m6A‐methylated circFANCB Promotes Gastric Cancer Progression by Regulating Cellular Ferroptosis Through miR‐454‐3p/CEACAM5

ABSTRACT

Gastric cancer (GC) is a major global digestive malignancy with high incidence and mortality. Circular fanconi anemia complementation group B (circFANCB) expression is elevated in cancer, and its suppression suppresses tumor progression. However, research on the function of circFANCB in GC has not been reported yet. The role and mechanism of circFANCB in GC were investigated through circular RNA (circRNA) sequencing, database prediction, and experimental validation, including quantitative real‐time PCR (qRT‐PCR), ribonuclease R assay, nuclear‐cytoplasmic fractionation, 5‐ethynyl‐2′‐deoxyuridine assay, flow cytometry, wound healing assay, Transwell assay, tube formation assay, western blot, dual‐luciferase reporter assay, and enzyme‐linked immunosorbent assay. Furthermore, its mechanism was further explored in vivo using a xenograft nude mouse model and immunohistochemistry. Online prediction tools, methylation‐dependent RNA immunoprecipitation (MeRIP), and RNA immunoprecipitation (RIP) assays were employed to investigate whether methyltransferase‐like protein 3 (METTL3) targeted circular FANCB (circFANCB) via m6A methylation. In GC, circFANCB expression was upregulated. Knockdown of circFANCB inhibited cell proliferation, migration, invasion, and angiogenesis while promoting cell death. CircFANCB increased CEACAM5 expression by binding to miR‐454‐3p. CEACAM5 suppressed ferroptosis in GC cells. Rescue experiments confirmed that circFANCB inhibited ferroptosis by upregulating CEACAM5, thereby promoting GC cell proliferation. Animal models demonstrated that circFANCB knockdown inhibited GC tumor growth. Additionally, METTL3 enhanced circFANCB expression through m6A methylation. Rescue experiments further confirmed that METTL3 promotes the malignant phenotype of GC cells by upregulating circFANCB. The m6A‐modified circFANCB promotes the malignant progression of GC by regulating ferroptosis via the miR‐454‐3p/CEACAM5 axis, providing a theoretical basis for potential clinical therapeutic strategies.

This study reveals that METTL3 regulates circFANCB expression by inducing m6A modification. CircFANCB modulates cellular ferroptosis by regulating the miR‐454‐3p/CEACAM5 axis, thereby promoting the growth and metastasis of gastric cancer cells.

1. Introduction

Gastric cancer (GC) is one of the most serious tumors worldwide, exhibiting an elevated incidence rate and lethality rate that ranks among the highest for malignant tumors [1, 2]. The low survival rate of GC patients is primarily attributed to factors such as delayed diagnosis, tumor metastasis, intratumoral heterogeneity, and chemotherapy resistance, all of which contribute to poor overall prognosis [3]. Currently, there is no standardized treatment regimen for GC. For patients with locally advanced disease, perioperative chemotherapy can effectively improve survival rates and quality of life; meanwhile, the addition of novel anticancer agents, such as targeted therapies and immunotherapies, provides more options for comprehensive treatment. The malignant progression of tumors and resistance to existing therapies are largely attributable to the evasion of programmed cell death by cancer cells [4]. Therefore, in‐depth elucidation of the intrinsic molecular mechanisms by which GC cells resist cell death holds urgent clinical significance for the development of novel targeted therapeutic strategies.

Ferroptosis is a novel form of cell death driven by iron‐dependent lipid peroxidation. Inducing ferroptosis is a promising cancer treatment strategy [5]. Recent studies have shown that ferroptosis is closely associated with the initiation, progression, and drug resistance of GC [6, 7]. During GC progression, tumor cells often evade ferroptosis by upregulating the antioxidant defense system to counteract the accumulation of reactive oxygen species (ROS) [8]. Studies have demonstrated that non‐coding RNA networks can reshape the sensitivity of GC cells to ferroptosis by coordinately regulating key molecules involved in ferroptosis, thereby influencing tumor progression. For example, Lin et al. showed that circTFRC protected GC cells from ferroptosis and promoted tumor progression by interacting with stearoyl‐CoA desaturase 1 mRNA [9]. Although some understanding of the metabolic characteristics of ferroptosis has been achieved, the precise post‐transcriptional and epigenetic regulatory networks that govern ferroptosis in GC have not yet been fully elucidated.

Non‐coding RNAs, particularly circular RNAs (circRNAs), are emerging as key breakthroughs in investigating GC progression and as highly promising diagnostic biomarkers due to their unique covalently closed loop structures and high stability [10]. CircRNAs function primarily as molecular sponges or competitive endogenous RNAs (ceRNAs), which sequester specific microRNAs (miRNAs) and thereby relieve the post‐transcriptional repression of their target genes [11]. Among them, circular fanconi anemia complementation group B (circFANCB) has been shown to be overexpressed in cervical cancer, where its suppression impedes malignant progression, suggesting a potential oncogenic role in tumorigenesis [12]. However, the functional mechanisms of circFANCB in GC remain unexplored.

MicroRNA‐454‐3p (miR‐454‐3p) is deeply involved in tumorigenesis and progression. It has been reported to promote cell proliferation and metastasis in multiple cancer types, including in lung cancer [13], colorectal cancer [14], breast cancer [15], cervical cancer, and neuroblastoma [16], while inhibiting tumor growth in pancreatic cancer, non‐small cell lung cancer (NSCLC) [17], and bladder cancer [18]. This suggests that it may serve as a key hub in the regulatory network of tumor progression; however, its critical role in the proliferation and metastasis of GC remains to be explored.

Carcinoembryonic antigen‐related cell adhesion molecule 5 (CEACAM5) is a highly glycosylated cell surface glycoprotein [19]. Its overexpression has been demonstrated to contribute to the pathogenesis of multiple malignancies through several mechanisms. High expression of CEACAM5 in breast cancer is associated with reduced patient survival rates [20]. Furthermore, CEACAM5 can serve as a biomarker for GC diagnosis [21]. Suppression of CEACAM5 by miR‐498 has been shown to attenuate malignant behaviors, including tumor growth and metastasis [22]. These findings collectively highlight CEACAM5 plays a key regulatory role in tumor progression.

N6‐methyladenosine (m6A), one of the mechanisms of epigenetic regulation, is dynamically regulated by a coordinated system comprising methyltransferases, demethylases, and m6A‐binding proteins [23]. Functioning as a master regulator of post‐transcriptional processes, m6A fine‐tunes RNA metabolism and function through its reversible chemical marks, playing pivotal roles in both physiological processes and tumor pathogenesis [24]. Previous studies have established methyltransferase‐like 3 (METTL3), the core catalytic subunit of the m6A methyltransferase complex, as a key promoter of tumor progression across cancer types. For instance, METTL3‐mediated m6A modification of basic leucine zipper and W2 domains 2 drives myeloma development [25]. In cervical cancer, METTL3 upregulates NIMA‐related kinase 2 through m6A‐dependent mechanisms to accelerate malignancy [26]. Furthermore, in GC, METTL3 promotes tumorigenesis by installing m6A marks on long non‐coding RNA small nucleolar RNA host gene 3, thereby activating the miR‐186‐5p/CyclinD2 axis [27]. Moreover, the biological consequences of m6A modification are predominantly dictated by specific m6A “reader” proteins, such as the YTH domain‐containing family. These readers recognize m6A sites and determine the fate of target RNAs by modulating their stability, splicing, or translation [28]. This study identified m6A modification sites on circFANCB and predicted the binding sites between METTL3 and circFANCB. It remains unclear whether circFANCB is regulated by the m6A methyltransferase METTL3, thereby influencing the malignant progression of GC.

This study aims to investigate the role of the circFANCB‐miR‐454‐3p‐CEACAM5 axis in GC across molecular, cellular, and in vivo levels, thereby providing a theoretical basis for clinical treatment.

2. Materials and Methods

2.1. CircFANCB Circular Structure and Sanger Sequencing

The circular structure of circFANCB was identified from a circular RNA database and was formed by the back‐splicing of several exons of the FANCB gene. Its back‐splice junction was confirmed by Sanger sequencing, which utilized dideoxynucleotides (ddNTPs) to terminate DNA chain elongation. The resulting fragments were separated by capillary electrophoresis, and the sequence was decoded from the fluorescent peak data.

2.2. Human Samples

Tumor tissues (n = 64) and matched adjacent normal tissues (n = 64) were collected from 64 GC patients at The People's Hospital of Beilun District. Written informed consent was obtained from all participants, and the study protocol was approved by the People's Hospital of Beilun District Ethics Committee (Approval NO.K2025‐035‐F2), and complied with the Declaration of Helsinki.

2.3. Cell Culture and Treatment

The normal gastric epithelial cell line GES‐1 and the human GC cell lines AGS and HGC‐27 were obtained from the Shanghai institute for life sciences cell center (Shanghai, China). All cells were cultured in RPMI 1640 (Gibco, Grand Island, NY, USA) complete medium supplemented with 10% fetal bovine serum (Gibco) and 1% penicillin/streptomycin (Gibco).

2.4. Quantitative Real‐Time PCR (qRT‐PCR)

TRIzol Reagent (Abcam, Cambridge, UK) was used to isolate total RNA, and cDNA was generated with the PrimeScript RT Reagent Kit (Takara, Shiga, Japan). cDNA was mixed with SYBR Green qPCR Mix (Takara) and amplified in an ABI 7500 PCR system to detect expression levels of target mRNAs and miRNAs. U6 was used as the internal control for miRNA quantification, while glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) acted as the internal control for mRNAs and circRNAs quantification. The primer sequences used were:

CircFANCB Forward: GTCGCGGCTGCTTTATCAGA

CircFANCB Reverse: CCCATTTAGCAGCAACTTTTAGGT

miR‐454‐3p Forward: CTATCTTAGTGCAATATTGCTTATA

miR‐454‐3p Reverse: CTGGTGTCGTGGAGTCGG

CEACAM5 Forward: TCTAACCCATCCCCGCAGTA

CEACAM5 Reverse: TTATTGCGGCCAGTAGCCAA

GAPDH Forward: GTCAAGGCTGAGAACGGGAA

GAPDH Reverse: AAATGAGCCCCAGCCTTCTC

YTH N6‐methyladenosine RNA binding protein F1 (YTHDF1) Forward: CGTGGACACCCAGAGAACAA

YTHDF1 Reverse: TAGCTGGACAGGTAGGGGTC

YTH N6‐methyladenosine RNA binding protein F2 (YTHDF2) Forward: GGCAGTGGGTTCGGTCATAA

YTHDF2 Reverse: GGACCGAAGCTTCTCCAACA

YTH N6‐methyladenosine RNA binding protein F3 (YTHDF3) Forward: GCCACTAGCGTGGATCAGAG

YTHDF3 Reverse: GCTGTTACTCTGATTTGTCTGGC

U6 Forward: CTCGCTTCGGCAGCACATA

U6 Reverse: CGAATTTGCGTGTCATCCT

2.5. RNnase R Digestion Assay

The extracted total RNA was treated with RNase R (Thermo Fisher, Shanghai, China) and the corresponding reaction buffer, and incubated at 37°C for 15–30 min. The reaction was immediately terminated upon completion, and the RNA was purified using an RNA purification kit (Yeasen, Shanghai, China). The resulting RNA was then reverse‐transcribed into cDNA, which was subjected to qRT‐PCR analysis.

2.6. Nuclear‐Cytoplasmic Fractionation Assay

GC cells were treated using the NE‐PER Nuclear/Cytoplasmic Isolation Kit (Thermo Fisher). The supernatant constituted the cytoplasmic fraction, while the pellet represented the nuclear fraction. TRIzol reagent (Abcam) was added to both cytoplasmic and nuclear fractions, and total RNA was extracted following standard operating procedures. Equal volumes of nuclear and cytoplasmic RNA were used for reverse transcription to synthesize cDNA. Expression levels of circFANCB in nuclear and cytoplasmic fractions were detected separately via qRT‐PCR.

2.7. Small Interfering RNA (siRNA) and Overexpression Vector Establishment

The siRNAs targeting circFANCB (si‐circFANCB) and CEACAM5 (si‐CEACAM5), the overexpression vectors of CEACAM5 (pcDNA‐CEACAM5), and the corresponding negative controls (si‐NC and pcDNA‐NC) were constructed and synthesized by by GenePharma (Shanghai, China). Transfection into cells was performed using Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, California, USA) according to the manufacturer's specifications.

2.8. Construction of Anti‐miRNA Molecules

The anti‐miRNA oligonucleotide (anti‐miR‐454‐3p) was synthesized by GenePharma. Anti‐miR‐454‐3p was transfected into cells using Lipofectamine 2000 transfection reagent (Invitrogen).

2.9. 5‐ethynyl‐2′‐deoxyuridine (EdU) Incorporation Assay

Proliferation of GC cells was measured with the Click‐iT EdU cell proliferation imaging kit (Thermo Fisher). Briefly, the medium was exchanged for fresh medium pre‐warmed medium containing EdU working solution, followed by incubation in an incubator. Cells were fixed and permeabilized with Triton X‐100 solution (Beyotime, Shanghai, China). The permeabilization solution was then removed, and the Click‐iT reaction cocktail was added to each well, ensuring complete coverage of the cells. The nuclei were counterstained with 4′,6‐diamidino‐2‐phenylindole prior to mounting. Finally, observe under a fluorescence microscope (Olympus, Tokyo, Japan).

2.10. Cell Death Assay

Cell death was analyzed using the Annexin V‐FITC apoptosis detection kit (Beyotime). GC cells were collected, gently resuspended in Annexin V‐FITC binding buffer, and then mixed with Annexin V‐FITC and propidium iodide staining solution. The mixture was incubated in the dark for 15 min before undergoing cell death analysis via flow cytometry (BD Biosciences, San Jose, California, USA).

2.11. Wound Healing Assay

A pipette tip was used to create a scratch in the GC cell. The wound areas were clearly marked and imaged using an inverted microscope (Olympus). After 0 and 24 h, images were captured again at the same marked locations. The acquired images were subsequently analyzed and quantified using ImageJ software.

2.12. Transwell Assay

GC cells resuspended in serum‐free medium were seeded into the upper chamber coated with Matrigel (BD Biosciences). The lower chamber of the Transwell was then filled with 600 μL of complete medium. After 24 h of incubation, cells were fixed and stained with 0.1% crystal violet (Beyotime) for 10 min. After phosphate buffer solution (Thermo Fisher) rinses, cells were subjected to microscopic examination (Olympus) for quantification.

2.13. Tube Formation Assay

To form the matrix layer, a thin layer (50–100 μL) of Matrigel (Yeasen Biotechnology, Shanghai, China) was applied to pre‐chilled 96‐well plates and polymerized by incubation at 37°C for 1 h. Human umbilical vein endothelial cells (HUVECs) (Shanghai institute for life sciences cell center) were harvested by trypsinization (Thermo Fisher), resuspended, and seeded onto the polymerized Matrigel at a density of 1.0–2.0 × 10⁴ cells per well. Following transfection with si‐circFANCB, cells were cultured for an additional 12 h. Tubular structures were imaged using an inverted microscope (Olympus), with 3–5 randomly selected fields captured per well to ensure clear visualization of the capillary‐like networks.

2.14. Database Analysis

CEACAM5 expression in GC and its correlation with ferroptosis‐related genes were obtained from the gene expression profiling interactive analysis (GEPIA) database (http://gepia.cancer‐pku.cn/). M6A methylation sites were assessed using the sequence‐RNA adenosine methylation prediction (SRAMP) website (http://www.cuilab.cn/sramp). METTL3 binding sites on circFANCB were predicted using the RNA‐binding protein suite (RBPsuit) website (http://www.csbio.sjtu.edu.cn/bioinf/RBPsuite/).

2.15. Western Blot

Following treatment, cells were lysed with RIPA buffer (Beyotime) to extract total protein, which was then denatured by boiling. After electrophoresis, transfer, and blocking, the polyvinylidene difluoride (PVDF) membrane (Beyotime) was probed with the primary antibody: anti‐CEACAM5 (1:1000, 14‐0669‐82, Thermo Fisher), anti‐METTL3 (1:1000, ab195352, Abcam), anti‐acyl‐CoA synthetase long‐chain family member 4 (ACSL4; 1:2000, PA5‐27137, Invitrogen), anti‐glutathione peroxidase 4 (GPX4; 1:500, MA5‐32827, Invitrogen), anti‐β‐actin (1:10000, ab6276, Abcam), and anti‐GAPDH (1:10000, ab181602, Abcam). Subsequently, the PVDF membrane was incubated with a secondary antibody (1:2000, ab6721, Abcam) for 1.5 h and visualized for western blot signals.

2.16. Dual‐Luciferase Reporter Assay

The complementary sequence of miR‐454‐3p was amplified and cloned into the pmirGLO reporter vector (Promega, Madison, Wisconsin, USA) to construct wild‐type luciferase reporter vectors (circFANCB‐wt and CEACAM5 3' untranslated region (UTR)‐wt). Corresponding mutant vectors (circFANCB‐mut and CEACAM5 3'UTR‐mut) were also generated. The luciferase reporter vectors were then co‐transfected with either miR‐454‐3p or miR‐NC into cells. Subsequently, luciferase activity was quantified using the dual‐luciferase reporter assay system (Promega).

2.17. Measurement of Intracellular Fe²⁺

After cell lysis, the supernatant was collected by centrifugation. The Ferrozine working solution (Solarbio, Beijing, China) was added to the mixture, followed by 15–30 min incubation, and the absorbance was recorded at 562 nm.

2.18. ROS Assay

After cell collection, following resuspension in a 1:1000 dilution of 2’,7’‐dichlorodihydrofluorescein diacetate (Beyotime) in serum‐free medium and a 30 min incubation, the cells were washed three times with PBS. Fluorescence intensity was then analyzed using flow cytometry.

2.19. Malondialdehyde (MDA) Detection and Glutathione (GSH) Assay

MDA and GSH levels in GC cells were measured using the MDA enzyme‐linked immunosorbent assay (ELISA) Kit (Elabscience, Wuhan, China) and the GSH ELISA Kit (Elabscience). Serially diluted standards, standard/sample diluent, and test samples were added to the designated standard, blank, and sample wells, respectively. The reaction was terminated, and the absorbance was immediately measured at 450 nm to quantitatively determine the reaction product.

2.20. Mouse Models

BALB/c nude mice (male, 4–6 weeks old, weighing 18–20 g) were purchased from Shanghai SLAC Laboratory Animal Co. Ltd. (Shanghai, China). The mice were randomly divided into two groups (n = 6 per group). GC cells transfected with sh‐circFANCB or sh‐NC (GenePharma) and labeled with pLenti‐luciferase virus (Fenghui Biotech, Changsha, China) were subcutaneously injected into the mice. A successful xenograft tumor model was established when palpable masses developed 5 days post‐inoculation. Tumor location, size, and metastasis were monitored using a bioluminescence imaging system. Tumor volume in mice was measured regularly at 5‐day intervals. After 25 days, the mice were euthanized, and tumors were excised for further analysis. All animal experiments were conducted in accordance with institutional guidelines and approved by the People's Hospital of Beilun District Ethics Committee (Approval NO.K2025‐032‐F1). Animal procedures followed the ARRIVE guidelines and 3Rs principle.

2.21. Immunohistochemistry (IHC)

Following fixation, tumor tissues were embedded in paraffin, sectioned, and incubated with diluted primary antibodies specific for Ki67 (1:200, ab16667, Abcam), CEACAM5 (1:200, 14‐0669‐82, Thermo Fisher), solute lecarrier family 7 member 11 (SLC7A11) (1:500, ab307601, Abcam), ACSL4 (1:500, PA5‐27137, Invitrogen), and GPX4 (1:200, MA5‐32827, Invitrogen) in a humidified chamber at 4°C for 12 h. A secondary antibody (1:1000, ab6721, Abcam) was applied to the sections and incubated for 60 min. Following another PBS wash, color development was performed using 3,3’‐diaminobenzidine (Gene Tech, Shanghai, China). The levels of Ki67, CEACAM5, and SLC7A11 in tumor tissues were evaluated based on staining intensity.

2.22. Methylated RNA Immunoprecipitation (MeRIP)

The MeRIP assay was performed using the riboMeRIP m6A transcriptome analysis kit (Ribobio, Guangzhou, China). RNA was randomly fragmented into segments of less than 200 nt using RNA fragmentation reagents. Protein A/G agarose beads were washed twice with IP buffer and then coated with an m6A‐specific antibody. The immunoprecipitation samples were incubated with the antibody‐bead complexes in IP buffer at 4°C for 2–4 h. The agarose‐antibody‐RNA complexes were subsequently washed with pre‐chilled IP buffer. Bound RNA was eluted and purified using elution buffer incubated at 55°C. Results were analyzed by qRT‐PCR.

2.23. RNA Binding Protein Immunoprecipitation (RIP)

Detection was performed using the BeyoRIP RIP Kit (Beyotime). Protein A/G agarose beads were washed with NT2 wash buffer and then incubated with the corresponding primary antibodies‐either anti‐IgG antibody (1:50, B900620, Proteintech, Wuhan, China) or anti‐METTL3 antibody (1:50, ab195352, Abcam)‐for 30 min to form antibody‐bead complexes. The complexes were subsequently incubated with cell lysates for 4 h on a shaking platform. Eluted RNA was subsequently purified and analyzed by qRT‐PCR.

2.24. Statistical Approach

Data from three independent experiments are presented as mean ± standard deviation. An unpaired t‐test and analysis of variance were applied to assess statistical significance. GraphPad Prism 8.0 software (GraphPad Software, San Diego, CA, USA) was employed for the visualization of experimental data, with statistical significance set at p < 0.05.

3. Results

3.1. CircFANCB Shows Elevated Expression in GC and Predicts Poor Prognosis

To investigate the function of circFANCB in GC, its circular structure was first identified and validated by Sanger sequencing (Figure 1A,B). Comparing the levels of circFANCB in GC tissues (Tumor, N = 64) versus adjacent normal tissues (Normal, N = 64), circFANCB expression was elevated in GC tissues (Figure 1C). Kaplan‐Meier curves revealed that patients with high circFANCB expression exhibited lower 5‐year overall survival rates compared to those with low circFANCB expression. (Figure 1D). Furthermore, circFANCB levels were significantly elevated in patients with stage III disease relative to those with stages I + II (Figure 1E), and were also higher in metastatic cases compared with non‐metastatic ones (Figure 1F). In GC cell lines (AGS, HGC‐27), circFANCB expression was significantly higher than in human normal gastric mucosal cells (GES‐1) (Figure 1G), further validating circFANCB upregulation in GC at the cellular level. Subsequently, the circular stability of circFANCB was assessed using an RNase R digestion assay (Figure 1H,I), and its predominant cytoplasmic localization was verified through nuclear‐cytoplasmic fractionation experiments (Figure 1J,K). Based on the above findings, circFANCB is highly expressed in GC and is closely associated with tumor progression, metastasis, and poor prognosis.

Figure 1.

circFANCB is upregulated in GC. (A) Circular structure of circFANCB. (B) Sequencing peak diagram of the circFANCB circular structure obtained via Sanger sequencing. (C) qRT‐PCR detection of circFANCB expression in GC tissues and adjacent normal tissues (N = 64). (D) Kaplan–Meier survival curve analysis of the correlation between circFANCB expression levels and 5‐year survival rates in GC patients (N = 32). (E, F) Scatter plots showing the correlation between circFANCB expression levels and GC patient staging and metastasis. (G) qRT‐PCR detection of circFANCB expression in human normal gastric mucosal cells (GES‐1) and GC cell lines (AGS, HGC‐27). (H, I) RNase R treatment assay for circFANCB circular stability. (J, K) Nucleoplasmic fractionation assay for circFANCB cellular localization. U6: Small nuclear RNA, localized to the nucleus, serving as a nuclear fraction reference; GAPDH mRNA: localized to the cytoplasm, serving as a cytoplasmic fraction reference. ns, no significance, ***p < 0.001.

3.2. Knockdown of CircFANCB Suppresses Malignant Progression of GC

To elucidate the functional role of circFANCB in GC, it was knocked down in AGS and HGC‐27 cell lines (Figure 2A). A series of functional assays were then conducted to assess its impact on diverse cellular processes, such as proliferation, cell death, migration, and invasion. EdU staining and Annexin V/PI staining revealed that circFANCB knockdown significantly reduced the proliferation capacity of AGS and HGC‐27 cells (Figure 2B) and promoted cell death in these cell lines (Figure 2C). Furthermore, cell scratch and transwell assays confirmed that circFANCB knockdown suppressed migration and invasion in GC cells (Figure 2D,E). Moreover, tube formation assays revealed that the conditioned medium from circFANCB‐knockdown GC cells suppressed angiogenesis in HUVECs (Figure 2F). These findings demonstrate that circFANCB knockdown significantly suppresses the malignant phenotype of GC cells.

Figure 2.

Knocking down CircFANCB inhibits the malignant phenotype of GC. Knockdown of circFANCB was performed in GC cells (AGS, HGC‐27). (A) qRT‐PCR detection of circFANCB knockdown efficiency. (B) EdU assay to assess the proliferation of AGS and HGC‐27 cells. Red fluorescence labels proliferating cells; blue fluorescence (DAPI) labels all cell nuclei; the merged panel shows the overlay image. (C) Flow cytometry analysis of cell death in AGS and HGC‐27 cells after Annexin V/PI staining. (D) Scratch assay to assess the migration ability of AGS and HGC‐27 cells. (E) Transwell assay to evaluate the invasion capacity of AGS and HGC‐27 cells. (F) Tube formation assay to detect angiogenesis in HUVEC cells. **p < 0.01, and ***p < 0.001.

3.3. CircFANCB Promotes CEACAM5 Expression Via miR‐454‐3P

CircRNAs frequently exert their biological functions through the competitive ceRNA mechanism. Within this regulatory framework, circRNAs compete with mRNAs for binding to shared miRNAs, thereby modulating miRNA availability and their post‐transcriptional regulation of downstream target genes [29]. To further investigate the mechanism of circFANCB, prediction analysis using the Starbase database revealed complementary sequences between circFANCB and miR‐454‐3p. Notably, these binding sites were also present in the 3'UTR of CEACAM5, a downstream target gene of miR‐454‐3p (Figure 3A). Analysis of the GEPIA database revealed that CEACAM5 was highly expressed in GC tissues (Figure 3B). Consistent with this, qRT‐PCR and western blot analyzes confirmed that CEACAM5 was significantly upregulated at both mRNA and protein levels in GC tissues (Figure 3C,D). In GC tissues, circFANCB expression exhibited a significant positive correlation with CEACAM5 mRNA (Figure 3E). At the cellular level, comparative analysis of CEACAM5 protein expression in GES‐1 versus AGS and HGC‐27 cells revealed significantly elevated levels of this protein in the cancer cell lines (Figure 3F). In contrast, miR‐454‐3p expression was significantly downregulated in GC tissues (Figure 3G). Moreover, circFANCB expression was significantly negatively correlated with miR‐454‐3p levels, and miR‐454‐3p expression also showed a marked negative correlation with CEACAM5 mRNA expression (Figure 3H,I). To validate their interaction, miR‐454‐3p was overexpressed in GC cells, with its overexpression efficiency confirmed by qRT‐PCR (Figure 3J). Moreover, dual‐luciferase reporter assays demonstrated that miR‐454‐3p specifically bound to the wild‐type sequences of circFANCB and CEACAM5 3'UTR in AGS and HGC‐27 cells, respectively (Figure 3K–N). Furthermore, functional rescue experiments were conducted by transfecting GC cells with a miR‐454‐3p inhibitor to suppress its activity, and the knockdown efficiency was validated (Figure 3O). Knockdown of circFANCB in GC cells significantly reduced CEACAM5 protein expression; while co‐transfection with anti‐miR‐454‐3p partially restored CEACAM5 protein expression (Figure 3P). This rescue experiment conclusively demonstrates that circFANCB sequesters miR‐454‐3p, and thus relieves its repression on CEACAM5 to upregulate its expression.

Figure 3.

CircFANCB promotes CEACAM5 expression by antagonizing miR‐454‐3p activity. (A) Sequences of circFANCB (wild‐type wt, mutant mut), miR‐454‐3p, and CEACAM5 3′UTR (wild‐type wt, mutant mut). Binding regions among the three are predicted based on base complementary pairing (indicated by vertical lines). (B) Analysis of the GEPIA database revealed that CEACAM5 was upregulated in GC. Red regions indicate GC group; gray indicates normal group. T denotes tumor tissue (T = 408); N denotes normal tissue (N = 211). (C) qRT‐PCR detection of CEACAM5 mRNA expression levels in GC. (N = 64). (D) Western blot detection of CEACAM5 protein expression levels in GC. (E) Correlation analysis of CEACAM5 and circFANCB expression in GC. (F) Western blot detection of CEACAM5 protein expression in GC cell lines. (G) qRT‐PCR was performed to detect the expression of miR‐454‐3p in GC and adjacent normal tissues. (H, I) Correlation analysis of circFANCB and miR‐454‐3p, as well as miR‐454‐3p and CEACAM5 mRNA. (J) qRT‐PCR detection of miR‐454‐3p overexpression efficiency after transfection of miR‐454‐3p mimetic in AGS and HGC‐27 cells. (K–N) Dual‐fluorescence assay detecting the binding of miR‐454‐3p with CircFANCB and CEACAM5. (O) qRT‐PCR detection of miR‐454‐3p suppression efficiency after anti‐miR‐454‐3p transfection in AGS and HGC‐27 cells. (P) Western blot analysis of CEACAM5 protein expression in AGS and HGC‐27 cells following transfection with si‐circFANCB or/and anti‐miR‐454‐3p. ns, no significance, **p < 0.01, and ***p < 0.001.

3.4. CEACAM5 Suppresses Ferroptosis in GC Cells

Analysis of the GEPIA database revealed a positive correlation between CEACAM5 and SLC7A11, a key regulator of ferroptosis, in GC tissues (Figure 4A), suggesting a potential link between CEACAM5 and ferroptotic pathways. To investigate the functional role of CEACAM5 in GC, knockdown and overexpression experiments of CEACAM5 were performed in AGS and HGC‐27 cell lines (Figure 4B,C). Furthermore, ferroptosis inhibitors and inducers were employed to investigate whether CEACAM5 modulates cellular susceptibility to ferroptosis. At the level of iron metabolism, CEACAM5 knockdown increased intracellular Fe²⁺ levels, whereas its overexpression decreased them; these effects were partially reversed by the ferroptosis inhibitor ferrostatin‐1 or the inducer Erastin (Figure 4D,E). In terms of oxidative stress, CEACAM5 knockdown increased lipid ROS and MDA levels but reduced GSH expression. Overexpression of CEACAM5 led to the opposite phenotypes. These changes were also reversible upon treatment with ferroptosis modulators (Figure 4F–K). Western blot analysis further confirmed that CEACAM5 knockdown significantly downregulated GPX4 and upregulated ACSL4, whereas the ferroptosis inhibitor Fer‐1 reversed these changes (Figure 4L). Conversely, CEACAM5 overexpression significantly upregulated GPX4 and downregulated ACSL4, and these effects were reversed by the ferroptosis activator Erastin (Figure 4M). Furthermore, cell death was assessed by flow cytometry. CEACAM5 knockdown enhanced cell death and its overexpression attenuated it, alterations that were reversed by ferroptosis modulators (Figure 4N,O). These results demonstrate that CEACAM5 modulates iron homeostasis, lipid peroxidation, and cell death in GC cells by regulating ferroptosis sensitivity.

Figure 4.

CEACAM5 inhibits ferroptosis in GC cells. (A) GEPIA database showed the expression correlation between CEACAM5 and SLC7A11. (B, C) Western blot analysis of CEACAM5 knockdown and overexpression efficiency after transfecting AGS or HGC‐27 cells with si‐CEACAM5 or pcDNA‐CEACAM5. (D–O) AGS or HGC‐27 cells were transfected with si‐CEACAM5 or pcDNA‐CEACAM5 and simultaneously treated with ferroptosis inhibitor (Ferrostatin‐1) or inducer (Erastin). Fe²⁺ concentration was measured by colorimetric assay (D, E), cellular ROS levels were assessed by flow cytometry (F, G), MDA and GSH levels were detected by ELISA (H–K), protein levels of GPX4 and ACSL4 were determined by Western blot (L, M), and the cell death was analyzed by flow cytometry (N, O). *p < 0.05, **p < 0.01, and ***p < 0.001.

3.5. CircFANCB Facilitates GC Cell Growth by Upregulating CEACAM5 to Suppress Ferroptosis

In GC cells with circFANCB knockdown, CEACAM5 was overexpressed to investigate whether circFANCB regulates GC cell function and ferroptosis through CEACAM5 (Figure 5A). A systematic evaluation was then conducted of cell proliferation, cell death, migration, invasion, angiogenesis, and ferroptosis‐related indicators. Results revealed that CEACAM5 overexpression reversed the circFANCB knockdown‐induced inhibition of cell proliferation (Figure 5B) and mitigated its pro‐death effects (Figure 5C). Concurrently, CEACAM5 restoration also reestablished the impaired cell migration, invasion, and angiogenesis capabilities caused by circFANCB knockdown (Figure 5D–F). Regarding ferroptosis‐related markers, knockdown of circFANCB increased the levels of Fe²⁺, ROS, and MDA, while it decreased the level of GSH in GC cells. These alterations were reversed by co‐transfection with pcDNA‐CEACAM5 (Figure 5G–K). Western blot analysis revealed that circFANCB knockdown significantly downregulated GPX4 and upregulated ACSL4, whereas concomitant CEACAM5 overexpression partially reversed these changes (Figure 5L,M). Rescue experiments confirm that circFANCB suppresses ferroptosis and promotes the malignant phenotype of GC cells by upregulating CEACAM5.

Figure 5.

CircFANCB promotes GC cell growth by suppressing ferroptosis through upregulation of CEACAM5. AGS and HGC‐27 cells were co‐transfected with pcDNA‐CEACAM5 after transfection with si‐circFANCB. (A) Western blot analysis of CEACAM5 protein expression in each group of AGS and HGC‐27 cells. (B) EdU assay for cell proliferation in AGS and HGC‐27 cells. (C) Flow cytometry was used to detect cell death in AGS and HGC‐27 cells. (D) Wound healing assay for cell migration in AGS and HGC‐27 cells. (E) Transwell assay for cell invasion in AGS and HGC‐27 cells. (F) Tube formation assay for HUVEC angiogenesis. (G) The Fe²⁺ concentration was measured by a colorimetric assay. (H, I) Intracellular ROS levels were assessed by flow cytometry. (J, K) Oxidative stress‐related indicators, MDA and GSH, were measured using ELISA. (L, M) Protein levels of GPX4 and ACSL4 were detected by Western blot. *p < 0.05, **p < 0.01, and ***p < 0.001.

3.6. CircFANCB Promotes GC Tumorigenesis Via miR‐454‐3p/CEACAM5

Building upon previous molecular and cellular findings that circFANCB regulates GC cell function and ferroptosis through the miR‐454‐3p/CEACAM5 axis, its role in tumorigenesis was further evaluated using an in vivo model. Knockdown of circFANCB in mouse tumor models significantly reduced the volume and weight of xenograft tumors (Figure 6A–C). Successful downregulation of circFANCB in tumor tissues was confirmed by qRT‐PCR analysis (Figure 6D). Furthermore, IHC staining showed that circFANCB knockdown led to downregulated expression of CEACAM5, the proliferation marker Ki67, and the anti‐ferroptosis proteins SLC7A11 and GPX4 in tumor tissues, while the expression of the pro‐ferroptosis protein ACSL4 was upregulated (Figure 6E). These results indicate that circFANCB knockdown suppresses tumor proliferation and activates ferroptosis in vivo.

Figure 6.

Knocking down circFANCB inhibits GC tumor growth. Subcutaneous injection of GC cells stably transfected with sh‐circFANCB or sh‐NC established a xenograft tumor model in nude mice (n = 6). (A) Detection of tumor volume changes. (B) Tumor tissues from the sh‐NC and sh‐circFANCB groups were photographed. (C) Tumor weight measurement. (D) qRT‐PCR analysis of circFANCB expression in tumor tissues. (E) IHC detection of Ki67, CEACAM5, SLC7A11, ACSL4, and GPX4 expression in tumor tissues. ***p < 0.001.

3.7. METTL3 Promotes the Expression of circFANCB by Mediating m6A Methylation

Through predictive analysis via the SRAMP website, circFANCB was found to contain m6A modification sites (Figure 7A). MeRIP assay results demonstrated that compared with the IgG group, the enrichment of circFANCB was markedly increased in the m6A group, indicating that circFANCB underwent m6A methylation modification (Figure 7B). Subsequently, the potential binding sites between circFANCB and the m6A methyltransferase METTL3 were predicted using the online database RBPsuit (Figure 7C). It is speculated that circFANCB may be regulated by the m6A methyltransferase METTL3. RIP assays further confirmed that the enrichment level of circFANCB was higher in the METTL3 group, demonstrating a direct interaction between METTL3 and circFANCB (Figure 7D). Knockdown of METTL3 was performed in AGS and HGC‐27 cells, and the knockdown efficiency was verified by western blot (Figure 7E). Further MeRIP assays confirmed that METTL3 knockdown reduced the m6A modification of circFANCB (Figure 7F,G). Additionally, METTL3 knockdown led to a significant decrease in circFANCB expression in AGS and HGC‐27 cells, suggesting that METTL3 regulated circFANCB expression stability by enhancing its m⁶A modification (Figure 7H). To identify the m⁶A reader factors responsible for regulating circFANCB, YTHDF1, YTHDF2, and YTHDF3 were knocked down in AGS cells. The results showed that knockdown of YTHDF1 or YTHDF3 did not significantly alter circFANCB expression, whereas YTHDF2 knockdown significantly reduced circFANCB expression (Figure 7I–K). Western blot analysis demonstrated that pcDNA‐METTL3 significantly upregulated METTL3 protein expression levels in both AGS and HGC‐27 cells (Figure 7L). Overexpression of METTL3 increased circFANCB expression, and this effect was partially reversed by simultaneous YTHDF2 knockdown (Figure 7M). Collectively, these findings indicate that METTL3‐mediated m⁶A modification of circFANCB enhances its stability in a YTHDF2‐dependent manner.

Figure 7.

METTL3 enhances the expression of circFANCB via m6A methylation. (A) SRAMP website prediction of circFANCB m6A methylation sites. Red vertical lines mark regions with significantly enriched m6A peaks. (B) MeRIP validation of m6A methylation modification in circFANCB. IgG (negative control antibody), m6A (m6A‐specific antibody). (C) RBPsuit website predicted binding sites between circFANCB and the m6A methyltransferase METTL3. (D) RIP assay detected circFANCB binding to METTL3. IgG (negative control antibody), METTL3 (METTL3‐specific antibody). (E) Knockdown of METTL3 in AGS and HGC‐27 cells; Western blot analysis of METTL3 knockdown efficiency. (F, G) Knockdown of METTL3 in AGS and HGC‐27 cells; MeRIP assay to detect METTL3‐mediated m6A methylation of circFANCB. (H) Knockdown of METTL3 in AGS and HGC‐27 cells; qRT‐PCR to detect circFANCB expression. (I–K) circFANCB expression levels following knockdown of YTHDF1, YTHDF2, or YTHDF3. (L) Western blot validation of METTL3 overexpression efficiency. (M) AGS and HGC‐27 cells were transfected with pcDNA‐NC, pcDNA‐METTL3, pcDNA‐METTL3 + si‐NC, or pcDNA‐METTL3 + si‐YTHDF2, and the expression levels of circFANCB were detected by qRT‐PCR. ns, no significance, ***p < 0.001.

3.8. METTL3 Promotes the Malignant Phenotypes of GC Cells Via circFANCB

To verify whether METTL3 affects the malignant phenotypes and ferroptosis of GC cells by regulating circFANCB, rescue experiments were performed in AGS and HGC‐27 cells by knocking down METTL3 while overexpressing circFANCB. The results showed that METTL3 knockdown significantly reduced circFANCB expression, whereas circFANCB overexpression effectively restored its expression level (Figure 8A). METTL3 knockdown significantly inhibited GC cell proliferation, promoted cell death, and suppressed cell migration, invasion, and angiogenic capacity. Notably, concomitant circFANCB overexpression reversed these effects induced by METTL3 knockdown (Figure 8B–G). Furthermore, ferroptosis‐related assays revealed that after METTL3 knockdown, intracellular Fe²⁺, ROS, and MDA levels were significantly increased, while GSH content was markedly decreased (Figure 8H–L). Western blot analysis further demonstrated that the expression of the ferroptosis inhibitory protein GPX4 was downregulated, whereas the expression of the ferroptosis‐promoting protein ACSL4 was upregulated. Importantly, circFANCB overexpression significantly reversed these ferroptosis‐related alterations (Figure 8M,N). Collectively, these rescue experiments confirm that METTL3 promotes the malignant biological behavior of GC cells and inhibits ferroptosis by upregulating circFANCB.

Figure 8.

METTL3 regulates GC cell phenotypes and ferroptosis via circFANCB. AGS and HGC‐27 cells were transfected with si‐NC, si‐METTL3, si‐METTL3+Vector, or si‐METTL3 + OE‐circFANCB. (A) qRT‐PCR was performed to detect the expression level of circFANCB. (B) EdU assay for cell proliferation in AGS and HGC‐27 cells. (C) Flow cytometry was used to detect cell death in AGS and HGC‐27 cells. (D) Wound healing assay for cell migration in AGS and HGC‐27 cells. (E) Transwell assay for cell invasion in AGS and HGC‐27 cells. (F‐G) Tube formation assay for HUVEC angiogenesis. (H) The Fe²⁺ concentration was measured by a colorimetric assay. (I‐J) Intracellular ROS levels were assessed by flow cytometry. (K‐L) Oxidative stress‐related indicators, MDA and GSH, were measured using ELISA. (M‐N) The protein levels of GPX4 and ACSL4 were detected by Western blot. *p < 0.05, **p < 0.01, and ***p < 0.001.

4. Discussion

CircRNAs play significant biological roles in tumorigenesis and cancer progression. On one hand, they serve as ideal non‐invasive biomarkers for detection; on the other hand, circRNAs mediate molecular communication between primary tumor sites and distant metastatic locations [30]. Evidence suggests that circFANCB is significantly upregulated in cervical cancer, and suppression of circFANCB expression impedes malignant progression of the disease [12]. Consistent with this pro‐tumorigenic feature, the present study confirmed that circFANCB is also highly expressed in GC. Given that GC is a highly heterogeneous malignancy prone to therapeutic resistance, it is crucial to elucidate novel mechanisms of death evasion. Ferroptosis is a form of regulated cell death driven by the iron‐dependent accumulation of lipid peroxides [6]. It has been reported that many drug‐resistant or metastatic cancer cells exhibit high susceptibility to ferroptosis [31, 32]. Previous studies have shown that signal transducer and activator of transcription 3 (STAT3) acts as a key regulator of ferroptosis; suppression of STAT3 promotes ferroptotic cell death and restores chemosensitivity in gastric tumors [33]. Furthermore, Zhao et al. demonstrated that knockdown of mucin 1 significantly enhanced cellular sensitivity to ferroptosis induction and suppressed intrahepatic cholangiocarcinoma growth in vivo [34]. In line with these findings, experimental data confirmed that knockdown of circFANCB significantly relieved the inhibition of ferroptosis and effectively suppressed the malignant phenotype and tumor growth of GC cells.

According to the competitive ceRNA hypothesis, circRNAs and mRNAs can competitively bind to miRNAs through shared miRNA response elements, thereby reciprocally regulating their expression and influencing tumorigenesis and cancer progression [35]. Several studies have constructed circRNA‐miRNA‐mRNA networks in GC. For instance, in NSCLC, circ6834 functions as a tumor suppressor by acting as a molecular sponge for miR‐873‐5p. This sponge activity leads to the upregulation of thioredoxin interacting protein expression, thereby inducing apoptosis. These concerted actions collectively suppress tumor progression [36]. The investigation of the circFANCB‐miR‐454‐3p‐CEACAM5 axis in GC revealed that CEACAM5 was highly expressed in GC cells. CEACAM5 is recognized as a metastasis‐driving oncogene, whose overexpression reinforces an epithelial gene expression signature and promotes tumor growth at metastatic sites [22]. Mechanistically, circFANCB functioned as a competitor to miR‐454‐3p, attenuated its repression of the downstream target CEACAM5, and thereby promoted ferroptosis and malignant phenotype in GC cells.

METTL3, the core “writer” of m6A methylation, serves as a key node in the epitranscriptomic regulatory network. By precisely modulating the stability, translation, and splicing of downstream genes, it plays a fundamental role in the initiation, progression, metastasis, and therapy resistance of multiple cancers [37]. In GC research, METTL3 has been identified as a key therapeutic factor [38, 39]. Yu et al. reported that METTL3‐mediated m6A methylation enhanced signal transducer and activator of transcription 5A stability, thereby regulating krüppel‐like factor 4 to promote GC progression [40]. Additionally, METTL3 acts as a therapeutic target and potential prognostic biomarker by stabilizing DEK proto‐oncogene oncogene mRNA expression to facilitate GC cell proliferation and migration [41]. Consistent with these findings, the present data demonstrated that circFANCB contained m6A methylation sites and established that METTL3 regulated its expression through this modification. Importantly, it was further elucidated that this regulatory effect is dependent on the m6A reader protein YTHDF2. While YTHDF2 is classically recognized for targeting m6A‐modified mRNAs for degradation, the results revealed an unconventional role in which YTHDF2 knockdown decreased circFANCB expression. This finding suggests that YTHDF2 may unexpectedly enhance the stability of circFANCB. It is hypothesized that the circular structure of circFANCB might sterically hinder the typical YTHDF2‐mediated decay machinery, or alternatively, YTHDF2 might recruit other stabilizing factors specific to circular RNAs. Future investigations are required to fully unravel the intricate structural basis by which YTHDF2 dictates the fate of m6A‐modified circFANCB in GC.

Although this study has preliminarily revealed the role of m6A‐methylated circFANCB in GC via the miR‐454‐3p/CEACAM5 axis, several limitations remain, including a limited clinical sample size, a relatively narrow exploration of the ceRNA network, and an unclear understanding of the deep molecular mechanism by which CEACAM5 regulates ferroptosis. To address these shortcomings, future research plans include conducting multicenter, large‐cohort clinical validation to clarify its biomarker value, employing multi‐omics approaches to map a more comprehensive epitranscriptomic‐ceRNA regulatory network, and deeply investigating the core molecular network underlying ferroptosis resistance, thereby providing a basis for clinical treatment.

In conclusion, this study reveals that METTL3 regulates circFANCB expression by inducing m6A modification. CircFANCB modulates cellular ferroptosis by regulating the miR‐454‐3p/CEACAM5 axis, thereby promoting the growth and metastasis of GC cells. This provides novel therapeutic targets and strategies for GC research.

Author Contributions

Haifeng Wang: project administration, supervision, writing – original draft. Yan Li: resources, validation. Tiantian Sun: formal analysis. Man Su: formal analysis. Wei Dai: software. Zhihui Jin: data curation. Hangchen Shen: data curation. Changyu Jin: methodology. Yan Gao: conceptualization, investigation, funding acquisition, visualization, writing – review and editing.

Ethics Statement

The study protocol was approved by the People's Hospital of Beilun District Ethics Committee (Approval NO.K2025‐035‐F2), and complied with the Declaration of Helsinki.

All animal experiments were conducted in accordance with institutional guidelines and approved by the People's Hospital of Beilun District Ethics Committee (Approval NO.K2025‐032‐F1). Animal procedures followed the ARRIVE guidelines and 3Rs principle.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

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