3,3’-Diindolylmethane blocks the dissemination of gastric cancer cells in the peritoneal cavity through the MFGE8-MAPK/ERK signaling pathway

Abstract

Purpose

Milk fat globule epidermal growth factor 8 (MFGE8) has been demonstrated to have the potential to facilitate the onset and advancement of different forms of cancer. Studies have shown that 3,3’-Diindolylmethane (DIM) has the ability to suppress GC. The precise function and operational mechanism of MFGE8 and DIM in relation to gastric cancer peritoneal dissemination (GCPD) remain poorly understood.

Methods

The role of MFGE8 in GC and GCPD was confirmed through the use of bioinformatics and patient tissues in this study. Malignant biological behavior and changes in protein expression were detected using CCK-8, colony formation, adhesion assay, transwell, and western blotting techniques. To assess the in vivo effects of MFGE8 and DIM, mice models were used.

Results

Based on our findings, MFGE8 could serve as a therapeutic target for GC and GCPD, while also stimulating the aggressive biological behavior of GC cells. DIM could be an effective inhibitor of the oncogenic properties of GC cells. Subsequent studies revealed that DIM was capable of inhibiting the activation of Mitogen-activated protein kinase (MAPK)/Extracellular signal-regulated kinase (ERK) signaling pathway by blocking MFGE8, leading to the ultimate suppression of GCPD by impeding the epithelial-mesenchymal transition procedure.

Conclusion

MFGE8 had the potential to be regarded as a marker that could provide prognostic information for GC and GCPD. Furthermore, MFGE8-MAPK/ERK pathway might be involved in the mechanism of DIM inhibiting GCPD.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12935-025-04137-7.

Introduction

Gastric carcinoma (GC) remains a significant worldwide health issue, contributing to over 1 million newly diagnosed instances and approximately 769,000 fatalities in the year 2020. This places GC in the top five for both morbidity and mortality worldwide [1]. Peritoneal dissemination is a prevalent type of metastasis in GC, affecting as many as 14% of newly diagnosed GC patients. It is also the primary site of recurrence after radical gastrectomy for GC [2]. The prognosis for individuals diagnosed with peritoneal dissemination is grim, as there is currently no treatment proven to effectively extend survival beyond a median of only 3–6 months [2, 3]. Therefore, it is essential to find reliable therapeutic targets and related therapeutic drugs for gastric cancer peritoneal dissemination (GCPD).

In vivo, milk fat globule epidermal growth factor 8 (MFGE8), a secreted glycoprotein, is expressed universally [4]. MFGE8 was initially identified in the mammalian milk fat globule [5, 6]. This glycoprotein contains an arginine-glycine-aspartate motif that mediates specific binding to integrin receptors, through which it exerts its diverse biological functions [7–9]. Research has demonstrated that MFGE8 is intricately involved in the onset, advancement, and spread of different tumors [9–30]. Experimental evidence demonstrates that MFGE8 suppression markedly attenuates proliferative capacity, invasive potential, and migratory behavior in breast carcinoma models, with concomitant downregulation of matrix metalloproteinases (MMPs) and epithelial-mesenchymal transition (EMT)-associated biomarkers [29]. In colorectal cancer systems, genetic ablation of MFGE8 has been shown to promote mesenchymal-epithelial reversion while suppressing metastatic dissemination, paralleled by reduced enzymatic activities of MMP2 and MMP9 [30]. Similarly, MFGE8 deficiency in melanoma models significantly compromises neoplastic cell viability, EMT progression, and invasive properties [14]. Mechanistic investigations have further revealed that MFGE8-mediated activation of the IL-6/JAK/STAT3 axis regulates EMT dynamics during gastric carcinoma progression [31]. Despite these advances, the pathophysiological contributions of MFGE8 to GCPD pathogenesis remain completely unexplored [31–34].

3,3’-Diindolylmethane (DIM), a naturally occurring phytochemical, can be found in cruciferous vegetables [35]. Research has indicated that DIM exhibits potent tumor cell-killing activity while demonstrating minimal toxicity to healthy cells. Given its outstanding safety profile, DIM has gained significant interest in the field of anti-tumor drug development [36–38]. Numerous studies have indicated that DIM possesses the ability to govern the process of DNA damage repair, regulate nuclear transcription factors, hinder cell proliferation, induce cell apoptosis, and prevent tumor angiogenesis, migration, and invasion. In hepatocellular carcinoma, DIM activates the caspase-dependent apoptotic pathway via endoplasmic reticulum stress and the unfolded protein response while suppressing EMT [39]. DIM demonstrates antitumor activity in malignant melanoma cells, potentially through PTEN/Akt signaling pathway-mediated activation of the mitochondrial apoptosis pathway [40]. In colorectal cancer cells, DIM reduces MDM2 expression levels, thereby inhibiting proliferation and inducing cell cycle arrest and apoptosis [41]. DIM reverses EMT in esophageal squamous cell carcinoma by regulating the aryl hydrocarbon receptor through inhibition of the RhoA/ROCK1-mediated COX2/PGE₂ signaling pathway [42]. Long-term low-dose DIM administration inhibits proliferation, migration, and metastasis of nasopharyngeal carcinoma cells by blocking ERK signaling pathway activation [43]. DIM suppresses adhesion, migration, and invasion of ovarian cancer cells via downregulation of MMP2 and MMP9 expression [44]. In gastric cancer, DIM exerts antitumor effects through multiple mechanisms: Induction of ferroptosis via the BAP1-IP3R axis [45]; STIM1-mediated store-operated calcium entry leading to cell death [35]; inhibition of proliferation and induction of apoptosis via the TRAF2-p38 signaling pathway [46]; potentiation of paclitaxel’s antitumor effects through the Akt/FOXM1 signaling cascade [47]; suppression of cell growth via Hippo signaling pathway activation [48]; activation of the aryl hydrocarbon receptor pathway, promoting its nuclear translocation, inducing apoptosis, delaying cell cycle progression, and inhibiting proliferation [49]. However, the precise function of DIM in GCPD and the biological pathways underlying this function remain obscure and require further study.

This study aims to investigate the role and underlying mechanisms of MFGE8 in GCPD, as well as the effects and potential mechanisms of DIM in this context. Our findings may offer further evidence for target prediction and clinical treatment of GCPD.

Materials and methods

Bioinformatics analysis

We selected several large publicly available gastric cancer datasets (GSE54129, GSE62254, and GSE21328) for difference analysis. Subsequently, we explored the prognostic value of MFGE8 using GSE62254, GSE15459 GSE84426 GSE84433 and GSE54129, excluding samples with incomplete clinical information (especially survival time). Data sets GSE21328, GSE62254 (300 patients with GC), GSE19826 (12 pairs of GC and adjacent normal tissues), GSE15459 (200 patients with GC), GSE84426 (76 patients with GC), GSE84433 (357 patients with GC) and GSE54129 (111 patients with GC and 21 healthy subjects) were retrieved from the GEO database and subsequently analyzed using the GEO2R algorithm for differential gene expression between different groups. The list of 500 genes most strongly associated with GC overall survival (OS) was retrieved from the GEPIA 2 database (http://gepia2.cancer-pku.cn/#survival). RNAseq data from the STAR process for the GC dataset project were retrieved and collated from the TCGA database (375 patients with GC) (https://portal.gdc.cancer.gov). The data extracted for further analysis included TPM format data and clinical data. The prognostic data were derived from a Cell article [50]. The structures of the ligand DIM and the receptor MFGE8 were obtained from the PubChem (https://pubchem.ncbi.nlm.nih.gov/) and Uniprot (https://www.uniprot.org/) databases, respectively.

Molecular docking

The structures of the ligand DIM and the receptor MFGE8 were downloaded from the PubChem (https://pubchem.ncbi.nlm.nih.gov/) and UniProt (https://www.uniprot.org/) databases, respectively. The ligand structure was optimized in Chem3D, while water molecules and residual ligands were removed from the receptor using PyMOL. Hydrogen atoms were added to the receptor structure using AutoDockTools, and the active pockets of both the ligand and receptor were identified. Molecular docking was performed via the command prompt, and the results were visualized in PyMOL.

Individuals and specimens

The Ethics Review Committee of the First Hospital of China Medical University granted approval for this study, ensuring strict adherence to the ethical guidelines outlined in the Helsinki Declaration. From January 2015 to December 2016, a total of 90 samples of gastric cancer tissue were collected from patients who underwent radical gastrectomy at the Department of General Surgery. To ensure the reliability of the study, patients were required to meet the following criteria: (1) Diagnosis of GC validated through histological examination. (2) Absence of any other concurrent malignant tumors or significant diseases that could potentially affect the patient’s prognosis. (3) Not having received neoadjuvant chemotherapy. From medical records and pathology reports, demographic and clinical characteristics including sex, age, date of initial diagnosis, tumor grade, and pathological stage were collected. Patients were followed up every 6 months.

IHC

GC tissue sections were deparaffinized and hydrated for antigen repair and treated with an endogenous peroxidase blocker (MXB, China) for 10 min and a nonspecific stain blocker (MXB, China) for 20 min. Following a 1-hour incubation at 37℃ with the primary antibody, biotin-labeled sheep anti-mouse/rabbit IgG polymer (MXB, China) was applied for a 10-minute period, after which Streptomyces antibiotin protein-peroxidase (MXB, China) was employed for an additional 10 min. Next, the DAB kit (MXB, China) was used for color development. Finally, staining of GC tissues was observed under a microscope.

Two independent evaluators used a semi-quantitative method to analyze immunoreactivity. For further analysis, five representative areas were chosen at random from the microscope’s field of view. The immunostaining score was calculated by considering both the percentage of cells that showed positive staining and the intensity of the staining. The scoring scheme for the proportion of affirmative cells was established in the subsequent manner: 0 for < 9%, 1 for 10% to 25%, 2 for 26% to 50%, 3 for 51% to 75%, and 4 for 76% to 100%. The scoring system for staining intensity used the following scale: absence of staining scored as 0; light yellow assigned a value of 1; yellow scored as 2; and brown was assigned a score of 3. The ultimate immunoreactive score was determined by multiplying the ratio of scores for positive cells with the scores for staining intensity.

Reagents and the cultivation of cells

In this study, the establishment of the high-potential peritoneal dissemination cell line MKN-45P was achieved by using MKN-45 cell line derived from human poorly differentiated gastric adenocarcinoma. The cell line was built on previously published research protocols [32, 51]. In short, MKN-45 cells were subcutaneously injected into BALB/c nude mice. Once subcutaneous tumors were established, these tumor cells were extracted and injected into the abdominal cavities of the mice. After tumor formation in the abdominal cavity of nude mice, tumor cells were extracted and re-injected into the abdominal cavity of nude mice. The procedure was performed five times, resulting in successful isolation and culture of the MKN-45P cell line. MKN-45 and MKN-45P cells were maintained in RPMI 1640 supplemented with 10% fetal bovine serum (BioInd, Israel) and 1% penicillin-streptomycin solution (Beyotime, China). HMrSV5 cell line was purchased from aoruicell (Shanghai, China). For HMrSV5 cells, DMEM high glucose medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin solution was used for culturing. The cell incubator, which was filled with 5% CO₂, kept all cell lines at a temperature of 37℃.

Recombinant Human MFG-E8 Protein (rhMFGE8) was acquired from R&D Systems (Minneapolis, USA). In this study, MKN-45 cells were pretreated with 100ng/ml rhMFGE8 for 24 h. DIM (HY-15758) was acquired from MedChemExpress (Shanghai, China). MFGE8 siRNAs were acquired from Sangon Biotech (Shanghai, China). The MFGE8-specific sense and antisense RNA oligonucleotides were as follows: si-1/sense siRNA, ACAGCCUUAAUGGACACGAAUTT, and antisense siRNA, AUUCGUGUCCAUUAAGGCUGUTT and si-2/sense siRNA CUACAGUAAUGACAGUGCGAATT and antisense siRNA UUCGCACUGUCAUUACUGUAGTT.

Antibody catalog

Primary antibody against MFGE8 (1:200 for IHC) was acquired from SANTA CRUZ (Texas, USA). Sigma-Aldrich (Darmstadt, Germany) provided the primary antibody against MFGE8 (1:1000 for western blotting). Primary antibodies against GAPDH (1:20000 for western blotting), Vimentin (1:2000 for western blotting and 1:5000 for IHC), MMP9 (1:1500 for western blotting and 1:200 for IHC) and HRP-conjugated Affinipure Goat Anti-Rabbit IgG(H + L) (1:5000 for western blotting) were acquired from Proteintech (Wuhan, Hubei, China). Cell Signaling Technology (Danvers, USA) provided the primary antibodies for western blotting against p-ERK (1:2000), ERK (1:1000), and E-cadherin (1:1000), as well as for IHC against E-cadherin (1:400) and N-cadherin (1:125). We obtained the N-cadherin primary antibody (1:2000 for western blotting) from Abmart (Shanghai, China).

Colony formation

Six well plates were used to culture MKN-45 and MKN-45P cells at a density of 1000 cells per well. After 8–14 days, the culture was terminated when most colonies consisted of more than 50 cells. Samples were treated with a 4% paraformaldehyde solution (Beyotime, China) and then dyed with 0.1% crystal violet (Solarbio, China).

Adhesion assay

Peritoneal mesothelial cells HMrSV5 were cultured overnight in 24-well plates to establish a monolayer of peritoneal mesothelial cells. GC cells were stained with DiD cell-labeling solution for 20 min. Vybrant™ DiD cell-labeling solution (Thermo Fisher, China) is a dye-delivery solution that can be added directly to normal culture medium to uniformly label cells in suspension or adherent cultures for cell fusion, cell attachment, and migration research applications. The stained cells were then seeded onto peritoneal mesothelial cells at a density of 1 × 10⁵ cells per well. After co-incubation for 6 h, nonadherent cells were washed off using sterile PBS. Photographs were taken at 200× fluorescence and white light field.

Transwell

The wells in the upper chamber (Corning, USA) were filled with a density of 3 × 10⁵ GC cells [52]. To perform the invasion test, the chambers were covered with Matrigel (Corning, USA) diluted at a ratio of 1:10. After a 72-hour co-incubation period, the cells were treated with methanol for fixation and subsequently stained with Wright-Giemsa Stain Kit (Nanjing Jiancheng Bioengineering Institute, China). Photographs were taken using a 100× magnification microscope.

Western blotting

Following mixing with RIPA lysis buffer (Beyotime, China) on ice for 30 min, cells were lysed using an ultrasonic cell disruptor under ice-cold conditions. The lysate was subsequently centrifuged at 12,000 rpm for 30 min at 4 °C. Protein concentration was quantified using a BCA Protein Assay Kit (Beyotime, China) with standardized protocols. Protein samples were added to SDS-PAGE gel (Beyotime, China) for electrophoresis followed by membrane transfer. After a 30-minute block, the membranes were left to incubate with primary antibodies overnight at a temperature of 4 °C. The following day, the membranes underwent an incubation period of one hour at room temperature in conjunction with the secondary antibody. Finally, immunoblots were developed using Super ECL Detection Reagent (Yeasen, China).

CCK-8

A total of 5000 GC cells were spread out in each well of 96 well plates for seeding. The original medium was replaced after 24 h incubation with different concentrations of DIM. 10µL of CCK-8 reagent (APExBIO, USA) was added after 24 and 48 h, respectively, followed by an incubation of 2–4 h. Using a microplate reader, the sample’s absorbance was measured at a particular wavelength of 450 nm.

GCPD model

A total of twenty-four female BALB/c nude mice, with an age range of 4 to 6 weeks, were raised in an environment free from specific pathogens. The mice were injected with 200µL of a serum-free suspension containing MKN-45P cells into their abdominal cavity. The concentration of cells was 1 × 10⁷ cells/ml. After three days of injection of MKN-45P cells, the 24 mice were randomly assigned into four groups (n = 6 each): DIM group (10 mg/kg, i.p.), rhMFGE8 group (20ug/kg, i.p.), DIM + rhMFGE8 group, and control group. The mice were injected every other day and sacrificed on day 15. The animal tests were carried out in strict conformity with the ethical standards sanctioned by the Ethics Committee of the First Hospital of China Medical University.

Statistical analysis

Statistical analyses and visualizations were performed using R and GraphPad Prism 8.0.1 software. All experimental procedures were performed in triplicate with independent replicates. We used a chi-square test or continuous correction chi-square test to investigate the correlation between MFGE8 expression and demographic characteristics and clinicopathological parameters in GC patients. We compared the survival curves using a log-rank test. For analyzing prognostic data, we used both univariate and multivariate Cox proportional regression risk models. Multivariate Cox regression analysis included univariate factors with a significant P < 0.1 level. Statistical significance was considered for P < 0.05. Data comparison between two groups was conducted using either an independent samples t-test or a paired t-test through statistical analysis. Statistical analysis was conducted on multiple data sets using one-way ANOVA and multiple hypothesis testing.

Results

Online datasets and our tissue samples identified the MFGE8 as a biomarker for unfavorable prognosis in patients with GC

Initially, online datasets were utilized to screen for potential biomarkers for GCPD. The GSE54129 dataset yielded 1592 genes that exhibited higher expression levels in GC specimens than in normal gastric mucosa. In the GSE21328 dataset, 1005 genes were identified with elevated expression in MKN-45P cells, relative to MKN-45 cells. The GSE62254 dataset revealed a set of 3301 genes that were significantly upregulated in GC tissues with peritoneal dissemination, compared to those without peritoneal dissemination. Additionally, the top 500 differential survival genes in GC were obtained from the GEPIA 2 database. The aforementioned gene sets were then overlapped via a Venn diagram, resulting in the identification of four genes (Fig. 1A). Among these four genes, MFGE8 exhibited the largest logFC value and the smallest adjusted P value in the GSE21328 dataset (Fig. 1A). Data from two publicly available datasets also showed that MFGE8 expression was upregulated in gastric cancer tissues (Fig. 1B). Therefore, we selected MFGE8 as the target gene for our subsequent investigations.

Fig. 1.

Online datasets and our tissue samples identified the MFGE8 as a biomarker for unfavorable prognosis in patients with GC. A, GEO and GEPIA 2 databases were used to screen for GCPD biomarkers. B, Data from GSE19826 and GSE54129 indicated that MFGE8 expression was upregulated in gastric cancer tissues. C, The independent samples t-test was conducted using the TCGA-STAD dataset to compare the levels of MFGE8 mRNA in normal and GC tissues (** P<0.01). D, The TCGA-STAD dataset was used to compare MFGE8 mRNA levels in paired paracancerous and GC tissues (paired t-test, * P<0.05). E, F, G, The relationship between MFGE8 and OS, DSS, and PFI was investigated using the TCGA-STAD dataset. H, I, J, K, The prognostic value of MFGE8 was validated in four external datasets

Subsequently, we evaluated the expression levels and prognostic implications of MFGE8 in the TCGA-STAD dataset. Our study revealed that MFGE8 expression levels in GC tissues were higher than in normal tissues, as shown in Fig. 1C. The same trend was observed when comparing paired paracancerous tissues with GC tissues, as depicted in Fig. 1D. Furthermore, the survival analysis indicated that individuals exhibiting elevated levels of MFGE8 had poorer OS (HR = 1.402, P = 0.041, Fig. 1E), disease-specific survival (DSS) (HR = 1.539, P = 0.041, Fig. 1F), and progression-free interval (PFI) (HR = 1.505, P = 0.022, Fig. 1G) compared to individuals with lower MFGE8 expression. Table 1 provides a summary of the TCGA-STAD patient demographic characteristics, with a notable correlation between MFGE8 expression and tumor histological grade (P = 0.020). Finally, we validated the prognostic value of MFGE8 in four external GEO datasets (Fig. 1H, I, J, K).

Table 1.

The TCGA-STAD dataset provided information on the demographic characteristics of GC patients

Variables	N	MFGE8 expression		Chi-square value	P value
		High n(%)	Low n(%)
Total	375	188	187
Age				3.811	0.051
≤ 65	164	92 (56.1%)	72 (43.9%)
>65	207	95 (45.9%)	112 (54.1%)
Gender				0.031	0.859
Male	241	120 (49.8%)	121 (50.2%)
Female	134	68 (50.7%)	66 (49.3%)
T stage				3.001	0.083
T1-2	99	42 (42.4%)	57 (57.6%)
T3-4	268	141 (52.6%)	127 (47.4%)
N stage				0.135	0.714
N0-1	208	102 (49%)	106 (51%)
N2-3	149	76 (51%)	73 (49%)
M stage				1.046	0.307
M0	330	167 (50.6%)	163 (49.4%)
M1	25	10 (40%)	15 (60%)
pStage				0.000	1.000
Ⅰ+Ⅱ	164	82 (50%)	82 (50%)
Ⅲ+Ⅳ	188	94 (50%)	94 (50%)
Grade				5.405	0.020
G1 + G2	147	63 (42.9%)	84 (57.1%)
G3	219	121 (55.3%)	98 (44.7%)

In parallel, demographic data for participants in the GSE62254 dataset can be found in Table 2. It was found that the expression level of MFGE8 was significantly correlated with various pathological factors such as T stage (P < 0.001), N stage (P = 0.048), M stage (P = 0.002), tumor stage (P = 0.002), liver metastasis (P = 0.033), peritoneal seeding (P < 0.001), ascites (P = 0.007), and other sites of metastasis (P = 0.033) (Fig. 2).

Table 2.

The GSE62254 dataset provided information on the demographic characteristics of GC patients

Variables	N	MFGE8 expression		Statistic	P value
		High n (%)	Low n (%)
Total	300	150	150
Age				0	1.000
≤ 65	172	86 (50%)	86 (50%)
>65	128	64 (50%)	64 (50%)
Gender				0.015	0.903
Male	199	100 (50.3%)	99 (49.7%)
Female	101	50 (49.5%)	51 (50.5%)
T stage				11.092	< 0.001
T1-2	186	79 (42.5%)	107 (57.5%)
T3-4	114	71 (62.3%)	43 (37.7%)
N stage				3.916	0.048
N0-1	169	76 (45%)	93 (55%)
N2-3	131	74 (56.5%)	57 (43.5%)
M stage				9.158	0.002
M0	273	129 (47.3%)	144 (52.7%)
M1	27	21 (77.8%)	6 (22.2%)
pStage				9.954	0.002
Ⅰ+Ⅱ	127	50 (39.4%)	77 (60.6%)
Ⅲ+Ⅳ	173	100 (57.8%)	73 (42.2%)
Liver metastasis				4.545	0.033
N	264	126 (47.7%)	138 (52.3%)
Y	36	24 (66.7%)	12 (33.3%)
Peritoneal seeding				10.930	< 0.001
N	246	112 (45.5%)	134 (54.5%)
Y	54	38 (70.4%)	16 (29.6%)
Ascites				7.291	0.007
N	253	118 (46.6%)	135 (53.4%)
Y	47	32 (68.1%)	15 (31.9%)
Bone metastasis				0.585	0.444
N	293	148 (50.5%)	145 (49.5%)
Y	7	2 (28.6%)	5 (71.4%)
Other sites of metastasis				4.521	0.033
N	237	111 (46.8%)	126 (53.2%)
Y	63	39 (61.9%)	24 (38.1%)

Fig. 2.

A, Univariate and multivariate Cox regression were used to analyze GC patients in TCGA-STAD, GSE62254, and self-contained cohort data sets. B, A total of 90 GC tissues were examined for MFGE8 protein expression via IHC. High expression tissue is shown on the left and low expression tissue is shown on the right (Scale bar: 200×: 300 μm, 400×:100 μm). C, The Kaplan-Meier analysis showed notable variations in survival results among GC patients who indicated high or low levels of MFGE8 (P = 0.016)

Prognostic value of MFGE8 in gastric cancer

Figure 2 A displays the results of the multivariate Cox regression analysis conducted on the TCGA-STAD dataset, indicating that increased MFGE8 expression independently contributed to the risk of GC patients. Figure 2 A also displays the results of the multivariate Cox regression analysis conducted on the GSE62254 dataset, indicating that increased MFGE8 expression independently contributed to the risk of GC patients.

Our study performed IHC analysis on GC tissues. Table 3 shows the classification of the 90 tissues into groups based on their expression of MFGE8, with an equal distribution of 45 tissues in both the high and low expression groups. Table 3 displays noteworthy associations between MFGE8 expression and T stage (P < 0.001), N stage (P = 0.014), and tumor stage (P < 0.001) in the findings. The Kaplan-Meier analysis presented a significant finding, indicating that patients with a high level of MFGE8 expression exhibited a lower OS rate than those with a low level of MFGE8 expression (HR = 2.329, P = 0.016, Fig. 2 C). Univariate Cox regression analysis demonstrated that MFGE8 expression might affect the prognosis of our GC patients as shown in Fig. 2A. The results indicated that MFGE8 could potentially function as a biomarker and target for therapeutic intervention in GCPD.

Table 3.

Demographic characteristics of GC patients

Variables	N	MFGE8 expression		Chi-square value	P value
		High n (%)	Low n (%)
Total	90	45	45
Age				0	1.000
≤ 65	58	29 (50%)	29 (50%)
>65	32	16 (50%)	16 (50%)
Gender				0.055	0.814
Male	65	33 (50.8%)	32 (49.2%)
Female	25	12 (48%)	13 (52%)
T stage				26.235	< 0.001
T1-2	38	7 (18.4%)	31 (81.6%)
T3-4	52	38 (73.1%)	14 (26.9%)
N stage				6.016	0.014
N0-1	68	29 (42.6%)	39 (57.4%)
N2-3	22	16 (72.7%)	6 (27.3%)
pStage				17.273	< 0.001
Ⅰ+Ⅱ	57	19 (33.3%)	38 (66.7%)
Ⅲ+Ⅳ	33	26 (78.8%)	7 (21.2%)
Grade				0	1.000
G1 + G2	32	16 (50%)	16 (50%)
G3	58	29 (50%)	29 (50%)

MFGE8 promoted malignant biological behavior of GC cells

The experiments showed that MKN-45P cells had stronger proliferation, adhesion, migration, and invasion abilities than MKN-45 cells (Fig. 3A, G), indicating that MKN-45P cells established by us might have stronger peritoneal dissemination ability than its parent MKN-45 cells. Western blotting analysis showed higher expression of MFGE8 in GC cells when compared to GES-1 cells, as illustrated in Fig. 3B. Furthermore, MFGE8 expression in MKN-45P cells was found to be higher than in MKN-45 cells, indicating a potential relationship between MFGE8 and the enhanced peritoneal dissemination potential of MKN-45P cells.

Fig. 3.

MFGE8 promoted malignant biological behavior of GC cells. A, Proliferation, adhesion, migration, and invasion abilities were compared between MKN-45 and MKN-45P cells. B, MFGE8 protein levels in GES-1, MKN-45, and MKN-45P cell lines were analyzed using western blotting. C, Proliferation, adhesion, migration, and invasion abilities were compared between control and rhMFGE8 group in MKN-45 cells. D, The effectiveness of MFGE8 knockdown was verified by western blotting. E, Proliferation, adhesion, migration, and invasion abilities were compared between control and MFGE8 knock-down groups. (Scale bar: adhesion assay: 500 μm, transwell: 1000 μm). F, Correlation of MFGE8 with four scores. G, Number of colonies formed in the three group comparisons (** P<0.01, *** P<0.001)

To assess the influence of MFGE8 on GC cells, rhMFGE8 was applied to MKN-45 cells. Subsequently, treatment resulted in increased proliferation, adhesion, migration, and invasion (Fig. 3C). Simultaneously, we successfully suppressed MFGE8 expression in MKN-45P cells, as shown in Fig. 3D. Consequently, there was a decrease in cell proliferation, adhesion, migration, and invasion, as depicted in Fig. 3E. These results suggested that MFGE8 might promote malignant biological behavior in GC cells. Invasion, EMT, Metastasis and proliferation scores were calculated by GSVA, and pearson correlation analysis of MFGE8 with these scores was calculated. The results showed that MFGE8 showed a strong correlation with them (Fig. 3F).

MFGE8 activated MAPK/ERK signaling pathway to promote EMT in GC

MFGE8 has been shown to participate in multiple physiological and pathological processes by regulating signaling pathways. In order to provide more clarity on the regulatory mechanism of MFGE8 in GC, we conducted a correlation analysis between MFGE8 and other genes in the TCGA-STAD dataset. Figure 4A shows the genes most strongly associated with MFGE8. A total of 4112 genes with cor > 0.3 and P < 0.05 were included in the KEGG enrichment analysis together with MFGE8 (Fig. 4B). GSEA was employed to further enrich MFGE8-related signaling pathways (Fig. 4C). In both study groups, it was found that the mitogen-activated protein kinase (MAPK) signaling pathway ranked within the top 10 signaling pathways according to the results. GSEA analysis also showed that MFGE8 had a promoting effect on EMT (Fig. 4D).

Fig. 4.

MFGE8 activated MAPK/ERK signaling pathway to promote EMT in GC. A, Co-expression heatmap demonstrating the most relevant genes to MFGE8. B, C KEGG and GSEA were used to enrich MFGE8-related signaling pathways in GC. D, GSEA shows that MFGE8 promotes EMT. E, F, G, H, The TCGA-STAD dataset was utilized to investigate the relationship between MFGE8 expression levels and EMT-related proteins. I, Western blotting was used to detect changes in MAPK/ERK and EMT-related protein expression after treatment with rhMFGE8 or siRNAs

In the TCGA-STAD dataset, we conducted additional research on the connection between MFGE8 and proteins associated with epithelial-mesenchymal transition (EMT). MFGE8 expression level showed a negative correlation with E-cadherin (Fig. 4E), while displaying a positive correlation with N-cadherin (Fig. 4F), Vimentin (Fig. 4G), and MMP9 (Fig. 4H).

Treatment with rhMFGE8 resulted in elevated phosphorylation of extracellular signal-regulated kinase (ERK) in MKN-45 cells, as demonstrated by western blotting analysis (Fig. 4I). Furthermore, there was a decrease in the amount of E-cadherin protein, while the levels of N-cadherin, Vimentin, and MMP9 proteins were observed to be elevated. After knockdown of MFGE8 expression in MKN-45P cells, phosphorylation of ERK was decreased. In addition, upregulation of E-cadherin and downregulation of N-cadherin, Vimentin, and MMP9 were observed. The findings indicated that MFGE8 had the potential to induce EMT in GC by activating the MAPK/ERK signaling pathway.

DIM was an anti-GC drug targeting MFGE8

In the TCGA-STAD dataset, a single-gene differential analysis of MFGE8 identified 72 genes with log2(FC) greater than 2 and P.adj less than 0.05. Compounds inhibiting MFGE8 in GC were screened using these genes in the CMap database. As revealed by the findings, six compounds were most likely to inhibit MFGE8 in GC (Fig. 5A). By reviewing relevant literature, we selected DIM, which has been extensively studied in GC but not explored in GCPD, as the research object (Fig. 5B). Next, molecular docking revealed that DIM and the MFGE8 protein domain formed a hydrogen-bonded link with a docking score of − 6.0 kcal/mol (Fig. 5C). Western blotting confirmed that DIM could inhibit MFGE8 in GC cells (Fig. 5D). The inhibitory effect of DIM on GC cell proliferation was assessed through the CCK-8 assay, demonstrating a dose- and time-dependent impact (Fig. 5E).

Fig. 5.

DIM was an anti-GC drug targeting MFGE8. A, The top six compounds screened in the CMap database that were most likely to inhibit MFGE8 in GC. B, Chemical structure of DIM. C, Molecular docking between MFGE8 and DIM. D, Western blotting was utilized to measure MFGE8 expression following treatment with DIM in GC cells. E, Various concentrations of DIM were administered to GC cells, and the CCK-8 assay was utilized to assess cell proliferation (* P<0.05, ** P<0.01, *** P<0.001, **** P<0.0001)

MFGE8-MAPK/ERK pathway might be involved in the mechanism of DIM inhibiting GCPD

The findings from the colony formation (Fig. 6A), adhesion (Fig. 6B) and transwell assay (Fig. 6C, D) showed that DIM could inhibit malignant biological behavior of GC cells. In addition, DIM could reverse rhMFGE8-mediated promotion of GC cell malignant biological behavior (Fig. 6A, B, C, D), indicating that DIM could inhibit GCPD by inhibiting MFGE8.

Fig. 6.

MFGE8-MAPK/ERK pathway might be involved in the mechanism of DIM inhibiting GCPD. A, GC cell proliferation was detected by colony formation in four groups. “Combined” refers to DIM combined with rhMFGE8. The DIM concentrations applied to MKN-45 and MKN-45P cells were based on their respective 24-hour EC50 values (approximately 120µM for MKN-45 and 100 µM for MKN-45P) (* P<0.05, ** P<0.01, **** P<0.0001). B, The adhesion assay was utilized to evaluate the adhesion ability of GC cells on HMrSV5 cells across four groups. C, Four groups were tested for GC cell migration using transwell. D, Four groups were tested for GC cell invasion using transwell. E, Western blotting was employed to measure changes in MAPK/ERK and EMT-related protein expression in four groups. (Scale bar: adhesion assay: 500 μm, transwell: 1000 μm)

In previous results, MFGE8 was found to regulate EMT in GC via the MAPK/ERK signaling pathway, and DIM was found to inhibit MFGE8 in GC. We then confirmed that DIM could inhibit ERK protein phosphorylation and EMT in GC cells by western blotting, and could reverse the increase in ERK protein phosphorylation and EMT induced by rhMFGE8 (Fig. 6E). This suggested that MFGE8-MAPK/ERK pathway might be involved in the mechanism of DIM inhibiting GCPD.

DIM suppressed GCPD by targeting MFGE8 in vivo

In order to provide additional evidence for the in vivo impacts of MFGE8 and DIM on GCPD, a GCPD mice model was created by utilizing MKN-45P cells. Treatment with DIM reduced the number of abdominal metastasis and reversed the increase in abdominal metastasis caused by rhMFGE8 (Fig. 7A). In addition, immunohistochemical results showed that DIM could inhibit MFGE8 and EMT, and reverse the promotion of EMT caused by rhMFGE8 (Fig. 7B). These results further confirmed that DIM could inhibit MFGE8 in GC and reverse EMT in GC cells, thereby inhibiting GCPD.

Fig. 7.

DIM suppressed GCPD by targeting MFGE8 in vivo. A, Four groups of mice were injected with MKN-45P cells and exhibited abdominal metastases. B, IHC was employed to determine the expression of MFGE8 and EMT-related proteins (Scale bar: 5 μm) (* P<0.05, ** P<0.01, *** P<0.001, **** P<0.0001)

Discussion

Peritoneal dissemination is common in metastatic or recurrent GC, which is a form of metastasis that is inoperable and lacks effective treatment [53]. Despite the availability of drugs that specifically target molecules like HER2, VEGFR, and PD-1, as well as immune checkpoint inhibitors, these therapies are not effective for all patients [53]. Therefore, the exploration of molecules and drugs associated with GCPD is a crucial aspect of improving the therapeutic efficacy of GC.

To identify novel biomarker candidates for GCPD, we conducted an extensive online database screening, which led to the identification of MFGE8. A recent investigation has demonstrated that MFGE8 exhibits marked overexpression in gastric carcinoma, with elevated expression levels showing statistically significant correlation with diminished overall survival outcomes [34]. Through a comprehensive bioinformatics analysis of gene expression data retrieved from TCGA and GEO databases, in addition to IHC analysis of collected GC patient tissues, it was confirmed that MFGE8 protein expression was upregulated in GC tissues, and increased MFGE8 expression was correlated with poor patient prognosis and an increased risk of peritoneal dissemination. The results indicated that MFGE8 had the potential to be utilized as both a diagnostic biomarker and a therapeutic target for GCPD.

The progression of peritoneal dissemination can be divided into a series of interconnected steps, which are summarized as follows: cancer cells detachment from the primary tumor; migration of these cells to the peritoneal cavity; engagement of cancer cells with peritoneal mesothelial cells; and tumor expansion accompanied by the formation of blood vessels [54–56]. Indeed, the essence of GCPD is the process of cellular adhesion, migration, invasion, EMT, and angiogenesis [57]. Preclinical evidence has revealed that MFGE8 functionally augments cellular proliferation, potentiates EMT program, and drives tumorigenic growth in gastric carcinoma through IL-6/JAK/STAT3 axis activation [31]. Similarly, our study showed that MFGE8 promoted proliferation, adhesion, migration, and invasion of GC cells. MFGE8 has been found to activate MAPK in luminal and myoepithelial cells in developing epithelial ducts [58], and to promote ERK phosphorylation in liver injury [59], diabetic atherosclerosis formation [60], and human differentiated neutrophils [61]. Additionally, MFGE8 has been shown to trigger EMT in breast cancer [29], colorectal cancer [30], and melanoma [14]. In this study, we observed that MFGE8 might contribute to the progression of EMT in GC by altering phosphorylation levels of the MAPK/ERK signaling pathway, which was supported by both bioinformatics analysis and western blotting. Past studies have shown that the MAPK/ERK signaling pathway is implicated in the regulation of EMT in GC [62]. Our study further refined the mechanism by which MFGE8 promotes GCPD. Nevertheless, the mechanism by which MFGE8 triggers the MAPK/ERK signaling pathway requires further investigation.

In addition, we identified DIM as a compound inhibiting MFGE8 through CMap screening. Previous research has indicated that DIM may exhibit anti-GC effects [35, 45–49, 63, 64]. Experimental studies have delineated that DIM triggers apoptosis and autophagy in gastric carcinoma cells through STIM1-regulated store-operated calcium entry [35]. Additionally, DIM exerts potent anti-neoplastic effects by activating intrinsic apoptotic pathways via TRAF2-dependent p38 MAPK hyperphosphorylation [46]. Notably, systematic interrogation of kinase cascades confirms the pivotal involvement of MAPK signaling axis in mediating DIM’s therapeutic efficacy. Although multiple investigations have been conducted, there is no available information regarding the impact of DIM on GCPD. Thus, we first demonstrated by molecular docking that DIM could bind to MFGE8. Moreover, the efficacy of DIM on MFGE8 and GCPD was further validated through both cell and animal experiments, elucidating its underlying mechanism. The study findings suggested that MFGE8-MAPK/ERK pathway might be involved in the mechanism of DIM inhibiting GCPD. Emerging evidence from published pharmacological studies has established the biphasic oncological behavior of DIM, wherein 30µM exerts tumor-suppressive effects on gastric cancer cells while subtherapeutic levels (1–10µM) paradoxically enhance neoplastic progression [65]. Notably, our empirical screening of concentrations (all > 20µM) converged with existing reports describing antitumor efficacy at higher doses, thereby independently validating the therapeutic window for suppressing both primary tumor expansion and peritoneal metastasis in gastric malignancies. This biphasic pharmacological behavior—spanning pro-oncogenic to tumor-suppressive effects—mandates rigorous dose stratification in preclinical development, particularly when considering DIM’s potential repurposing for adjuvant intraperitoneal therapies.

In conclusion, the results of this investigation indicated that MFGE8 was highly expressed in GC, potentially serving as a biomarker and therapeutic target for GCPD. The MAPK/ERK signaling pathway was regulated by MFGE8, playing a vital part in promoting GCPD. On the other hand, MFGE8-MAPK/ERK pathway might be involved in the mechanism of DIM inhibiting GCPD. This study provided more evidence for the target prediction and clinical treatment of GCPD.

Author contributions

Jiaqing Liu: Writing - original draft, Writing - review & editing, Data curation, Methodology, Software, Validation, Visualization. Yuzhe Zhang: Writing - original draft, Data curation, Methodology, Software, Visualization. Lirong Yan: Data curation, Methodology, Validation, Visualization. Fang Li: Methodology, Validation, Visualization. Aoran Liu: Methodology. Chunjiao Yang: Methodology. Yi Yang: Methodology, Supervision. Yanke Li: Project administration. Ye Zhang: Funding acquisition, Project administration, Resources, Supervision.

Funding

This work was supported by National Natural Science Foundation of China [No. 82073244]; Shenyang Youth Science and Technology Innovation Talent Project [RC200267].

Data availability

Not applicable.

Declarations

Ethics approval and consent to participate

This study was approved by the Human Ethics Review Committee of the First Hospital of China Medical University and followed the Helsinki Declaration. All patients provided written informed consent. The animal experiments in this study have passed the animal welfare and ethical review of China Medical University application for laboratory.

Competing interests

The authors declare no competing interests.