The human gastric mucosal epithelial cell (GES-1; cat. no. AW-CNH199), human GC cell lines HGC27 (cat. no. AW-CNH127) and MKN74 (cat. no. AW-CCH277), mouse forestomach carcinoma (MFC; cat. no. AW-CCM404) and tohoku hospital pediatrics-1 (THP-1; AW-CCH098) were purchased from Abiowell (https://www.abiowell.com/). All plasmids were purchased from Honor Gene Co., Ltd. These included TBX15 silencing (si-TBX15; cat. no. HG-SH152380), TBX15 overexpression (oe-TBX15; cat. no. HG-HO152380), matrix metalloproteinase 14 (MMP14) overexpression (oe-MMP14; cat. no. HG-HO004995), METTL3 silencing (si-METTL3#1-3; cat. no. HG-SH019852), MMP14-wt (cat. no. HG-YO020961) and MMP14-mut (cat. no. HG-YO020961M). Male C57BL/6 mice (6 weeks) were ordered from Hunan SJA Laboratory Animal Co., Ltd. TRIzol (Thermo Fisher Scientific, Inc.), mRNA reverse transcription kit (CWBIO), and UltraSYBR Mixture (CWBIO) were used for reverse transcription-quantitative PCR (RT-qPCR) analysis. 5-Ethynyl-2′-deoxyuridine (EdU) kit (cat. no. C10310) was purchased from Guangzhou RiboBio Co., Ltd. Annexin V-APC assay kit (Nanjing KeyGen Biotech Co., Ltd.) and TUNEL assay kit (cat. no. 40306ES50; Yeasen Biotechnology) (https://www.yeasen.com/) were used to detecT cell death. The inducible nitric oxide synthase (iNOS; cat. no. CSB-E08326m), IL-1β (cat. no. CSB-E08054m), TNF-α (cat. no. CSB-E04741m), Arg-1 (cat. no. CSB-EL002005MO), IL-10 (cat. no. CSB-E04594m) and CD206 (cat. no. CSB-EL014782MO) kits were purchased from Wuhan Huamei Biotech Co., Ltd. The m6A assay kit (cat. no. P-9005; EpiGentek) was used to analyze m6A detection. MeRIP-qPCR kit (cat. no. 17-700) and Actinomycin D (cat. no. A4262) were purchased from MilliporeSigma. A dual luciferase assay kit (cat. no. E1910) was purchased from Promega Corporation. All antibody information is shown in Table SI.

The RNA sequencing profiles and clinical data for 33 cancerous from the Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/) and Genotype-Tissue Expression (GTEx) cohorts (GTEx Analysis Release V8, https://gtexportal.org) were retrieved from the University of California Santa Cruz Xena database (https://xenabrowser.net/datapages/). Univariate Cox regression (uniCox) and Kaplan-Meier analyses were performed. The R packages 'survminer' and 'survival' visualized the results, including overall survival (OS), disease-specific survival (DSS), disease-free interval (DFI) and progression-free interval (PFI) of patients with GC. JASPAR online software (https://jaspar.elixir.no/) was used to predict potential downstream targets of transcription factors. SRAMP online software (http://www.cuilab.cn/sramp) predicted m6A modification sites on RNA. RM2Target (http://rm2target.canceromics.org/#/home) predicted m6A-associated enzymes that bind to target genes.

GC tissue samples (n=10) and corresponding paracancerous tissue samples were collected. All samples were obtained from patients with GC in Zhuzhou Hospital Affiliated to Xiangya School of Medicine, Central South University between May 2024 and May 2025. The procedures followed in the present study were approved (approval no. ZZCHEC2022070-01) by the Ethics Committee (Zhuzhou, China) and performed followed Declaration of Helsinki Guideline. Written informed consent was obtained from each enrolled subject. The baseline information statistics of patients are shown in Table SII.

Cells were cultured in RPMI-1640 containing 10% fetal bovine serum (both from Gibco; Thermo Fisher Scientific, Inc.) and 1% penicillin/streptomycin. The cell incubator environment was maintained at 37°C with a humidified atmosphere containing 5% CO₂.

To explore the effects of TBX15, MMP14 and METTL3 on GC cells, plasmid intervention experiments of si-TBX15, oe-TBX15, oe-MMP14 and si-METTL3#1-3 were performed. The corresponding plasmids and negative control (si-NC and oe-NC) were transfected into cells. si-TBX15#1-3 sequences are as follows: si-TBX15#1, 5′-GAC AAT AAA AGA TAC AGA TAT-3′; si-TBX15#2, 5′-GGG GTG AAA ACG TTC AAC TTT-3′; si-TBX15#3, 5′-GAC ATA TAC CCA AGA ACA AGA-3′. si-METTL3#1-3 sequences are as follows: si-METTL3#1, 5′-GAG TTG ATT GAG GTA AAG CGA-3′; si-METTL3#2, 5′-TAG AGC TAT TAA ATA CTA CAA-3′; si-METTL3#3, 5′-GAC GAA TTA TCA ATA AAC ACA-3′. si-NC sequence is 5′-TGC GGG GCT AGG GTC CAA CGG-3′.

A co-culture system of GC cell-TAM cells was constructed. The THP-1 was resuspended in RPMI-1640 medium and seeded in 6-well culture plates at a density of 7.5×10⁵ cells/well. THP-1 cells were treated with phorbol 12-myristate 13-acetate (PMA; 320 nmol/l) for 6 h and then co-cultured with tumor cells for 24 h. After washing the cells with PBS, they were resuspended in serum-free RPMI-1640 medium. Macrophages were harvested, and flow analysis was performed.

A co-culture system of MFC cell-CD8⁺ T cells was established. Spleen-derived CD8⁺ T cells were isolated using MojoSort™ Mouse CD8⁺ T cell Isolation kit (BioLegend, Inc.). Isolated CD8⁺ T cells were activated with Ultra-LEAF™ purified anti-mouse CD3/CD28 (BioLegend, Inc.) for 3 days following the manufacturer's protocol. IL-2 (10 ng/ml) was added to RPMI-1640 medium for experiments. Transfected MFC cells were co-incubated with activated CD8⁺ T cells for 48 h. The ratio of tumor cells to CD8⁺ T cells was 1:1-1:20. CD8⁺ T cells were collected and analyzed for granzyme B (GZMB), perforin (PFP), CD100, IFN-γ and TNF-α expression.

A total of 48 C57BL/6 mice (male, 6 weeks old, ~20 g) were maintained under pathogen-free conditions for 1 week. Mice were raised on a 12 h/12 h light/dark cycle (22~25°C) with ad libitum access to water and food. All animal experiments were conducted in accordance with animal care, animal welfare and ethics. It has been approved by the Institutional Animal Care and Use Committee (IACUC) of the Second Xiangya Hospital, Central South University (approval no. 2022722).

Experiment 1: Mice were randomly divided into 3 groups (n=6): Control group, sh-NC group and sh-TBX15 group. Mice in the sh-NC and sh-TBX15 groups were injected with 1×10⁶ MFC cells (100 µl, resuspended in PBS) that had been transfected with sh-NC and sh-TBX15 plasmids. At room temperature, 3 µg of sh-NC and sh-TBX15 plasmids were transfected into MFC cells by using Lipofectamine™ 2000 (Invitrogen; Thermo Fisher Scientific, Inc.). At 48 h after transfection, cells were collected immediately for injection. The injection site was the right underarm. MFC cells injected in the Control group were not treated. The body weight and tumor were observed twice a week, and the tumor volume was measured with vernier calipers. The tumor volume was calculated by the formula V=(L × W²)/2, where V was the volume (mm³), L was the maximum diameter (mm), and W was the minimum diameter (mm). The experiment was finished 21 days after tumor seeding, and animals were euthanized by intraperitoneal injection of 200 mg/kg sodium pentobarbital solution. Samples were received and images were captured after euthanasia. sh-TBX15 sequence is 5′-GAC AAT AAA AGA TAC AGA TAT-3′ and sh-NC sequence is 5′-TGC GGG GCT AGG GTC CAA CGG-3′.

Experiment 2: Mice were randomly divided into 5 groups (n=6): Control group, sh-NC group, sh-METTL3 group, sh-METTL3 + oe-NC group, sh-METTL3 + oe-TBX15 group. Consistent with the aforementioned intervention procedure, MFC cells subjected to sh-METTL3 and oe-TBX15 interventions were injected into the right armpit of mice. After 21 days, mice were euthanized by intraperitoneal injection of sodium pentobarbital solution (200 mg/kg) and tumors were collected. sh-METTL3 sequence is 5′-GAC GAA TTA TCA ATA AAC ACA-3′. sh-NC sequence is 5′-TGC GGG GCT AGG GTC CAA CGG-3′.

To detect gene expression in tissues and cells, qPCR analysis was performed (38). Total RNA was extracted from GC tissues, mouse tumor bodies, and cells using TRIzol reagent, and the concentration was determined by ultraviolet spectrophotometer. cDNA was synthesized using an mRNA reverse transcription kit according to the manufacturer's instructions. qPCR was applied to detect the mRNA levels using UltraSYBR Mixture. PCR reaction conditions were 95°C for 10 min, and 95°C for 15 sec, and 60°C for 30 sec, for 40 cycles. Melting curve analysis was performed with a temperature gradient of 60-95°C. The primers were designed by primer5 (https://premierbiosoft.com/) and synthesized by the TsingKe Biological Technology. The relative mRNA levels of the target genes were determined by the 2^−ΔΔCq method (39) and normalized to β-actin (as an endogenous control). All primer sequences are included in Table SIII.

To determine the protein levels of TBX15, MMP14, Granzyme B, Perforin, CD100, IFN-γ, TNF-α and METTL3, western blotting was performed. Referring to a previous study (40), total protein was extracted from GC tissues and cells using the RIPA lysis buffer (Beyotime Institute of Biotechnology). A BCA assay kit was performed to detect protein concentrations. Equal amounts of protein (15-20 µg) were separated on a 10% SDS-PAGE gel and then transferred to a nitrocellulose membrane. After 5% skim milk was used for incubation and blocking for 90 min at room temperature, freshly prepared specific primary antibodies were used to incubate the membranes at 4°C overnight. After incubation with the secondary antibody (90 min, room temperature) was completed, ECL chemiluminescence solution was added to the membrane, followed by imaging using a gel imaging system (ChemiScope6100, Clinx). β-actin was applied to normalize the target protein levels. Relative expression level of the target protein=gray value of target protein/gray value of internal reference (β-actin).

After the mouse tumor and GC tissues were fixed in paraffin for 24 h at room temperature, the samples were sectioned (2 µm) (25). After deparaffinization with xylene, it was then rehydrated with graded ethanol series. The sections were immersed in 0.01 M sodium citrate antigen repair solution (pH 6.0) and heated in a microwave oven for thermal antigen repair. Incubation with 1% periodate for 15 min was used to inactivate endogenous enzymes. After incubation with specific primary antibodies (4°C overnight) anti-TBX15, anti-Ki67, anti-PD-L1, anti-CD100, anti-IFN-γ and anti-TNF-α, sections were subjected to secondary antibody incubation for 30 min at 37°C. After DAB color development, hematoxylin counterstain was performed, after which section images were randomly acquired. Data are presented as the mean IOD and positive rate. Positive rate was calculated as follows: (Number of positive cells/the number of total cells) ×100%.

To detect the proliferative activity of GC cells, Cell Counting Kit-8 (CCK-8) assay (Dojindo Laboratories, Inc.) and 5-Ethynyl-2′-deoxyuridine (EdU) analysis were performed. For CCK-8 assay, 10 µl CCK-8 reagent was added to prepared cells (5×10³ cells) and incubated for 4 h at 37°C, followed by detection of absorbance values (OD values) in each well at 480 nm.

The EdU cell proliferation kit was performed. Cells were seeded into a culture medium containing EdU reagent and incubated overnight. Cells were fixed by incubation with 4% paraformaldehyde for 30 min at room temperature. After incubation with 100 µl of 1X Apollo^® reaction solution for staining, 1X Hoechst33342 reaction solution was added for staining (30 min at room temperature). Images were captured using a fluorescence microscope. Cell proliferation rate was calculated as follows: (Number of Edu-positive cells/total number of cells) ×100%.

To analyze the level of apoptosis, an Annexin V-APC assay kit was performed (25). Prepared cells were harvested. The Annexin V-APC and Propidium Iodide reagents were added. Reaction was performed at room temperature and in the dark for 10 min before analysis using flow cytometry (CytoFLEX; Beckman Coulter, Inc.). Data were analyzed with the use of CytExpert software (version 2.5; Beckman Coulter, Inc.). Apoptotic rate (%) was calculated as follows: Early apoptotic rate (%) + late apoptotic rate (%).

To analyze the polarization of M1 and M2 macrophages, CD86 and CD206 positive cells were identified using flow cytometry. After digestion and washing of the cells, anti-CD86 or anti-CD206 antibodies were added. FCM was used for analysis after 30 min of staining in the dark. Percentage of positive cells (%) was calculated as follows: (Cells expressing a marker/targeT cell population) ×100%.

GC cells were seeded in 6-well culture plates until a monolayer was formed. A micropipette tip (200 µl) was used to generate the wound. Sterile PBS was used to wash and remove suspended cells. Image of the scratch was captured at 0 h. After incubation in serum-free Dulbecco's Modified Eagle Medium (DMEM) medium for 24 and 48 h, images were captured again for recording. A total of 3 visual fields were taken at each time point.

Cell invasion ability was determined by performing Transwell assays (41) with 8-µm-pore membranes. Stably transfected GC cells (2×10⁶ cells/well) were seeded into the upper chamber, and 10% fetal bovine serum culture medium was added to the lower chamber. After culturing at 37°C for 48 h, cells on the upper surface of the filter were removed by wiping with a cotton ball. Next, the filter was fixed with 4% paraformaldehyde for 20 min and stained with 0.1% crystal violet for 5 min at room temperature. A total of three randomly selected fields were counted under a light microscope.

To detect apoptosis in the tumor tissue, TUNEL staining was performed using a TUNEL assay kit according to the manufacturer's instructions. DAPI working solution (5 µg/ml) was used to stain the nuclei. A fluorescence microscope was used to view the images. A total of 6 fields of view were randomly selected for assessment. Data are presented in the form of positive rates. Positive rate was calculated as follows: (Number of positive cells/the number of total cells) ×100%.

After total protein extraction from mouse tumor bodies, the levels of iNOS, IL-1β, TNF-α, Arg-1, IL-10 and CD206 were measured by commercial kits according to the manufacturer's instructions. An absorbance (OD) value at 450 nm was used to quantify protein levels. Final sample concentration equals measured concentration/protein concentration.

A dual luciferase assay kit was used to detect targeting between TBX15 and MMP14 according to manufacturer's instructions. The MMP14-wt, MMP14-mut and TBX15 overexpression plasmid were transiently transfected into cells using Lipofectamine™ 2000 (Invitrogen; Thermo Fisher Scientific, Inc.). A reference reporter plasmid containing Renilla luciferase (pRL-TK) was co-transfected into the cells to regulate the transfection efficiency. After 48 h, cells were harvested and lysed for 15 min. Cell lysates were obtained for analysis of luciferase activity. The MMP14-mut sequence is: 5′-ACC ACA CT-3′.

ChIP-qPCR was performed to analyze the targeting relationship between TBX15 and MMP14 promoters. Immunoprecipitation was performed using anti-TBX15 and IgG. qPCR was performed to quantify the immunoprecipitated DNA. The primers used for ChIP-qPCR are shown in Table SI.

The m6A assay kit was performed to detect the global m6A levels. The cells and tissues were harvested and total RNA was extracted. The working solution of anti-m6A was used to incubate the samples for 30 min. The reaction was performed with the developer solution at room temperature in the dark for 10 min. After the terminator stop solution for 10 min, the m6A level was quantified by observing the absorbance value (OD values) at 450 nm. m6A (ng)=[Sample OD -NC OD)/Slope, m6A %=(m6A Amount (ng)]/S ×100%. Slope is the slope of the standard curve, and S is the amount of input sample RNA in ng (200 ng).

A MeRIP-qPCR kit was applied to detect the m6A levels of the target genes (25). Total RNA was extracted from GC cells. Samples were coupled to protein A/G magnetic beads (40 µl) overnight in IP buffer containing anti-m6A and anti-IgG. m6A-modified TBX15 was enriched. qPCR was performed to detect TBX15 levels with m6A modification.

METTL3 and TBX15 mRNA binding was identified. The prepared cells were harvested and total protein was extracted. The TBX15 mRNA probe, which was used as a bait factor, was bound to streptavidin magnetic beads. Total protein was mixed with magnetic beads after probe binding and incubated for 60 min. After elution, proteins were collected and assayed for METTL3 protein expression.

GC cells were cultured in a culture medium containing Actinomycin D (5 µg/ml) for specific periods of time (42). Cells were collected for total RNA extraction. The stability of TBX15 mRNA was analyzed by measuring the mRNA level of TBX15 by qPCR.

All experimental data were presented as the mean ± standard deviation (SD). GraphPad Prism (version 9.0.0; Dotmatics) was employed to perform data analysis. The Shapiro-Wilk test was employed to analyze the normality of the data. The two-tailed t-test (data comparison between two groups), one-way ANOVA, and two-way ANOVA (data comparison between multiple groups) were applied. Post hoc tests for multiple group comparisons were performed using Tukey's multiple comparisons test. P<0.05 was considered to indicate a statistically significant for comparison between groups.

Based on TCGA and GTEx pan-cancer data, TBX15 expression was analyzed. TBX15 was enriched in cancer tissues in 21 tumors, including GC (TCGA-STAD) (Fig. S1A). In TCGA-STAD, TBX15 was significantly highly expressed in both unpaired and paired GC samples (Fig. 1A). Pan-cancer prognostic data for TBX15 is presented in Fig. S1B. TBX15 was a high-risk factor for GC (TCGA-STAD) in OS, DSS and PFI analyses (OS, P=0.048; DSS, P=0.007; PFI, P=0.003). Patients with GC with high TBX15 predicted shorter survival time (OS, P=0.013; DSS, P=0.00088; DFI, P=0.034; PFI, P<0.0001; Fig. 1B). Next, the TBX15 levels were analyzed in the collected GC tissues [TBX15 mRNA: 3.62±0.95; TBX15 (1G): 3.92±0.20; TBX15 (1H): 34.58±2.22] and their paracancerous tissues [TBX15 mRNA: 1.18±0.23; TBX15 (1D): 1.00±0.13; TBX15 (1E): 9.80±1.80] (Fig. 1C-E). TBX15 was highly expressed in GC. The aforementioned results proved that TBX15, which is associated with poor prognosis in patients with GC, was enriched in GC.

To further explore the involvement of TBX15 in GC cells, the expression of TBX15 was identified in GC cells (HGC27 and MKN74). GES-1 (TBX15 mRNA: 1.00±0.05; TBX15: 0.09±0.01) was used as control. TBX15 was highly expressed in HGC27 (TBX15 mRNA: 3.30±0.67; TBX15: 0.39±0.03) and MKN74 (TBX15 mRNA: 3.17±0.58; TBX15: 0.43±0.03) cells (Fig. 2A). The TBX15-silenced cells were successfully constructed in HGC27 and MKN74 (Figs. 2B and S2A). Proliferation ability decreased after silencing TBX15 (Fig. 2C and D; Fig. S2B and C). TBX15 silencing promoted GC cell apoptosis (Figs. 2E and S2D). The effect of TBX15 silencing intervention on GC cell migration was analyzed. At 24 and 48 h, the cell migration distance in the si-TBX15 group (HGC27: 934.1±253.1; MKN74: 934.1±177.2) was lower than the si-NC group (HGC27: 788.8±386.5; MKN74: 836.4±251.8) (Figs. 2F and S2E). Transwell was performed to analyze the invasion ability of GC cells. TBX15 silencing inhibited cell invasion (Figs. 2G and S2F). These data proved that knockdown of TBX15, which is highly expressed in HGC27 and MKN74, inhibited the proliferation, migration and invasion of GC cells.

Next, the effect of TBX15 on GC development was analyzed by tumor burden. MFC cells with or without sh-TBX15 intervention were subcutaneously injected into mice to induce tumor formation. The tumor volume (maximum diameter: 14.78 mm, maximum volume: 1181.46 mm³) and weight decreased after TBX15 silencing (Fig. 3A-C). TBX15 silencing reduced tumor burden in mice. The apoptotic rate increased, and Ki67 levels decreased (Fig. 3D-F). These evidence demonstrated that TBX15 silencing could inhibit GC development in vivo.

MMP14 is a membrane-anchored matrix metalloproteinase. Based on the patient data in TCGA-STAD, the expression of TBX15 and MMP14 was predicted to be positively correlated (Fig. 4A). MMP14 mRNA and protein levels increased in GC tissues, and TBX15 was positively correlated with MMP14 (Fig. 4B and C). JASPAR online prediction found that the transcription factor TBX15 might target the MMP14 promoter (Fig. 4D). A dual luciferase reporter gene was used to verify the activity of the MMP14 promoter. Overexpression of TBX15 increased MMP14 promoter activity (Fig. 4E). ChIP-PCR assay showed that TBX15 interacted with the MMP14 promoter (Fig. 4F). Overexpression of TBX15 was found to promote MMP14 mRNA and protein expression (Fig. 4G and H). These results suggested that the transcription factor TBX15 may target the MMP14 promoter to regulate its transcriptional activity.

The relationship between TBX15 and immune cells was further explored. In the GC microenvironment, macrophages were enriched, while CD8⁺ T cell infiltration level was significantly reduced (43). Macrophages play a crucial role in the innate immune system. Under different microenvironmental stimuli, macrophages can transform into two different subtypes (M1 and M2), which have completely different molecular phenotypes and functional characteristics. TAMs are macrophages that differentiate under the influence of various factors in the tumor microenvironment (TME) and have the characteristics and functions of M2-type macrophages. To analyze the role of TBX15 and MMP14 in GC immune escape, a co-culture system of GC cells and macrophages was constructed. After intervention with TBX15 silencing and MMP14 overexpression, GC cells were then co-cultured with macrophages (Fig. 5A; Fig. S3A and B). TBX15 silencing promoted the levels of M1-type TAM cells and decreased the M2-type levels (Fig. 5B and C; Fig. S3C and D). The gating strategy of Fig. S4A and B was used to gate the cell subsets. M1 macrophage markers (iNOS, IL-1β and TNF-α) were increased after TBX15 silencing, and the opposite was observed for M2 macrophage markers (Arg-1, IL-10 and CD206) (Fig. 5D and E; Fig. S3E and F). These results suggested that TBX15 silencing promotes TAMs toward an M1 proinflammatory phenotype. Overexpression of MMP14 decreased the levels of iNOS, IL-1β and TNF-α and increased Arg-1, IL-10 and CD206. This indicated that overexpression of MMP14 reduced the regulatory effect of TBX15 silencing on macrophage polarization.

In addition, CD8⁺ T cell-associated cytokines GZMB, PFP, CD100, IFN-γ and TNF-α were identified in the co-culture system constructed by MFC cells and CD8⁺ T cells. In the TBX15-silenced co-culture system, GZMB, PFP, CD100, IFN-γ and TNF-α levels increased. Silencing TBX15 enhanced CD8⁺ T cell activation and antitumor immunity. Overexpression of MMP14 reduced the levels of GZMB, PFP, CD100, IFN-γ, and TNF-α (Fig. 5F and G).

In summary, TBX15 participates in the immune escape process of GC by promoting macrophage M2 polarization and inhibiting CD8⁺ T cell activation.

The SRAMP online website predicted that TBX15 might undergo m6A modification, and RM2Target predicted that TBX15 would bind to the m6A methyltransferase METTL3 (Fig. S5A and B). In GC, METTL3 expression and total m6A levels were increased (Fig. 6A-C). METTL3 expression and total m6A level increased in HGC27 (METTL3 mRNA: 4.02±0.34; METTL3: 0.44±0.03; m6A%: 0.12±0.01) and MKN74 (METTL3 mRNA: 3.68±0.38; METTL3: 0.48±0.02; m6A%: 0.12±0.01) compared with GES-1 (METTL3 mRNA: 1.00±0.11; METTL3: 0.04±0; m6A%: 0.07±0.02) (Fig. 6D-F). To further explore the regulatory relationship between METTL3 and TBX15, METTL3 silenced cells were constructed (Figs. 6G and S5C). After silencing METTL3, the m6A level of TBX15 was decreased (Figs. 6H and S5D). Results proved that METTL3 binds to TBX15 mRNA (Figs. 6I and S5E). METTL3 silencing reduced TBX15 mRNA stability (Figs. 6J and S5F). These results indicated that silencing METTL3 inhibits the m6A level of TBX15 and reduces TBX15 mRNA stability.

To explore whether METTL3 regulates GC cell function and development through the TBX15/MMP14 signaling axis, METTL3 silencing and TBX15 overexpression cells were constructed. METTL3 silencing inhibited MMP14 expression, and TBX15 overexpression attenuated the inhibitory effect of METTL3 silencing on MMP14 expression (Figs. 7A and S6A). METTL3 silencing decreased proliferation activity and increased apoptotic rate of GC cells (Fig. 7B-D; Fig. S6B-D). Overexpression of TBX15 increased proliferation and decreased apoptosis. Overexpression of TBX15 reduced the inhibitory effect of METTL3 silencing on migration (Figs. 7E and S6E). Under the intervention conditions of METTL3 silencing and TBX15 overexpression, cell invasion was altered, consistent with migration (Figs. 7F and S6F).

Next, the role of METTL3/TBX15/MMP14 signaling axis in GC cell immune escape was further analyzed. Silencing METTL3 promoted the polarization of TAMs to M1 macrophages in the co-culture system of GC cell-macrophages (Fig. 7G and H; Fig. S6G and H; Fig. S7A and B). Levels of iNOS, IL-1β and TNF-α increased, and Arg-1, IL-10 and CD206 decreased (Fig. 7I and J; Fig. S6I and J). Overexpression of TBX15 reduced the effect of METTL3 silencing. In the co-culture system of MFC cell-CD8⁺ T cells, the expression of GZMB, PFP, CD100, IFN-γ and TNF-α increased after METTL3 silencing. Overexpression of TBX15 inhibited the regulation of METTL3 silencing on CD8⁺ T cell-related cytokines GZMB, PFP, CD100, IFN-γ and TNF-α (Fig. 7K). METTL3 regulated proliferation, migratory activity and immune escape process through TBX15/MMP14 signaling axis in GC.

The effect of METTL3/TBX15/MMP14 signaling axis on MFC cells' ability to induce tumorigenesis in mice was further analyzed. MFC cells with METTL3 silencing and TBX15 overexpression were injected subcutaneously into mice. After 21 days, tumor volume and weight were reduced after METTL3 silencing. Overexpression of TBX15 attenuated the effect of METTL3 silencing, and the tumor weight and volume (maximum diameter: 14.99 mm, maximum volume: 1,385.54 mm³) increased (Fig. 8A-C). Levels of METTL3, TBX15 and MMP14 were reduced after METTL3 silencing. Overexpression of TBX15 promoted TBX15 and MMP14 levels but had no significant regulatory effect on METTL3 (Fig. 8D and E). In the tumor, the apoptotic rate increased and the expression of Ki67 decreased after METTL3 silencing. Overexpression of TBX15 decreased the apoptotic rate and increased the expression of Ki67 (Fig. 8F and G). Similarly, the expression of M1 macrophage markers (iNOS, IL-1β and TNF-α) was increased after METTL3 silencing, and overexpression of TBX15 reduced the effect of METTL3 silencing. Levels of M2 macrophage markers (Arg-1, IL-10 and CD206) were opposite (Fig. 8H). Overexpression of TBX15 inhibited the regulation of METTL3 silencing on CD8⁺ T cell-related cytokines GZMB, PFP, CD100, IFN-γ and TNF-α (Figs. 8I and S8). The aforementioned results suggested that silencing METTL3 inhibits tumor development, via the TBX15/MMP14 signaling axis.