The METTL3-IGF2BP3 axis drives osteosarcoma progression by enhancing ID1 mRNA stability

Abstract

Objective

Osteosarcoma (OS) is a highly aggressive malignant bone tumor. While ID1 plays a critical role in OS progression, the underlying mechanisms remain unclear. This study investigates the METTL3-IGF2BP3 axis-mediated N6-methyladenosine (m⁶A) modification in regulating ID1 mRNA stability and its functional implications in OS.

Methods

RNA-binding proteins (RBPs) associated with ID1 were predicted using the ENCORI database. Differentially expressed RBP genes (RBP-DEGs) were screened via the GSE253548 dataset, and core RBP-DEGs were identified using machine learning algorithms. m⁶A modification sites on ID1 mRNA were predicted by SRAMP, while the interaction between METTL3/IGF2BP3 and ID1 was analyzed via the RM2Target database. Functional experiments including CCK-8, colony formation, wound healing, and Transwell assays were performed to assess OS cell proliferation, migration, and invasion. Mechanistic insights were validated through MeRIP-qPCR, RIP-qPCR, RNA stability assays, and dual-luciferase reporter experiments. Additionally, a subcutaneous xenograft mouse model of OS was established to evaluate tumor growth, with tumor volume/weight and Ki67 expression monitored.

Results

Bioinformatics analysis identified METTL3 and IGF2BP3 as core regulators of ID1, with elevated ID1, IGF2BP3, and METTL3 expression observed in OS cells. Mechanistically, IGF2BP3 enhanced ID1 mRNA stability by binding to m⁶A-modified ID1 transcripts, while METTL3 catalyzed ID1 mRNA m⁶A modification to strengthen IGF2BP3-ID1 interaction. Cellular assays demonstrated that the METTL3-IGF2BP3 axis significantly promoted OS cell proliferation, migration, and invasion via m⁶A-dependent ID1 stabilization. In vivo studies further confirmed that this axis accelerated OS tumor growth by upregulating ID1 expression.

Conclusion

The METTL3-IGF2BP3 axis facilitates OS progression by enhancing ID1 mRNA stability and expression, highlighting its potential as a therapeutic target.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12891-026-09527-0.

Introduction

Osteosarcoma (OS), a mesenchymal cell-derived primary bone malignancy defined by aberrant osteoid and immature bone formation, predominantly affects adolescents and young adults [1]. Clinical presentation typically includes progressive localized pain, swelling, and restricted joint mobility, which often mimicking benign musculoskeletal conditions and contributing to delayed diagnosis [2]. Characterized by aggressive local invasion and early systemic metastases, this disease is associated with severe impairment of limb function and substantially diminished quality of life [3]. Despite advancements in multimodal therapies incorporating neoadjuvant chemotherapy and limb-salvage surgery, many patients develop chemoresistance-driven relapse and metastasis, resulting in a dismal 5-year survival rate of < 30% [4, 5]. This persistent clinical crisis highlights the urgent need to elucidate the molecular mechanisms driving OS progression, which may unlock novel therapeutic strategies to combat tumor progression and improve clinical outcomes.

N6-methyladenosine (m⁶A), the most prevalent internal chemical modification of eukaryotic mRNAs, governs post-transcriptional regulation by dynamically controlling RNA stability, splicing, and translation efficiency, which are increasingly implicated in oncogenesis and cancer progression including OS [6–8]. This reversible modification is orchestrated by three conserved protein classes: writers (methyltransferases), erasers (demethylases), and readers (m⁶A-binding proteins) [9]. The m⁶A methyltransferase complex (MTC), with methyltransferase-like 3 (METTL3) as its catalytic subunit, installs m⁶A marks on target transcripts to modulate their fate [10], while readers, e.g., insulin-like growth factor 2 mRNA-binding protein 3 (IGF2BP3), decode methylation signals to execute context-specific functions [11]. In oncogenic contexts, METTL3 has been reported to be involved in tumorigenesis of various malignancies including hepatocellular carcinoma and breast cancer by stabilizing proto-oncogenic mRNAs [12, 13]. This functional role is particularly pronounced in OS, where studies report METTL3 as either a driver of metastasis or a tumor promoter [14, 15]. Similarly, the m⁶A reader IGF2BP3 enhances transcript stability in glioblastoma, lung adenocarcinoma and OS by forming protective complexes with m⁶A-modified metastasis-associated genes, thus regulating tumor progression [16–18]. However, the mechanistic interplay between METTL3-mediated m⁶A deposition and IGF2BP3-dependent stabilization, specifically their coordinated regulation of pro-metastatic transcripts in OS, remains uncharted, representing a pivotal gap in epitranscriptomic oncology.

Inhibitor of DNA binding 1 (ID1), a member of the helix-loop-helix (bHLH) transcription factor family, drives tumor cell proliferation, invasion, and angiogenesis [19, 20]. Although ID1 is aberrantly upregulated in OS and correlates with poor prognosis [21], the mechanisms sustaining its overexpression are unknown. Intriguingly, bioinformatic analyses reveal conserved m⁶A motifs within the ID1 mRNA 3’UTR, suggesting potential m⁶A-mediated post-transcriptional regulation, which is yet to be experimentally validated. On this basis, we hypothesize that the METTL3-IGF2BP3 axis orchestrates m6A-dependent stabilization of ID1 mRNA, thereby fueling OS aggressiveness.

This study integrates multi-omics profiling and functional validation to decipher the METTL3-IGF2BP3-ID1 regulatory axis in OS, elucidate how m⁶A modification governs ID1 mRNA stability and protein expression, and evaluate the therapeutic potential of targeting this axis in preclinical OS models. Our findings will unveil a previously unrecognized epitranscriptomic circuit driving OS progression and provide a rationale for targeting m⁶A machinery in OS treatment.

Methods and materials

OS-related GEO dataset download and bioinformatics analysis

The OS-related dataset GSE253548 was downloaded from the GEO database (https://www.ncbi.nlm.nih.gov/gds/?term=). This dataset has a relatively large sample size, including 50 OS tumor tissue samples and 40 normal bone tissue samples, providing an appropriate basis for comparative analysis of differential expression between tumor and normal tissues. Differentially expressed genes (DEGs) between Normal and Tumor groups were identified using the R package “limma” with stringent thresholds (|log₂(fold change)| >1 and P-value < 0.05). Subsequently, the ENCORI database facilitated prediction of RNA-binding proteins (RBPs) potentially regulating ID1. The “venn” package in R was then used to determine the intersection of DEGs and predicted RBPs, screening out RBP-related DEGs that might regulate ID1 (RBP-DEGs). Machine learning algorithms, including least absolute shrinkage and selection operator (LASSO), support vector machine-recursive feature elimination (SVM-RFE), and random forest, were applied to these RBP-DEGs using the “glmnet”, “e1071”, “kernlab”, “caret”, and “randomForest” packages in R for in-depth analysis. The intersection of genes identified by LASSO, SVM-RFE, and random forest was extracted via the “venn” package to further filter OS-specific RBP-DEGs, ultimately identifying the core RBP-DEGs with the most significant differential expression in OS.

The m⁶A modification sites in ID1 mRNA were predicted using the SRAMP website (http://www.cuilab.cn/m6asiteapp/old). The RM2Target database (http://rm2target.canceromics.org/#/home) was utilized to assess the potential interactions and mechanisms between the key m⁶A “reader” insulin-like growth factor 2 mRNA-binding protein 3 (IGF2BP3), “writer” methyltransferase-like 3 (METTL3), and ID1. Spearman correlation scatter plots of gene expression were generated on the Xiantao Academic platform (https://www.xiantaozi.com/).

Cell culture

The human osteoblast line hFOB1.19 (CL0140, FENGHUISHENGWU) was maintained in Dulbecco’s modified Eagle’s medium/nutrient mixture F-12 (DMEM/F12, 12634028, ThermoFisher), while human OS cell lines MG-63 (CL0217, FENGHUISHENGWU) and U2OS (SNL-054, SUNNCELL) were cultured in minimum essential medium (MEM, PM150410, Procell) and McCoy’s 5 A medium (16600082, ThermoFisher), respectively. All media were uniformly supplemented with 10% fetal bovine serum (FBS, F0193, Sigma-Aldrich) and 1% penicillin-streptomycin (TMS-AB2, Sigma-Aldrich). hFOB1.19 cells were cultured in an incubator at 34 °C with 5% CO₂, whereas all other cell lines were maintained at 37 °C with 5% CO₂. The culture medium was refreshed every 2–3 days.

Cell treatment and grouping

Lentiviruses for knocking down ID1 (shID1), IGF2BP3 (shIGF2BP3), and METTL3 (shMETTL3), as well as those for overexpressing ID1 (oe-ID1), IGF2BP3 (oe-IGF2BP3), and METTL3 (oe-METTL3), along with negative control lentiviruses (shNC, oe-NC), were purchased from VectorBuilder (Guangzhou, China). MG-63 cells were infected with lentiviruses at 60% confluence for 48 h, followed by neomycin selection to establish stable OS cell lines. Successful gene knockdown or overexpression was verified by reverse transcription-quantitative polymerase chain reaction (RT-qPCR).

Grouping: (1) To clarify the role of ID1 in the malignant progression of OS cells, the cells were divided into two groups: shNC group and shID1 group. (2) To evaluate the effect of IGF2BP3 on ID1 expression and OS cell malignant behavior, cells were divided into six groups: oe-NC group, oe-IGF2BP3 group, shNC group, shIGF2BP3 group, shIGF2BP3 + oe-NC group, and shIGF2BP3 + oe-ID1 group. (3) To investigate the effect of METTL3 on ID1 m6A modification and OS cell malignant behavior, cells were divided into six groups: oe-NC group, oe-METTL3 group, shNC group, shMETTL3 group, oe-METTL3 + shNC group, and oe-METTL3 + shID1 group. (4) To clarify the impact of the METTL3–IGF2BP3 axis on ID1-mediated OS cell malignant behavior, cells were divided into four groups: oe-NC group, oe-METTL3 group, oe-METTL3 + shNC group, and oe-METTL3 + shIGF2BP3 group.

To assess the effect of METTL3 methyltransferase activity on ID1 m6A modification, MG-63 cells stably infected with shMETTL3 were transfected with wild-type METTL3 overexpression plasmid (pcDNA-METTL3; #160250), catalytically inactive METTL3 mutant overexpression plasmid (pcDNA-METTL3-APPA; #160251), or control empty vector (pcDNA; #208051) using Lipofectamine 3000 (L3000001, Thermo Fisher) according to the manufacturer’s instructions. All plasmids were purchased from Addgene (Massachusetts, USA). Based on the transfected plasmids, cells were divided into three groups: pcDNA group, pcDNA-METTL3 group, and pcDNA-METTL3-APPA group.

To further evaluate the effect of the METTL3–IGF2BP3–ID1 axis on OS cell malignant behavior, MG-63 cells were treated with either the METTL3 inhibitor STM2457 (200 µM; HY-134836, MedChemExpress, Monmouth Junction, NJ, USA) or the selective IGF2BP3 inhibitor I3IN-002 (5 µM; HY-174228, MedChemExpress). According to the inhibitor treatment, cells were divided into three groups: Control group, STM2457 group, and I3IN-002 group.

Cell Counting Kit-8 (CCK8) assay

MG-63 cell proliferation was quantified using the CCK-8 assay. Infected cells were plated in 96-well plates (5 × 10³ cells/well) and allowed to adhere overnight. At 0, 12, 24, and 48 h, CCK-8 reagent (10 µL/well; C0038, Beyotime) was added and incubated for 2 h. At the end of the incubation, the optical density (OD) of the cells was measured at 450 nm using an Infinite 200 PRO microplate reader (TECAN, Switzerland).

Colony formation assay

Infected MG-63 cells were plated into 6-well plates at a density of 1 × 10³ cells/well and cultured for 14 days until colonies were clearly visible. Cells were fixed with 4% paraformaldehyde (PFA, P0099, Beyotime) for 20 min, thoroughly rinsed with phosphate-buffered saline (PBS), and stained with crystal violet staining solution (C0121, Beyotime) for 10 min. Finally, colonies were imaged using a light microscope (Olympus, Japan) and counted per well.

Wound healing assay

Confluent OS cell monolayers (90% confluence) were subjected to standardized wound generation using a 200-µL sterile pipette tip along a predefined linear axis. Following gentle PBS washes to remove debris, cells were maintained in serum-free medium to minimize proliferation-driven closure. Images of identical fields were acquired at 0/24-hour intervals using an inverted microscope (Olympus Japan), and wound healing rates were quantified through ImageJ.

Transwell invasion assay

Invasion capacity was assessed using a Transwell plate (CLS3470, Corning). The lower compartment received 500 µL complete medium containing 10% FBS, while 200 µL MG-63 cell suspension (5 × 10⁵ cells/mL) in serum-free medium was seeded onto the 60-µL Matrigel-coated upper chamber. Following 48-hour incubation, transmigrated cells were fixed with 4% PFA for 30 min. After PBS washing, cells were stained with crystal violet for 10 min, and imaged under an optical microscope (Olympus, Japan), with invasive cell counts quantified using ImageJ.

MeRIP-qPCR assay

Methylated RNA immunoprecipitation-quantitative polymerase chain reaction (MeRIP-qPCR) was conducted using an m⁶A-RNA methylation quantification kit (Bes5203-2, BersinBio). Briefly, cellular RNA was fragmented according to the kit protocol, and subjected to immunoprecipitation with 4 µg anti-m⁶A antibody at 4 °C for 4 h. Pre-blocked Protein A/G magnetic beads were then introduced to capture antibody-RNA complexes for 1 h at 4 °C. Enriched RNA was eluted from the beads and reverse-transcribed for quantitative assessment of ID1 mRNA enrichment via RT-qPCR.

RIP-qPCR assay

RNA immunoprecipitation (RIP) was performed using the BeyoRIP™ RIP assay kit (P1801S, Beyotime) to investigate IGF2BP3-ID1 mRNA interactions. Following lysis of infected MG-63 cells, supernatants were incubated overnight at 4 °C with pre-blocked Protein A/G agarose beads conjugated to anti-IGF2BP3 antibody (81805-1-RR, Proteintech), anti-METTL3 (#86132, Cell Signaling Technology, Danvers, MA, USA), or control IgG (A7016, Beyotime). Protein-RNA complexes were eluted with elution buffer, with one portion used for WB analysis and the other for RNA isolation and purification. Finally, qRT-PCR was performed to evaluate ID1 mRNA enrichment.

RNA stability assay

To determine ID1 mRNA stability, infected MG-63 cells were exposed to 4 µg/mL actinomycin D (HY-17559, MedChemExpress), a transcriptional elongation inhibitor, for defined durations (0/2/4/6 hours). Total RNA was isolated at each timepoint, followed by RT-qPCR quantification of residual ID1 mRNA levels normalized to GAPDH.

Dual-luciferase reporter gene assay

To investigate the regulatory role of m⁶A modifications on ID1 expression, coding DNA sequences (CDS) of wild-type (ID1-WT) and m⁶A site-mutated (ID1-MUT) ID1 were cloned into the pmirGLo Fluc vector to construct dual-luciferase reporter gene plasmids. Forty-eight hours after transfecting cells with these plasmids, luciferase activity was measured using the Dual-Glo Luciferase System (Promega, USA).

RT-qPCR assay

Total RNA was isolated from cells using TRIzol™ reagent (B511311-0100, Sangon Biotech) and reverse-transcribed into complementary DNA (cDNA) with the CellAmp™ Direct TB Green RT-qPCR Kit (3735A, Takara). Next, qPCR amplification was performed on a LightCycler 96 System (Roche, Switzerland). With glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as an internal reference, relative mRNA expression levels were calculated via the 2^−ΔΔCt method. All primer pairs were synthesized by Sangon Biotech (Shanghai, China), as listed in Table 1.

Table 1.

Primer sequences for RT-qPCR

Gene	Species	Primer sequence
ID1	Human	F 5’-GTTGGAGCTGAACTCGGAATCC-3’
		R 5’-ACACAAGATGCGATCGTCCGCA-3’
IGF2BP3	Human	F 5’-GCTCTATCAGTCGGTGCCATCATC-3’
		R 5’-GCCTTGAACTGAGCCTCTGGTG-3’
METTL3	Human	F 5’-AGATGGGGTAGAAAGCCTCCT-3’
		R 5’-TGGTCAGCATAGGTTACAAGAGT-3’
GAPDH	Human	F 5’-AGATCCCTCCAAAATCAAGTGG-3’
		R 5’-GGCAGAGATGATGACCCTTTT-3’

ID1 inhibitor of DNA binding 1, IGF2BP3 insulin-like growth factor 2 mRNA-binding protein 3, METTL3 methyltransferase-like 3, N6-adenosine-methyltransferase complex catalytic subunit, GAPDH glyceraldehyde-3-phosphate dehydrogenase, F forward, R reverse

Animal preparation

Four-six-week-old specific pathogen-free (SPF)-grade BALB/c nude mice were provided by Hunan SJA Laboratory Animal Co., Ltd. (Hunan, China). All animal procedures were approved by the Animal Ethics Committee of Hunan Evidence-based Biotechnology Co., Ltd (No: 2025064) and strictly followed the Guide for the Care and Use of Laboratory Animals. Mice were acclimated in an SPF environment for 1 week before experiments. The animals were maintained under the following conditions: temperature of 22 ± 4 °C, relative humidity of 60 ± 10%, a 12-hour light/dark cycle, with ad libitum access to food and water.

Subcutaneous xenograft model establishment and grouping

Thirty-six BALB/c nude mice were randomly divided into six groups (n = 6 per group) based on the administration of stably infected MG-63 cells: shNC + oe-NC, shMETTL3 + oe-NC, shMETTL3 + oe-ID1, shIGF2BP3 + oe-NC, shIGF2BP3 + oe-ID1, with an additional repeated shNC + oe-NC group serving as a methodological control.

To establish OS xenograft models, each mouse received a subcutaneous injection of 5 × 10⁶ MG-63 cells. All animals were maintained under SPF conditions with free access to food and water. Tumor length and width were measured weekly, with tumor volumes calculated using the formula: volume (mm³) = length × width² × 0.5. Following a five-week observation period post-implantation, mice were euthanized by intraperitoneal administration of an overdose of sodium pentobarbital (150 mg/kg; P3761, Sigma-Aldrich). Subcutaneous tumors were collected, weighed, and photographed.

Immunohistochemistry (IHC)

Tumor specimens harvested from mouse models underwent standardized immunohistochemical processing through sequential histological preparations. Freshly excised tissues were immersion-fixed in 4% PFA for 24 hours, followed by paraffin embedding, and sectioning into 4-µm-thick slices. After routine deparaffinization, rehydration, and antigen retrieval, tissue sections were subjected to nonspecific binding blockade via 30-minute room temperature incubation with 5% bovine serum albumin (BSA; V900933, Sigma-Aldrich). Sections were subsequently probed with anti-Ki67 primary antibody (1 : 1,000 dilution, 28074-1-AP, Proteintech) for 2 hours at 37°C, followed by 30-minute incubation with horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG secondary antibody (1 : 50 dilution, A0208, Beyotime). Chromogenic development was achieved using 3,3’-diaminobenzidine (DAB; 36202ES01, YEASEN), with counterstained sections digitally imaged using an optical microscope (Olympus, Japan). Finally, the percentage of Ki67-positive areas was analyzed using ImageJ.

Western Blot (WB)

Total proteins from OS cells and tumor tissues were extracted with pre-cooled radioimmunoprecipitation assay (RIPA) buffer (20101ES60, YEASEN) and quantified using a bicinchoninic acid (BCA) protein assay kit (20201ES76, YEASEN). Equal protein samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes (IEVH85R, Millipore), which were blocked with 5% DAB for 1 h at room temperature. After that, membranes were incubated overnight at 4 °C with primary antibodies against ID1 (1 : 1,000, 18475-1-AP, Proteintech), IGF2BP3 (1 : 10,000, 81805-1-RR, Proteintech), METTL3 (1 : 1,000, #86132, Cell Signaling Technology), and GAPDH (1 : 10,000, 10494-1-AP, Proteintech). Following washing three times with Tris-buffered saline with Tween-20 (TBST), incubation was continued with HRP-conjugated goat anti-rabbit IgG secondary antibody (1 : 1,000, A0208, Beyotime) for 2 h at room temperature. Protein bands were visualized using enhanced chemiluminescence (ECL; BL523B, Biosharp) on a SCG-W3000 PLUS chemiluminescence imaging system (Servicebio, Wuhan, China), and relative protein expression was quantified with ImageJ using GAPDH as a loading control.

Statistical analysis

Quantitative data were presented as mean ± standard error of the mean (SEM) and statistically analyzed using GraphPad Prism 10.1.2. Differences between two groups were evaluated by independent samples t-test, while comparisons among three or more groups were performed using one-way analysis of variance (ANOVA) followed by Tukey’s honestly significant difference (HSD) post-hoc test. A P-value < 0.05 was considered statistically significant.

Results

Suppression of ID1 attenuates malignant behaviors in OS cells

Analysis of the OS-related dataset GSE253548 revealed abnormally elevated ID1 expression in tumor tissues (Fig. 1A). Consistently, ID1 was significantly upregulated in human OS cell lines (MG63 and U2OS) as confirmed by RT-qPCR and WB analyses (Fig. 1B-C), suggesting ID1 as a potential driver gene in OS pathogenesis. Subsequently, we selected MG-63 cells with the highest ID1 expression for further investigation. To validate this hypothesis, ID1 was knocked down using shRNA lentiviral infection, with shID1-2 demonstrating the highest knockdown efficiency and thus selected for subsequent experiments (Fig. 1D). CCK-8 assays and colony formation experiments revealed that ID1 silencing markedly suppressed OS cell proliferation (Fig. 1E-F). Furthermore, wound healing and Transwell invasion assays demonstrated significant inhibition of OS cell migration and invasion capabilities upon ID1 knockdown (Fig. 1G-H).

Fig. 1.

ID1 modulates malignant behaviors in OS cells. A: ID1 expression profile in the GSE253548 dataset. B: RT-qPCR analysis of ID1 mRNA levels across cell groups. C: Western blot detection of ID1 protein expression. D: Validation of ID1 knockdown efficiency by RT-qPCR. E: CCK-8 assay assessing cellular proliferation. F: Colony formation capacity quantified across experimental groups. G: Scratch assay evaluating cell migration ability, scale bar = 400 μm. H: Transwell invasion assay with representative images, scale bar = 200 μm. Data are presented as mean ± SEM. ***P < 0.001, ****P < 0.0001. Statistical analyses were performed using independent-samples t-test or one-way ANOVA followed by Tukey’s post hoc test. All experiments were performed in triplicate.

These findings collectively indicate that ID1 serves as an oncogenic driver in OS, and its suppression effectively mitigates malignant behaviors in OS cells.

IGF2BP3 may regulate ID1 expression via m⁶A recognition

Differential analysis of the GSE253548 dataset identified 2,730 DEGs, including 2,702 upregulated and 28 downregulated genes in OS tissues (Fig. 2A). To investigate the regulatory mechanisms of ID1, RBPs interacting with ID1 were predicted using the ENCORI database and intersected with DEGs, yielding 16 RBP-DEGs potentially regulating ID1 (Fig. 2B).

Fig. 2.

Identification of OS-specific RNA-binding proteins regulating ID1 and mechanistic predictions. A: Volcano plot of DEGs in the GSE253548 dataset. B: Venn diagram intersecting ENCORI-predicted RBPs with DEGs. C: Feature selection by LASSO regression. D: SVM-RFE algorithm-derived feature genes. E: Variable importance ranking from random forest analysis. F: Venn diagram overlapping feature genes identified by LASSO, SVM-RFE, and random forest. G: IGF2BP3 expression profile in the GSE253548 dataset. H: RT-qPCR analysis of IGF2BP3 mRNA levels. I: Western blot detection of IGF2BP3 protein expression. J: SRAMP-predicted m⁶A modification site on ID1 mRNA. K: Secondary structure of the m⁶A-modified site at 782 nt on ID1 mRNA. L: MeRIP-qPCR quantification of ID1 m⁶A modification levels. M: RM2Target-predicted m⁶A-mediated interaction between IGF2BP3 and ID1. N: Spearman correlation of IGF2BP3 and ID1 expression. Data are presented as mean ± SEM. **P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical analyses were performed using independent-samples t-test or one-way ANOVA followed by Tukey’s post hoc test. All experiments were performed in triplicate.

Subsequently, machine learning algorithms were applied to identify candidate OS-specific RBP-DEGs governing ID1 regulation. LASSO regression identified 5 feature genes (Fig. 2C), SVM-RFE selected 10 (Fig. 2D), and random forest analysis prioritized 12 genes, with IGF2BP3 exhibiting the highest variable importance (Fig. 2E). Integration of these results revealed 5 OS-specific RBP-DEGs: IGF2BP3, CDK1, YTHDF2, AQR, and CSTF2 (Fig. 2F). Among these, IGF2BP3 displayed the most pronounced differential expression and was significantly upregulated in OS tissues (Fig. 2G). Consistent with this, our experiments also found overexpressed IGF2BP3 in OS cells (Fig. 2H-I), suggesting its potential role in binding ID1 mRNA and promoting OS progression.

Notably, IGF2BP3 is an established m⁶A reader protein that regulates gene expression by recognizing and binding to m⁶A-modified mRNAs. Using the SRAMP database, we predicted a high-confidence m⁶A modification site at position 782 nt on ID1 mRNA (Fig. 2J), with its secondary structure illustrated in Fig. 2K. As expected, m⁶A modification levels of ID1 mRNA were significantly elevated in OS cells (Fig. 2L). RM2Target database analysis further predicted an m⁶A-dependent interaction between IGF2BP3 and ID1 (Fig. 2M). Spearman correlation analysis confirmed a positive correlation between IGF2BP3 and ID1 expression (R = 0.481, P < 0.001; Fig. 2N).

These results suggest that IGF2BP3 may regulate ID1 expression by binding to its m⁶A modification site, highlighting a novel epigenetic mechanism in OS pathogenesis.

IGF2BP3 stabilizes ID1 mRNA via m⁶A recognition to promote malignant progression in OS cells

We established IGF2BP3-overexpressing and knockdown OS cell models using lentiviral infection, with RT-qPCR confirming infection efficiency (Fig. 3A-B). For subsequent knockdown experiments, shIGF2BP3-2 demonstrating the highest silencing efficiency was selected. IGF2BP3 overexpression significantly upregulated ID1 expression, while IGF2BP3 knockdown markedly downregulated ID1 at both mRNA and protein levels (Fig. 3C-D). RIP-qPCR analysis further confirmed direct binding between IGF2BP3 and ID1 mRNA (Fig. 3E).

Fig. 3.

IGF2BP3 regulates ID1 to modulate malignant behaviors in OS Cells. A-B: RT-qPCR analysis of IGF2BP3 expression in engineered cells. C: RT-qPCR detection of ID1 mRNA levels. D: Western blot analysis of ID1 protein expression. E: RIP-qPCR demonstrating IGF2BP3-ID1 mRNA interaction, with RIP efficiency validated by Western blot. F: Schematic of ID1-WT/MUT CDS and their dual-luciferase reporter constructs. G: Dual-luciferase reporter gene assay results. H: RT-qPCR for ID1 mRNA expression in cells at different time points (0, 2, 4, 6 h) after actinomycin D treatment. I: Western blot for IGF2BP3 and ID1 expression in cells. J: CCK-8 assay assessing cellular proliferation. K: Colony formation capacity quantified across experimental groups. L: Scratch assay evaluating cell migration ability, scale bar = 400 μm. M: Transwell invasion assay with representative images, scale bar = 200 μm. Data are presented as mean ± SEM. ns P > 0.05, ****P < 0.0001. Statistical analyses were performed using independent-samples t-test or one-way ANOVA followed by Tukey’s post hoc test. All experiments were performed in triplicate.

To investigate the m⁶A-dependency of this interaction, we constructed dual-luciferase reporter plasmids containing ID1-WT or ID1-MUT (782 nt m⁶A site-mutated) CDS based on SRAMP predictions (Fig. 3F). Dual-luciferase reporter gene assays revealed that shIGF2BP3 significantly reduced ID1-WT reporter activity compared to shNC controls, while showing no effect on ID1-MUT constructs (Fig. 3G), indicating m⁶A-dependent binding. RNA stability assays demonstrated that IGF2BP3 overexpression enhanced ID1 mRNA stability, whereas IGF2BP3 silencing accelerated its degradation (Fig. 3H), collectively establishing IGF2BP3 as an m⁶A-dependent stabilizer of ID1 mRNA to upregulate ID1 expression.

To assess whether IGF2BP3 regulates malignant behaviors in OS cells through ID1, we performed co-infection with shIGF2BP3 and oe-ID1 lentiviruses. WB confirmed successful ID1 restoration in shIGF2BP3 + oe-ID1 cells without affecting IGF2BP3 expression (Fig. 3I). Functional assays revealed that IGF2BP3 knockdown significantly inhibited OS cell proliferation (CCK-8 and colony formation, Fig. 3J-K), migration (wound healing assay, Fig. 3L), and invasion (Transwell, Fig. 3M). Notably, ID1 overexpression effectively rescued these inhibitory effects.

In summary, IGF2BP3 promotes OS malignancy by recognizing m⁶A-modified ID1 mRNA to enhance its stability and expression.

The METTL3-IGF2BP3 axis stabilizes ID1 mRNA in an m⁶A-dependent manner

METTL3, a key m⁶A “writer” protein, showed significantly elevated expression in OS tissues (Fig. 4A). RT-qPCR and WB analyses confirmed its upregulation in human OS cells (Fig. 4B-C). RM2Target database analysis predicted METTL3 might mediate m⁶A modification of ID1 (Fig. 4D). Spearman correlation analysis revealed a strong positive correlation between METTL3 and ID1 expression (R = 0.471, P < 0.001) (Fig. 4E), suggesting METTL3-mediated m⁶A regulation of ID1.

Fig. 4.

The METTL3-IGF2BP3 axis mediates m⁶A-dependent regulation of ID1 expression and stability. A: METTL3 expression in GSE253548 dataset. B-C: METTL3 expression in cell lines via RT-qPCR (B) and Western blot (C). D: RM2Target prediction of METTL3-ID1 interaction. E: Spearman correlation between METTL3 and ID1 expression. F-G: METTL3 expression validation in engineered cells. H: MeRIP-qPCR analysis of ID1 m6A modification levels. I: RIP-qPCR analysis of the interaction between METTL3 and ID1 mRNA, with the efficiency of the RIP assay validated by Western blot. J: RIP-qPCR detection of IGF2BP3-ID1 mRNA interaction. K: MeRIP-qPCR analysis of ID1 m6A modification levels in MG-63 cells with endogenous METTL3 knockdown. L: RIP-qPCR analysis of the interaction between IGF2BP3 and ID1 mRNA in MG-63 cells with endogenous METTL3 knockdown. M-N: ID1 mRNA (M) and protein (N) expression under indicated treatments. O: RNA stability assay under actinomycin D treatment (0, 2, 4, 6 h). Data are presented as mean ± SEM. ns P > 0.05, ***P < 0.001, ****P < 0.0001. Statistical analyses were performed using independent-samples t-test or one-way ANOVA followed by Tukey’s post hoc test. All experiments were performed in triplicate.

To validate this hypothesis, we established METTL3-overexpressing and knockdown OS cell models using lentiviral infection, with RT-qPCR confirming infection efficiency (Fig. 4F-G). shMETTL3-1 demonstrating optimal silencing efficiency was selected for subsequent experiments (Fig. 4G). MeRIP-qPCR analysis showed that METTL3 overexpression (oe-METTL3) significantly increased m⁶A modification levels of ID1 mRNA, while METTL3 knockdown (shMETTL3) reduced them (Fig. 4H). Meanwhile, RIP-qPCR further confirmed that METTL3 bound to ID1 mRNA (Fig. 4I). These findings indicate that METTL3 catalyzes the m⁶A modification of ID1 mRNA through direct RNA binding.

Notably, silencing METTL3 led to a reduction in the binding between IGF2BP3 and ID1 mRNA (Fig. 4J), indicating that METTL3 promotes this interaction by increasing m6A modification of ID1 mRNA. Furthermore, as shown in Fig. 4K-L, in cells with endogenous METTL3 knockdown, overexpression of wild-type METTL3 (pcDNA-METTL3) restored the m6A modification level of ID1 mRNA and enhanced its binding to IGF2BP3, whereas the catalytically inactive mutant plasmid (pcDNA-METTL3-APPA) failed to produce a similar effect. These results suggest that the methyltransferase activity of METTL3 is essential for the IGF2BP3-dependent regulation of ID1.

We further investigated the METTL3-IGF2BP3 axis’s functional impact. RT-qPCR and WB demonstrated that oe-METTL3 significantly upregulated ID1 expression compared to oe-NC controls, while co-treatment with shIGF2BP3 reversed this effect (oe-METTL3 + shIGF2BP3 vs. oe-METTL3 + shNC) (Fig. 4M-N). RNA stability assays confirmed that IGF2BP3 knockdown abolished oe-METTL3-induced stabilization of ID1 mRNA (Fig. 4O).

These findings establish that the METTL3-IGF2BP3 axis enhances ID1 mRNA stability and upregulate ID1 expression through m⁶A-dependent mechanisms.

The METTL3-IGF2BP3 axis promotes malignant progression in OS cells by enhancing ID1 mRNA stability

To investigate the functional role of the METTL3-IGF2BP3 axis in OS malignancy, we assessed cellular proliferation, migration, and invasion. CCK-8 and colony formation assays demonstrated that METTL3 overexpression significantly enhanced OS cell proliferation compared to oe-NC controls (Fig. 5A-B). Scratch wound healing and Transwell assays further revealed that oe-METTL3 markedly increased migratory and invasive capacities of OS cells (Fig. 5C-D).

Fig. 5.

The METTL3-IGF2BP3 axis mediates ID1-dependent malignant behaviors in OS cells. A: CCK-8 assay assessing cell proliferation. B: Colony formation assay quantifying clonogenic capacity. C: Scratch wound healing assay evaluating migration, scale bar = 400 μm. D: Transwell invasion assay, scale bar = 200 μm. E: Cell proliferation evaluated by CCK8 assay. F: Colony formation assay to assess colony numbers. G: Scratch assay to evaluate the migration ability of cells, scale bar = 400 μm. H: Transwell assay to assess the invasion ability of cells, scale bar = 200 μm. Data are presented as mean ± SEM. **P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical analyses were performed using independent-samples t-test or one-way ANOVA followed by Tukey’s post hoc test. All experiments were performed in triplicate.

To verify IGF2BP3’s mediatory role in METTL3-driven malignancy, we silenced IGF2BP3 in the oe-METTL3 background. Notably, IGF2BP3 knockdown (oe-METTL3 + shIGF2BP3) significantly attenuated the pro-tumorigenic effects of METTL3 overexpression on proliferation, migration, and invasion compared to oe-METTL3 + shNC controls (Fig. 5A-D). Similarly, ID1 knockdown (oe-METTL3 + shID1) suppressed malignant behaviors, mirroring the phenotypic effects observed in oe-METTL3 + shIGF2BP3 cells (Fig. 5E-H).

Furthermore, pharmacological inhibition experiments demonstrated that treatment with the METTL3 inhibitor STM2457 or the selective IGF2BP3 inhibitor I3IN-002 significantly downregulated ID1 expression and promoted ID1 mRNA degradation (Fig. 6A–C). In addition, both inhibitors effectively suppressed proliferation, migration, and invasion of MG-63 cells (Fig. 6D–G).

Fig. 6.

Effects of METTL3 and IGF2BP3 inhibitors on ID1-mediated malignant behaviors of OS cells. A: RT-qPCR analysis of ID1 expression in each group of cells. B: WB analysis of ID1 expression in each group of cells. C: RT-qPCR analysis of ID1 mRNA levels at different time points (0, 2, 4, 6 h) following actinomycin D treatment. D: CCK-8 assay assessing cell proliferation. E: Colony formation assay evaluating colony numbers in each group. F: Wound healing assay assessing cell migration ability; scale bar = 400 μm. G: Transwell assay assessing cell invasion ability; scale bar = 200 μm. Data are presented as mean ± SEM. ****P < 0.0001. Statistical analyses were performed using one-way ANOVA followed by Tukey’s post hoc test. All cell experiments were repeated three times.

These findings collectively demonstrate that the METTL3-IGF2BP3 axis drives OS malignancy through stabilization of ID1 mRNA.

The METTL3-IGF2BP3 axis promotes in vivo tumor growth via ID1

Finally, we investigated the role of the METTL3-IGF2BP3-ID1 axis in OS progression using in vivo models. WB analysis revealed that METTL3 knockdown (shMETTL3 + oe-NC) significantly reduced both METTL3 and ID1 expression in tumor tissues compared to shNC + oe-NC controls, while ID1 overexpression (oe-ID1) restored ID1 levels without affecting METTL3 expression (Fig. 7A). Treatment with shIGF2BP3 significantly inhibited the expression of both IGF2BP3 and ID1. Compared to shIGF2BP3 + oe-NC group, shIGF2BP3 + oe-ID1 treatment upregulated ID1 expression without altering IGF2BP3 levels (Fig. 7B).

Fig. 7.

The METTL3-IGF2BP3 axis mediates ID1-dependent tumor growth in vivo. A: Western blot analysis of METTL3 and ID1 expression in tumor tissues from each group of mice. B: Western blot analysis of IGF2BP3 and ID1 expression in tumor tissues from each group of mice. C-D: Changes in tumor volume in each group of mice. E-F: Representative images of tumors from each group of mice. G-H: Tumor weights in each group of mice. I-J: Representative immunohistochemical images of Ki67 staining and quantification of Ki67-positive area percentage in tumor tissues from each group of mice, scale bar = 100 μm. N = 6. Data are presented as mean ± SEM. ns P > 0.05, ***P < 0.001, ****P < 0.0001. Statistical analyses were performed using one-way ANOVA followed by Tukey’s post hoc test.

In vivo functional studies demonstrated that METTL3 or IGF2BP3 silencing markedly inhibited tumor growth, as evidenced by significant reductions in tumor volume and weight (Fig. 7C-H). Notably, ID1 overexpression reversed these inhibitory effects. Ki67 immunohistochemical staining further confirmed that METTL3/IGF2BP3 knockdown substantially decreased tumor proliferation with reduced Ki67 levels, while ID1 restoration rescued the proliferative capacity (Fig. 7I-J).

These findings conclusively demonstrate that the METTL3-IGF2BP3 axis drives OS tumorigenesis in vivo through ID1 regulation.

Discussion

Our multi-omics analysis and functional validation robustly confirmed that ID1 was significantly overexpressed in OS tissues and cell lines, establishing ID1 as a critical driver of OS progression. Functionally, ID1 knockdown significantly suppressed OS cell proliferation, migration, and invasion, underscoring its oncogenic role and therapeutic potential in OS. This finding aligns with previous studies showing that ID1 acts as an oncogene across multiple cancers, promoting cell stemness and metastatic potential through different pathways [22–24]. For instance, in non-small cell lung cancer, ID1 enhances cell invasion by activating the RIP3/MLKL pathway [25]. In OS, our findings suggest that ID1 exerts similar oncogenic functions by inhibiting cell differentiation and accelerating cell cycle progression.

To explore the regulatory mechanisms of ID1 in OS progression, our integrative approach identified IGF2BP3 as an m⁶A reader that binds to the m⁶A modification site at position 782 nt of ID1 mRNA, thereby significantly enhancing ID1 mRNA stability to upregulate its expression. This mechanism is consistent with recent research on the IGF2BP family, revealing that IGF2BP2/3 maintain c-Myc mRNA stability by recognizing its m⁶A sites, promoting tumor progression [26, 27]. Strikingly, our dual-luciferase reporter gene assay demonstrated that mutation of the m⁶A modification site abrogated the binding of IGF2BP3 to ID1, confirming the m⁶A-dependent nature of their interaction. Moreover, the positive correlation between IGF2BP3 and ID1 expression, together with rescue experiments showing that ID1 overexpression counteracted the inhibitory effects of IGF2BP3 knockdown, firmly supports the conclusion that IGF2BP3 promotes OS malignancy by stabilizing m6A-modified ID1 mRNA. These findings are supported by prior bioinformatics and clinical evidence implicating IGF2BP3 as a key driver in high-risk or metastatic OS [28, 29].

Moreover, we unveiled a cooperative METTL3-IGF2BP3-ID1 axis that amplifies OS malignancy. As a key writer of m⁶A modification, METTL3 was found to be highly expressed in OS and positively correlated with ID1. Functional assays further showed that METTL3 increased the m⁶A modification level of ID1 mRNA, thereby promoting its binding to IGF2BP3 and enhancing ID1 stability. This METTL3-IGF2BP3-ID1 axis exemplifies a feed-forward loop in epitranscriptomic regulation, during which METTL3-mediated m⁶A deposition creates binding platforms for IGF2BP3, which in turn stabilizes ID1 mRNA to amplify its oncogenic output. Such cooperative interplay between writers and readers may represent a generalizable mechanism for sustaining oncogene addiction in aggressive cancers. Similar collaborative axes have been reported in other cancers. For example, the METTL3-IGF2BP2 axis affects colorectal cancer progression in an m⁶A modification-dependent manner [30]. In esophageal cancer, METTL3/IGF2BP2 promotes the malignant progression by activating the PIK3CA/AKT pathway [31]. Importantly, this is the first study to validate such an axis in OS, and our in vivo experiments confirmed that silencing METTL3 or IGF2BP3 inhibited tumor growth, an effect that was reversed by ID1 overexpression. Together, these findings highlight the central role of the METTL3-IGF2BP3-ID1 pathway in OS progression.

Our study systematically identifies and validates a novel epitranscriptomic regulatory mechanism underlying OS progression. Through integrated multi-omics analysis and functional experiments, we demonstrated that METTL3 and IGF2BP3 cooperatively stabilize ID1 mRNA via m6A modification, thereby promoting OS cell proliferation, migration, and invasion. This finding expands the current understanding of m6A-mediated regulation in OS, which has previously focused primarily on single regulators, such as YTH N6-methyladenosine RNA binding protein 1 (YTHDF1), shown to promote OS progression by modulating the m⁶A level of CCR4-NOT transcription complex subunit 7 (CNOT7) [32]. In contrast, our work highlights a writer–reader synergistic mechanism in which METTL3-mediated m6A deposition facilitates IGF2BP3 recognition, leading to enhanced ID1 stability and oncogenic output. The identification of this METTL3-IGF2BP3 axis as a key regulator of ID1 adds a new dimension to the m6A regulatory network in OS and provides potential therapeutic insights for targeting aberrant RNA modification in this malignancy.

This study still has several limitations. First, although we retrieved and analyzed the publicly available TARGET-OS dataset, it contains only tumor tissue samples from OS patients and lacks normal bone tissue samples. Consequently, the differential expression of METTL3, IGF2BP3, and ID1 between tumor and normal tissues could not be assessed. Moreover, our findings were primarily validated in OS cell lines and nude mouse models, without clinical verification in human OS specimens or correlation with patient outcomes, which limits the translational significance of the study. Second, the upstream regulatory mechanisms of METTL3 and IGF2BP3, such as transcriptional or non-coding RNA-mediated regulation, were not explored, as the current study focused mainly on the functional characterization and molecular regulation of the METTL3-IGF2BP3-ID1 axis.

In future research, we plan to collect clinical OS samples with matched follow-up data to validate the expression patterns and prognostic relevance of the METTL3-IGF2BP3-ID1 axis. Furthermore, integrative multi-omics analyses and molecular experiments will be employed to identify the upstream regulators of METTL3 and IGF2BP3, thereby constructing a more comprehensive epitranscriptomic regulatory network in OS. Overall, this study elucidates the molecular mechanism by which the METTL3-IGF2BP3 axis enhances ID1 mRNA stability through m⁶A modification, thereby promoting the malignant behaviors of OS cells. This pathway not only provides a new theoretical explanation for OS tumorigenesis but also identifies potential targets for therapeutic strategies targeting the m⁶A modification network.

Conclusion

Our study elucidates a pivotal epitranscriptomic circuit wherein METTL3 catalyzes m⁶A deposition on ID1 mRNA, enabling IGF2BP3-mediated stabilization to fuel OS progression (Fig. 8). This writer-reader partnership exemplifies how coordinated RNA modifications sustain oncogene addiction, offering dual therapeutic nodes (METTL3/IGF2BP3) for disrupting ID1-driven malignancy. The conserved RRACH motif in ID1’s 3’UTR further suggests broad applicability of targeting this axis across ID1-dependent cancers. By integrating computational biology with mechanistic validation, we bridge the gap between m⁶A dysregulation and OS pathogenesis, providing a framework for RNA-centric therapeutic development.

Fig. 8.

Mechanistic diagram of METTL3-IGF2BP3-mediated ID1 in promoting osteosarcoma progression

Authors’ contributions

Rongbing Shu, Qiuxin Cheng and Min Zhao: research design; experiment work; and writing of the manuscript. Zhuanyi Yu, Huaqiang Zhou and Mingchao Lin: data analysis and resource. Minghong Shi, Jianhe Chen and Jingxiang Chen: manuscript review, and project administration. All authors read and approved the final manuscript.

Funding

None.

Data availability

The data supporting this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

All animal procedures were approved by the Animal Ethics Committee of Hunan Evidence-based Biotechnology Co., Ltd (No: 2025064) and strictly followed the Guide for the Care and Use of Laboratory Animals.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.