FMRP promotes gastric cancer progression via m6A-dependent stabilization of DVL2 and activation of the Wnt/PCP-JNK pathway Authors: Xiaoyu Hou, Jinlan Gao, Xuesha Xing, Qi Liu, Ming Yi, Chen Liu, Qi Chu, Mengdie Zou, Meirong Hu, Yang Luo Published: 2026 ABSTRACT BackgroundN6-methyladenosine (m6A) modification has emerged as a critical regulator in gastric cancer progression. Fragile X messenger ribonucleoprotein 1 (FMRP), an RNA-binding protein with tumorigenic potential, remains poorly characterized in gastric cancer.MethodsFMRP expression in gastric cancer was assessed through bioinformatic analyses and tissue microarray-based immunohistochemistry. Its biological functions were examined through loss-of-function experiments in vitro and in vivo. Mechanistic investigations, including RNA interaction, m6A site mutation, and RNA stability analyses, were performed to identify downstream targets and pathways regulated by FMRP.ResultsBioinformatic analyses and tissue microarray-based immunohistochemistry showed that FMRP was significantly upregulated in gastric cancer tissues. FMRP depletion suppressed tumor growth and metastasis. Subsequently, Dishevelled segment polarity protein 2 (DVL2) was identified as a candidate downstream target of FMRP. DVL2 was also significantly upregulated in gastric cancer tissues, and its expression positively correlated with FMRP expression. Mechanistically, FMRP bound to DVL2 mRNA and enhanced its stability in an m6A-dependent manner. Among the candidate sites tested, mutation of 1271A attenuated FMRP-mediated regulation of DVL2 reporter activity, supporting the involvement of this site. Functionally, FMRP promoted activation of the noncanonical Wnt/planar cell polarity (PCP) signaling pathway and enhanced the proliferative, migratory, and invasive capacities of gastric cancer cells via DVL2.ConclusionsThese results define FMRP as an oncogenic driver in gastric cancer that operates via the DVL2/Wnt/PCP axis. Targeting this pathway may provide a potential therapeutic strategy for gastric cancer.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12967-026-08834-3. Introduction Global annual estimates by the International Agency for Research on Cancer indicate that gastric cancer results in approximately 1.1 million new cases and 770,000 deaths, making it the fifth most common cancer and the fourth leading cause of cancer-related deaths worldwide [1]. More than 80% of patients receive a diagnosis at advanced stages because specific molecular markers for early detection are lacking [2, 3]. Numerous risk factors, such as Helicobacter pylori infection, poor dietary habits, and previous surgical history, are well-documented contributors to gastric cancer development [4–6]. However, the mechanisms underlying gastric cancer progression and the potential therapeutic targets remain to be elucidated and require further exploration. Recent studies have identified N6-methyladenosine (m6A) RNA regulation as a crucial factor in carcinogenesis [7, 8]. Methyltransferases (m6A “writers”) and demethylases (m6A “erasers”) are involved in the reversible modification of m6A RNA. Meanwhile, m6A “readers” bind to these modified sites to regulate RNA functions such as stability, splicing, translation, degradation, and pri-miRNA processing [9, 10]. m6A dysregulation is significantly associated with the development of several cancers, including glioblastoma, breast, gastric, and colorectal cancers [11–14]. Recent research indicates that RNA m6A modification levels are elevated in gastric cancer tissues relative to healthy gastric tissues [15, 16]. However, the downstream relationship between RNA m6A methylation and gastric cancer progression and metastasis remains poorly understood. Encoded by the Fragile X Messenger Ribonucleoprotein 1 (FMR1) gene, FMR1 protein (FMRP) is a classical RNA binding protein (RBP) widely distributed in various tissues [17]. FMRP functions as an m6A reader protein, regulating target gene expression by binding to m6A modification sites on their mRNAs [18]. Several studies have concluded that FMRP regulates its target RNA stability, nuclear export, and translation [19–22]. FMR1 gene expression levels are increased in breast cancer, colorectal cancer, hepatocellular carcinoma, and melanoma [23–25]. Recent studies indicate that FMRP overexpression in primary breast tumors facilitates lung metastasis by influencing target mRNAs related to epithelial-mesenchymal transition, and affects colorectal cancer progression through the regulation of the RIPK/MLKL necroptosis pathway [26, 27]. Despite its established role as a central regulator of tumor progression, the function and underlying mechanism of FMRP in gastric cancer have yet to be elucidated. This study demonstrated that FMRP knockdown inhibited the proliferation, migration, and invasion of gastric cancer cells both in vitro and in vivo. Mechanistically, FMRP interacted with DVL2 mRNA and enhanced its stability in an m6A-dependent manner, thereby activating the noncanonical Wnt/planar cell polarity (PCP) pathway and promoting gastric cancer cell proliferation, migration, and invasion. Our study provides new insights into the mechanisms of gastric cancer tumorigenesis, highlighting a pivotal role for FMRP, and establishes a molecular basis that underscores its potential as a target for novel diagnostic and therapeutic strategies. Materials and methods Cell culture The human gastric cancer cell lines HGC-27, SGC-7901, MGC-803, and AGS, the immortalized gastric epithelial cell line GES-1, and the human embryonic kidney cell line HEK-293 were obtained from the Peking Union Medical College Cell Resource Center (Beijing, China). HEK-293, GES-1, and AGS cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM), whereas HGC-27, SGC-7901, and MGC-803 cells were cultured in RPMI-1640 medium; both media were obtained from Gibco (Grand Island, NY, USA). Culture media were enriched with 10% fetal bovine serum (21HY, TBD), penicillin, and streptomycin, and maintained at 37 °C with 5% CO2. Isolation of RNA, reverse transcription and qPCR analysis Total RNA was extracted following the manufacturer’s protocol using the Total RNA Extraction Reagent (R401-01, Vazyme). Reverse transcription utilized 1 μg of total RNA with HiScript II Q Select RT SuperMix for qPCR+gDNA wiper reagent (R233-01, Vazyme). qPCR was conducted with ChamQ Universal SYBR qPCR Master Mix (Q711-02, Vazyme) and analyzed using a Roche LightCycler 480 Real-Time PCR system, following the manufacturer’s guidelines. The qPCR primer sequences for human ACTB, FMR1, DVL2, VIM, and HPRT1 genes were obtained from PrimerBank, a public database for PCR primers [28]. ACTB was used as the internal reference gene, and relative mRNA expression was calculated using the 2−ΔΔCt method [29]. Experiments were independently conducted in triplicate, with primer sequences detailed in Supplemental Table S1. Protein extraction and western blot analysis Cells were lysed in RIPA buffer supplemented with protease (B14001) and phosphatase (B15001) inhibitor cocktails from Bimake, USA. Protein concentration was determined using a BCA assay (A53225, Thermo Scientific, USA). Equal amounts of protein samples were separated via 8% or 10% SDS‒PAGE. The membranes were incubated overnight at 4 °C with specific primary antibodies, followed by incubation with horseradish peroxidase-conjugated secondary antibodies. The bands were identified utilizing a Hypersensitive ECL kit (WLA006a, Wanleibio). Images were visualized with a DNR bioimaging system (DNR Bio-Imaging Systems, Jerusalem, Israel) and quantified using ImageJ software (National Institutes of Health). All antibodies used are summarized in Supplemental Table S2. Plasmid construction The 3×FLAG-tagged DVL2 expression plasmid (oe-FLAG-DVL2) was constructed via PCR amplification and subsequently subcloned and inserted into the p3×FLAG-CMV-14 plasmid. Human cDNA from HEK-293 cells was used to amplify the ORF of DVL2. The primer sequences were as follows: forward primer 5′-CCGGAATTCATGGCGGGTAGCAGCACTGG-3′ and reverse primer 5′-CGGGGTACCCACATAACATCCACAAAGAAC-3′. The PCR product was inserted into the EcoR I and Kpn I linearized p3×FLAG-CMV-14 plasmid using Solution I DNA ligase. Lastly, the vectors were amplified using Trans5α chemically competent cells (CD201-01, Transgene). Transient transfection Transient transfection was performed using jetPRIME reagent (Polyplus, France) following the manufacturer’s guidelines. siRNA Three independent small interfering RNAs (siRNAs) targeting human FMR1 (designated siFMR1-1, siFMR1-2, and siFMR1-3) and a negative control siRNA (siNC) were synthesized by Sangon Biotech (Shanghai, China). The target sequence of siFMR1#1 (5’-CGUGAAUGGAGUACCCUAATT-3’) was obtained from the MIT/ICBP siRNA Database (http://web.mit.edu/sirna/) [30], whereas the additional siRNAs were designed and synthesized by Sangon Biotech. All siRNA sequences are listed in Supplementary Table S3. Lentiviral transduction and preparation of stable FMRP-knockdown cells Lentiviruses carrying shRNA targeting human FMR1 (shFMR1) or control shRNA (shNC) were sourced from OBiO Technology (Shanghai, China). The FMR1-targeting sequence used for shFMR1 was identical to that of siFMR1#1. Cells were transduced with lentivirus following the manufacturer’s guidelines. Two days after transduction, puromycin-resistant cells were selected using 1 μg/mL puromycin (ST551, Beyotime) to generate stable FMRP-knockdown cells. Cell proliferation assay Gastric cancer cells were seeded in 96-well plates following transfection with siRNA. HGC-27 and SGC-7901 cells were seeded at densities of 3.5 × 10^3 and 3 × 10^3 cells per well, respectively, in a total volume of 100 μL of medium per well. Cell proliferation was evaluated at specified intervals using a CCK8 kit (A311-01-AA, Vazyme) following the manufacturer’s guidelines. In summary, 10 μL of CCK8 solution was added to each well, followed by a 2-hour incubation at 37 °C. Cell viability was assessed by measuring the optical density at 450 nm with a TECAN Infinite M200 microplate reader. Each group underwent experiments conducted in triplicate. Colony formation assay A total of 400 cells per well were seeded in six-well plates and cultured for 12 days. Colonies were fixed with 4% paraformaldehyde and subsequently stained using 1% crystal violet (C0121, Beyotime). Each group underwent experiments conducted in triplicate. Cell cycle analysis Cells were collected, rinsed with PBS, and fixed overnight in cold 70% ethanol at −20 °C. Cellular DNA was stained using a cell cycle staining solution with propidium iodide (WLA010a; Wanleibio) at 4 °C in darkness for 30 minutes. The cell cycle distribution was assessed with a FACS Calibur and analyzed using Cell Quest acquisition software (Canto II, BD Biosciences, USA). Each group underwent experiments conducted in triplicate. Wound healing assays Cells were seeded (1.5 × 10^6 HGC-27 cells per well or 1 × 10^6 SGC-7901 cells per well) and incubated overnight in 6-well plates until reached confluency. The confluent cell monolayer was scratched with a micropipette tip, followed by replacing the medium containing 1% FBS. Wound closure was visualized and recorded at 0, 24, and 48 h under a microscope. The migration rate is determined by dividing the migration distance by the original distance. Each group underwent experiments conducted in triplicate. Cell migration assay Cell migration assays utilized 24-well Transwell inserts with 8 μm pore membranes (3422, Corning Inc., Lowell, MA, USA). Serum-starved cells (3 × 10^4) were seeded into the upper inserts in 200 μL of serum-free RPMI-1640 medium. In contrast, the lower chambers were filled with 600 μL of medium containing 20% FBS and then incubated for 24 h at 37 °C. Migrated cells were fixed using 4% paraformaldehyde and stained with 1% crystal violet. The number of migrated cells on the lower surface of the filters was quantified under a microscope. The experiments were conducted three times. Cell invasion assay Matrigel-Transwell invasion assays were conducted using Matrigel-coated 24-well Transwell inserts with 8 μm pore membranes (3422, Corning Inc., Lowell, MA, USA). The Matrigel Basement Membrane Matrix (356234, BD Biosciences) was diluted in serum-free RPMI-1640 medium at a 1:6 ratio. The subsequent steps followed the aforementioned protocol for the cell migration assay. Immunofluorescence analysis Cells were sequentially fixed in 4% PFA for 20 minutes, permeabilized with 0.3% Triton X-100 in PBS for 15 minutes, and blocked with 1% sheep serum in PBS for 15 minutes before immunofluorescence analysis. The cells were incubated overnight at 4 °C with an anti-β-catenin antibody (17565–1-AP, Proteintech, USA, 1:500), followed by a 1-hour incubation in the dark with Rhodamine (TRITC)-conjugated goat anti-rabbit IgG (SA00007-2, Proteintech, USA, 1:300). DAPI was used to counterstain the cell nuclei at room temperature for 10 minutes. Images were captured using an EVOS M5000 Fluorescence Imaging System (AMF5000, Thermo Scientific, USA). Experimental animals, xenograft assay and metastasis assay BALB/c nude mice were procured from Huafukang Biotechnology and housed under pathogen-free conditions. All animal experiments were performed in accordance with the institutional guidelines for the care and use of laboratory animals and were approved by the Institutional Animal Care and Use Committee of China Medical University (approval nos. CMU2021517 and CMU2021518). In xenograft assays, ten 5-week-old female BALB/c nude mice were randomly assigned to two groups of five each. Mice received a subcutaneous injection of 2 × 10^6 SGC-7901 cells (either shNC or shFMR1) in 0.2 mL of serum-free RPMI-1640 medium into the upper back. Tumor size was assessed every three days after the tumors became substantial. The formula used to calculate tumor volume was V = (width^2 × length) / 2. Thirty days post-injection, the inoculated mice were euthanized, and their tumors were excised for analysis. In the metastasis nude mouse assay, twelve 5-week-old female BALB/c nude mice were randomly assigned to two groups of six mice each. 1 × 10^6 SGC-7901 cells (shNC or shFMR1) suspended in 0.1 mL of serum-free RPMI-1640 medium were injected into nude mice via the lateral tail vein. Forty days post-injection, the mice were euthanized and dissected. The metastatic foci in the lung tissue were quantified via H&E staining. RIP‒qPCR Protein A/G magnetic beads (HY-K0202, MCE) were incubated with 5 μg of either anti-FMRP (13755–1-AP, Proteintech, USA) or anti-rabbit IgG antibodies (A7016, Beyotime) for 1 hour at room temperature. The HGC-27 cell lysate was incubated with antibody-coated magnetic beads overnight at 4 °C. The magnetic bead-protein‒RNA complexes were washed five times using 1×NT2 buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1 mM MgCl2, and 0.05% NP-40) and then incubated in proteinase K digestion buffer (1×NT2 buffer, 1% SDS, 10 mg/mL proteinase K) at 55 °C with shaking for 30 minutes. RNA extraction was performed via the phenol-chloroform method, followed by reverse transcription using HiScript II Q Select RT SuperMix for qPCR with gDNA wiper reagent (R233-01, Vazyme). Interactions between FMRP and DVL2 mRNA were assessed using qPCR, and results were normalized to input controls. m6A-methylated RNA immunoprecipitation (MeRIP)-qPCR MeRIP-qPCR assay was performed to determine m6A levels. Total RNA was isolated from HGC-27 cells using Vazyme’s Total RNA Extraction Reagent (R401-01) following the provided protocol. Protein A/G magnetic beads (HY-K0202, MCE) were coated with 10 μg of either anti-m6A antibody (#202003, Synaptic Systems) or anti-rabbit IgG (A7016, Beyotime) in an IP buffer (20 mM Tris pH 7.5, 140 mM NaCl, 1% NP-40, and 2 mM EDTA) with RNase/protease inhibitors for 1 hour. 100 μg of total RNA was incubated with the antibody-bound beads for 2 hours at 4 °C with rotation. The m6A-modified RNA was eluted from the dynabeads and precipitated using 5 μg of glycogen, 0.1 volumes of 3 M sodium acetate, and 2.5 volumes of 100% ethanol, then stored at −80 °C overnight. m6A enrichment was determined using qPCR analysis and normalized relative to the input. RNA stability assay HGC-27 and SGC-7901 cells were incubated overnight in 6-well plates. The cells were incubated with 5 μg/mL Actinomycin D (APExBIO, USA) for 0, 2, or 4 hours, after which total RNA was extracted and analyzed using qPCR. RNA levels at various time points were calculated and normalized to the expression level of β-actin. Dual-luciferase reporter assay DVL2 wild-type reporter constructs containing the candidate m6A site regions were generated by cloning the corresponding DVL2 sequences into the pmirGLO dual-luciferase reporter vector. Mutant reporter constructs were generated by site-directed mutagenesis of the candidate m6A motif(s). All constructs were verified by Sanger sequencing. For the dual-luciferase reporter assay, HGC-27 cells were seeded into 24-well plates and co-transfected with DVL2 wild-type or mutant reporter plasmids together with siNC or siFMR1 using jetPRIME transfection reagent according to the manufacturer’s instructions. After transfection, firefly and Renilla luciferase activities were measured using a dual-luciferase reporter assay system. Firefly luciferase activity was normalized to Renilla luciferase activity. Each experiment was performed with at least three independent biological replicates. For dual-luciferase reporter assays involving siRNA treatment and reporter genotype, statistical significance was analyzed using two-way ANOVA followed by appropriate multiple-comparison tests. Human tissue microarray and immunohistochemistry (IHC) A commercial human tissue microarray containing 80 pairs of gastric cancer (GC) tissues and matched adjacent normal gastric tissues (Cat No. STC1602) was acquired from Shanghai Superbiotek Pharmaceutical Technology Co., Ltd. (Shanghai, China). Immunohistochemical staining was performed using a commercial immunohistochemistry kit (Servicebio, Wuhan, China) according to the manufacturer’s instructions. The stained sections were digitally quantified using QuPath software. Due to tissue core loss or unevaluable staining, 79 paired samples were evaluable for FMRP and 70 paired samples were evaluable for DVL2. For survival analysis, the optimal cutoff value for FMRP H-score was determined using maximally selected rank statistics implemented in the surv_cutpoint() function of the R package survminer. Patients were subsequently classified into FMRP-high and FMRP-low expression groups according to this cutoff. The correlation between FMRP and DVL2 protein expression was assessed using Spearman’s rank correlation analysis based on paired semi-quantitative H-scores obtained from 70 gastric cancer tissue samples. All tests were two-sided, and P < 0.05 was considered statistically significant. This study was approved by the Institutional Ethics Committee of China Medical University (approval no. [2026]165), and the study was performed in accordance with the principles of the Declaration of Helsinki. Public database The transcriptional profiles of FMR1 in normal and various tumor tissues were examined using the GEPIA website (http://gepia2.cancer-pku.cn) [31], the TNMplot website (https://tnmplot.com) [32], the UALCAN website (n = 515; http://ualcan.path.uab.edu) [33], and GSE27342 dataset from the GEO database (http://www.ncbi.nlm.nih.gov/geo/) [34, 35]. DNA variations in gastric adenocarcinoma tissue samples were examined using three datasets retrieved from the TCGA database (PanCancer Atlas, n = 440; Nature2014, n = 295; Firehose Legacy, n = 478; https://tcga-data.nci.nih.gov/tcga), accessed via the cBioPortal website (https://www.cbioportal.org) [36, 37]. “FMRP” was used as a keyword to retrieve CLIP-seq data for FMRP-binding mRNAs via the Starbase website (http://starbase.sysu.edu.cn/) [38]. To provide additional support for the selection of DVL2 as a downstream candidate of FMRP, a focused bioinformatic analysis was performed using public RNA-binding, epitranscriptomic, and transcriptomic datasets. Putative FMRP-binding transcripts were obtained from the ENCORI database based on available CLIP-seq evidence [39] (Table S4). Genes with experimentally supported m6A modification were collected from a public m6A-seq dataset [40] (Table S5). In addition, FMR1/Fmr1 knockout-responsive genes were extracted from two independent RNA-seq datasets, GSE107434 and GSE121809 [18, 41] (Table S6 and Table S7). The overlapping genes among FMRP-binding transcripts, m6A-modified genes, and FMR1/Fmr1 knockout-responsive genes were identified by Venn analysis and were considered candidate genes potentially associated with FMRP-mediated regulation. Functional enrichment analysis of the overlapping genes was performed using Metascape, and pathways with P < 0.05 were considered significantly enriched. Statistical analysis GraphPad Prism 10 (San Diego, CA, USA) was used for data analysis and presentation. A two-tailed unpaired t-test was used for comparing two groups, while analysis of variance (ANOVA) was applied for comparisons involving multiple groups. Results from three independent experiments are expressed as means ± SD, and P < 0.05 was considered statistically significant (*P < 0.05, ** P< 0.01, *** P < 0.001, ****P < 0.0001, ns, not significant). Results FMR1 expression is elevated in gastric cancer cells To investigate the expression pattern of FMR1 across different human cancers, pan-cancer analysis was performed using the TNMplot and GEPIA databases. The study found increased mRNA expression levels of FMR1 in various human cancers, such as gastric cancer (Fig. 1A, B). The expression of FMR1 in gastric cancer tissues was further analyzed and compared with that in healthy gastric tissues using the UALCAN and TNMplot databases. As anticipated, the results indicated that the expression level of FMR1 was significantly higher in gastric cancer tissue compared to healthy gastric tissue (Fig. 1C,D). Additionally, FMR1 gene expression patterns in gastric cancer versus paired adjacent normal gastric tissues was analyzed using GeneChip data from the TNMplot database. Paired analysis confirmed the upregulation of FMR1 expression in gastric cancer tissues relative to adjacent normal gastric tissues (Fig. 1E). Furthermore, the frequency and types of genetic alterations in FMR1 were investigated in a cohort of 1213 gastric cancer patient samples based on DNA-seq data using the cBioPortal database. Interestingly, the frequency of FMR1 gene variations among patients with gastric adenocarcinoma was approximately 4–5%. These variations include mutations, amplifications, and deep deletions, as illustrated in Fig. 1F. Amplification, signifying an increase in DNA fragment copy number, was the most prevalent type of FMR1 genetic alteration. Moreover, FMR1 copy number alterations were positively correlated with FMR1 mRNA expression levels in gastric cancer samples (Fig. 1G), suggesting that FMR1 gene copy number alterations may contribute to its upregulation in gastric cancer samples. To further validate the clinical relevance of FMRP expression, we performed immunohistochemical staining using a gastric cancer tissue microarray containing paired tumor and adjacent normal tissues. Consistent with the database analyses, FMRP protein expression was significantly increased in gastric cancer tissues compared with matched adjacent tissues (Fig. 1H,I). Kaplan–Meier survival analysis showed a trend toward poorer overall survival in patients with high FMRP expression, although the difference did not reach statistical significance (log-rank P = 0.0938; Fig. 1J). These findings clarify the upregulation of FMR1 expression in gastric cancer and suggest its potential involvement in tumor initiation and progression.Fig. 1FMR1 expression is elevated in gastric cancer tissues. (A) Differential expression of FMR1 across cancers according to RNA-Seq data retrieved from the TNMplot database (https://tnmplot.com/). Significant differences are marked in red font. (B) FMR1 expression profiles across 31 different types of cancer tissues and adjacent healthy tissues from the TCGA were analyzed using the GEPIA website (http://gepia.cancer-pku.cn/). (C-D) Differences in FMR1 expression between gastric cancer tissues and healthy gastric tissues using data from (C) The UALCAN database (http://ualcan.path.uab.edu/) and (D) The TNMplot database. (E) Expression analysis of FMR1 mRNA in paired tumor and adjacent normal gastric tissues using TNMplot GeneChip data. (F) The frequency of FMR1 gene variation in gastric cancer tissues was analyzed using the cBioportal database (https://www.cbioportal.org/). (G) The correlation between mRNA expression levels and copy number values of FMR1 in the TCGA STAD dataset. (H) Representative immunohistochemical staining of FMRP in paired adjacent non-tumour and tumour tissues at low (×8) and high (×40) magnification. Brown staining indicates FMRP expression, and nuclei were counterstained with haematoxylin. Scale bars, 200 μm (low magnification) and 40 μm (high magnification). (I) Paired comparison of FMRP immunohistochemical H-scores between adjacent non-tumour (normal) and tumour tissues. Lines connect matched samples. (n = 79; P = 0.0488, two-sided exact Wilcoxon matched-pairs signed-rank test). (J) Kaplan–Meier analysis of overall survival in patients stratified into FMRP-low (n = 42) and FMRP-high (n = 37) groups. Tick marks indicate censored observations. P = 0.0938, log-rank test. In panels A, C, D, E and I, statistical significance is indicated by *P < 0.05 and ****P < 0.0001 FMR1 expression is elevated in gastric cancer tissues. (A) Differential expression of FMR1 across cancers according to RNA-Seq data retrieved from the TNMplot database (https://tnmplot.com/). Significant differences are marked in red font. (B) FMR1 expression profiles across 31 different types of cancer tissues and adjacent healthy tissues from the TCGA were analyzed using the GEPIA website (http://gepia.cancer-pku.cn/). (C-D) Differences in FMR1 expression between gastric cancer tissues and healthy gastric tissues using data from (C) The UALCAN database (http://ualcan.path.uab.edu/) and (D) The TNMplot database. (E) Expression analysis of FMR1 mRNA in paired tumor and adjacent normal gastric tissues using TNMplot GeneChip data. (F) The frequency of FMR1 gene variation in gastric cancer tissues was analyzed using the cBioportal database (https://www.cbioportal.org/). (G) The correlation between mRNA expression levels and copy number values of FMR1 in the TCGA STAD dataset. (H) Representative immunohistochemical staining of FMRP in paired adjacent non-tumour and tumour tissues at low (×8) and high (×40) magnification. Brown staining indicates FMRP expression, and nuclei were counterstained with haematoxylin. Scale bars, 200 μm (low magnification) and 40 μm (high magnification). (I) Paired comparison of FMRP immunohistochemical H-scores between adjacent non-tumour (normal) and tumour tissues. Lines connect matched samples. (n = 79; P = 0.0488, two-sided exact Wilcoxon matched-pairs signed-rank test). (J) Kaplan–Meier analysis of overall survival in patients stratified into FMRP-low (n = 42) and FMRP-high (n = 37) groups. Tick marks indicate censored observations. P = 0.0938, log-rank test. In panels A, C, D, E and I, statistical significance is indicated by *P < 0.05 and ****P < 0.0001 Targeting FMRP expression impairs gastric cancer progression To select appropriate gastric cancer cell lines for functional studies, we assessed FMR1 mRNA and protein expression in four gastric cancer cell lines and the normal human gastric epithelial cell line GES-1. FMR1 expression was significantly upregulated at both mRNA and protein levels in HGC-27 and SGC-7901 cells compared with GES-1 cells (Fig. 2A–C). However, no significant difference was observed between MGC-803 and AGS cells. To further explore the impact of FMRP on gastric cancer cells and reduce the possibility of sequence-specific off-target effects, three independent siRNAs targeting FMR1 were evaluated in SGC-7901 and HGC-27 cells. Following transfection with these siRNAs, FMR1 expression was reduced at both the mRNA and protein levels (Fig. 2D–F and Supplementary Fig. S1A-C). Based on its robust knockdown efficiency, siFMR1#1 and siFMR1#2 were selected for further phenotypic validation. Subsequent CCK-8 and colony formation assays revealed that FMRP knockdown significantly inhibited cell proliferation (Fig. 2G–I). Additionally, FMRP knockdown induced G1-phase cell-cycle arrest (Fig. 2J). Wound healing and Transwell migration assays revealed that reducing FMRP expression suppressed gastric cancer cell migration (Fig. 2K–N and Supplementary Fig. S1D-G). Furthermore, in the Matrigel Transwell invasion assay, FMRP knockdown impaired gastric cancer cell invasive capacity (Fig. 2O,P). In summary, these findings collectively suggest that FMRP knockdown inhibits gastric cancer cell proliferation, migration, and invasion in vitro.Fig. 2FMRP knockdown inhibits gastric cancer cell proliferation, migration and invasion in vitro. (A) Relative mRNA expression of FMR1 in four gastric cancer cell lines (HGC-27, SGC-7901, AGS and MGC-803) and the normal gastric epithelial cell line GES-1 quantified by RT‒qPCR. Expression levels were normalized to those of GES-1 cells, with β-actin serving as the internal control. Data are shown as mean ± SD (n = 4 for the AGS group and n = 3 for other groups). (B-C) Expression of FMRP in gastric cancer cell lines. (B) Representative western blot images. (C) Densitometric quantification of the blots shown in (B) using ImageJ. (D-F) Validation of FMRP knockdown efficiency by siFMR1#1–#3. (D) FMR1 mRNA levels were measured by RT-qPCR and normalized to β-actin. (E-F) Representative western blot images of FMRP expression in HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1–#3. (G) CCK-8 cell proliferation assay for HGC-27 and SGC-7901 cells transfected with siNC, siFMR1#1, or siFMR1#2. (H-I) Colony formation assays of HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1. (H) Representative images. (I) Quantification of colony numbers. (J) Cell cycle analysis of HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1 via flow cytometry. (K-L) Wound healing assays of HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1. (K) Representative images (original magnification, 100×; scale bar, 200 μm). (L) Quantification of wound closure. (M-N) Transwell migration assays of HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1. (M) Representative images (original magnification, 100×; scale bar, 200 μm). (N) Quantification of migrated cells. (O-P) Matrigel invasion assays of HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1. (O) Representative images (original magnification, 100×; scale bar, 200 μm). (P) Quantification of invaded cells. Unless otherwise indicated, data are presented as the mean ± SD of three independent experiments. *P < 0.05, **P < 0.01, ****P < 0.0001; ns, not significant FMRP knockdown inhibits gastric cancer cell proliferation, migration and invasion in vitro. (A) Relative mRNA expression of FMR1 in four gastric cancer cell lines (HGC-27, SGC-7901, AGS and MGC-803) and the normal gastric epithelial cell line GES-1 quantified by RT‒qPCR. Expression levels were normalized to those of GES-1 cells, with β-actin serving as the internal control. Data are shown as mean ± SD (n = 4 for the AGS group and n = 3 for other groups). (B-C) Expression of FMRP in gastric cancer cell lines. (B) Representative western blot images. (C) Densitometric quantification of the blots shown in (B) using ImageJ. (D-F) Validation of FMRP knockdown efficiency by siFMR1#1–#3. (D) FMR1 mRNA levels were measured by RT-qPCR and normalized to β-actin. (E-F) Representative western blot images of FMRP expression in HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1–#3. (G) CCK-8 cell proliferation assay for HGC-27 and SGC-7901 cells transfected with siNC, siFMR1#1, or siFMR1#2. (H-I) Colony formation assays of HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1. (H) Representative images. (I) Quantification of colony numbers. (J) Cell cycle analysis of HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1 via flow cytometry. (K-L) Wound healing assays of HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1. (K) Representative images (original magnification, 100×; scale bar, 200 μm). (L) Quantification of wound closure. (M-N) Transwell migration assays of HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1. (M) Representative images (original magnification, 100×; scale bar, 200 μm). (N) Quantification of migrated cells. (O-P) Matrigel invasion assays of HGC-27 and SGC-7901 cells transfected with siNC or siFMR1#1. (O) Representative images (original magnification, 100×; scale bar, 200 μm). (P) Quantification of invaded cells. Unless otherwise indicated, data are presented as the mean ± SD of three independent experiments. *P < 0.05, **P < 0.01, ****P < 0.0001; ns, not significant FMRP promotes gastric cancer cell growth and metastasis in vivo To investigate the role of FMRP in gastric cancer in vivo, we generated stable SGC-7901 and HGC-27 cell lines with FMRP knockdown using lentiviruses carrying FMR1-targeting shRNA (shFMR1) or control shRNA (shNC), followed by puromycin selection. Lentiviral transduction efficiency and FMRP knockdown were validated by GFP expression, RT-qPCR, and western blotting (Fig. 3A–D). A xenograft model was established in nude mice to evaluate the effect of FMRP knockdown on gastric tumor growth. FMRP knockdown significantly suppressed SGC-7901 tumor growth in vivo, as evidenced by reduced tumor volume and weight compared with the control group (Fig. 3E–G). To further investigate the role of FMRP in metastasis, stable FMRP-knockdown or control SGC-7901 cells were injected into nude mice via the lateral tail vein. Lungs were then examined for metastatic foci, which were observed in the control group but rarely detected in the FMRP-knockdown group (Fig. 3H–J). In summary, FMRP promotes the growth and metastasis of gastric cancer cells in vivo.Fig. 3Knockdown of FMRP inhibits tumor growth and metastasis in vivo. (A) Transfection efficiency of shNC or shFMR1 lentivirus in HGC-27 and SGC-7901 cells, assessed by EGFP expression (original magnification, 100×; scale bar, 200 μm). (B-D) Validation of FMR1 knockdown efficiency in HGC-27 and SGC-7901 cells transduced with shNC or shFMR1 lentivirus. (B) FMR1 mRNA levels were measured by RT-qPCR and normalized to β-actin. (C) Representative western blot images of FMRP expression. (D) Densitometric quantification of the blots shown in (C) using ImageJ. (E-G) FMRP knockdown suppresses gastric tumor growth in vivo. SGC-7901 cells stably expressing a control shRNA (shNC) or shRNA targeting FMR1 (shFMR1) were subcutaneously injected into the flanks of nude mice (n = 5). (E) Representative images of dissected tumors at endpoint. (F) Tumor growth curves measured at indicated time points. (G) Tumor weights at endpoint. (H-J) FMRP knockdown suppresses gastric cancer metastasis in vivo. (H) Representative images of lungs from nude mice injected with shNC or shFMR1 SGC-7901 cells. (I) Representative H&E-stained sections of lung metastatic nodules (scale bar, 500 μm). (J) Quantification of metastatic foci per lung (n = 6). Data are presented as mean ± SD. *P< 0.05, **P < 0.01 compared with shNC group Knockdown of FMRP inhibits tumor growth and metastasis in vivo. (A) Transfection efficiency of shNC or shFMR1 lentivirus in HGC-27 and SGC-7901 cells, assessed by EGFP expression (original magnification, 100×; scale bar, 200 μm). (B-D) Validation of FMR1 knockdown efficiency in HGC-27 and SGC-7901 cells transduced with shNC or shFMR1 lentivirus. (B) FMR1 mRNA levels were measured by RT-qPCR and normalized to β-actin. (C) Representative western blot images of FMRP expression. (D) Densitometric quantification of the blots shown in (C) using ImageJ. (E-G) FMRP knockdown suppresses gastric tumor growth in vivo. SGC-7901 cells stably expressing a control shRNA (shNC) or shRNA targeting FMR1 (shFMR1) were subcutaneously injected into the flanks of nude mice (n = 5). (E) Representative images of dissected tumors at endpoint. (F) Tumor growth curves measured at indicated time points. (G) Tumor weights at endpoint. (H-J) FMRP knockdown suppresses gastric cancer metastasis in vivo. (H) Representative images of lungs from nude mice injected with shNC or shFMR1 SGC-7901 cells. (I) Representative H&E-stained sections of lung metastatic nodules (scale bar, 500 μm). (J) Quantification of metastatic foci per lung (n = 6). Data are presented as mean ± SD. *P< 0.05, **P < 0.01 compared with shNC group DVL2 is a candidate downstream target of FMRP in gastric cancer cells To identify candidate downstream targets of FMRP, we performed an integrative bioinformatic analysis of public RNA-binding, epitranscriptomic, and transcriptomic datasets. High-confidence FMRP-binding transcripts were intersected with experimentally supported m6A-modified transcripts and genes responsive to FMR1/Fmr1 knockout in two independent RNA-seq datasets. This analysis identified 179 overlapping candidate genes, including DVL2 (Supplementary Fig. S2A and Table S8). Functional enrichment analysis of these overlapping genes revealed enrichment of Wnt signaling-related pathways (Supplementary Fig. S2B). Mapping the 179 genes onto the KEGG Wnt signaling pathway identified DVL2 and APC as the only candidates corresponding to core pathway components. Since DVL2 functions as a central scaffold downstream of Frizzled receptors and mediates both canonical Wnt/β-catenin and noncanonical Wnt/PCP signaling. we next examined whether DVL2 was regulated by FMRP in gastric cancer cells. As displayed in Fig. 4A–C and Supplementary Fig. S2C, FMRP knockdown significantly reduced DVL2 expression at both mRNA and protein levels in HGC-27 and SGC-7901 cells. The interaction between FMRP and DVL2 mRNA was validated using RIP‒qPCR in HGC-27 cells. Anti-FMRP antibodies significantly enriched endogenous DVL2 mRNA compared with the IgG control (Fig. 4D–F). Vimentin (VIM) mRNA was used as a positive control, whereas hypoxanthine phosphoribosyl transferase 1 (HPRT1) mRNA was used as a negative control, respectively [26]. We next evaluated the clinical relevance of DVL2 using a gastric cancer tissue microarray containing paired tumor and adjacent normal tissues. Immunohistochemical staining demonstrated that DVL2 protein expression was significantly increased in gastric cancer tissues compared with matched adjacent normal tissues (Fig. 4G,H). Moreover, correlation analysis of tissue microarray specimens revealed a significant positive association between FMRP and DVL2 protein expression in gastric cancer tissues (Fig. 4I). Consistent with these clinical findings, analysis of the GEPIA database also demonstrated a significant positive correlation between FMR1 and DVL2 mRNA expression in STAD (Fig. 4J), as well as in other cancer types including BRCA, COAD, LUAD, and PRAD (Supplementary Fig. S2D-G). Collectively, these findings support DVL2 as a clinically relevant downstream target of FMRP in gastric cancer. In addition, Kaplan–Meier Plotter analysis demonstrated that elevated DVL2 mRNA expression was significantly associated with poor prognosis in gastric cancer (Fig. 4K). These findings suggest that FMRP may promote the proliferation, migration, and invasion of gastric cancer cells by targeting DVL2.Fig. 4DVL2 is a potential target of FMRP in gastric cancer cells. (A) RT-qPCR analysis of FMR1 and DVL2 mRNA expression levels in FMRP-knocked down (shFMR1) HGC-27 and SGC-7901 cells. (B-C) Expression level of the DVL2 protein in HGC-27 and SGC-7901 cells with FMRP knockdown (shFMR1). (B) Representative western blot images. (C) Densitometric quantification of DVL2 protein levels normalized to β-actin and expressed relative to the shNC group. (D-E) RIP-qPCR assays were performed to determine the association between FMRP and DVL2 mRNA. (D) IP efficiency of the FMRP antibody was validated by western blotting. (E) Relative enrichment of DVL2 mRNA associated with FMRP was analyzed using RT-qPCR. (F) DVL2 mRNA enrichment in IgG and RIP-FMRP groups, presented as percentage of input. HPRT1 served as a negative control, whereas VIM served as a positive control, respectively. Data are presented as mean ± SD of three independent experiments. *P < 0.05, **P < 0.01, ns; not significant. (G) Representative immunohistochemical staining of DVL2 in paired adjacent non-tumour and gastric tumour tissues at low (×8) and high (×40) magnification. Brown staining indicates DVL2 expression, and nuclei were counterstained with haematoxylin. Scale bars, 200 μm (low magnification) and 40 μm (high magnification). (H) Paired comparison of DVL2 immunohistochemical H-scores between adjacent non-tumour (normal) and tumour tissues. Lines connect matched samples (n = 70). P = 0.0384, two-sided exact Wilcoxon matched-pairs signed-rank test. (I) Correlation between FMRP and DVL2 H-scores in gastric cancer tissues (spearman rₛ = 0.3993, 95% CI: 0.1745–0.5846, two-sided P = 0.0006, n = 70). (J) Correlation between FMR1 and DVL2 mRNA expression in gastric cancer tissues analyzed using GEPIA. (K) Kaplan–Meier analysis of overall survival according to DVL2 mRNA expression using Kaplan–Meier plotter (HR = 1.31, 95% CI: 1.10–1.55; log-rank P = 0.0021) DVL2 is a potential target of FMRP in gastric cancer cells. (A) RT-qPCR analysis of FMR1 and DVL2 mRNA expression levels in FMRP-knocked down (shFMR1) HGC-27 and SGC-7901 cells. (B-C) Expression level of the DVL2 protein in HGC-27 and SGC-7901 cells with FMRP knockdown (shFMR1). (B) Representative western blot images. (C) Densitometric quantification of DVL2 protein levels normalized to β-actin and expressed relative to the shNC group. (D-E) RIP-qPCR assays were performed to determine the association between FMRP and DVL2 mRNA. (D) IP efficiency of the FMRP antibody was validated by western blotting. (E) Relative enrichment of DVL2 mRNA associated with FMRP was analyzed using RT-qPCR. (F) DVL2 mRNA enrichment in IgG and RIP-FMRP groups, presented as percentage of input. HPRT1 served as a negative control, whereas VIM served as a positive control, respectively. Data are presented as mean ± SD of three independent experiments. *P < 0.05, **P < 0.01, ns; not significant. (G) Representative immunohistochemical staining of DVL2 in paired adjacent non-tumour and gastric tumour tissues at low (×8) and high (×40) magnification. Brown staining indicates DVL2 expression, and nuclei were counterstained with haematoxylin. Scale bars, 200 μm (low magnification) and 40 μm (high magnification). (H) Paired comparison of DVL2 immunohistochemical H-scores between adjacent non-tumour (normal) and tumour tissues. Lines connect matched samples (n = 70). P = 0.0384, two-sided exact Wilcoxon matched-pairs signed-rank test. (I) Correlation between FMRP and DVL2 H-scores in gastric cancer tissues (spearman rₛ = 0.3993, 95% CI: 0.1745–0.5846, two-sided P = 0.0006, n = 70). (J) Correlation between FMR1 and DVL2 mRNA expression in gastric cancer tissues analyzed using GEPIA. (K) Kaplan–Meier analysis of overall survival according to DVL2 mRNA expression using Kaplan–Meier plotter (HR = 1.31, 95% CI: 1.10–1.55; log-rank P = 0.0021) FMRP regulates the stability of DVL2 mRNA in an m6A-dependent manner To determine the m6A modification status of DVL2 mRNA in gastric cancer cells, a gene-specific m6A assay was conducted. The results demonstrated significant enrichment of m6A in DVL2 mRNA (Fig. 5A). To further assess whether the interaction between FMRP and DVL2 mRNA is m6A dependent, HGC-27 cells were treated with cycloleucine, a global m6A inhibitor, followed by RIP-qPCR to assess the enrichment of DVL2 mRNA by FMRP. As depicted in Fig. 5B–D, the enrichment of DVL2 mRNA by FMRP was significantly lower in cycloleucine-treated HGC-27 cells compared to untreated HGC-27 cells. RNA stability assay revealed that FMRP knockdown significantly accelerated the degradation of DVL2 mRNA in HGC-27 and SGC-7901 cells compared to controls (Fig. 5E,F), indicating that FMRP plays an important role in maintaining DVL2 mRNA stability. In conclusion, these findings demonstrate that FMRP interacts with DVL2 mRNA and maintains its stability in an m6A-dependent manner.Fig. 5FMRP regulates the stability of DVL2 mRNA in an m6A-dependent manner. (A) Validation of DVL2 mRNA m6A modification. In HGC-27 cells by MeRIP-qPCR. (B-D) Effect of cycloleucine on the association of FMRP and DVL2 mRNA in HGC-27 cells, analyzed by RIP-qPCR assays. (B) Validation of FMRP antibody IP efficiency by western blotting. (C) Relative enrichment of DVL2 mRNA associated with FMRP in HGC-27 cells treated with or without cycloleucine. (D) DVL2 mRNA enrichment in IgG and RIP-FMRP groups, presented as percentage of input. (E-F) DVL2 mRNA stability was analyzed in (E) HGC-27 and (F) SGC-7901 cells following FMRP knockdown (shFMR1) in the presence of actinomycin D. (G) Schematic representation of the DVL2 mRNA structure and the candidate m6A sites at 706A/790A, 1271A and 2038A. The corresponding wild-type (WT) sequences and mutant (MUT) reporter constructs are shown. (H) Dual-luciferase reporter assays evaluating the effect of FMR1 knockdown on WT and MUT DVL2 reporters encompassing the 1271A candidate site. Firefly luciferase activity was normalized to Renilla luciferase activity. Data are presented as the mean ± SD of three independent experiments. ns, not significant; *P < 0.05 and **P < 0.01 FMRP regulates the stability of DVL2 mRNA in an m6A-dependent manner. (A) Validation of DVL2 mRNA m6A modification. In HGC-27 cells by MeRIP-qPCR. (B-D) Effect of cycloleucine on the association of FMRP and DVL2 mRNA in HGC-27 cells, analyzed by RIP-qPCR assays. (B) Validation of FMRP antibody IP efficiency by western blotting. (C) Relative enrichment of DVL2 mRNA associated with FMRP in HGC-27 cells treated with or without cycloleucine. (D) DVL2 mRNA enrichment in IgG and RIP-FMRP groups, presented as percentage of input. (E-F) DVL2 mRNA stability was analyzed in (E) HGC-27 and (F) SGC-7901 cells following FMRP knockdown (shFMR1) in the presence of actinomycin D. (G) Schematic representation of the DVL2 mRNA structure and the candidate m6A sites at 706A/790A, 1271A and 2038A. The corresponding wild-type (WT) sequences and mutant (MUT) reporter constructs are shown. (H) Dual-luciferase reporter assays evaluating the effect of FMR1 knockdown on WT and MUT DVL2 reporters encompassing the 1271A candidate site. Firefly luciferase activity was normalized to Renilla luciferase activity. Data are presented as the mean ± SD of three independent experiments. ns, not significant; *P < 0.05 and **P < 0.01 To obtain more direct evidence that specific candidate m6A sites within DVL2 mRNA contribute to FMRP-mediated regulation, we performed site-directed mutagenesis. Based on experimentally validated m6A sites from the m6A-Atlas database and high-confidence predictions from SRAMP, three candidate m6A site regions within DVL2 mRNA were selected for site-directed mutagenesis, including two experimentally validated sites located at chr17:7226293 (2038A) and chr17:7228987 (1271A), as well as one predicted candidate region containing adjacent high-confidence sites (706A and 790A). Wild-type and mutant DVL2 reporter constructs were generated and co-transfected with siNC or siFMR1, followed by dual-luciferase reporter assays. Compared with siNC, FMRP knockdown reduced the luciferase activity of the wild-type DVL2 reporter. Mutation of the 1271A m6A site attenuated the regulatory effect of FMR1 knockdown on DVL2 reporter activity (Fig. 5G,H). By contrast, mutation of the other candidate m6A sites did not significantly alter the response of the DVL2 reporter to FMRP knockdown (Supplementary Fig. S2H,I). These findings suggest that the 1271A candidate site may contribute, at least in part, to FMRP-mediated regulation of DVL2 reporter activity. Together, these results further support the notion that FMRP regulates DVL2 mRNA through an m6A-dependent mechanism. Knock down of FMRP inhibits the noncanonical Wnt/PCP pathway Next, we investigated the molecular mechanism by which FMRP knockdown suppresses gastric cancer progression. Given the dysregulation of DVL2 protein functions in regulating canonical Wnt/β-catenin and noncanonical Wnt/PCP signaling pathways, we investigated the role of FMRP in these pathways using HGC-27 and SGC-7901 cells. Initially, the impact of FMRP on β-catenin expression and phosphorylation in gastric cancer cells was assessed. However, our findings suggested that knockdown of FMRP did not impact the expression or phosphorylation levels of β-catenin (Fig. 6A,B). Considering that β-catenin function in the canonical Wnt/β-catenin pathway depends on its nuclear localization, an immunofluorescence assay was conducted to examine the location of β-catenin in gastric cancer cells. No significant difference was observed between FMRP knockdown and control cells (Fig. 6C). Taken together, these findings indicate that FMRP does not influence the canonical Wnt/β-catenin pathway in gastric cancer cells. Next, the effect of FMRP on the noncanonical Wnt/PCP pathway was examined. The results demonstrated that FMRP knockdown inhibited the phosphorylation of JNK and c-Jun, key markers of noncanonical Wnt/PCP pathway activation (Fig. 6D,E). These results indicate that FMRP promotes the noncanonical Wnt/PCP pathway in gastric cancer cells.Fig. 6FMRP knockdown inhibits the Wnt/PCP-JNK signaling pathway. (A-B) FMRP knockdown does not affect β-catenin expression or phosphorylation in cells transfected with siNC or siFMR1#1. (A) Representative western blot images. (B) Densitometric quantification of FMRP and total β-catenin normalized to β-actin and p-β-catenin normalized to total β-catenin. (C) Representative images illustrating the localization of β-catenin (red) in HGC-27 and SGC-7901 cells with FMRP knockdown (shFMR1) compared with cells transfected with the negative control (shNC). Nuclei were stained with DAPI (blue). (Original magnification, 100×; scale bar, 100 μm). (D-E) FMRP knockdown inhibits JNK and c-Jun phosphorylation in gastric cancer cells transfected with siNC or siFMR1#1. (D) Representative western blot images of phosphorylated and total JNK and c-Jun in HGC-27 and SGC-7901 cells transfected with siFMR1#1 or the negative control siNC. (E) Densitometric quantification of p-JNK normalized to total JNK and p-c-Jun normalized to total c-Jun. The above data are presented as mean ± SD of three independent experiments. *P < 0.05, **P < 0.01, ***P <0.001, ****P <0.0001; ns, not significant FMRP knockdown inhibits the Wnt/PCP-JNK signaling pathway. (A-B) FMRP knockdown does not affect β-catenin expression or phosphorylation in cells transfected with siNC or siFMR1#1. (A) Representative western blot images. (B) Densitometric quantification of FMRP and total β-catenin normalized to β-actin and p-β-catenin normalized to total β-catenin. (C) Representative images illustrating the localization of β-catenin (red) in HGC-27 and SGC-7901 cells with FMRP knockdown (shFMR1) compared with cells transfected with the negative control (shNC). Nuclei were stained with DAPI (blue). (Original magnification, 100×; scale bar, 100 μm). (D-E) FMRP knockdown inhibits JNK and c-Jun phosphorylation in gastric cancer cells transfected with siNC or siFMR1#1. (D) Representative western blot images of phosphorylated and total JNK and c-Jun in HGC-27 and SGC-7901 cells transfected with siFMR1#1 or the negative control siNC. (E) Densitometric quantification of p-JNK normalized to total JNK and p-c-Jun normalized to total c-Jun. The above data are presented as mean ± SD of three independent experiments. *P < 0.05, **P < 0.01, ***P <0.001, ****P <0.0001; ns, not significant FMRP regulates gastric cancer cell proliferation, migration and invasion via DVL2 To determine whether FMRP promotes gastric cancer cell proliferation, migration, and invasion through DVL2, we performed rescue experiments. HGC-27 cells with stable FMRP knockdown (shFMR1) were transfected with a DVL2 overexpression plasmid (oe-FLAG-DVL2) or the corresponding negative control plasmid (oe-NC). Restoration of DVL2 expression partially rescued the phosphorylation levels of JNK and c-Jun (Fig. 7A–B). Moreover, the colony formation, migration, and invasion capacities of FMRP-knockdown cells were partially rescued upon the restoration of DVL2 protein expression (Fig. 7C–H). Taken together, these findings demonstrate that FMRP promotes malignant phenotypes in gastric cancer cells at least in part by stabilizing DVL2 mRNA and subsequently activating the noncanonical Wnt/PCP pathway.Fig. 7FMRP knockdown inhibited the proliferation, migration, and invasion of gastric cancer cells via DVL2. (A-B) DVL2 overexpression partially rescued JNK and c-Jun phosphorylation in FMRP-knockdown HGC-27 cells. (A) Representative western blot images of phosphorylated and total JNK and c-Jun. (B) Densitometric quantification of FMRP and DVL2 normalized to β-actin, p-JNK normalized to total JNK, and p-c-Jun normalized to total c-Jun. (C-D) DVL2 overexpression partially rescues colony formation in FMRP-knockdown HGC-27 cells. (C) Representative images of colony formation assays. (D) Quantification of colon numbers. (E-F) DVL2 overexpression partially rescues migration in FMRP-knockdown HGC-27 cells. (E) Representative images of Transwell migration assays (original magnification, 100×; scale bar, 200 μm). (F) Quantification of migrated cells per field. (G-H) DVL2 overexpression partially rescues invasion in FMRP-knockdown HGC-27 cells. (G) Representative images of Matrigel invasion assays (original magnification, 100×; scale bar, 200 μm). (H) Quantification of invaded cells per field. Data are presented as mean ± SD of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 FMRP knockdown inhibited the proliferation, migration, and invasion of gastric cancer cells via DVL2. (A-B) DVL2 overexpression partially rescued JNK and c-Jun phosphorylation in FMRP-knockdown HGC-27 cells. (A) Representative western blot images of phosphorylated and total JNK and c-Jun. (B) Densitometric quantification of FMRP and DVL2 normalized to β-actin, p-JNK normalized to total JNK, and p-c-Jun normalized to total c-Jun. (C-D) DVL2 overexpression partially rescues colony formation in FMRP-knockdown HGC-27 cells. (C) Representative images of colony formation assays. (D) Quantification of colon numbers. (E-F) DVL2 overexpression partially rescues migration in FMRP-knockdown HGC-27 cells. (E) Representative images of Transwell migration assays (original magnification, 100×; scale bar, 200 μm). (F) Quantification of migrated cells per field. (G-H) DVL2 overexpression partially rescues invasion in FMRP-knockdown HGC-27 cells. (G) Representative images of Matrigel invasion assays (original magnification, 100×; scale bar, 200 μm). (H) Quantification of invaded cells per field. Data are presented as mean ± SD of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 Discussion Gastric cancer is one of the most aggressive malignancies, hallmarked by rapid progression to advanced stages and highly metastatic properties [42]. The prognosis remains worse for patients diagnosed with gastric cancer due to the absence of early symptoms [43]. Recent studies have pointed out that m6A modification of RNA is involved in regulating gastric cancer progression [12, 15]. The expression of FMRP, an m6A reader protein, was elevated in gastric cancer tissues relative to normal gastric tissues. Knock down of endogenous FMRP in gastric cancer cells significantly inhibited tumor growth in vivo and in vitro, suggesting that FMRP facilitates gastric cancer progression. Mechanistic studies revealed that FMRP interacts with DVL2 mRNA and regulates its mRNA stability in an m6A-dependent manner, thereby driving the noncanonical Wnt/PCP pathway and the proliferation, migration, and invasion of gastric cancer cells through DVL2 (Fig. 8).Fig. 8A schematic model illustrating the proposed mechanism by which FMRP stabilizes DVL2 mRNA in an m6A-dependent manner, activates the Wnt/PCP-JNK signaling pathway, and promotes gastric cancer progression. The figure was generated using Figdraw A schematic model illustrating the proposed mechanism by which FMRP stabilizes DVL2 mRNA in an m6A-dependent manner, activates the Wnt/PCP-JNK signaling pathway, and promotes gastric cancer progression. The figure was generated using Figdraw Fragile X syndrome (FXS), the leading inherited intellectual disability, is mainly due to the silencing of the FMR1 gene [44]. A clinical report highlighted reduced glioblastoma invasion in a patient diagnosed with FXS [45]. Additionally, FXS patients have been reported to have a lower risk and incidence of cancer [26]. While some studies have described the impact of FMRP on the progression of glioblastoma, breast, gastric, and colorectal cancers, the specific mechanisms involved are yet to be clarified [46]. It is worthwhile emphasizing that FMRP is not only an RNA-binding protein but also an m6A modification reader protein [20, 41]. To identify the target genes of FMRP, candidate genes were screened through integrated analysis of public RNA-binding, epitranscriptomic, and transcriptomic datasets. Then, RIP-qPCR was performed to validate the interaction between FMRP and the potential target mRNAs. The results showed that DVL2 is a potential target of FMRP in gastric cancer, and the knockdown of endogenous FMRP inhibited both the mRNA and protein expression of DVL2. The RNA stability assay demonstrated that FMRP knockdown decreased the stability of DVL2 mRNA in gastric cancer cells. Correlation analysis also revealed a positive association between DVL2 and FMR1 expression in gastric cancer tissues. Overall, these results indicate that FMRP governs the expression of DVL2 by binding to its mRNA. To investigate whether the binding of the FMRP protein to DVL2 mRNA is m6A modification dependent, an m6A antibody was used to enrich DVL2 mRNA, and an MeRIP-qPCR assay was performed. Importantly, the results confirmed that DVL2 mRNA undergoes m6A modification in gastric cancer cells. Furthermore, cycloleucine treatment significantly reduced the ability of FMRP to enrich DVL2 mRNA in gastric cancer cells. These results suggest that FMRP mediates DVL2 expression in an m6A-dependent manner. To obtain more direct evidence supporting the m6A dependency of this regulatory axis, we further performed site-specific mutagenesis of candidate m6A sites within DVL2 mRNA. Two experimentally validated high-confidence m6A sites from m6A-Atlas and one predicted candidate region from SRAMP were selected for reporter-based functional validation. The dual-luciferase reporter assays showed that mutation of the 1271A candidate site attenuated the regulatory effect of FMR1 knockdown on DVL2 reporter activity, whereas mutation of the other candidate sites produced no significant effect, suggesting that 1271A contributes to FMRP-mediated regulation of DVL2. These findings provide additional evidence that the FMRP-DVL2 regulatory relationship is associated with m6A-modified regions within DVL2 mRNA. It should be noted that the reporter assay mainly reflects the functional contribution of the selected candidate m6A site(s) to FMRP-mediated regulation of DVL2 reporter activity. Because RNA-binding proteins may recognize multiple sites or broader RNA structural contexts, mutation of a single site may not completely abolish the regulatory effect of FMRP. Therefore, our results support the involvement of the 1271A candidate site in FMRP-dependent regulation of DVL2, while further studies such as CLIP-based mapping or RIP analysis using wild-type and mutant DVL2 transcripts would help to define the precise binding sites more directly. DVL2 is a key component of both the canonical Wnt/β-catenin and noncanonical Wnt/PCP pathways [47, 48]. Dysregulation of canonical Wnt/β-catenin pathway activity was detected in the brains of Fmr1 knockout mice, suggesting that FMRP is involved in the regulation of the canonical Wnt/β-catenin pathway [49]. Given that activation of the canonical Wnt/β-catenin pathway promotes the occurrence and progression of different cancers [50, 51], the present study investigated the involvement of FMRP in regulating the canonical Wnt/β-catenin pathway through DVL2 in gastric cancer cells. There is no evidence that FMRP influences the expression or nuclear localization of β-catenin, the primary regulator of the canonical Wnt/β-catenin pathway, in gastric cancer cells. On the other hand, FMRP knockdown inhibited the phosphorylation of JNK and c-Jun, highlighting its effect on the noncanonical Wnt/PCP pathway in gastric cancer cells. The distinct effects of FMRP on Wnt signaling across different tissues are likely context dependent rather than reflecting an intrinsic preference for a specific Wnt branch. As an m6A reader and RNA-binding protein, FMRP regulates signaling by controlling the stability of tissue-specific target transcripts [41]. Although DVL2 functions as a common signaling hub for both canonical and non-canonical Wnt pathways, its downstream output is determined by the cellular signaling context, including the composition of Wnt ligands, receptors, co-receptors, and interacting proteins [47, 48]. In our gastric cancer model, FMRP depletion reduced JNK and c-Jun phosphorylation without affecting β-catenin expression or nuclear localization, suggesting that stabilization of DVL2 preferentially enhances non-canonical Wnt/PCP signaling in this cellular context. We therefore speculate that the tissue-specific transcriptome and signaling environment determine the downstream consequences of FMRP-mediated DVL2 regulation. Further studies are required to define the molecular determinants governing pathway selection in different tumor types. Notably, DVL2 overexpression partly rescued JNK and c-Jun phosphorylation in FMRP-knockdown gastric cancer cells. Furthermore, restoring DVL2 expression partially rescued the inhibitory effects of FMRP knockdown on gastric cancer cell proliferation, migration, and invasion. Therefore, DVL2 mRNA may not be the sole target through which FMRP regulates gastric cancer progression. From a therapeutic perspective, the FMRP–DVL2 axis identified in this study may provide a potential molecular basis for future targeted intervention in gastric cancer. Recent advances in gastric cancer treatment have increasingly emphasized biomarker-guided therapeutic strategies, including established or emerging approaches targeting HER2, CLDN18.2, FGFR2/FGFR2b, VEGFR2, and immune checkpoints [52]. In this context, interference with FMRP-mediated RNA regulation or DVL2-associated signaling may represent a direction worthy of further investigation. Nevertheless, the translation of such molecular findings into therapeutic strategies requires effective and tumor-selective delivery systems. Nanomedicine and delivery platforms have been explored as strategies to improve drug accumulation, enhance targeting efficiency, and reduce systemic toxicity in gastric cancer [53]. Consistently, Wang and Liao reviewed recent progress in microrobot-based tumor drug delivery and discussed its potential value in improving local drug delivery, tumor targeting, and biological barrier penetration [54]. Although this field remains exploratory, these emerging delivery technologies may offer useful perspectives for future gastric cancer therapy. In addition, functional peptides and bioactive substances may serve as complementary tools for modulating tumor-related cellular pathways. Miao et al. summarized co-assembly strategies of natural plant compounds and discussed their potential role in improving bioavailability, biocompatibility, biodegradability, and pharmacological efficacy [55]. Therefore, future studies may examine whether functional peptides, natural bioactive substances, or optimized delivery platforms can modulate FMRP/DVL2-related signaling or its downstream pathways. However, these possibilities remain speculative and require further experimental validation. A recent study demonstrated that FMRP promotes gastric cancer progression by stabilizing c-MYC protein through inhibition of proteasomal degradation [56]. Interestingly, although MYC mRNA has been reported to undergo m6A modification in gastric cancer, that study did not support an m6A-dependent RNA regulatory mechanism for FMRP. Together with our findings, these observations suggest that FMRP promotes gastric cancer through multiple downstream effectors and distinct regulatory mechanisms, including post-translational stabilization of c-MYC protein and m6A-dependent stabilization of DVL2 mRNA. In conclusion, this study identified a new function of FMRP-mediated stabilization of DVL2 mRNA in promoting the noncanonical Wnt/PCP pathway and gastric cancer progression in an m6A-dependent manner. These findings provide new insights into the occurrence and progression of gastric cancer and position FMRP as a potential therapeutic target for gastric cancer. Conclusions Overall, this study demonstrates that FMRP functions as an oncogene in gastric cancer. Specifically, FMRP knockdown inhibited the proliferation, migration, and invasion of gastric cancer cells. Mechanistically, FMRP interacted with the DVL2 mRNA in an m6A-dependent manner and maintained its stability, thereby regulating the activity of the noncanonical Wnt/PCP pathway through DVL2 and concurrently promoting the proliferative, migratory, and invasive capabilities of gastric cancer cells. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4 Supplementary Material 5 Supplementary Material 5 Supplementary Material 6 Supplementary Material 6 Supplementary Material 7 Supplementary Material 7