristic adherent, spindle-shaped morphology with grey, opaque appearance. Subsequent co-culture with MKN-45 cells induced differentiation into TAMs, identifiable by an adherent, spindle-shaped morphology with elongated pseudopods and clustered or colony-like growth patterns (Figure 4A).

Figure 4 In vitro tumor-associated macrophage (TAM) model exhibits pro-tumor M2 phenotype associated with protein phosphatase 2A (PP2A) enzyme activity. (A) Morphology of THP-1 cells, M0 macrophages, and TAMs under phase-contrast microscopy. Scale bar: 200 μm. (B and C) Western blotting analysis of iNOS, Arg-1, TNF-α, and IL-10 protein expression. Quantified relative protein levels with statistical analysis. (D–F) Flow cytometry analysis of M1/M2 macrophage proportions. (G) PP2A enzyme activity across groups. Data are presented as means ± SD (n = 3 independent experiments). *P < 0.05, **P < 0.01, ***P < 0.001.

Molecular characterization confirmed the phenotypic shift. Compared to undifferentiated macrophages (Control group), TAMs (Co-culture group) exhibited significantly upregulated expression of Arg-1 and IL-10 (M2-associated proteins), alongside significantly downregulated iNOS and TNF-α (M1-associated proteins) (Figure 4B and C). A significantly decreased proportion of CD86⁺ cells and a significantly increased proportion of CD206⁺ cells in the Co-culture group versus Control (Figure 4D–F), confirming successful TAMs model establishment. Notably, while PP2Ac protein expression showed no significant difference between groups (Figure 4B and C), PP2A enzyme activity was significantly reduced in the Co-culture group compared to Control (Figure 4G), suggesting PP2A activity modulation is involved in TAM polarization within the GC microenvironment.

YJD Inhibits GC Progression in vitro by Modulating TAMs

CCK-8 assays assessed YJD’s impact on macrophage viability and its indirect effect on MKN-45 cell proliferation via macrophage-secreted factors. Low concentrations of YJD (0–2 mg/mL) showed no significant cytotoxicity towards macrophages themselves (Figure 5A). However, treatment of MKN-45 cells with conditioned medium from YJD-exposed macrophages resulted in a dose- and time-dependent inhibition of MKN-45 proliferation (Figure 5B). These results indicate low-dose YJD exerts anti-tumor effects against GC cells mediated by macrophages without direct macrophage suppression. Consequently, subsequent experiments utilized 1 mg/mL (YJD-L, low dose) and 2 mg/mL (YJD-H, high dose) YJD.

Figure 5 YJD inhibits GC progression in vitro by modulating TAMs. (A) Viability of macrophages treated with YJD (24/48 h). (B) Viability of MKN-45 cells exposed to conditioned medium from YJD-treated macrophages (24/48 h). (C and D) Wound healing assay of MKN-45 cell migration. Scale bar: 500 μm. (E and F) Western blotting analysis of E-cadherin, N-cadherin, Bax, and Bcl-2 in MKN-45 cells. Data are presented as means ± SD (n = 3 independent experiments). *P < 0.05, **P < 0.01, ***P < 0.001.

YJD significantly inhibited the migratory capacity of MKN-45 cells in co-culture compared to the Co-culture group alone, with YJD-H exhibiting a more pronounced effect (Figure 5C and D). Compared to Co-culture, the YJD-H group exhibited significantly upregulated pro-apoptotic protein Bax and epithelial marker E-cadherin, alongside significantly downregulated anti-apoptotic protein Bcl-2 and mesenchymal marker N-cadherin (Figure 5E and F). Additionally, to evaluate the generalizability and translational relevance of our findings, we conducted validation experiments in the AGS cell line. YJD demonstrated a consistent trend with that observed in the MKN-45 model, including the inhibition of gastric cancer cell proliferation, induction of apoptosis-related protein alterations, and promotion of macrophage polarization toward the M1 phenotype under co-culture conditions (Figure S2).

These findings demonstrate that YJD effectively reverses EMT and promotes apoptosis in MKN-45 GC cells in vitro through TAM modulation, thereby inhibiting GC cell proliferation and migration.

YJD Enhances PP2A Activity, Suppresses PI3K/AKT Axis, Modulates NF-κB Signaling, and Reprograms TAM Polarization

Building upon network pharmacology projections, we hypothesize that YJD exerts its anti-GC effects by coordinately regulating macrophage polarization through modulation of PP2A enzyme activity and the PI3K/AKT/NF-κB axis. While PP2Ac protein levels remained unchanged across groups (Figure 6C and D), PP2A enzyme activity was significantly elevated in YJD-treated groups compared to Co-culture, particularly in the YJD-H group (Figure 6H). Analysis of downstream signaling showed YJD-H significantly reduced the p-PI3K/PI3K and p-AKT/AKT ratios, while significantly increasing the p-NF-κB/NF-κB ratio (Figure 6C and D). YJD treatment, especially YJD-H, significantly upregulated iNOS and TNF-α (Figure 6A and B) and increased the proportion of CD86⁺ cells (Figure 6E and F) compared to Co-culture. Conversely, YJD-H significantly downregulated Arg-1 and IL-10 (Figure 6A and B) and decreased the proportion of CD206⁺ cells (Figure 6E and G). These results indicate YJD reprograms TAMs towards the M1 phenotype and away from the M2 phenotype, potentially through activating PP2A enzyme activity and modulating the PI3K/AKT/NF-κB axis. ELISA analysis showed that, compared with the Co-culture group, the level of the M1‑associated inflammatory factor TNF‑α was significantly increased, while that of the M2‑associated inflammatory factor IL‑10 was significantly decreased in the YJD‑H group (Figure S3).

Figure 6 Continued.

Figure 6 YJD enhances PP2A activity, suppresses PI3K/AKT signaling, modulates NF-κB, and reprograms TAM polarization. (A and B) Western blotting of iNOS, Arg-1, TNF-α, and IL-10 in macrophages. (C and D) Western blotting of p-PI3K/PI3K, p-AKT/AKT, p-NF-κB/NF-κB, and PP2Ac in macrophages. (E–G) Flow cytometry of M1 (CD86⁺CD206−) and M2 (CD86−CD206⁺) macrophage proportions. (H) PP2A enzyme activity. Data are presented as means ± SD (n = 3 independent experiments). *P < 0.05, **P < 0.01, ***P < 0.001.

YJD Suppresses Gastric Tumor Growth in vivo by Regulating TAM Polarization

To validate YJD’s in vivo efficacy, a GC xenograft model was established with four treatment groups (Figure 7A). YJD doses were set at the human equivalent dose (23.4 g/kg, YJD-H) and half-equivalent dose (11.7 g/kg, YJD-L). Both YJD-L and YJD-H, alongside the positive control 5-Fu, significantly reduced tumor weight and volume compared to the Control group (Figure 7B, C and E). YJD treatment also effectively mitigated tumor-induced body weight loss (Figure 7D). Compared with the control group, tumor cells in both 5-Fu and YJD treatment groups exhibited vacuolar degeneration, eosinophilic cytoplasm, indistinct cell membranes, and evident nuclear pyknosis. Notably, more extensive necrotic areas were observed in the YJD-H group (Figure 8H). IHC confirmed reduced Ki-67 expression in 5-Fu and YJD groups, with YJD-H showing greater improvement (Figure 8H). Histological examination of spleen, liver, and kidney tissues revealed no apparent pathological alterations (Figure 8I), demonstrating good in vivo safety.

Figure 7 YJD suppresses tumor growth in GC xenograft models. (A) In vivo experimental scheme. Created in BioRender. Peipei, Y. (2026) https://BioRender.com/nx5f5yc. (B) Representative tumor images from mice treated with YJD (11.7/23.4 g/kg), 5-Fu (25 mg/kg), or control (n = 6). (C) Tumor volumes. (D) Body weight changes during treatment. (E) Tumor weights. Data are presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 versus the control group.

Figure 8 Continued.

Figure 8 YJD inhibits GC growth in vivo by regulating TAM polarization. (A and B) Western blotting of E-cadherin, N-cadherin, Bax, Bcl-2, and PP2Ac.(C) PP2A enzyme activity in tumors. (D and E) Flow cytometry of M2 (F4/80⁺CD206⁺) TAM proportion in tumors. (F) Immunofluorescence of DAPI, F4/80, and CD86 in tumor sections. Scale bar: 50 μm. (G) Immunofluorescence of DAPI, F4/80, and CD206. Scale bar: 50 μm. (H) Hematoxylin and eosin (H&E) and immunohistochemical (IHC) staining of Ki-67, F4/80, CD206 in tumors. Scale bar: 50 μm. (I) H&E staining of liver, spleen, kidney. Scale bar: 100 μm. Data are presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Western blotting showed YJD-H significantly upregulated Bax and E-cadherin while downregulating Bcl-2 and N-cadherin compared to Control (Figure 8A and B), confirming in vivo inhibition of EMT and promotion of apoptosis. Mechanistically, while PP2Ac levels remained unchanged, PP2A enzyme activity was significantly higher in YJD-L and YJD-H groups versus Control (Figure 8A–C). A significant decrease in the proportion of M2-type TAMs in YJD-treated groups (Figure 8D and E). Immunofluorescence showed increased F4/80⁺CD86⁺ cells and decreased F4/80⁺CD206⁺ cells in YJD groups, especially YJD-H, compared to Control (Figure 8F and G). IHC further confirmed YJD suppressed F4/80 and CD206 expression in tumor tissues (Figure 8H). These in vivo results validate that YJD inhibits GC progression by activating PP2A enzyme activity and reducing M2 TAMs within the TME.

Rescue Experiments Confirm PP2A Activation is Crucial for YJD-Mediated TAM Reprogramming

To confirm the central role of PP2A activation, we inhibited PP2A activity using LB-100.^45–49 Compared to YJD-H, the YJD-H+LB-100 group exhibited significantly increased p-PI3K/PI3K and p-AKT/AKT ratios and significantly decreased p-NF-κB/NF-κB ratio, despite unchanged PP2Ac levels (Figure 9C and D). PP2A enzyme activity was significantly reduced in YJD-H+LB-100 group (Figure 9H).

Figure 9 Continued.

Figure 9 Rescue experiments confirm PP2A activation is essential for YJD-mediated TAM reprogramming. (A and B) Western blotting analysis of iNOS, Arg-1, TNF-α, IL-10 in macrophages. (C and D) Western blotting analysis of p-PI3K/PI3K, p-AKT/AKT, p-NF-κB/NF-κB ratios and PP2Ac. (E–G) Flow cytometry of M1 (CD86⁺CD206−) and M2 (CD86−CD206⁺) macrophage proportions. (H) PP2A enzyme activity. Data are presented as means ± SD (n = 3 independent experiments). *P < 0.05, **P < 0.01, ***P < 0.001.

Critically, LB-100 treatment reversed YJD’s reprogramming effect on TAMs: compared to YJD-H, the YJD-H+LB-100 group showed significantly downregulated iNOS and TNF-α, significantly upregulated Arg-1 and IL-10 (Figure 9A and B), a decreased proportion of CD86⁺ cells, and an increased proportion of CD206⁺ cells (Figure 9E–G). These rescue experiments definitively establish that PP2A enzyme activation is the key functional mechanism by which YJD reprograms TAMs.

Discussion

Within the immunosuppressive TME, TCM can reduce M2-type TAMs in tumor tissues, inhibit TAM recruitment and activation of tumor cells, reprogram macrophage polarization, and enhance immune responses, exerting anti-tumor effects. This tumor-suppressing mechanism aligns with the TCM concept of “consolidating healthy qi and eliminating pathogenic factors.” YJD, originating from the historically documented “Gui Shao Liu Jun Zi”, has been previously confirmed by our group to effectively inhibit GC growth, improve patient quality of life,²³ and suppress drug resistance.^24–26 Building on this foundation, the present study provides a novel mechanistic understanding by systematically elucidating, for the first time, how YJD reprograms the immunosuppressive TME. We identify the activation of PP2A enzyme activity as a pivotal and previously unrecognized mechanism through which YJD inhibits GC progression.

To predict the molecular mechanism of YJD’s anti-GC action via TAM modulation, we employed network pharmacology for precise mechanistic guidance. Network pharmacology, based on systems biology theory, utilizes complex biological network models to predict drug mechanisms, reflecting TCM’s holistic perspective through its focus on the integrity, systemic nature, chemical components, targets, and disease interactions of herbal formulas. Its development offers novel avenues for elucidating the multi-target mechanisms of TCM.^50,51 Integrating our prior research, we identified 41 YJD active components. Network pharmacology predicted potential targets, and PPI network analysis identified ten hub genes: EGFR, AKT1, STAT3, HSP90AA1, SRC, TNF, BCL2, IL1B, HSP90AB1, and JUN. Among these, TNF and IL1B are closely associated with macrophages, BCL2 is an apoptosis regulator, and PPP2CA exhibited significant network connectivity. Notably, the prediction highlighted PP2A (PPP2CA) as a node connecting to several key pathways, which informed our subsequent hypothesis. TNF is encoded by a protein-coding gene primarily secreted by macrophages, binds its receptors to regulate inflammation, immunity, apoptosis, and impacts GC proliferation, metastasis, invasion, and EMT.^52–55 IL1B encodes a cytokine family member produced by activated macrophages as a pro-protein, controlling cellular functions including proliferation, differentiation, survival, and apoptosis.⁵⁶ Meta-analyses confirm IL1B polymorphisms are significantly associated with GC risk.^57–59 BCL2 protein family members are key regulators of apoptosis, making them attractive therapeutic targets for cancer drug development.⁵⁸ PPP2CA encodes the PP2A catalytic α subunit isoform, crucial for regulating PP2A activity; its genetic variants have been linked to GC susceptibility in Chinese populations.⁵⁹

GO and KEGG enrichment analyses indicated that YJD likely regulates TAM polarization against GC through pathways such as “Pathways in cancer” and the “PI3K-Akt signaling pathway.” Given the critical role of the NF-κB signaling pathway in tumor biology and M1 macrophage polarization, and the known crosstalk between PI3K/Akt and NF-κB pathways, both modulated by PP2A enzyme activity,³⁸ we formulated a focused hypothesis centered on PP2A as a regulatory nexus.

Based on the above analysis, we hypothesized that YJD may exert its anti-gastric cancer effects by modulating PP2A activity, thereby coordinately regulating the PI3K/AKT/NF-κB axis and influencing macrophage polarization. To validate this hypothesis, we selected PPP2CA as a key experimental target. Subsequent experiments confirmed that YJD indeed significantly enhanced PP2A enzyme activity without changing its protein expression level, supporting the proposed mode of action.

This discovery underscores a key unique aspect of our study: YJD’s action depends on modulating the functional activity of PP2A, rather than altering its expression level. This represents a distinct form of molecular intervention that aligns with and provides a fresh mechanistic explanation for the multi-target, system-regulating paradigm of TCM.

What are the key active components mediating YJD’s anti-GC effects via TAMs? Construction of the “active component-target-pathway” network identified the core pharmacodynamic substances: Nobiletin, Zedoalactone A, trans-Caffeic acid, Germacrone-4,5-epoxide, Methylnissolin-3-O-glucoside, Alizarin, and Sinensetin. Strong binding affinity between these 7 key YJD components and the 11 hub targets. The Nobiletin-TNF pair exhibited the strongest binding energy (−8.5 kcal/mol). Among these components, all except Zedoalactone A and Methylnissolin-3-O-glucoside have documented anti-tumor activities. Nobiletin, a ubiquitous flavonoid from citrus plants with antioxidant, anti-inflammatory, and hepatoprotective properties, significantly inhibits GC by inducing autophagy-dependent cell death⁶⁰ and apoptosis with cell cycle arrest,⁶¹ making it a promising anti-cancer candidate. Trans-Caffeic acid (distinct from its cis isomer) is a natural phenolic compound whose potent anti-inflammatory, immunomodulatory, and antioxidant effects are significant in GC treatment, primarily by affecting Ca2+ homeostasis and apoptosis,⁶² and reducing proteasome function to enhance sensitivity to doxorubicin and cisplatin.⁶³ Germacrone-4,5-epoxide, a natural compound from turmeric, exerts anti-GC effects by inducing cell cycle arrest and apoptosis,⁶⁴ and promoting autophagosome formation.⁶⁵ Alizarin, derived from Rubia cordifolia, inhibits GC cells at low concentrations.⁶⁶ Sinensetin, from citrus with strong anticancer, antioxidant, and vasorelaxant activities.⁶⁷ The computational predictions outlined above provided a focused direction for subsequent experimental inquiry. The decisive evidence for the proposed mechanism, however, derives from the functional assays and rescue experiments presented in the following sections, which directly test the predicted involvement of PP2A and the PI3K/AKT/NF‑κB axis in macrophage reprogramming.

The high plasticity of TAMs within the TME makes them ideal therapeutic targets. This study established an in vitro tumor cell-macrophage co-culture system, assessing polarization direction via TAM marker expression. Polarization states are widely identified by markers including transmembrane glycoproteins, enzymes, growth factors, hormones, cytokines, and their receptors.⁶⁸ Known M1 markers include iNOS, CD86, CD80, TNF-α, and IFN-γ; M2 markers include Arg-1, IL-10, CD163, and CD206.⁶⁹ Morphological observation, WB, and flow cytometry confirmed that TAMs, compared to undifferentiated macrophages, exhibited altered morphology and a bias towards the M2 phenotype, successfully establishing the model. Previous studies also confirm that the TME favors M2 polarization, making M2 TAMs the predominant subtype. Crucially, this model revealed significantly reduced PP2A enzyme activity, suggesting PP2A activity modulation is involved in TAM polarization within the GC TME.

We demonstrated that YJD exerts anti-GC effects by modulating TAM phenotype: it significantly induces M1 polarization while inhibiting GC cell proliferation and migration, and reversing EMT. This reveals the modern biological connotation of YJD’s therapeutic principle of “consolidating healthy qi and eliminating pathogenic factors” – reprogramming TAMs to improve the tumor immune microenvironment, achieving the synergistic effect of “inhibiting the pro-tumor microenvironment (eliminating pathogenic factors) and enhancing anti-tumor immunity (consolidating healthy qi).” Mechanistic studies revealed that YJD’s reprogramming effect on TAMs critically depends on the specific modulation of PP2A enzyme activity. PP2A is recognized as a significant tumor suppressor, with dysregulated expression in various cancers.⁷⁰ Recent reports also link it closely to GC development and progression, suggesting it influences macrophage activation, maintains M1-M2 conversion, and impacts disease progression.^71,72 However, research on PP2A’s role in macrophage polarization, particularly GC TAMs, is limited. Our study found that while PP2Ac protein expression remained unchanged across groups, YJD significantly enhanced PP2A enzyme activity. This enhancement was closely linked to the dynamic balance of the PI3K/AKT/NF-κB axis: YJD, by enhancing PP2A activity, significantly suppressed PI3K/AKT phosphorylation while activating the NF-κB pathway. Although previous studies noted the importance of the PI3K/AKT/NF-κB axis in TAM polarization,^73,74 the key mediating role of PP2A was undefined. We are the first to confirm that YJD, via the PP2A signaling hub, coordinately modulates the PI3K/AKT/NF-κB axis. This mechanism, acting through enzyme activity modulation rather than protein expression, differs from traditional PP2A regulation approaches and provides a new molecular explanation for the “multi-target” action of TCM formulas. This work bridges tumor immunology with TCM pharmacology by pinpointing a specific enzymatic activity (PP2A) as a central target for immune microenvironment reprogramming, offering a novel molecular perspective on TCM-mediated GC therapy.

Using an in vivo GC xenograft model, we further validated that YJD inhibited tumor growth. Safety assessment showed YJD mitigated tumor-induced weight loss without major organ toxicity. Finally, rescue experiments using the specific PP2A inhibitor LB-100 definitively confirmed the central role of PP2A enzyme activity.

This study bridges tumor immunology and TCM, offering a novel micro-perspective on the mechanism of TCM pattern differentiation and treatment for GC. Although network pharmacology and molecular docking offered a systems‑view of YJD’s potential multi‑target activity, the decisive evidence stems from experimental validation in both cellular and animal models, which collectively support the conclusion that YJD suppresses gastric cancer progression by polarizing tumor‑associated macrophages via PP2A activation. Beyond elucidating the PP2A‑dependent mechanism, our findings also highlight translational implications. The comparable tumor suppression but reduced body weight loss of YJD relative to 5‑FU, along with its ability to reprogram the immunosuppressive TME, positions it as a promising candidate for further development—potentially as an adjunct therapy to enhance efficacy or reduce adverse effects of conventional treatments. Future clinical translation would benefit from standardizing YJD preparation, validating PP2A activity as a pharmacodynamic biomarker, and evaluating its synergy with chemotherapy or immunotherapy. Several limitations of this work should be acknowledged. Firstly, the in vitro co-culture model cannot fully replicate the complex cellular composition and dynamic interactions of the in vivo tumor microenvironment (TME). Secondly, the lack of a complete adaptive immune system in nude mice may limit the assessment of YJD’s full immunomodulatory profile. Additionally, the subcutaneous xenograft model does not recapitulate gastric tissue-specific biology; orthotopic or syngeneic immunocompetent models would provide more relevant insights into gastric TME and systemic immune responses. Consequently, while our study focused on and demonstrated YJD’s direct reprogramming effect on TAMs, its potential to elicit or coordinate a broader adaptive anti-tumor immune response remains an open question. Furthermore, while our study establishes the necessity of PP2A activation for the observed effects, several aspects of the underlying molecular mechanism warrant further precision. First, direct biochemical evidence of physical binding between YJD constituents and PP2A is pending, and we have not tested whether individual active components can independently reproduce the observed effects. Second, the detailed causal link between PP2A activation and the observed increase in NF-κB phosphorylation remains to be fully dissected, including the temporal sequence of these events. Third, the quantitative relationship between PP2A activity levels and phenotypic outcomes has not been mapped, as we used a single concentration of LB-100 rather than an activity gradient. Fourth, this study relied solely on the PP2A inhibitor LB-100 for functional rescue experiments and did not employ genetic perturbations or a second structurally distinct PP2A inhibitor to rule out potential off-target effects of LB-100.

To address these limitations and deepen the mechanistic understanding, future investigations should focus on several integrated fronts: (1) employing immunocompetent or humanized models with comprehensive immune profiling—such as multi‑color flow cytometry analysis of T cells, dendritic cells, and other immune populations— to validate findings in a physiologically relevant context; (2) further dissecting the causal role of PP2A via genetic perturbations (eg., PPP2CA siRNA/shRNA knockdown or CRISPR editing), a second structurally distinct PP2A inhibitor (eg., okadaic acid), and phosphoproteomic analysis of its substrates; (3) precisely mapping the downstream signaling by linking PP2A activation to IκBα/NF‑κB dynamics through time‑course experiments with shorter intervals (eg., 15, 30, 60, 120 minutes) and establishing a dose‑dependent quantitative PP2A activity–response relationship using a gradient of LB‑100 concentrations (eg., 0.5, 1, 2, 5 μmol/L); (4) deconvoluting the active constituents of YJD by systematically evaluating the seven prioritized compounds (Nobiletin, Zedoalactone A, trans‑Caffeic acid, Germacrone-4,5-epoxide, Methylnissolin‑3‑O‑glucoside, Alizarin, and Sinensetin) for their individual TAM‑polarizing effects and potential synergistic interactions, and validating direct target engagement through biophysical assays such as surface plasmon resonance (SPR) or isothermal titration calorimetry (ITC);and (5) exploring translational potential via synergy studies with immune checkpoint inhibitors. These concerted efforts will solidify the mechanistic framework and advance the clinical translation of YJD.

Conclusion

Our study demonstrated the anti-GC effects of YJD through TAM modulation both in vitro and in vivo. We elucidated the molecular mechanism by which YJD inhibits GC progression by regulating TAM polarization via modulation of PP2A enzyme activity and the PI3K/AKT/NF-κB axis. The favorable safety profile and significant efficacy demonstrated by YJD support further clinical evaluation in GC treatment. The specific regulatory network of this mechanism is illustrated in Figure 10. Beyond elucidating novel anti-tumor mechanisms of YJD, this work provides crucial translational foundations for implementing this TCM formulation in clinical GC management.

Figure 10 Molecular mechanism diagram. YJD inhibits gastric cancer by regulating TAM polarization through PP2A-mediated modulation of the PI3K/AKT/NF-κB axis. Created in BioRender. Peipei, Y. (2026) https://BioRender.com/wlodj64.