We retrieved RNA-Seq data, CNV data, and clinical data of GC from TCGA database. TCGA provides comprehensive data for more than 20,000 primary cancer samples spanning 33 different cancer types. Gene sets related to the EMT and angiogenesis were obtained from the Molecular Signatures Database (MsigDB), which offers resources of annotated gene sets for use. TCGA data were in count format and standardized as log${}_2$ (count+1) values.

The gene set variation analysis package in R was used to conduct ssGSEA analysis of the RNA-Seq data obtained from 375 tumor samples and 32 paracancer samples. Subsequently, the EMT and angiogenesis indexes for each sample were calculated. According to the median, the samples were segmented into low/high EMT and low/high angiogenesis index groups. Boxplots illustrating the EMT and angiogenesis indexes were generated using the ggplot2 in R. Correlation analyses between the EMT index and angiogenesis index were carried out using the cor function in R. Survival analyses were conducted using the survival package in R.

The DESeq2 package (Version 4.2) in R was applied for screening DEGs in the high and low angiogenesis index groups. DEGs meeting the criteria of p 0.05 and $|$log${}_2$FC$|$ $>$1 were deemed statistically significant. The EnhancedVolcano package (Version 1.11.3) in R was used to generate volcano plots illustrating the observed differences. Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) enrichment analyses were conducted using the clusterProfiler (version 4.2.2, https://yulab-smu.top/biomedical-knowledge-mining-book/ (docs) https://doi.org/10.1016/j.xinn.2021.100141 (paper)) package in R. Enrichment results were analyzed using the ggplot2 (version 3.3.6, https://ggplot2.tidyverse.org, https://github.com/tidyverse/ggplot2) package in R [23].

The CNV data and angiogenesis index of the samples were integrated to obtain the segment data of 381 samples. Copy number at the gene level was determined using GISTIC 2.0 (version 7, https://www.genepattern.org/modules/docs/GISTIC_2.0), with a threshold set to 0.3. Fisher’s exact test was employed to assess the hypothesis of gene copy number distribution between samples with high and low angiogenesis, and identify genes with significant differences in copy number between the low and high angiogenesis groups. Then the CNV-driven DEGs were obtained by Pearson’s correlation analysis for copy number and expression of genes. Combined with the clinical data, prognosis-related genes were identified by performing univariate Cox regression analysis of the CNV-driven DEGs using the survival package in R. PPI analysis was carried out using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) version 11.5, and K-M analysis was conducted using the Gene Expression Profiling Interactive Analysis database.

During GC resection, specimens of 1 cm${}^3$ were obtained from both the GC tissue and paracancer tissue. The surgically resected tissue specimens were fixed in 4% formaldehyde solution, and preserved as a wax block.

AGS GC cells (#CRL-1739) were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured in F-12K medium (#30-2004; ATCC) supplemented with 10% fetal bovine serum (FBS). Human umbilical vein endothelial cells (HUVECs) were purchased from ATCC (#PCS-100-013), and cultured in Dulbecco’s Modified Eagle Medium containing 10% FBS and 1% penicillin/streptomycin. GES-1 gastric epithelial cells were purchased from Cobioer Biosciences Co., Ltd. (#CBP60512; Nanjing, China) and cultured in DMEM (Cobioer) supplemented with 10% FBS. HGC-27 GC cells were purchased from Cobioer (#CBP60480) and inoculated in RPMI-1640 medium (Cobioer) supplemented with 10% FBS. All cells used in the experiments were cultured in an incubator at 37 °C with 5% CO${}_2$. All cell lines were validated by short tandem repeat profiling and tested negative for mycoplasma. The Ras-associated protein 1 (Rap1)/mitogen-activated protein kinase (MAPK) signaling pathway activator, staurosporine aglycone (SA), was purchased from Sigma (#S3939; St. Louis, MO, USA), and the cells were subjected to pre-treatment with 30 µmol/L SA for 2 h prior to the commencement of subsequent experiments.

Cells were seeded in 12-well plates (serum-free medium) at a density of 1.2 $\times$ 10${}^5$ per well and transfected with F2RL3 overexpressed plasmid using jetPRIME® Transfection Reagent (Polyplus, Illkirch-Graffenstaden, France). Cells were seeded in 12-well plates (serum-free medium) at a density of 1.2 $\times$ 10${}^5$ per well and transfected with small interfering RNA (siRNA) oligonucleotide targeting F2RL3 (Sangon Biotech, Shanghai, China) using jetPRIME® Transfection Reagent (Polyplus, Illkirch-Graffenstaden, France) for 6 h. Subsequently, medium supplemented with 10% FBS was added for incubation for 24 h, followed by the detection of F2RL3 expression by quantitative polymerase chain reaction (qPCR). After 72 h, F2RL3 protein expression was detected by Western blotting.

Proteins were extracted from the cell lysate supplemented with 1% protease inhibitor. The protein content in the sample was quantified using the bicinchoninic acid assay, followed by electrophoresis, electrotransfer of proteins to a membrane, and incubation with the following primary antibodies: F2RL3 (#ab137927; Abcam, Cambridge, MA, USA), E-cadherin (#ab40772; Abcam), vimentin (#ab92547; Abcam), Rap1 GTPase-activating-protein (Rap1GAP) (#ab32373; Abcam), Rap1 (#ab175329; Abcam), phosphorylated extracellular signal-regulated kinase 1/2 (p-ERK1/2) (#4370T; Cell Signaling Technology [CST], Danvers, MA, USA), ERK1/2 (#9194S; CST), p-p38 MAPK (#4511S; CST), p38 MAPK (#8690S; CST), and Snail (#3879S; CST). $\beta$-actin (#93473SF; CST) was selected as the internal reference gene. After the proteins were developed, their expression levels were quantified with ImageJ software. Three biological replicates were performed for all experiments.

Total RNA extraction was performed using the TRIzol Kit (Thermo Fisher Scientific, Waltham, MA, USA), and the extracted RNA was treated with DNase. The PrimeScript RT Reagent Kit (TaKaRa, Shiga, Japan) was used to obtain the cDNA, and the cDNA template was diluted with RNase-free water. As per the instructions of the SYBR Green PCR Kit (Thermo Fisher Scientific), qPCR was performed in a total reaction volume of 10 µL. The qPCR conditions included pre-denaturation at 95 °C for 15 min, followed by 40 cycles of denaturation at 94 °C for 15 s, annealing at 55 °C for 30 s, and annealing at 72 °C for 30 s. The primers sequences were as follows: F2RL3 forward primer, AGAAGAGGAGAGGACACAGAGACAC; F2RL3 reverse primer, CTTGGCATCGTGGCATCCCTTAG; $\beta$-actin forward primer, CAGATGTGGATCAGCAAGCAGGAG; and $\beta$-actin reverse primer, CGCAACTAAGTCATAGTCCGCCTAG. The 2${}^{-\mathrm{\Delta }\mathrm{\Delta }\text{Ct}}$ method was used to calculate the obtained results. Three biological replicates were performed for all experiments.

Tissue sections were prepared through slicing, dewaxing, and hydration processes. Antigen retrieval and inactivation were performed, followed by serum blocking. Then the sections were incubated with F2RL3 primary antibody (#25306-1-AP; Proteintech, Rosemont, IL, USA), followed by incubation with secondary antibody. Diaminobenzidine (DAB) color development was carried out, and cell nuclei were counterstained with hematoxylin. Finally, the samples were dehydrated, cleared, and observed, with images captured under a microscope. Three biological replicates were performed for all experiments.

Cells in logarithmic growth phase were used to prepare a cell suspension, cultured for 24 h, and then treated with Cell Counting Kit-8 (CCK-8) reagent (C0038; Beyotime, Beijing, China) for an additional 2 h. The absorbance at 450 nm of each cell suspension was detected using an enzymoleter. The upper compartment of the membrane at the bottom of the Transwell chamber was coated with Matrigel. After solidification, 100 µL of the cell suspension was added to the lower compartment, followed by the addition of medium to the lower compartment. Following 24 h of routine culture in the cell incubator and staining with crystal violet, cell invasion was assessed by observing the number of cells in the visual field under a microscope. For the scratch experiment, a trace gun was used to mark the central region of cell growth in single-layer adherent cells cultured in vitro dishes. Cells in the central part were removed, and culture continued with serum-free medium. Cell migration was evaluated by observing whether the surrounding cells migrated to the central scratch area under a microscope. Three biological replicates were performed for all experiments.

The supernatant of GC cells with F2RL3 knockdown was collected and used to culture HUVECs. Precooled Matrigel matrix was added to 24-well plates, followed by the addition of 500 µL cell suspension of HUVECs co-cultured with GC cells into each respective well. Tube formation outcomes were assessed following routine culture in a cell incubator for 2 to 6 h. Tube formation and number were recorded using NIS-Elements BR software (Nikon, Melville, NY, USA). Three biological replicates were performed for all experiments.

Statistical analyses of the bioinformatics analysis results and graphing were completed using R, and between-group comparisons were performed using the Wilcoxon rank-sum test. For the experimental results, statistical analyses and graphing were completed using Graphpad Prism (version 9.4.0, Boston, MA, USA), and between-group comparisons were conducted using t-tests. All experiments at the cellular level were performed with three biological replicates. p 0.05 was considered statistically significant.

The EMT and angiogenesis indexes of 375 tumor samples and 32 paracancer samples of GC were analyzed and calculated. Results indicated a higher EMT index (Fig. 1A) and angiogenesis index (Fig. 1B) of the tumor samples compared to the paracancer samples. Further, a notable positive correlation was identified between the EMT index and the angiogenesis index (p = 2.2 $\times$ 10${}^{-16}$, R = 0.67; Fig. 1C). In addition, according to survival analyses, a statistically significant difference was revealed in the survival time between low and high angiogenesis indexes, with a poorer prognosis noted for those with a high angiogenesis index (Fig. 1D).

Fig. 1.

Analyses of the epithelial-mesenchymal transition (EMT) and angiogenesis indexes of the gastric cancer (GC) samples. (A) The EMT index between tumor and paracancer samples. (B) Angiogenesis index between tumor and paracancer samples. (C) Correlation analyses of the EMT and angiogenesis index. (D) Survival analyses between the low and high angiogenesis groups.

Analysis of the differential expression of genes showed 573 DEGs between the high and low angiogenesis samples, including 496 upregulated and 77 downregulated genes (Fig. 2A). These DEGs were mainly enriched in the phosphoinositide 3-kinase/protein kinase B pathway, integrin binding, phagocytosis, signaling receptor activator activity, humoral immune response mediated by circulating immunoglobulin, and extracellular matrix–receptor interaction as demonstrated by GO and KEGG enrichment analyses (Fig. 2B–E).

Fig. 2.

Enrichment analyses of differentially expressed genes (DEGs). (A) Volcanic map. (B) Gene Ontology (GO)-Biological Process. (C) GO-Cellular Component. (D) GO-Molecular Function. (E) Kyoto Encyclopedia of Genes and Genomes (KEGG). PI3K-Akt, phosphoinositide 3-kinase-protein kinase B; ECM, Extracellular matrix; IL-17, interleukin-17; AGE-RAGE, advanced glycation end products-receptor for advanced glycation end products.

The gene-level copy number was calculated, and the distribution of gene copy number between samples with low and high angiogenesis was tested. Genes with significant differences in copy number between the high and low angiogenesis groups were identified. Then correlation analysis of gene copy number and expression level was conducted, leading to the identification of 37 CNV-driven DEGs (copy number was significantly different and positively correlated with the expression level). Combined with the clinical data, univariate Cox analyses of these 37 genes revealed 16 prognosis-related genes. Axis inhibition protein 2 (AXIN2) demonstrated a hazard ratio (HR) less than 1, indicating its role as a protective factor; that is, higher AXIN2 expression was correlated with a higher survival rate. Conversely, the HRs of other genes exceeded 1, indicating that these genes were risk factors, where higher gene expression was associated with lower survival rate (Fig. 3).

Fig. 3.

Forest map of the prognosis-related genes. HR, hazard ratio; AXIN2, Axis inhibition protein 2; TENM3, teneurin transmembrane protein 3; DUSP1, dual specificity phosphatase 1; FBN1, fibrillin 1; THBS1, thrombospondin-1; LBP, lipopolysaccharide binding protein; MRAP, Melanocortin 2 Receptor Accessory Protein; ERG, E-twenty six transcription factor; GREM1, Gremlin 1; ADAMTS1, A disintegrin and metalloproteinase with thrombospondin motifs 1; PTGFR, prostaglandin F receptor; ADH1B, alcohol dehydrogenase 1B (class I), beta polypeptide; F2RL3, F2R like thrombin or trypsin receptor 3; CCDC178, coiled-coil domain containing 178; CYTL1, cytokine like 1; PCDHGA3, protocadherin gamma subfamily A, 3.

The 16 identified genes underwent PPI analysis using the STRING database (Fig. 4A), and KEGG analysis was performed using clusterProfiler (Fig. 4B and Table 1). KEGG analyses revealed the enrichment of fibrillin-1/thrombospondin 1/gremlin 1 in the TGFb signaling pathway, AXIN2/melanocortin 2 receptor accessory protein in Cushing’s syndrome, and thrombospondin-1/F2RL3 in the Rap1 pathway (p 0.05). Currently, there is a lack of substantial advancements in studies on the mechanism of the Rap1 pathway in GC, and the role of F2RL3 participating in this pathway in GC remains unexplored. Therefore, an in-depth investigation on F2RL3 was conducted in this study. PPI analysis revealed an interaction between F2RL3 and prostaglandin F receptor (Fig. 4A). While F2RL3 expression did not significantly differ between the tumor and normal tissues (p $>$ 0.05; Fig. 4C), it was notably upregulated in advanced GC (stages I, III, and IV) compared to early-stage GC (stage I) (Fig. 4D). As shown by K-M analyses, higher expression of F2RL3 was associated with lower survival rates (Fig. 4E).

Fig. 4.

Screening of key prognostic genes. (A) Protein–protein interaction (PPI) analysis. (B) KEGG analysis. (C) Expression of F2RL3 between tumor and normal tissues. (D) Expression of F2RL3 in GC stages. (E) Kaplan-Meier (K-M) analysis of F2RL3.

Western blotting, qPCR, and immunohistochemistry were used for the detection of F2RL3 expression in GC and paracancer tissues, with the results showing significantly lower F2RL3 expression in paracancer tissue compared to GC tissue. At the same time, according to the TNM stage, we divided the clinical samples into pre-tumor (stage I) and pro-tumor (stage II, III, IV) samples, and found that the expression of F2RL3 was significantly increased in the pro-tumor samples compared with the pre-tumor samples (Fig. 5A–D). Therefore, we speculated that the expression of F2RL3 can promote the malignant progression of tumors.

Fig. 5.

F2RL3 expression level in human GC tissue. (A) F2RL3 expression detected by Western blotting (n = 6). (B) Bar chart showing F2RL3 expression (n = 6). (C) F2RL3 expression detected by quantitative polymerase chain reaction (qPCR) (n = 6). (D) F2RL3 expression detected by immunohistochemistry (n = 6), Scale: 100 µm. **p 0.01.

qPCR and western blotting were used to detect F2RL3 expression in normal human gastric epithelial cells (GES-1) and human GC cells (HGC-27 and AGS) (Fig. 6A–C). The results showed significantly lower F2RL3 expression in GES-1 cells than in AGS and HGC-27 cells. Additionally, AGS cells exhibited higher expression of F2RL3 compared to HGC-27 cells.

Fig. 6.

F2RL3 expression level in human GC cells. (A) F2RL3 expression detected by Western blotting. (B) Bar chart showing F2RL3 expression. (C) F2RL3 expression detected by qPCR (n = 3). *p 0.05, **p 0.01.

To determine the effect of F2RL3 in GC cells, the siF2RL3 plasmid was transfected into AGS and HGC-27 cells to knockdown F2RL3 expression (Fig. 7A–C). The siRNA with the highest knockdown efficiency (AGS-siF2RL3 b and HGC-27-siF2RL3 b) was selected for subsequent experiments. The CCK-8 assay demonstrated a notable reduction in the activity of GC cells after knocking down F2RL3 expression (Fig. 7D). The Transwell (Fig. 7E,F) and scratch (Fig. 7G,H) assays demonstrated a notable reduction in the metastasis and invasion of GC cells upon knockdown of F2RL3 expression. At the same time, the regulation of F2RL3 expression led to upregulation of the epithelial marker E-cadherin and downregulation of the interstitial markers Snail and vimentin (Fig. 7I,J), indicating that F2RL3 can regulate the EMT of GC cells.

Fig. 7.

Effects of F2RL3 knockdown on the EMT of GC cells. (A) F2RL3 expression detected by Western blotting. (B) Bar chart showing F2RL3 expression. (C) F2RL3 expression detected by qPCR. (D) GC cell activity detected by the CCK-8. (E) GC cell invasion detected by the Transwell, Scale: 100 µm. (F) Bar graph showing the invasion of GC cells. (G) GC cell migration detected by the scratch assay, Scale: 100 µm. (H) Bar graph showing GC cell migration. (I) Expression of E-cadherin, Snail, and vimentin detected by Western blotting. (J) Bar chart showing E-cadherin, Snail, and vimentin expression (n = 3). *p 0.05, **p 0.01.

To investigate the role of F2RL3 in angiogenesis, we knocked down F2RL3 expression in AGS and HGC-27 cells, cultured the cells, and then collected the cell culture supernatants. HUVECs were cultured with GC cell culture supernatants. The activity of HUVECs in the AGS-siF2RL3 and HGC27-siF2RL3 groups was significantly decreased (Fig. 8A), and its metastasis and invasion abilities (Fig. 8B–E) as well as angiogenesis (Fig. 8F,G) were significantly attenuated (p 0.05).

Fig. 8.

Effects of F2RL3 expression knockdown on angiogenesis. (A) The activity of Human Umbilical Vein Endothelial Cells (HUVECs) detected by the CCK-8. (B) The invasion of HUVECs detected by the Transwell, Scale: 100 µm. (C) Bar graph showing the invasion of HUVECs. (D) The migration of HUVECs detected by the scratch assay, Scale: 100 µm. (E) Bar graph showing the migration of HUVECs. (F) Tube-forming ability of HUVECs detected by the tube formation assay, Scale: 100 µm. (G) Bar graph showing the tube-forming ability of HUVECs (n = 3). *p 0.05, **p 0.01.

F2RL3 is known to regulate the EMT and angiogenesis in GC, but its molecular mechanism is not fully understood. Thus, we investigated the impact of F2RL3 on the Rap1/MAPK signaling pathway. Western blot analysis of the expression of Rap1/MAPK pathway-related proteins (Rap1GAP, Rap1, p-ERK1/2, ERK1/2, p-p38 MAPK, and p38 MAPK) revealed the upregulated expression of Rap1GAP in the AGS-siF2RL3 group, accompanied by the downregulated expression of Rap1, p-p38 MAPK, and p-ERK1/2. In the AGS-oeF2RL3 group, the expression of Rap1GAP was downregulated, while that of Rap1, p-p38 MAPK, and p-ERK1/2 was upregulated (Fig. 9A,B). These results indicate that F2RL3 can regulate the Rap1/MAPK pathway.

Fig. 9.

F2RL3 regulates the expression of Rap1/MAPK signaling pathway in GC cells. (A) Expression of proteins related to F2RL3 and the Rap1/MAPK signaling pathway detected by Western blotting. (B) Bar graph showing the expression of proteins related to F2RL3 and the Rap1/MAPK pathway (n = 3). **p 0.01.

F2RL3 was identified as a regulator of the Rap1/MAPK signaling pathway, prompting further investigation into its influence on the EMT process of GC cells. Groups were set as AGS, AGS-siControl, AGS-siF2RL3, and AGS-siF2RL3-Rap1/MAPK signaling pathway activator (SA). Following knockdown of F2RL3 expression in AGS cells, E-cadherin expression increased, while Snail and vimentin expression decrease (Fig. 10A,B). Meanwhile, the activity (Fig. 10C), metastasis, and invasion of GC cells (Fig. 10D–G) was significantly decreased (p 0.05). Further treatment with SA showed weakened effect of siF2RL3, along with the downregulated expression of E-cadherin and upregulated expression of Snail and vimentin (Fig. 10A,B). Meanwhile, there was increased cell activity (Fig. 10C), and enhanced metastasis and invasion (Fig. 10D–G). These results collectively indicate that F2RL3 can regulate the EMT of GC cells through the Rap1/MAPK signaling pathway.

Fig. 10.

Effects of the Rap1/MAPK signaling pathway on the EMT of GC cells. (A) Expression of F2RL3, E-cadherin, vimentin, and Snail detected by Western blotting. (B) Bar chart showing the expression of F2RL3, vimentin, E-cadherin, and Snail. (C) Activity of GC cells detected by the CCK-8. (D) Invasion of GC cells detected by the Transwell, Scale: 100 µm. (E) Bar graph showing the invasion of GC cells. (F) Migration of GC cells detected by the scratch assay, Scale: 100 µm. (G) Bar graph showing the migration of GC cells (n = 3) (*p 0.05, **p 0.01).

HUVECs were cultured with GC cell culture supernatants collected from the AGS, AGS-siControl, AGS-siF2RL3, AGS-siF2RL3-SA groups, and its activity (Fig. 11A), metastasis, invasion (Fig. 11B–E), and angiogenesis (Fig. 11F,G) were assessed. The results showed that the activity of HUVECs in the AGS-siF2RL3 group was decreased (p 0.05), accompanied by attenuated metastasis and invasion, along with significantly reduced angiogenesis. However, HUVECs in the AGS-siF2RL3-SA group showed significantly enhanced cell activity, metastasis, invasion, and angiogenesis (p 0.05). Together, these results underscore the regulatory role of F2RL3 in angiogenesis through the Rap1/MAPK signaling pathway.

Fig. 11.

Impact of the Rap1/MAPK signaling pathway on angiogenesis. (A) Activity of HUVECs detected by the CCK-8. (B) Invasion of HUVECs detected by the Transwell, Scale: 100 µm. (C) Bar graph showing the invasion of HUVECs. (D) Migration of HUVECs detected by the scratch assay, Scale: 100 µm. (E) Bar graph showing the migration of HUVECs. (F) Tube forming ability of HUVECs detected by the tube formation assay, Scale: 100 µm. (G) Number of HUVEC tubules (n = 3). **p 0.01.