The Role of IGF2BP3‐Mediated Immune Response Activation in Enhancing Immunotherapy of Koumine for Gastric Cancer

ABSTRACT

Resistance to immune checkpoint blockade in gastric cancer often arises from defective antigen presentation due to MHC‐I downregulation, yet strategies to restore MHC‐I expression remain limited. Here, using the natural alkaloid koumine as a chemical probe, we identified IGF2BP3 as a druggable regulator of MHC‐I antigen presentation. Network pharmacology, proteomics, surface plasmon resonance, and cellular thermal shift assays were used to identify and validate koumine's target. Genetic rescue experiments, loss‐of‐function studies, in vitro T‐cell co‐culture assays, and in vivo gastric cancer mouse models were performed to confirm the causal relationship. m6A‐seq and ChIP‐seq analyses delineated the downstream epigenetic pathway. Koumine binds IGF2BP3 and promotes its proteasome‐dependent degradation, which reduces m6A enrichment on EED mRNA, accelerates EED decay, lowers H3K27me3, and derepresses IRF transcription, thereby restoring MHC‐I expression and enhancing CD8⁺ T‐cell killing. Crucially, koumine enhanced the efficacy of anti‐PD‐1 therapy in gastric cancer mouse models, suppressing tumor growth and fostering an immunologically active tumor microenvironment characterized by CD8⁺ T‐cell infiltration and activation. This study identifies IGF2BP3 as an actionable target for improving antigen presentation and establishes koumine as a valuable chemical probe to dissect this pathway, with implications for immunotherapy sensitization beyond gastric cancer.

Koumine binds IGF2BP3 and promotes its proteasomal degradation, reducing m6A on EED mRNA, accelerating EED decay, lowering H3K27me3, and promoting IRF transcription, thereby restoring MHC‐I and enhancing CD8⁺ T cell killing.

Abbreviations

Act D

actinomycin D

BCA

bicinchoninic acid assay

BP

biological process

CETSA

cellular thermal shift assay

CHX

cycloheximide

DEGs

differentially expressed genes

DEPs

differentially expressed proteins

DMSO

dimethyl sulfoxide

EED

embryonic ectoderm development

EZH2

enhancer of zeste homolog 2

FBS

fetal bovine serum

FDR

false discovery rate

GAPDH

glyceraldehyde‐3‐phosphate dehydrogenase

GC

gastric cancer

GEO

Gene Expression Omnibus

GO

Gene Ontology

GSVA

Gene Set Variation Analysis

GZMB

granzyme B

H3K27me3

trimethylation of histone H3 at lysine 27

HE

hematoxylin and eosin

HRP

horseradish peroxidase

ICIs

immune checkpoint inhibitors

IFN‐γ

interferon gamma

IGF2BP3

insulin‐like growth factor 2 mRNA‐binding protein 3

IHC

immunohistochemistry

IRFs

interferon regulatory factors

KEGG

Kyoto Encyclopedia of Genes and Genome

KM

koumine

LDH

lactate dehydrogenase

MDA

malondialdehyde

MFI

mean fluorescence intensity

MHC‐I

major histocompatibility complex‐I

MOI

multiplicity of infection

OE

overexpression

OT‐I

ovalbumin‐specific T cell receptor transgenic T cells

PD‐1

programmed cell death protein 1

PI

propidium iodide

PPI

protein–protein interaction

PRC2

Polycomb repressive complex 2

RIPA

radioimmunoprecipitation assay buffer

SDS‐PAGE

sodium dodecyl sulfate‐polyacrylamide gel electrophoresis

sgRNA

single‐guide RNA

SPR

surface plasmon resonance

STAD

stomach adenocarcinoma

TCGA

the cancer genome atlas

TCM

traditional Chinese medicine

TES

transcription end site

TPM

transcripts per million

UMAP

uniform manifold approximation and projection

1. Introduction

Gastric cancer (GC) remains a leading cause of cancer‐related mortality worldwide [1, 2, 3], and conventional therapies show limited efficacy in advanced stages [3, 4, 5]. Although immunotherapy has revolutionized treatment for advanced GC, its clinical benefit is often constrained by inadequate tumor antigen presentation, which leads to insufficient T‐cell activation [2, 6, 7, 8, 9]. A key mechanism by which GC evades immune surveillance is the downregulation of major histocompatibility complex Class I (MHC‐I), which impairs tumor antigen presentation and subsequent CD8⁺ T‐cell‐mediated killing [10, 11, 12]. Consequently, restoring MHC‐I expression on GC cells is a promising strategy to enhance immunotherapy efficacy [13, 14] However, pharmacological agents capable of effectively and specifically restoring MHC‐I in tumors remain limited, and the regulatory networks governing MHC‐I expression are incompletely understood, representing a critical barrier to overcoming immunotherapy resistance.

One source of potential immunomodulatory agents is traditional chinese medicine (TCM), which has demonstrated significant synergistic and efficacy‐enhancing effects in comprehensive cancer therapy [15, 16], and has served as a chemical probe for dissecting intricate biological signaling pathways and identifying new druggable molecular targets. Koumine (KM, C₂₀H₂₂N₂O, molecular weight 306.1804) is the most abundant alkaloid monomer in the extracts of the plant Gelsemium elegans Benth [17]. Previous studies have confirmed that KM possesses diverse pharmacological activities, including antitumor, immunomodulatory, and analgesic effects [18, 19, 20]. Regarding its antitumor properties, KM has been shown to induce apoptosis in various cancer cell lines [21, 22, 23]. Notably, a recent study suggested that KM acts as a tumor immunomodulator and an immunotherapy sensitizer in GC [18], positioning it as a promising candidate for improving GC immunotherapy. However, current understanding of KM's role in GC immune regulation remains limited, and its precise mechanisms have yet to be fully elucidated.

Therefore, we employed KM as a chemical probe to investigate the mechanistic basis of MHC‐I regulation in GC. This study provides a mechanistic rationale for KM as an immunotherapy adjuvant and lays a theoretical foundation for developing targeted strategies to improve the efficacy of cancer immunotherapy.

2. Materials and Methods

2.1. Network Pharmacology Analysis

Potential KM targets were identified through the GeneCards, PharmMapper, SwissTargetPrediction, and TargetNet Databases. The SMILES notation of KM (PubChem CID: 44583834) was retrieved. The disease‐associated targets of GC were collected from GeneCards, Therapeutic Target database, and DrugBank. Overlapping targets were analyzed for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment.

2.2. Cell Culture

AGS and MKN45 cells (Cell Bank of the Chinese Academy of Sciences) were cultured in RPMI‐1640 medium (#11875093, GIBCO, USA) with 10% FBS (#26010074, GIBCO) at 37°C with 5% CO₂. Cell lines were authenticated by STR profiling and confirmed to be mycoplasma‐free.

2.3. Proteomic Profiling

Detailed information can be found in Data S1.

2.4. Molecular Docking

Molecular docking was used to predict the interactions between KM and insulin‐like growth factor 2 mRNA‐binding protein 3 (IGF2BP3). Docking was performed by importing the crystal structure into PyMOL (version 2.4.1, https://pymol.org/2/) for dehydration, hydrogenation, and ligand separation. AutoDock Vina (version 1.1.2) was used to construct a docking grid box around the protein binding site. Semi‐flexible docking was performed by comparing with the original ligands and intermolecular interactions. The grid box center and dimensions were defined to evaluate binding interactions. Results were visualized using PyMOL.

2.5. Surface Plasmon Resonance Analysis

Surface plasmon resonance (SPR) analysis was performed on a Biacore T200 instrument (GE, USA) with a CM7 sensor chip. Detailed information can be found in Data S1.

2.6. Cellular Thermal Shift Assay

Cellular thermal shift assay (CETSA) was performed to assess the interaction between KM and IGF2BP3. Briefly, a plasmid encoding Flag‐tagged human IGF2BP3 was transfected into HEK293T cells. After 48 h, cells were harvested and lysed in RIPA buffer. The lysates (protein concentration, 100 μg) were incubated with 20 μM KM or 0.1% DMSO for 30 min at room temperature, then heated at different temperatures for 3 min. After heating, lysates were centrifuged to remove aggregated proteins, and the soluble fraction was analyzed by western blot using an anti‐Flag antibody.

2.7. Cycloheximide Chase and MG132 Rescue Assay

To evaluate the protein stability and proteasome‐dependent degradation of IGF2BP3, cells were treated with cycloheximide (CHX) or MG132. For the CHX chase assay, cells were incubated with CHX at a final concentration of 50 μg/mL and harvested at the indicated time points. For proteasome inhibition, cells were treated with MG132 at a final concentration of 10 μM for the indicated times before collection.

2.8. Bioinformatics Analysis of Single‐Cell Public Database

Detailed information can be found in Data S1.

2.9. TCGA GC Data Analysis

Detailed information can be found in Data S1.

2.10. Immunofluorescence Staining and Quantification of Human GC Tissues

Ethical approval for the study was granted by the ethics committee of Fujian Medical University Union Hospital (protocol number 2021KJCX036), and all patients provided written informed consent. A total of 40 formalin‐fixed, paraffin‐embedded primary GC tissue samples were obtained from patients who underwent curative gastrectomy at Fujian Medical University Union Hospital between December 2025 and January 2026. Clinicopathological information was collected from medical records. Detailed information can be found in Data S1.

2.11. Construction of IGF2BP3‐Overexpressing Cells

Detailed information can be found in Data S1.

2.12. CRISPR‐Cas9‐Mediated Knockout of IGF2BP3

Detailed information can be found in Data S1.

2.13. Quantitative Real‐Time‐PCR Assay

According to the manufacturer's instructions, total mRNA was extracted and reverse transcribed using SuperScript II Reverse Transcriptase (#18064014, Invitrogen, USA). The PCR mixture was prepared following SYBR Green‐based qRT‐PCR protocols (#FP201‐02, Tiangen Biotech, China) with GAPDH as the endogenous control. Each sample was tested in triplicate in 96‐well plates. Detailed primer sequences are shown in Table S1.

2.14. Western Blot Analysis

Cells were lysed in RIPA buffer, and protein concentrations were determined by BCA assay. Equal amounts of protein (30 μg) were resolved by SDS‐PAGE, transferred to nitrocellulose membranes, and immunoblotted with the indicated primary antibodies (overnight at 4°C) and corresponding HRP‐conjugated secondary antibodies. Antibodies are listed in Table S2. Bands were visualized and quantified using ImageJ.

2.15. Co‐Culture With CD8 ⁺ T Cells and Cytotoxicity Assays

CD8⁺ T cells were purchased from Milecell Bio (China) and activated for 48 h with anti‐CD3/CD28 Dynabeads (#70912–200, Beaver Bio, China) in the presence of recombinant human IL‐2 (20 U/mL, #90267ES60, Yeasen). The pre‐treated GC cells (from all four groups, 1 × 10⁴ cells per well) were co‐cultured with the activated CD8⁺ T cells at GC:T ratios of 1:0 (GC cells only), 1:1, and 1:10 in 96‐well plates for 24 h.

To establish an antigen‐specific T‐cell killing model, MFC cells were engineered to stably express ovalbumin (OVA) and H‐2Kᵇ. Detailed information can be found in Data S1. After co‐culture, tumor cell killing was evaluated by flow cytometry and ELISA assay. The levels of effector molecules, including interferon gamma (IFN‐γ), malondialdehyde (MDA) and lactate dehydrogenase (LDH) released by the cells were further detected using commercial ELISA kits (#ml002277, #ml53201, #ml107163, Enzyme‐linked Bio, China) strictly following the manufacturers' instructions.

2.16. Apoptotic Assay

Annexin‐V‐APC/Propidium iodide (PI) double staining was used to detect cell apoptosis in each group according to the manufacturer's instructions. Briefly, cells with a density of 2 × 10⁴ cells per well were cultured for 48 h. Cells were collected, washed, and resuspended in 1 × loading buffer containing 5 μL of Annexin V‐APC and PI each. The percentage of apoptosis was analyzed by flow cytometry (Beckman CytoFlex, USA).

2.17. Xenograft Tumor Model and Histopathology

All animal experiments were approved by the Ethics Committee of Fujian Medical University (protocol number IACUC FJMU 2023‐Y‐0270). Four‐week‐old male inbred 615 strain mice (Shanghai Laboratory Animal Research Center, average weight 22 ± 2 g, animal license number: SYXK [Shanghai] 2019–0004) were housed in specific pathogen‐free conditions under a 12‐h light/dark cycle in a controlled environment at 22°C–25°C and 45%–50% humidity. Detailed information can be found in Data S1.

2.18. m6A and ChIP Sequencing Analysis

Following a 48‐h treatment with either 100 μg/mL KM or DMSO, total RNA was subjected to m6A‐seq, m6A‐modified RNA fragments were immunoprecipitated using an m6A‐specific antibody (Magna MeRIP Kit, #17–10,499, Sigma‐Aldrich, USA) and sequenced. Chromatin was subsequently fragmented and subjected to immunoprecipitation with ChIP‐validated H3K27me3‐specific antibodies (#ab192985, Abcam, USA) and SimpleChIP kit (#56383, CST, USA). Detailed information can be found in Data S1.

2.19. Chromatin Immunoprecipitation Followed by Quantitative PCR

Chromatin immunoprecipitation followed by quantitative PCR (ChIP‐qPCR) was performed to evaluate the enrichment of H3K27me3 at the promoter regions of IRF1, IRF6, and IRF9. Detailed information can be found in Data S1 and Table S3.

2.20. Actinomycin D Chase Assay

AGS and MKN45 cells were seeded in 6‐well plates at 5 × 10⁵ cells per well and allowed to attach overnight. After pre‐treatment, actinomycin D (Act D, #A1410, Sigma, USA) was added to a final concentration of 5 μg/mL to block new transcription. Total RNA was harvested at 0, 4, 8, 12, and 24 h after Act D addition.

2.21. Statistical Analysis

All statistical analyses were performed using GraphPad Prism 8.0, IBM SPSS Statistics (version 24), and R (version 4.2.1; for bioinformatics). Data are presented as mean ± standard deviation (SD) from at least three independent biological replicates. Normality was assessed using the Shapiro–Wilk test. For comparisons between two groups of parametric data, an unpaired two‐tailed Student's t‐test was used. For comparisons among multiple groups, one‐way or two‐way analysis of variance was applied followed by Bonferroni's post hoc test for multiple comparisons. For nonparametric data, the Mann–Whitney U test (two groups) or Kruskal–Wallis test with Dunn's correction (multiple groups) was used. A p‐value < 0.05 was considered statistically significant.

3. Results

3.1. KM Promotes the Antigen‐Presenting Ability of GC Cells

GO and KEGG pathway enrichment analyses were performed based on network pharmacology (Figure 1A). Biological process (BP) terms revealed significant enrichment for immune response and cell cycle regulation functions, such as “antigen processing and presentation” (Figure 1B). KEGG pathway analysis highlighted several key immune‐oncology pathways, including “antigen processing and presentation” and “PD‐L1 expression and PD‐1checkpoint pathway in cancer” (Figure 1C). These computational findings suggest that KM may exert its antitumor effects in GC by enhancing antigen presentation.

FIGURE 1.

KM promotes antigen presentation in GC cells. (A) Structure of KM. (B, C) GO and KEGG pathway enrichment analysis of KM in treatment for GC. (D) Quantification of DEPs in AGS and MKN45 cells treated with KM (100 μg/mL, 48 h). (E) Volcano scatter plot of DEPs. (F, G) GO and KEGG pathway enrichment analysis of DEPs. Quantitative proteomic analysis was performed on three independent biological replicates per condition. DEPs with |log₂fold change| ≥ 1 and FDR < 0.05 were considered significant.

Quantitative proteomics revealed that KM treatment induced substantial proteomic alterations in both GC cell lines, with IGF2BP3 consistently downregulated (Figure 1D,E and S1A,B). Multiple key proteins were identified based on PPI analysis (Figure S1C). GO and KEGG enrichment analyses of the differentially expressed proteins (DEPs) in both AGS and MKN45 cells converged on immune‐related processes, most notably “immune system process”, “MHC Class I biosynthetic process”, and “antigen processing and presentation”(Figure 1F,G).

To exclude the possibility that the observed effects were due to nonspecific cytotoxicity, we performed CCK‐8 assays (Figure S2). KM at concentrations up to 100 μg/mL did not significantly reduce cell viability, confirming that this concentration is sub‐cytotoxic and suitable for mechanistic studies.

3.2. KM Directly Binds IGF2BP3 and Promotes Its Downregulation

Venn diagram analysis identified IGF2BP3 as a candidate target (Figure 2A). KM was predicted to dock into the IGF2BP3 protein with favorable binding energy (Figure 2B). This direct interaction was experimentally validated by SPR, which revealed a specific binding affinity (K_D = 18.4 ± 1.2 μM), and by lysate‐based CETSA using overexpressed Flag‐IGF2BP3 in 293 T cells, which showed that KM binding stabilized IGF2BP3 against thermal denaturation (Figure 2C,D). In addition, while KM treatment did not alter IGF2BP3 mRNA levels (Figure 2E), it led to a significant downregulation of IGF2BP3 protein expression in a dose‐ and time‐dependent manner (Figure 2F). The CHX chase assay indicated that KM treatment significantly accelerated IGF2BP3 degradation in both AGS and MKN45 cells (Figure 2G). Co‐treatment with MG132 significantly blocked the KM‐induced reduction of IGF2BP3 protein levels, restoring IGF2BP3 to near‐control levels (Figure 2H).

FIGURE 2.

Km directly targets IGF2BP3 protein. (A) Venn diagram of the KM target and DEPs. (B) Molecular docking of KM with IGF2BP3. (C) Surface plasmon resonance (SPR) binding analysis. (D) CETSA analysis. Representative western blot (upper) and quantification (lower) are shown. (E) Real‐time PCR analysis of the expression of IGF2BP3 in AGS and MKN45 cells treated with KM for 48 h. (F) IGF2BP3 protein expression. Quantification of band intensity normalized to β‐Actin is shown (mean ± SD, n = 3 independent biological replicates). Statistical analysis was performed using two‐way ANOVA with Bonferroni's post hoc test. *p < 0.05, **p < 0.01, and ***p < 0.001 indicated statistically significant differences.

3.3. IGF2BP3 Acts as a Key Suppressor of Antigen Presentation

Single‐cell RNA‐seq analysis of GC tissues revealed that malignant epithelial cells formed distinct clusters and exhibited significant transcriptomic alterations compared to normal epithelial cells (Figure 3A and S3A,B). Differential gene expression analysis of these epithelial cells identified IGF2BP3 as one of the most significantly upregulated genes in GC (Figure 3B). Crucially, pathway enrichment analysis of IGF2BP3‐correlated genes revealed a strong negative association with antigen processing and presentation pathways (Figure 3C).

FIGURE 3.

IGF2BP3 is a key suppressor of antigen presentation. (A) UMAP visualization of single‐cell RNA‐seq data. (B) Volcano plot and dot plot of the differential expression of key genes. (C) GO‐BP enrichment analysis of upregulated genes. (D) Sankey diagram of cell–cell communication. (E) Immunofluorescence and correlation analysis between IGF2BP3 and MHC‐I MFI in human GC tissues. (F) Immune infiltration correlated with IGF2BP3 expression. (G) KEGG enrichment for high IGF2BP3 expression. Statistical significance was determined by two‐way ANOVA followed by Bonferroni's post hoc test. *p < 0.05 indicated statistically significant differences.

Further quantification affirmed a significant increase in interaction strength, especially in the macrophage‐to‐epithelial cell axis (Figure S3C,D). The Sankey diagram revealed that the interactions between immune cells and epithelial cells were markedly intensified, and IGF2BP3 was identified as a key signaling molecule within these enhanced networks (Figure 3D, S3E).

A significant negative correlation between IGF2BP3 MFI and MHC‐I MFI (R ² = 0.5548, Pearson R = −0.7448, p < 0.0001) was observed in GC tissue samples (Figure 3E, S4). TCGA analysis showed that IGF2BP3 mRNA expression was significantly elevated in STAD samples compared to normal tissues (Figure S3F). Deconvolution analysis using ssGSEA revealed that IGF2BP3 expression was significantly related to immune infiltration, including monocytes and M0/M1 macrophages (Figure 3F). The differentially expressed genes correlated with IGF2BP3 were primarily enriched in antigen processing and presentation, immune responses, and metabolic pathways (Figure 3G).

3.4. IGF2BP3 Is the Functional Target of KM in Restoring MHC‐I Expression

IGF2BP3‐overexpressing (OE) cell lines were constructed in both AGS and MKN45 cells (Figure 4A–C). KM upregulated MHC‐I in control cells but not in IGF2BP3‐OE cells (Figure 4D,E). Notably, while KM significantly upregulated all three classical MHC‐I genes, the magnitude of upregulation was consistently greater for HLA‐B and HLA‐C than for HLA‐A. Flow cytometry confirmed that KM increased the proportion of MHC‐I‐positive cells in a manner dependent on IGF2BP3 inhibition (Figure 4F).

FIGURE 4.

IGF2BP3 mediates KM‐induced MHC‐I upregulation. (A) Representative images of fluorescence after transfection of lentivirus into AGS and MKN45 cells (scale bar: 100 μm). (B‐C) IGF2BP3 mRNA (B) and protein (C) levels in overexpression (OE)‐control and OE‐IGF2BP3 of AGS and MKN45 cells. (D) The mRNA levels of HLA‐A, HLA‐B, and HLA‐C in OE‐control and OE‐IGF2BP3 of AGS and MKN45 cells treated with KM (100 μg/mL) or DMSO for 48 h. (E) The protein expression of MHC‐I in OE‐control and OE‐IGF2BP3 of AGS and MKN45 cells treated with KM (100 μg/mL) or DMSO for 48 h. (F) Flow cytometry analysis and quantification of MHC‐I positive ratio. Data are represented by mean ± SD (n = 3 independent biological replicates). Statistical significance was determined by two‐way ANOVA followed by Bonferroni's post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 indicates a statistically significant difference.

In addition, loss‐of‐function studies using CRISPR‐Cas9 were performed. IGF2BP3 mRNA and protein levels were substantially elevated in multiple GC cell lines compared to normal gastric epithelial GES‐1 cells (Figure S5A,B). Knockout of IGF2BP3 in AGS and MKN45 cells was validated by loss of protein expression (Figure S5C). Notably, IGF2BP3 knockout alone upregulated MHC‐I, and KM did not further increase MHC‐I in knockout cells (Figure S5D).

3.5. KM Potentiates CD8 ⁺ T Cell‐Mediated Cytotoxicity in an IGF2BP3‐Dependent Manner

KM treatment significantly sensitized GC cells to CD8⁺ T‐cell‐mediated cytotoxicity in a dose‐dependent manner. This sensitization effect was profoundly attenuated in IGF2BP3‐OE cells (Figure 5A,B). Consistent with the apoptosis data, KM significantly increased the secretion of IFN‐γ (a master T‐cell activation cytokine) in a manner dependent on the GC: T‐cell ratios (Figure 5C). In addition, we observed concurrent rises in MDA and LDH levels, indicating enhanced tumor cell membrane damage and lipid peroxidation, respectively (Figure 5D,E). Overexpression of IGF2BP3 largely abolished KM's ability to elevate IFN‐γ, LDH, and MDA levels. Instead, IGF2BP3 knockout significantly enhanced T‐cell‐mediated apoptosis, as well as IFN‐γ, LDH, and MDA levels, in a GC: T‐cell ratio‐dependent manner (Figure S5E,F).

FIGURE 5.

KM enhances CD8⁺ T cell cytotoxicity via IGF2BP3. OE‐Control or OE‐IGF2BP3 GC cells were pre‐treated with KM or DMSO (48 h) and co‐cultured with CD8⁺ T cells at indicated ratios. (A‐B) Apoptosis staining and quantification. (C‐E) IFN‐γ, MDA, and LDH levels in supernatants. Data are presented as mean ± SD (n = 3 independent biological replicates). Statistical significance was analyzed by two‐way ANOVA with Bonferroni's multiple comparisons test. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 indicate a statistically significant difference.

As shown in Figure S6A, IGF2BP3 knockout was confirmed in these cells. KM treatment upregulated MHC‐I in sgCtrl cells, and IGF2BP3 knockout alone phenocopied this effect (Figure S6B). Co‐culture with OT‐I T cells revealed that both KM treatment and IGF2BP3 knockout significantly increased target cell apoptosis, as well as IFN‐γ, LDH, and MDA levels in a GC:T ratio‐dependent manner (Figure S6C,D).

3.6. KM Enhances the Efficacy of Anti‐PD‐1 Therapy to Suppress Tumor Growth and Augment Immunity

While KM or anti‐PD‐1 monotherapy moderately suppressed tumor growth, their combination resulted in markedly greater inhibition of tumor progression, as well as reduced tumor volumes and weights (Figure 6B–E). IHC analysis confirmed that KM treatment, both alone and in combination, effectively downregulated IGF2BP3 protein levels and upregulated MHC‐I protein levels within the tumors. Consistent with enhanced T‐cell activation, a substantial increase in the intratumoral concentration of IFN‐γ was observed in the combination group (Figure 6F). Furthermore, tumors from the combination group exhibited a robust infiltration of CD8⁺ T cells, a significantly elevated proportion of which were Granzyme B (GZMB)‐positive, indicating a shift towards a highly activated, cytotoxic T‐cell phenotype (Figure 6G).

FIGURE 6.

KM synergizes with anti‐PD‐1 therapy in vivo. (A) Schematic representation of study design. (B) In vivo optical imaging and quantification of the fluorescence signal of tumor‐bearing mice in four groups after four weeks of tumor growth, the tumor growth curve (C), the image of tumors (D), and tumor weights (E) were shown. (F) Representative HE staining images of MFC tumors (400 ×, scale bar: 50 μm), IGF2BP3, MHC‐I, and IFN‐γ expression based on IHC staining in MFC tumors (400 ×, scale bar: 50 μm). (G) Immunofluorescence of GZMB/CD8 in MFC tumors (400 ×, scale bar: 100 μm). DAPI, stained nucleus; GZMB, granzyme B. Data are presented as mean ± SD (n = 6 biologically independent mice per group). Statistical significance was determined by two‐way ANOVA followed by Bonferroni's post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 indicate a statistically significant difference.

3.7. KM Destabilizes EED mRNA Stability to Alleviate H3K27me3‐Dependent Repression of MHC‐I

m6A‐seq revealed that KM significantly reduced m6A enrichment on the EED transcript (Figure 7A–C). The EED transcript track showed a marked decrease in m6A signal after KM treatment (Figure 7D), consistent across both AGS and MKN45 cells (Figure 7E). Actinomycin D chase assays indicated that KM accelerated the decay of EED mRNA, indicating that destabilization of the transcript contributes to the loss of m6A signal (Figure 7F). Similarly, IGF2BP3 knockout also accelerated EED mRNA decay (Figure 7G). Downregulation of EED protein was subsequently validated without altering the levels of enhancer of zeste homolog 2 (EZH2), confirming specific impairment of PRC2 complex activity (Figure 7H). Consistent with this, IGF2BP3 knockout phenocopied the effects of KM on EED (Figure 7I).

FIGURE 7.

m6A sequencing and validation in KM‐treated GC cells. (A, B) m6A‐seq global distribution and peak annotation. (C) MA plot showing reduced m6A enrichment on EED after KM. (D) Genome browser tracks of EED m6A peaks. (E) m6A enrichment of EED. (F) EED mRNA stability assessed by Act D chase assay in AGS and MKN45 cells pretreated with DMSO or KM (100 μg/mL, 48 h). (G) EED mRNA stability assessed by Act D chase assay in sgCtrl and sgIGF2BP3 cells. (H) Western blot of EED and EZH2 proteins in AGS and MKN45 cells treated with KM or DMSO for 48 h. (I) Western blot of EED in AGS, MKN45, and MFC cells with IGF2BP3 knockout. Data are represented by mean ± SD (n = 3 independent biological replicates). m6A‐seq was performed on two independent biological replicates per condition. Statistical significance was determined by two‐way ANOVA followed by Bonferroni's post hoc test. Peaks with FDR < 0.05 and |log₂fold change| ≥ 1 were considered differential. **p < 0.01 indicates a statistically significant difference.

Furthermore, KM treatment reduced H3K27me3 levels in both AGS and MKN45 cells without altering total histone H3. IGF2BP3 knockout similarly reduced H3K27me3, and KM did not further decrease H3K27me3 in sgIGF2BP3 cells (Figure 8A), confirming that the effect on H3K27me3 is mediated by IGF2BP3. ChIP‐seq analysis revealed that IRFs were enriched in the H3K27me3 methylated regions (Figure 8B), and this result was further confirmed by ChIP‐qPCR (Figure 8C). The KM‐induced increase in IRF mRNAs was abrogated upon IGF2BP3 overexpression (Figure 8D), whereas IGF2BP3 knockout increased IRF mRNA levels to a similar extent as KM (Figure 8E). Similar results were obtained in MFC cells (Figure 8F).

FIGURE 8.

KM upregulates IRFs transcription. (A) Western blot analysis of H3K27me3 and total histone H3 in AGS, MKN45, and MFC cells treated with KM (Upper panel) or with IGF2BP3 knockout (Lower panel). (B) Heatmap of H3K27me3 ChIP‐seq. (C) ChIP‐qPCR validation of H3K27me3 enrichment at the transcription start site regions of IRF1, IRF6, and IRF9. (D) RT‐qPCR analysis of IRF1, IRF6, and IRF9 mRNA levels in AGS and MKN45 cells with IGF2BP3 overexpression (OE). (E) RT‐qPCR analysis of IRFs mRNA levels in AGS and MKN45 cells with IGF2BP3 knockout. (F) RT‐qPCR analysis of Irfs mRNA levels in MFC cells with IGF2BP3 knockout. Data are represented by mean ± SD (n = 3 independent biological replicates). ChIP‐seq was performed on two independent biological replicates per condition. Statistical significance was determined by two‐way ANOVA followed by Bonferroni's post hoc test. Peaks with FDR < 0.05 and |log₂fold change| ≥ 1 were considered differential. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 indicates a statistically significant difference.

4. Discussion

In this study, we used the natural alkaloid KM as a chemical probe to uncover a potential immunomodulatory mechanism: KM restores MHC‐I expression by targeting IGF2BP3, enhances CD8⁺ T‐cell cytotoxicity, and sensitizes tumors to anti‐PD‐1 therapy in vivo (Figure 9). Our findings establish IGF2BP3 as a druggable regulator of antigen presentation in GC and as a novel, actionable node for overcoming immunotherapy resistance, with potential applicability beyond GC.

FIGURE 9.

Proposed mechanism of KM action. KM degrades IGF2BP3 via the proteasome, reducing m6A on EED mRNA, accelerating EED decay, and impairing PRC2 to relieve H3K27me3 repression. This derepresses IRFs, restoring MHC‐I and enhancing CD8⁺ T cytotoxicity and anti‐PD‐1 response. Black lines: Inhibition; red arrows: Activation; dashed lines: Associations not yet directly proven.

KM exhibits a broad spectrum of pharmacological properties and relatively low toxicity (LD₅₀ of 100 mg/kg i.p. in mice) [17, 24]. While previous work showed its immunomodulatory effects in GC [18], the underlying mechanism remained elusive. Our data suggest that KM enhances antigen presentation by upregulating MHC‐I. Loss of MHC‐I antigen presentation is a common cancer immune evasion mechanism [25, 26]. Cancers can evade this immunity through multiple mechanisms [27], and loss of MHC antigen presentation represents one of the most common [28]. This reversal of MHC‐I downregulation sensitizes GC cells to CD8⁺ T‐cell killing and explains how KM enhances the efficacy of immune checkpoint inhibitors (ICIs). Importantly, this mechanism is driven by target engagement rather than nonspecific effects, reinforcing the validity of IGF2BP3 as a therapeutic target.

As an m6A “reader” protein [29, 30], IGF2BP3 is a direct target of KM, and its degradation is required for MHC‐I upregulation. Mechanistically, our functional data indicate that IGF2BP3 supports the stability of EED mRNA, a core component of the PRC2/EZH2 complex [31]. EZH2 catalyzes the trimethylation of H3K27, leading to chromatin compaction and transcriptional repression [32]. In our GC cell lines, direct ChIP‐qPCR did not detect significant H3K27me3 enrichment at MHC‐I promoters. However, we found that IRF1, IRF6, and IRF9 genes reside within H3K27me3‐marked chromatin regions, and KM upregulated their mRNA. Given that IRFs are master transcription factors that directly activate MHC‐I gene expression [33], we propose that KM restores MHC‐I expression primarily by derepressing IRF transcription by inhibiting IGF2BP3, providing a mechanistic explanation for the observed immunomodulatory effects. Crucially, this establishes IGF2BP3 as a conserved epigenetic gatekeeper of antigen presentation, relevant across multiple cancer types where MHC‐I downregulation drives immune evasion. Interestingly, we observed that KM upregulated HLA‐B and HLA‐C more robustly than HLA‐A. This heterogeneity is consistent with known differences in MHC‐I gene regulation, where HLA‐A often exhibits higher basal promoter methylation or distinct transcription factor binding affinities in certain cancer types [34, 35]. Nevertheless, the functional consequence remained significant, suggesting that even partial restoration of HLA‐A contributes to the overall effect.

The combination benefit of KM and anti‐PD‐1 underscores its translational potential in vivo. Anti‐PD‐1 monotherapy alone modestly reduced IGF2BP3 protein levels, and the combination with KM further enhanced this effect. We propose an indirect mechanism whereby anti‐PD‐1 therapy activates CD8⁺ T cells to secrete IFN‐γ, which upregulates IRF1 through JAK‐STAT1 signaling [36]. IRF1 represses MYC transcription by binding to the interferon‐stimulated response element in the MYC promoter [37]. Since IGF2BP3 is a direct transcriptional target of MYC, reduced MYC leads to decreased IGF2BP3 synthesis [38, 39]. Thus, the combination synergistically reduces IGF2BP3 by targeting both its synthesis and stability. This dual suppression model further validates IGF2BP3 as a critical node where therapeutic intervention can be amplified by combining it with standard ICIs.

Several limitations of the current study should be noted. First, the binding affinity between KM and IGF2BP3 was approximately 18 μM. While this value is higher than typical nanomolar‐affinity drug‐target interactions, it falls within the range observed for many bioactive natural products that exert multi‐target characteristics in the context of natural product pharmacology and the specific mechanism of action [40, 41, 42, 43]. The moderate affinity underscores the power of using KM as a chemical probe to reveal a tractable target, and future structure–activity relationship studies may yield KM analogs with improved IGF2BP3 affinity. Second, the CETSA data were obtained using overexpressed Flag‐IGF2BP3 in cell lysates rather than intact cells with endogenous protein. Future studies using intact‐cell CETSA on endogenous IGF2BP3 would provide additional confirmation. Third, future studies should include CD8⁺ depletion and RIP‐qPCR to validate in vivo dependency and direct physical binding between IGF2BP3 and EED mRNA. Additionally, while we used two independent sgRNA‐derived clones for each cell line and observed consistent results, we cannot formally exclude the possibility of clonal effects. Future studies using pooled knockout populations or additional independently derived clones would further strengthen the conclusions. Finally, while our analysis of 40 GC specimens reveals a significant inverse association between IGF2BP3 and MHC‐I, this finding should be interpreted as a biological association rather than clinical validation of IGF2BP3 as a predictive biomarker. Prospective studies with ICI‐treated patient cohorts are needed to assess its predictive value.

Despite these limitations, our findings establish IGF2BP3 as an actionable target for enhancing antigen presentation in GC. The combination benefit of KM with anti‐PD‐1 highlights its promise as an immunotherapy adjuvant, warranting further development of selective IGF2BP3 inhibitors and exploration of combination strategies.

Author Contributions

Qiu Hongqiang: investigation, methodology, funding acquisition, writing – original draft. Lin Hailing: methodology, investigation, data curation. Lou Luqi: formal analysis, resources, visualization. Wang Linyao: software, resources, visualization. Lin Tian: validation, project administration. Shengjie Zhu: project administration, methodology. Zhang Dongmei: methodology, investigation, data curation. Chen Rong: software, visualization. Mao Yunyi: software, visualization. Wancai Que: conceptualization, formal analysis, supervision. Yu Cheng: conceptualization, supervision, funding acquisition, writing – review and editing.

Funding

This work was supported by Joint Funds for the Innovation of Science and Technology, Fujian Province (No. 2021Y9038 and 2024Y9255), Fujian Provincial Health Technology Project (No. 2025GGA015), and Fujian Province Natural Science Foundation of China (2026J001587).

Disclosure

Animal Studies: Animal experiments were performed according to the guidelines for the Care and Use of Laboratory Animals of the committee of Fujian Medical University, and the approval of the animal research protocol by the ethics committee of Fujian Medical University (protocol number IACUC FJMU 2023‐Y‐0270).

Ethics Statement

Approval of the research protocol by the ethics committee of Fujian Medical University Union Hospital (protocol number 2021KJCX036), and all patients provided written informed consent.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: Supporting information.

Table S1: Primer sequences.

Table S2: Antibody information.

Table S3: Promoter ChIP‐qPCR primers.

Figure S1: Differential gene expression and pathway analysis in AGS and MKN45 cells with KM.

Figure S2: CCK‐8 assay was used to detect proliferation of AGS (upper panel) and MKN45 cells.

Figure S3: Cellular communication and signaling analysis in GC and normal epithelial cells. (A) Dot.

Figure S4: Representative immunofluorescence staining of IGF2BP3 and MHC‐I in human gastric.

Figure S5: IGF2BP3 expression in gastric cancer cell lines and functional consequences on MHC‐I.

Figure S6: Antigen‐specific CD8+ T‐cell killing in the OT‐I system. (A) Validation of IGF2BP3.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.