Magnesium ions facilitate osteogenic differentiation and intervertebral fusion via m6A methylation of RhoA mRNA

Abstract

Background

Magnesium-based implants facilitate bone regeneration via degradation. However, the epigenetic mechanisms, particularly N6-methyladenosine (m6A) modification regulated by Mg²⁺, remain incompletely understood. This study investigated the role of Mg²⁺ in osteogenic differentiation through the METTL3-RhoA axis and evaluated its potential in intervertebral fusion.

Methods

The optimal Mg²⁺ concentration was identified using MC3T3-E1 cells. Methylated RNA immunoprecipitation sequencing (MeRIP-seq) and MeRIP-PCR were employed to identify m6A target genes. Functional assays (knockdown, overexpression, and rescue) validated the METTL3-YTHDF1-RhoA pathway. A rat tail intervertebral fusion model with magnesium implants was used to assess in vivo effects.

Results

Treatment with 4 mM Mg²⁺ significantly enhanced osteogenic activity and increased METTL3 levels. Mechanistically, METTL3 promoted m6A methylation of RhoA mRNA, which was subsequently bound by YTHDF1, enhancing translation and activating the RhoA/ROCK pathway. In vivo, magnesium implants accelerated fusion and improved trabecular bone quality; however, these effects were inhibited by METTL3 or RhoA inhibitors.

Conclusion

Mg²⁺ enhances osteogenic differentiation through the METTL3-YTHDF1-RhoA/ROCK pathway.

The translational potential of this article

This study provides an epigenetic framework for optimizing magnesium-based orthopedic implants and suggests that targeting the m6A-RhoA axis could improve spinal fusion outcomes.

Graphical abstract

Graphical Abstract: Magnesium ions upregulate METTL3 expression, enhancing m6A modification on RhoA mRNA. The m6A reader YTHDF1 recognizes and binds to the modified sites, promoting RhoA translation. This activates the RhoA/ROCK signaling pathway, ultimately driving osteogenic differentiation, bone remodeling, and intervertebral fusion.

1. Introduction

Bone is a complex organ capable of regeneration [1]. In orthopedic clinical practice, the fusion of damaged bone tissue can be facilitated through external fixation. Recent studies have demonstrated that magnesium-based materials exhibit excellent mechanical properties, good biocompatibility, and degradation products (Mg²⁺) that can induce osteogenesis, making them highly promising biodegradable materials for orthopedic implants [2]. However, early clinical applications of magnesium-based implants have faced challenges, including excessively high concentrations of degradation products and rapid mechanical degradation, potentially leading to fixation failure [3]. These characteristics limit their application in orthopedics, particularly in intervertebral fusion [4]. Therefore, understanding the effects of magnesium-based material degradation products on osteogenesis is critical.

Mg²⁺ is the primary degradation product of magnesium-based materials [5]. Mg²⁺ is an essential macronutrient in the human body, serving as a cofactor for numerous enzymes in cellular processes, thereby regulating energy metabolism, protein synthesis, and nucleic acid synthesis [6]. Osteoblasts play a pivotal role in bone fusion, a complex and dynamic physiological process requiring a balance between osteogenesis and bone resorption. While the regulation of osteoblasts to promote bone formation and mineralization is a key mechanism in bone fusion, the formation of bone alone is insufficient to ensure the integrity and functionality of the bone microstructure [7,8]. Therefore, elucidating the optimal concentration and function of Mg²⁺ during osteoblast differentiation is imperative.

N6-methyladenosine (m6A) modification is a key post-transcriptional mechanism that fine-tunes gene expression [9,10]. This process is implicated in various physiological and pathological processes, including development and metabolic regulation [11]. Recent studies have confirmed the association between m6A-regulated genes and bone diseases, such as osteoporosis [12], osteosarcoma [13], and intervertebral disc degeneration [14]. Mg²⁺ plays a critical role in the m6A modification process, potentially regulating the expression of methylation-related proteins, including METTL3, METTL14, WTAP, and FTO [15,16]. Therefore, an in-depth investigation of m6A epigenetic modifications in cells influenced by Mg²⁺ could help elucidate the mechanism by which Mg²⁺ promotes osteoblast differentiation.

Previous studies demonstrated that Mg²⁺ possesses osteogenic potential and can facilitate bone defect healing in animal models [17]. It has been proposed that Mg²⁺ mediates osteogenic gene transcription through the classic Wnt/β-catenin pathway, induces osteogenic differentiation via the TRPM7/PI3K pathway, upregulates osteogenic genes through the Notch pathway, and promotes osteoblast adhesion and proliferation by activating the FAK signaling pathway [[18], [19], [20], [21]]. However, it remains unclear whether Mg²⁺ can influence the epigenetic landscape of osteoblasts via m6A methylation.

This study systematically confirmed the promotive effect of Mg²⁺ on osteogenic differentiation, mineralization, and phenotypic expression of osteoblasts, with the most significant effect observed under culture conditions supplemented with 4 mmol/L magnesium. Mechanistic investigations revealed that Mg²⁺ upregulates METTL3 expression, leading to an overall increase in m6A RNA modification levels. Through m6A epitranscriptomic sequencing and experimental validation, RhoA was identified as a key downstream target gene in this magnesium-mediated regulatory process. Further molecular experiments confirmed that METTL3-mediated m6A modification enhances the binding and translation of RhoA mRNA by YTHDF1, thereby activating the RhoA/ROCK signaling pathway and promoting osteogenic differentiation and mineralization. Additionally, in vivo experiments using a rat tail vertebral fusion model with implanted magnesium slow-release plates demonstrated that magnesium treatment significantly enhances intervertebral bone fusion. In this process, METTL3 primarily facilitates bone matrix mineralization by elevating m6A modification levels, while the RhoA/ROCK pathway plays a key regulatory role in optimizing the morphology of trabecular bone structure. This study elucidates the critical role of the magnesium–m6A–RhoA axis in osteogenic regulation and provides a theoretical foundation for the application of magnesium-based materials in bone repair.

2. Material and methods

2.1. Osteogenic induction and magnesium environment construction

The basal osteogenic induction medium (OM) was prepared by supplementing α-MEM (#PM150421; Procell) with 100 nM dexamethasone, 10 mM β-glycerophosphate, and 50 μM ascorbic acid-2-phosphate. Mouse calvarial preosteoblasts (MC3T3-E1 cells) were purchased from Shanghai Genechem Co., Ltd. MC3T3-E1 cells (P3-P5) were seeded into 24-well plates and cultured for 14 days with medium replacement every two days. Four distinct treatment groups were established: Group A maintained the basal OM medium throughout the experiment, while Groups B, C, and D were supplemented with 2 mM, 4 mM, and 16 mM MgCl₂·6H₂O, respectively, within the basal OM medium. After a 24-h induction, Group A was transitioned to standard complete medium for continued culture, whereas Groups B, C, and D were maintained in complete medium with the corresponding concentrations of MgCl₂·6H₂O (2 mM, 4 mM, and 16 mM) to observe cell differentiation.

2.2. Cell treatments

MC3T3-E1 cells (P3-P5) were cultivated in α-MEM containing 10% fetal bovine serum (FBS; #164210; Procell), 100 U/mL penicillin, and 100 μg/mL streptomycin. Gene overexpression was performed using plasmids transfected into MC3T3-E1 cells with Lipofectamine™ 3000 according to the manufacturer's instructions. The empty vector negative control plasmid (NC), METTL3, RhoA, and YTHDF1 plasmids were used for overexpression. Small interfering (si) RNA was used to knock down genes, with si-NC, si-RhoA, si-METTL3, and si-YTHDF1 purchased from Shanghai GenePharma Co., Ltd. Assays were performed 48 h post-transfection. To inhibit METTL3's capacity to promote m6A modification, cells were treated with 10 μmol/L of the METTL3 functional inhibitor STM2457 (HY-134836, MedChemExpress) for 24 h [22].

2.3. Alkaline phosphatase (ALP) and alizarin red S (ARS) staining

After one week of osteogenic culture, osteogenic effects were evaluated using ARS and ALP staining. After washing, cells were fixed with paraformaldehyde and washed three times. ARS staining was performed using a 0.1% ARS staining solution (#C0138; Beyotime), and ALP staining was conducted according to the BCIP/NBT Alkaline Phosphatase Colour Development Kit (#C3206; Beyotime). Reactions were carried out in the absence of light, and staining results were observed.

2.4. Reverse transcription quantitative polymerase chain reaction (RT-qPCR)

Total RNA was extracted from the cells using the Trizol method (#R1100; Solarbio). After chloroform extraction, the mixture was centrifuged at 4 °C and 12,000 rpm for 15 min to separate the layers. The upper phase was collected, and RNA was precipitated using isopropanol. The precipitate was subjected to purification steps, including washing with anhydrous ethanol, centrifugation, and drying. The RNA was then dissolved in RNase-free water. RNA concentration and purity were assessed by spectrophotometry (Thermo Fisher Scientific, USA). For subsequent experiments, samples with a concentration of at least 50 ng/μL and an A260/A280 ratio close to 1.9 (acceptable range: 1.8–2.1) were selected. The reverse transcription system contained 500 ng RNA template, 4 μL 5 × PrimeScript™ RT Master Mix, and RNase-free water to adjust the volume. The system was incubated at 37 °C for 15 min to complete reverse transcription. Target gene-specific primers were designed based on the NCBI database (Table S1). The 20-μL reaction system included diluted cDNA as the template (#D7190M; Beyotime), 10 μL of 2 × SYBR Premix Ex Taq™ II (#D7633M; Beyotime), 2 μL of cDNA template, 1 μL of forward/reverse primers, and 6 μL of RNase-free water. The amplification was performed using a LightCycler 480 real-time quantitative PCR instrument (Roche, Basel, Switzerland), following a program of pre-denaturation, amplification cycles (annealing/extension), and melting curve analysis. Using a housekeeping gene as a reference, the relative expression levels of the target gene were calculated using the ΔΔCt method. After normalization, differences between treatment groups were analyzed.

2.5. Western blotting (WB)

For protein extraction, cells were lysed using RIPA buffer (AS1004, ASPEN). Protein concentration was determined using the BCA method (AS1086, ASPEN), with bovine serum albumin (BSA) as the standard and a 0–0.5 mg/mL gradient prepared. Protein samples were mixed with BCA working solution (reagent A:B = 50:1) and incubated at 37 °C for 30 min. The absorbance at 562 nm was measured using a spectrophotometer. Protein concentration was calculated based on the standard curve, and samples that met the requirements for subsequent experiments were selected. SDS-PAGE gels (#P0692; Beyotime) (separating and concentrating) were prepared, and 20 μg of protein was loaded per lane. Electrophoresis was conducted at a constant voltage of 200 V for 30 min, until the bromophenol blue reached the bottom of the gel. The gel strip containing the target molecular weight region was cut, and the protein was transferred to an ethanol-activated PVDF membrane (IPVH00010, Millipore) using the wet transfer method (380 mA constant current for 45 min). The membrane was blocked with a rapid blocking solution, followed by overnight incubation with a primary antibody (Table S2) at 4 °C, and incubation with an HRP-labeled secondary antibody at room temperature for 1 h. The membrane was washed with TBST to remove unbound antibodies, and ECL developer (AS1059, ASPEN) was added. Band signals were captured using a chemiluminescence imaging system. The gray values of the target protein and internal control were analyzed using ImageJ software to quantitatively calculate the relative expression levels.

2.6. Dot-blot

After isolating RNA from tissues or cells and determining its concentration, a specific quantity of RNA was pipetted onto N+ nylon membranes (GE Healthcare Systems). The membranes were allowed to air-dry naturally and then irradiated in a UV cross-linker for 10 min. The membranes were blocked with 5% skimmed milk at room temperature for 2 h. Subsequently, the N+ nylon membranes were incubated overnight with an anti-m6A antibody (202003, Synaptic Systems, 1:1000). Afterward, a rabbit secondary antibody (111-035-003, Jackson ImmunoResearch, 1:5000) was incubated with the membranes at room temperature for 1 h. The membranes were then washed three times with PBST for 5 min each and developed using a chemiluminescence imaging system. After development, the N+ nylon membranes were incubated with 0.01% methylene blue for 10 min, and photographs were taken under natural light as a control.

2.7. Detection of m6A methylation modification by enzyme-linked immunosorbent assay (ELISA)

Cells or tissues were harvested and homogenized at low temperature. After centrifugation, the supernatant was collected. Total RNA was extracted from each group, and the level of m6A methylation modification was detected according to the instructions of the ELISA kit (ab185912, abcam).

2.8. m6A epigenetic modification omics analysis

Following RNA extraction, the integrity of the RNA was assessed using agarose gel electrophoresis and spectrophotometry (A260/A280 ≈ 1.9) to confirm the absence of DNA contamination. Qualified samples were then sent to Shangpu Biotechnology Co., Ltd. (Shanghai, China) for further analysis. The RNA extraction process used 3–5 μg of total RNA and 2 μg of anti-m6A antibody, dissolved in IP buffer. This RNA-antibody solution was incubated at 4 °C for 2 h. After this, pre-blocked Dynabeads magnetic beads were added, and the RNA-antibody-magnetic bead complex was allowed to form over a 2-h period. The complex was thoroughly washed with IP buffer and low-salt wash buffer, followed by treatment with proteinase K-containing elution buffer at 50 °C for 1 h to release m6A-modified RNA. The RNA was purified through phenol-chloroform extraction and ethanol precipitation. The IP (enriched m6A RNA) and Sup (unmodified RNA) fractions were then separated. IP and Sup RNA were mixed with calibration control RNA and amplified using the Arraystar Super RNA Labeled Kit. The RNA was labeled with Cy5 (IP) and Cy3 (Sup) fluorescent dyes, purified, and the cRNA concentration and dye-binding efficiency (≥15 pmol dye/μg cRNA) were determined. Subsequently, equivalent amounts of Cy3/Cy5-labeled samples were mixed, fragmented at 60 °C, and hybridized with the m6A-mRNA and lncRNA chip at 65 °C for 17 h. Fluorescent signals were captured using an Agilent G2505C scanner, and the m6A modification sites and their abundance were analyzed based on chip probe binding intensity. Differential gene screening and bioinformatics analyses were then performed.

2.9. Prediction of m6A methylation modification sites of RhoA mRNA

In this study, the full-length mature mRNA sequence of the human RhoA gene was retrieved from the NCBI GenBank database. Sequence alignment was performed to verify the absence of intron contamination. On the SRAMP website, the "mature mRNA mode" and "general-purpose" prediction scope were selected, which led to the identification of several potential m6A methylation sites on the RhoA mRNA sequence, including high-confidence sites.

2.10. MeRIP-PCR

Total RNA was extracted from cells using TRIzol reagent and quantified with a NanoDrop microspectrophotometer. 600 μg of total RNA was mixed with 12 μg of anti-m6A antibody (ab151230, Abcam) in IP buffer (10 mmol/L pH 7.4 Tris-HCl, 150 mmol/L NaCl, 0.1% NP-40, and 40 U/μL RNAse inhibitor) and incubated at 4 °C with rotation for 2 h. To block, 15 μL of protein A (LSKMAGA10, Millipore) and 15 μL of protein G (LSKMAGG10, Millipore) magnetic beads were added to the IP buffer containing BSA (0.5 mg/mL) and incubated at 4 °C with rotation for 2 h. The IP mixture and blocked magnetic beads were then combined and incubated at 4 °C with rotation for 2 h. The mixture was eluted with m6A 5′-monophosphate sodium salt (6.7 mmol/L, sc-215524, Santa Cruz) dissolved in IP buffer, and the eluates were combined. The IPed RNA fragments were precipitated with ethanol, and first-strand complementary DNA was synthesized using a cDNA one-strand synthesis kit for subsequent fluorescence quantitative PCR verification.

2.11. RNA pull-down

Biotin-labeled RhoA sense and antisense strands were inserted into the pcDNA3.1(+) vector, resulting in the construction of a recombinant plasmid. Following XhoI digestion and agarose gel electrophoresis purification, the Thermo MEGAscript Kit was used for in vitro transcription to obtain DRAIC RNA. Simultaneously, the RNA underwent 3′ end biotin labeling using the Pierce RNA 3′ End Desulfurization and Biotinylation Kit. MC3T3-E1 cells were cultured under osteogenic conditions with a magnesium ion concentration of 4 mmol/L, and total protein was extracted from the cells. Next, 50 μL of Pierce streptavidin magnetic beads were washed sequentially with Wash Buffer, Tris-HCl (pH 7.5), and IP Buffer. Subsequently, 3 mg of protein was added to the beads, and the mixture was incubated at room temperature with vertical suspension for 45 min to remove non-specific binding. The supernatant was collected and mixed with 3 pmol of RhoA RNA (sense and antisense strands) for a 1-h incubation. No RNA was added to the control group. It is crucial to perform the magnetic bead washing step every 45 min to ensure optimal binding efficiency. After the incubation, the solution underwent seven washes with IP Buffer to remove unbound RNA and impurities. Finally, 1 × Loading Buffer was added, and the mixture was denatured at 95 °C for 10 min. Protein separation was achieved using SDS-PAGE electrophoresis at a constant voltage of 80 V for 1 h. The gel was cut and washed sequentially with distilled water, then stained with Coomassie Brilliant Blue staining solution (0.1 g Coomassie Brilliant Blue R-250, 50 mL methanol, 10 mL ice-cold acetic acid, and 40 mL distilled water) for 30 min. After staining, the gel was decolorized using a decolorizing solution (50 mL methanol, 10 mL ice-cold acetic acid, and 40 mL distilled water) with gentle shaking for 60 min until the background became transparent. Protein bands were captured using a gel imaging system, followed by analysis of the specificity of the interaction between RhoA RNA and its target proteins.

2.12. Assessment of RNA and protein stability

To evaluate the stability of RhoA mRNA and protein, MC3T3-E1 cells were treated with 5 μg/mL actinomycin D (Millipore, Germany) for 0, 3, 6, and 9 h. Total RNA was collected at each time point, and RhoA mRNA expression was quantified via RT-qPCR. For protein stability analysis, cells were treated with 100 μg/mL cycloheximide (CHX; Millipore, Germany) for the same durations. RhoA protein levels were assessed by Western blotting at the indicated time points. The half-lives of both mRNA and protein were estimated using established methodologies [23,24].

2.13. Polysome fractionation assay

To analyze the translational status of transcripts, MC3T3-E1 cells were incubated with 100 μg/mL CHX for 10 min at 37 °C to arrest translating ribosomes. Following incubation, cells were lysed in an ice-cold lysis buffer. Sucrose density gradients were prepared according to the manufacturer's protocol for the Gradient Station (BioComp Instruments, Fredericton, Canada) and subjected to ultracentrifugation. RNA from polysome fractions was extracted and further analyzed by RT-qPCR to assess mRNA distribution [14,24].

2.14. Construction of rat tail vertebra fusion model

In the pre-experimental group, five SD rats were anesthetized with isoflurane. A skin incision was made 5 mm below the ischium at the tail, and the tail vertebrae were carefully dissected layer by layer to expose the fifth and sixth vertebral joints. The intervertebral disc was excised using a surgical knife, and the remaining nucleus pulposus and cartilage endplates were removed using a drill. Autologous bone slurry was prepared to fill the intervertebral space, and the compatibility of pure magnesium plates of different sizes was tested. After suturing, the incision was irrigated with saline. Postoperatively, meloxicam (0.02 mg/day) and sodium penicillin (5000 U/day) were administered to prevent infection and provide analgesia. Rats were euthanized after two weeks to harvest bone tissue for assessing the modeling feasibility. The formal experimental group comprised 128 rats, determined by power analysis, and was divided into four categories: the control group, which received autologous bone slurry exclusively; the magnesium ion-releasing group, which received pure magnesium plates in conjunction with bone slurry; the magnesium ion-releasing + ROCK inhibitor group, which received magnesium plates, bone slurry, and a paraspinal muscle injection of Y-27632 inhibitor (10 μg/day); and the magnesium ion-releasing + METTL3 inhibitor group, which received magnesium plates, bone slurry, and a paraspinal muscle injection of STM2457 (10 μg/day). All groups underwent the same intervertebral fusion surgery, followed by daily paravertebral injections of the corresponding drugs for two weeks postoperatively. The control and magnesium ion-releasing groups received equal volumes of physiological saline. Animals were euthanized in batches at 2, 4, 6, and 8 weeks postoperatively (eight animals per group at each time point) to obtain fixed bone tissue from the surgical site for examination.

2.15. Extraction of tissue components

Tail vertebra bone samples from the surgical area, including the fusion site and 5 mm of bone tissue above and below, were carefully cleared of paravertebral muscles and connective tissue. Twenty milligrams of tissue were placed in a nuclease-free centrifuge tube containing 3 mm grinding beads and 500 μL of Buffer RL1. The tissue was ground at low temperature until homogenized, and the tube was then centrifuged at low speed to collect the pellet. Proteinase K was added to the homogenized solution, which was incubated at 56 °C. The solution was centrifuged, and genomic DNA was removed using a gDNA Eraser Spin Column. RNA was then precipitated with isopropanol, purified using an RNA Spin Column, washed with Buffer RW1/RW2, and eluted with nuclease-free water. RT-qPCR was performed to detect gene expression. The homogenate was frozen and reacted with Mg color development solution (ab102506, Abcam). The optical density at 520 nm was measured, and magnesium ion concentration was calculated using a standard curve derived from these measurements. The RNA was extracted from the homogenate, and methylation levels were detected according to the previously established experimental procedure (same as the m6A quantification method for cells).

2.16. Micro computed tomography (CT)

To assess implant degradation and new bone formation, high-resolution micro-CT scanning was performed (Skyscan1076, Bruker, Belgium; 79 kV tube voltage, 40 kV source voltage, 250 μA source current, and 240 ms exposure time). The bone-implant contact (BIC) region was scanned at 2, 4, 6, and 8 weeks post-surgery, and the images were reconstructed using NRecon software. The reconstructed images were imported into CTAn and CTVol software for analysis of implant degradation and new bone formation.

2.17. Haematoxylin & eosin (H&E) staining

Tissue samples were grouped and labeled using embedding frames, placed in a decalcification tank, and immersed in EDTA decalcification solution. Decalcification was performed at 25–30 °C in a constant-temperature shaking incubator (110–120 rpm). The decalcification solution was replaced every 2–3 days, and needle puncture was used to assess softening. Once softening was complete, the tissue was split to expedite the decalcification process. After decalcification, the tissue was rinsed with running water to remove residual EDTA. The samples were dehydrated using a series of graded ethanol and xylene solutions, followed by paraffin embedding. Sections were cut at 4 μm, mounted on slides in warm water at 40 °C, and fixed by drying at 60 °C. Following dewaxing, haematoxylin was applied for 3–5 min, counterstaining was done with eosin for 15 s (D12621, Xiya Reagent; H9627-25G, Sigma), followed by graded dehydration and sealing with a neutral binder. Microscopic examination revealed blue-stained cell nuclei and red-stained cytoplasm.

2.18. Masson staining

The procedure was carried out following the instructions provided in the Masson staining kit (G1340, Solarbio). After sealing the slides with neutral resin, they were photographed under a microscope. Microscopic analysis of tissue samples revealed collagen fibers, which appeared blue, and muscle fibers and red blood cells, which were stained red.

2.19. Goldner trichromatic staining

To assess osteoblast activity, histological analysis was performed using the Goldner trichromatic staining kit (G3550, Solarbio). Tissue sections were initially deparaffinized in xylene and rehydrated through a graded ethanol series. Staining was carried out in strict accordance with the manufacturer's protocol. According to the staining outcome, newly formed bone tissue appeared orange-red, osteoid was stained purple, mature bone exhibited green coloration, and cell nuclei were identified as blue-gray. Morphometric analysis of the orange-red-stained new bone areas was conducted utilizing Image Pro Plus software (version 6.0.1).

2.20. Tissue immunofluorescence staining

For immunofluorescence analysis of tissue sections, slides were dewaxed and subjected to heat-mediated antigen retrieval. After blocking with 10% donkey serum for 30 min at room temperature, the sections were incubated overnight at 4 °C with primary antibodies: anti-RhoA (1:200; Cell Signaling Technology, 2117S), anti-METTL3 (1:200; Cell Signaling Technology, 86132S), and anti-Runx2 (1:200; Cell Signaling Technology, 12556S). The following day, the sections were incubated at 37 °C for 1 h with fluorescent-conjugated secondary antibodies: goat anti-mouse IgG (1:500; Abcam, ab150113) or donkey anti-rabbit IgG (1:500; Abcam, ab150064). Nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI; Solarbio, C0065), and the slides were mounted with anti-fade mounting medium. Images were acquired using an inverted fluorescence microscope (Olympus IX71, Japan). For each specimen, at least three independent sections were examined to assess protein expression patterns in rat tail vertebrae tissues.

2.21. Statistical Analysis [25,26]

Data were analyzed using GraphPad Prism version 9.5.0 (GraphPad Software, La Jolla, CA, USA) and are presented as mean ± standard deviation (SD). For comparisons between two groups under conditions of normal distribution (assessed by the Shapiro–Wilk W test) and homogeneity of variance, an unpaired t-test was applied. Comparisons among multiple groups meeting these assumptions were performed using one-way analysis of variance (ANOVA). In cases where data did not follow a normal distribution or variances were unequal, the Wilcoxon rank-sum test was used for two-group comparisons, and the Kruskal–Wallis test was employed for multiple groups. Post-hoc pairwise comparisons were conducted using the LSD test. A two-sided significance level of α = 0.05 was adopted for all tests.

3. Results

3.1. Magnesium has a threshold effect on the differentiation of osteoblasts

To investigate the osteogenic differentiation of osteoblasts under different magnesium concentrations and establish the magnesium environment for subsequent osteogenesis-related mechanism research, MC3T3-E1 cells were cultured in conventional OM supplemented with 0 mmol/L, 2 mmol/L, 4 mmol/L, 8 mmol/L, and 16 mmol/L Mg²⁺. After culture, alkaline phosphatase (ALP) staining and alizarin red S (ARS) staining were performed to assess ALP activity and mineralization capacity. ALP staining results indicated that within the Mg²⁺ concentration range of 0 to 4 mmol/L, Mg²⁺ positively influenced the osteogenic differentiation of the cells, as evidenced by enhanced ALP activity. Specifically, the ALP activity in a 4 mmol/L Mg²⁺ environment was about 6.7 times that in a 0 mmol/L Mg²⁺ environment, suggesting vigorous osteogenic differentiation. In contrast, when Mg²⁺ concentrations exceeded 4 mmol/L, ALP activity significantly decreased, indicating that high Mg²⁺ concentrations inhibit osteogenic differentiation and that the therapeutic window for Mg²⁺ is relatively narrow (Fig. 1A and B). Notably, ALP is a key marker of early osteogenic differentiation, with high activity reflecting active differentiation into mature osteoblasts. ARS staining results revealed that at a magnesium ion concentration of 4 mmol/L, a substantial number of deep-red mineralized nodules were formed, with the quantity being approximately 3.6 times that in the magnesium-free environment, and the calcium salt deposition significantly increased. When Mg²⁺ concentrations ranged from 0 to 4 mmol/L, calcium salt deposition was visible in the red-stained cells, with a positive correlation between magnesium ion concentration and calcium salt deposition, as demonstrated by the deepening staining. However, when Mg²⁺ concentrations reached 8 mmol/L or higher, no significant calcium salt deposition was observed, suggesting reduced mineralization capacity (Fig. 1C and D). Calcium salt deposition is a key indicator of osteoblast differentiation and their functional role in bone tissue formation, implying that a magnesium ion concentration of 4 mmol/L is optimal for osteogenic mineralization in MC3T3-E1 cells. Conversely, concentrations exceeding 4 mmol/L may impair the cells' ability to undergo osteogenic differentiation and form mineralized matrices.

Fig. 1.

Different concentrations of magnesium ions can affect the osteogenic properties of MC3T3-E1 cells. (A-D) Following the addition of magnesium ions at different concentrations, MC3T3-E1 cells were subjected to ALP staining (A-B) and ARS staining (C-D). Subsequently, the staining results were quantitatively analyzed. (E) RT-qPCR was employed to assess the impact of different magnesium ion concentrations on the mRNA expression of osteogenesis - related genes in cells. (F-G) Western blotting was utilized to determine the influence of different magnesium ion concentrations on the expression of osteogenesis - related proteins in cells. (Data are presented as mean ± SD from three independent experiments (n = 3). Statistical differences were analyzed using one-way ANOVA. Post-hoc pairwise comparisons were conducted using the LSD test.).

RT-qPCR analysis demonstrated that the expression levels of osteogenesis-related genes, including Col1a1, BMP4, Runx2, OCN, and OPN, were significantly influenced by variations in Mg²⁺ concentration (Fig. 1E). In the 4 mmol/L Mg²⁺ treatment group, the expression of these genes peaked and was significantly upregulated compared to the Mg²⁺-free group, indicating that this concentration of Mg²⁺ effectively promotes osteogenic gene expression, in line with the enhanced ALP activity and improved mineralization observed previously. However, at a concentration of 16 mmol/L Mg²⁺, there was a marked decrease in osteogenic gene expression compared to the Mg²⁺-free group, suggesting that high Mg²⁺ concentrations may inhibit osteogenic differentiation. Western blotting analysis confirmed that the protein expression levels of Runx2, OCN, and OPN in the 4 mmol/L Mg²⁺ group were significantly higher than in the other groups, while the 16 mmol/L group showed significantly lower expression compared to the Mg²⁺-free group (Fig. 1F and G). The CCK-8 assays showed that 4 mmol/L Mg²⁺ did not inhibit cell proliferation, whereas 16 mmol/L showed significant cytotoxicity. This confirms that our chosen concentration (4 mmol/L) is within the safety window (Fig. S1). In conclusion, Mg²⁺ exerts a concentration-dependent effect on osteogenic-related gene and protein expression, with 4 mmol/L being the optimal concentration. Concentrations above this level, such as 16 mmol/L, exhibit an inhibitory effect.

3.2. Mg²⁺ enhances osteogenesis through upregulating METTL3

Dot blot fluorescence intensity analysis showed that the m6A methylation modification levels of mRNA in Mg²⁺-treated cells were significantly higher than those in the control group (Fig. 2A), indicating that Mg²⁺ promotes RNA methylation. ELISA further confirmed that the m6A modification content in the total RNA of Mg²⁺-treated cells was significantly higher than in the control group (Fig. 2B), supporting the positive regulatory effect of Mg²⁺ on m6A methylation. RT-qPCR results demonstrated a significant increase in METTL3 mRNA expression in the Mg²⁺ group (Fig. 2C and D), while the expression of other core regulatory genes, such as METTL14, ALKBH5, and FTO, remained unchanged. This suggests that the regulation of m6A modification by Mg²⁺ is primarily mediated by METTL3, rather than by other known methylation/demethylation enzymes.

Fig. 2.

Magnesium ions enhance osteogenesis through upregulating METTL3. (A) Dot blot experiments were conducted on MC3T3-E1 cells from different groups to determine the mRNA methylation levels and gray-scale analysis was performed. (B) The content of m6A modification in total RNA of MC3T3-E1 cells in different groups was determined by using the m6A RNA methylation quantification kit. (C-D) RT-qPCR was employed to assess the alterations in the mRNA levels of METTL3, ALKBH5, METTL14 and FTO within MC3T3-E1 cells following 4 mmol/L magnesium treatment. (E) The content of m6A modification in total RNA of MC3T3-E1 cells in different groups was determined by using the m6A RNA methylation quantification kit. (F) RT-qPCR was employed to assess the alterations in the mRNA levels of METTL3, Runx2 and OCN within MC3T3-E1 cells following different treatments. (G-H) Western blotting was utilized to measure the protein levels of METTL3, Runx2, and OCN in cells from different treatment groups. (I) MC3T3-E1 cells were subjected to ALP staining and ARS staining, and the results were subsequently quantified. (Data are expressed as mean ± SD (n = 3). Statistical significance was determined by Student's t-test (between two groups) or one-way ANOVA (among multiple groups)).

To explore the role of METTL3 in regulating osteogenic differentiation, a METTL3-overexpressing stable cell line (OE-METTL3 group) was established, and the METTL3 selective inhibitor STM2457 (METTL3-IN group) was used. RT-qPCR (Fig. 2F) and Western blotting analyses (Fig. 2G and H) showed that METTL3 mRNA and protein levels were significantly higher in the OE-METTL3 group than in the empty vector control group (OE-NC group). Concurrently, the transcription and translation levels of osteogenic genes Runx2 and OCN were elevated. ALP and ARS staining further confirmed that METTL3 overexpression significantly enhanced ALP activity and mineralization capacity (Fig. 2I), with effects surpassing those observed in the Mg²⁺-treated cells (Mg group). There were no significant differences in m6A modification levels or osteogenic gene expression between the Mg and METTL3-OE groups (Fig. 2E), suggesting that Mg²⁺ facilitates osteogenesis by enhancing METTL3 expression. In the Mg²⁺ environment, the addition of STM2457 to inhibit METTL3 enzyme activity led to a significant reduction in total m6A modification levels compared to the Mg group (Fig. S2A). However, the mRNA and protein expression of METTL3 remained unchanged (Fig. S2B–D), as STM2457 selectively inhibits the methyltransferase activity of METTL3 without affecting its transcription or translation. RT-qPCR and Western blotting analyses revealed that osteogenic genes Runx2 and OCN were significantly downregulated in the METTL3-IN group, and ALP activity and osteogenic capacity assays showed similar trends (Fig. S2E). These results confirm that the methyltransferase activity of METTL3 is essential for Mg²⁺-induced osteogenesis.

3.3. Mg facilitates osteogenesis through the regulation of RhoA

To identify the specific genes whose m6A modifications are regulated by the magnesium environment, mRNA methylation microarray assays were conducted. Total mRNAs were hybridized with probes targeting 48,161 mRNA loci, 8393 long non-coding RNA (lncRNA) loci, and 4087 mid-size non-coding RNA (ncRNA) loci (including quality control) that have been previously investigated. This approach provided the global transcriptome profiles of mRNAs and the epitranscriptome profiles of m6A modifications in MC3T3-E1 cells under magnesium conditions. Cluster analysis based on gene expression levels revealed 15,636 mRNAs. Compared to the control group, the magnesium-treated group displayed 555 mRNAs with increased expression and 187 mRNAs with decreased expression. In the magnesium environment, the mRNA levels of osteogenesis-related genes, including BMP4, Bglap3, Col1a1, Opn4, and Runx2, were significantly upregulated (Fig. 3A). Next, cluster analysis was performed to examine m6A modification levels. The magnesium environment group exhibited 682 significantly upregulated m6A-modified mRNAs and 167 significantly downregulated m6A-modified mRNAs. The overall mRNA methylation modification levels were elevated under magnesium-rich conditions (Fig. 3B). To identify genes with differences in both m6A modification levels and transcriptional levels, genes with p < 0.05 were selected for both m6A modification and mRNA expression in the cluster analysis (Fig. 3C). These genes may be regulated by m6A modifications, influencing transcript behavior and downstream functions. Using the unique fluorescence probe technology of methylation microarray detection, the percentage of m6A-modified and non-m6A-modified subpopulations within each transcript was calculated. This was done by comparing the fluorescence signal intensities of specific probes in RIP-enriched and m6A-RIP-enriched products. This value reflects the proportion of m6A-modified transcripts among the total transcripts. To identify specific downstream targets driving Mg²⁺-induced osteogenesis, a conjoint analysis of MeRIP-seq and RNA-seq data was performed using a four-quadrant plot. While several classical osteogenic genes, including Bmp4 and Runx2, showed upregulation, candidates were prioritized based on a “Hyper-methylation and Up-regulation” screening strategy (Fig. 3D). Among the overlapping genes, RhoA emerged as the top candidate for two key reasons. First, in terms of quantitative significance, RhoA exhibited one of the largest increases in both m6A methylation abundance and mRNA expression levels. Second, for biological relevance: Mg²⁺ is known to promote osteogenesis by regulating cell adhesion and cytoskeletal tension, and RhoA—a master regulator of the actin cytoskeleton—provides an intuitive molecular link between the extracellular ionic stimulus and the intracellular morphological changes observed. Based on these findings, the METTL3-m6A-RhoA axis was selected for further mechanistic testing.

Fig. 3.

Mg facilitate osteogenesis through the regulation of RhoA. (A) mRNA methylation microarray assays were conducted on cells from different groups. All probe sites exhibiting differences were meticulously recorded, and the corresponding transcripts were comprehensively analyzed. A heatmap was employed to visually represent all mRNAs with differential expression levels (left). Cluster analysis was then carried out on the assay results, where each data point represented a distinct transcript. The logarithm of the mRNA expression level was designated as the X - axis, and the negative logarithm of the p - value was set as the Y - axis. Using a threshold of p < 0.05 and |Fold Change| > 1.5, a total of 555 genes with significantly upregulated expression and 187 genes with significantly downregulated expression were identified. Select genes are labeled within the figure (right). (B) mRNA methylation microarray analyses were conducted on cells from different groups. All probe sites showing differential signals were precisely recorded, and the associated transcripts were subsequently analyzed. A heatmap was utilized to visualize all mRNAs presenting differences in m6A modification levels (left). Cluster analysis was then performed on the assay outcomes, with each data point representing a distinct transcript. The logarithm of the m6A modification level was designated as the X - axis, while the negative logarithm of the p - value was set as the Y - axis. Using a threshold of p < 0.05 and |Fold Change| > 1.5, a total of 682 genes with significantly upregulated m6A levels and 167 genes with significantly downregulated m6A levels were identified. A selection of these genes is labeled in the figure (right). (C) Select all genes with m6A modification level results showing p < 0.05 (irrespective of the value of Fold Change) and all genes with mRNA expression level results showing p < 0.05 (regardless of the magnitude of Fold Change) that were screened out from the cluster analysis. The intersection of these two gene sets was determined using a Venn diagram, yielding 633 potential genes. (D) Cluster analysis was performed on the potential genes screened from Figure C. Each data point in the analysis represents a distinct transcript. The logarithm of the m6A modification level was designated as the X - axis, while the inter - group difference (Diff) in the percentage of m6A - modified transcripts relative to total mRNA transcripts was set as the Y - axis. Using a threshold of Diff >5% and |Fold Change| > 1.5, a total of 97 genes were identified. These genes exhibit both an up - regulation in the percentage of m6A modification and an increase in the m6A modification level. A number of these genes are labeled in figure. (E) All differentially expressed genes were selected for GO analysis. The enrichment scores of genes under each GO term were calculated through GO analysis. After taking the logarithm, relevant GO classifications with high enrichment scores were sorted and listed in three dimensions: Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). (F) For the GO terms related to biological process selected in Figure E, a dot plot was constructed. In this plot, the p - value differences were visually represented by the color of the dots, the number of genes was shown by the dot size, and the enrichment scores were depicted on the X - axis to display potential pathway entries. (G) For the GO terms related to cellular components selected in Figure E, a dot plot was constructed. (H) For the GO terms related to molecular functions selected in Figure E, a dot plot was constructed. (I) The bar chart of KEGG analysis depicts the identified signaling pathways. The enrichment score is presented on the X - axis, the pathway name is shown on the Y - axis, and the length of each bar corresponds to the number of genes associated with that particular pathway. (J) Perform KEGG pathway analysis on genes of statistical significance identified in the clustering analysis. Screen for cellular pathways that contain a large number of genes with differential m6A modification levels and genes with differential mRNA expression levels. On the pathway map, genes with upregulated m6A modification levels are labeled in yellow, genes with upregulated mRNA modification levels only are labeled in red, and active genes that show no significant expression differences but have relatively high intracellular levels are labeled in green. The figure presented herein shows the pathway map of the staining of peripheral genes associated with the selected RhoA - ROCK pathway. (H) Select all the active genes with statistical significance in the cluster analysis, and draw a network relationship diagram of the important pathway genes related to the RhoA gene among them. In this diagram, dots represent genes and lines represent relationships. The size of the dots indicates the significance level, and the thickness of the lines represents the literature support for the interaction.

Gene Ontology (GO) analysis indicated that genes with altered mRNA modification levels under magnesium-rich conditions were primarily involved in pathways related to cell receptor binding, intracellular cytoskeletal structure, and cellular processes (Fig. 3E–H). Expression levels of genes associated with these pathways were upregulated, along with the increased expression of m6A-modified genes. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis revealed that genes with elevated methylation modification ratios and mRNA expression were predominantly enriched in ion channel-related pathways (Fig. 3I). Among these, RhoA and its downstream genes exhibited the most significant increases in both expression and m6A modification levels, with RhoA activity and its downstream signaling pathways being markedly elevated (Fig. 3J). Protein-protein interaction (PPI) network analysis showed that genes co-expressed with high expression and m6A modification levels included their interacting co-factors and several upregulated downstream signaling molecules, such as members of the ROCK family, myosin light chain (MLC), and key molecules involved in ion channels and bone development, including members of the TRPV channel family (Fig. 3K). These results suggest that the RhoA/ROCK pathway plays a pivotal role in ion regulation and bone tissue development under magnesium-rich conditions, indicating that Mg²⁺ may activate this pathway.

3.4. Magnesium facilitates bone formation via the METTL3/RhoA/ROCK axis

Based on the high levels and ratios of m6A modification of RhoA under magnesium ion influence, it was hypothesized that RhoA is a key pro-bone gene downstream of METTL3 in a magnesium-rich environment. To test this hypothesis, the compound STM2457 was used to inhibit METTL3 function in MC3T3-E1 cells under simulated magnesium conditions. Western blotting and RT-qPCR were employed to assess RhoA expression, downstream protein levels, and cellular m6A modification levels under these conditions. The results indicated that inhibition of METTL3 led to a decrease in RhoA protein expression, accompanied by a reduction in its downstream protein ROCK1 expression (Fig. 4A). Although STM2457 decreased the mRNA level of ROCK1, it had no effect on the mRNA level of RhoA (Fig. 4B).

Fig. 4.

Knocking down RhoA inhibits osteogenesis. (A) Western blotting was employed to assess the expression levels of METTL3, RhoA, and ROCK1 proteins in cells from different groups. (B) RT-qPCR was employed to assess the expression levels of METTL3, RhoA, and ROCK1 mRNA in cells from different groups. (C) Small interfering RNAs (siRNAs) were utilized to knockdown RhoA, and the knockdown efficiency was validated via Western blotting. (D) RT-qPCR was employed to assess the expression levels of METTL3, RhoA, Runx2, OCN and ROCK1 mRNA in cells from different groups. (E) Western blotting was employed to assess the expression levels of METTL3, Runx2, OCN and ROCK1 proteins in cells from different groups. (F) MC3T3-E1 cells were subjected to ALP staining and ARS staining, and the results were subsequently quantified. (Data are presented as mean ± SD from three independent experiments (n = 3). Statistical differences were analyzed using one-way ANOVA. Post-hoc pairwise comparisons were conducted using the LSD test.).

A stable RhoA-knockdown cell line (si-RhoA group) was established to investigate the role of the RhoA/ROCK pathway in Mg²⁺-induced osteogenic differentiation (Fig. 4C). RT-qPCR (Fig. 4D) and Western blotting (Fig. 4E) analyses revealed that both RhoA mRNA and protein expression were significantly decreased in the si-RhoA group compared to the si-NC group. Concurrently, the transcription and translation of osteogenic genes, such as Runx2 and OCN, were reduced. Additionally, the expression of ROCK1 and its downstream proteins decreased. ALP and ARS staining showed that RhoA knockdown led to a significant decrease in ALP activity and mineralization capacity (Fig. 4F). Inhibition of RhoA with siRNA resulted in decreased RhoA/ROCK pathway activity and a substantial reduction in the expression of Runx2, OCN, and osteogenic markers, compared to the Mg²⁺ group. These results confirm the positive regulatory role of RhoA in osteogenic differentiation.

To further validate the role of the METTL3/RhoA pathway in osteogenesis, siRNA was used to knock down METTL3 (Fig. S3A), followed by RhoA overexpression. RT-qPCR (Fig. S3B) and Western blotting (Fig. S3C) analyses demonstrated that Runx2, OCN, and ROCK1 expression was significantly downregulated compared to both the Mg²⁺ group and METTL3-overexpressing group. However, overexpression of RhoA significantly increased the levels of Runx2, OCN, and ROCK1 at both mRNA and protein levels. Additionally, knocking down METTL3 had no effect on RhoA expression, and overexpressing RhoA did not affect METTL3 expression. Assays of ALP activity and mineralization capacity showed similar results (Fig. S3D). METTL3 knockdown suppressed the osteogenic phenotype, while RhoA overexpression reversed this effect. This suggests that the RhoA/ROCK pathway is downstream of the Mg²⁺-METTL3 axis and serves as a mediator in converting m6A modification into the osteogenic phenotype. To ensure the translational relevance of this discovery, we verified the core “Mg²⁺-METTL3-RhoA” axis in human bone marrow mesenchymal stem cells (hBMSCs). The results of Western blotting (Fig. S4A–B) and RT-qPCR (Fig. S4C) showed that that 4 mM Mg²⁺ significantly upregulated METTL3 and RhoA expression in hBMSCs. Furthermore, METTL3 knockdown in hBMSCs suppressed Mg-induced osteogenic differentiation, consistent with our findings in MC3T3-E1 cells. This study confirms that Mg²⁺ activates METTL3 to enhance m6A modification, thereby promoting RhoA/ROCK pathway activity and driving the transcription and functional activation of key genes involved in osteogenic differentiation. However, METTL3 does not directly promote RhoA expression, indicating the involvement of another regulatory mechanism.

3.5. YTHDF1 promotes the translation of m6A-modified RhoA mRNA into protein

Based on previous bioinformatics analysis, the RhoA gene was identified as having an upregulated m6A modification level and modification percentage in the presence of Mg²⁺. Furthermore, associations between the RhoA/ROCK pathway, downstream osteogenic phenotypes, and upstream methylation modifications were uncovered. However, the mechanism by which m6A-modified RhoA exerts its function remains unclear. Therefore, it is critical to identify proteins that directly interact with m6A modification sites.

A comprehensive analysis of RhoA m6A-RIP samples was conducted using methylation chip fluorescent probes, leading to the identification of high-confidence m6A methylation sites (predicted by SRAMP) in the RhoA 3′UTR region. These sites were identified based on the GAACT sequence, which overlaps with the promoter-binding region (Fig. 5A). MeRIP-PCR validation showed that the m6A modification abundance at this site was significantly higher in Mg²⁺-induced cells compared to the control group. Moreover, mutating this site eliminated the m6A signal and removed intergroup differences, confirming that this sequence is a key site for Mg²⁺-dependent m6A modification (Fig. 5B). Further screening of TFBS databases identified RhoA 3′UTR-interacting proteins, and GO analysis highlighted factors related to m6A methylation regulation. This analysis identified YTHDF1 as the only candidate protein linked to m6A reading and RhoA/ROCK pathway activation (Fig. 5C). Given the established role of YTHDF1 in m6A-mediated post-transcriptional regulation, m6A modification of RhoA may enhance its mRNA stability or translation efficiency by recruiting YTHDF1, thereby activating the downstream RhoA/ROCK pathway to drive osteogenic differentiation.

Fig. 5.

YTHDF1 promotes translation of RhoA mRNA upon binding to it. (A) The methylation site sequences and their predicted reliability were screened out based on the probe sequence data of the m6A methylation detection chip through SRAMP analysis. (B) The MeRIP-PCR experiment was conducted to verify the accuracy of the predicted m6A sites. RhoA wild-type (RhoA-WT) and m6A site mutant (RhoA-MUT) primers were used to test the two groups of cells. (C) Utilizing the TFBS (Transcription Factor Binding Site) database, genes capable of binding to the modified sequences at the 3′UTR region were screened. These genes were then correlated with those associated with methylation modification regulation entries in the GO analysis. A Venn diagram was constructed to identify the intersection, thereby screening out potential candidate genes. (D) Following the pulldown assay, protein electrophoresis bands of the positive control group (Input), experimental group (Pulldown), and negative control group (NC) were subjected to Coomassie Brilliant Blue staining. (E) After the pulldown assay, Western Blotting analysis was performed on the proteins from each group to validate their enrichment levels. (F) RIP‐derived protein and RNA in MC3T3-E1 cells examined using Western blotting and RT-qPCR, respectively. (G) After treating MC3T3-E1 cells with actinomycin D and si-YTHDF1, mRNA was detected by qRT-PCR at 0, 3, 6, and 9 h, respectively. The Y-axis represents the relative level of mRNA compared to 0 h. (H) After treating MC3T3-E1 cells with actinomycin D and OE-YTHDF1, mRNA was detected by qRT-PCR at 0, 3, 6 and 9 h, respectively. (I-J) After treating MC3T3-E1 cells with CHX and si-YTHDF1, protein was detected by WB at 0, 3, 6, and 9 h, respectively. The Y-axis represents the relative level of protein compared to 0 h. (K-L) After treating MC3T3-E1 cells with CHX and OE-YTHDF1, protein was detected by WB at 0, 3, 6, and 9 h, respectively. (M) After treating MC3T3-E1 cells with si-YTHDF1, qRT-PCR analysis of RhoA mRNA distribution in different ribosome populations. (N) After treating MC3T3-E1 cells with OE-YTHDF1, qRT-PCR analysis of RhoA mRNA distribution in different ribosome populations. (O) YTHDF1's ability to recognize m6A modifications depends on m6A-binding pockets in the YTH domain. (P) RIP-derived protein and RNA in MC3T3-E1 cells examined using western blotting and RT-qPCR, respectively. (Q) WB was used to detect RhoA expression in the MC3T3-E1 cells. (Data are expressed as mean ± SD (n = 3). Statistical significance was determined by Student's t-test (between two groups) or one-way ANOVA (among multiple groups))

This study employed a combination of pulldown experiments and Western blotting analysis to investigate the interaction between RhoA and YTHDF1 in Mg²⁺-induced MC3T3-E1 cells. Binding was specific to the sense strand of RhoA RNA, but not the antisense strand. The analysis revealed a specific interaction between the two proteins. After enriching biotin-labeled RhoA using magnetic beads, a distinct band with a molecular weight corresponding to YTHDF1 (63 kD) was observed in the 55–75 kD range (Fig. 5D). Western blotting further confirmed that the enriched protein was YTHDF1 and did not contain non-specific proteins such as β-Actin (Fig. 5E). RIP analysis also confirmed that YTHDF1 interacts with RhoA mRNA (Fig. 5F). To further explore this interaction, MC3T3-E1 cells were treated with actinomycin D (a transcription inhibitor). Knocking down YTHDF1 led to a decrease in RhoA mRNA levels at various time points, similar to the reduction observed in control cells (Fig. 5G). Additionally, overexpression of YTHDF1 did not affect the decrease in RhoA mRNA levels (Fig. 5H), indicating that YTHDF1 does not influence RhoA mRNA stability. Furthermore, treatment with CHX, a protein translation inhibitor, revealed that YTHDF1 did not affect RhoA protein stability or its degradation rate (Fig. 5I–L). However, polysome profiling demonstrated that YTHDF1 knockdown led to a decrease in RhoA mRNA in the translation fractions (Fig. 5M), while YTHDF1 overexpression increased RhoA mRNA in the translation fractions (Fig. 5N), suggesting that YTHDF1 regulates RhoA translation.

As an m6A “reader,” YTHDF1 recognizes m6A modifications through binding pockets in its YTH domain. Mutations in K395 and Y397 abolished YTHDF1's ability to bind mRNA. Additionally, mutations were introduced into the YTH domain (Fig. 5O). YTHDF1-MUT could not recognize the m6A modification on RhoA mRNA, and the interaction between the two was nearly eliminated (Fig. 5P). Western blotting further showed that only YTHDF1-WT, not YTHDF1-MUT, could enhance RhoA protein levels in cells (Fig. 5Q). This study further assessed whether YTHDF1-MUT (Fig. S5A–B), which cannot recognize m6A modifications, and RhoA-MUT (Fig. S5C–D), which cannot be modified by m6A, could promote osteogenesis. The results showed that neither YTHDF1-MUT nor RhoA-MUT could promote osteogenesis, whereas YTHDF1-WT and RhoA-WT effectively promoted osteogenesis in cells. These results suggest that YTHDF1 can recognize and bind to m6A-modified RhoA mRNA, thereby enhancing its translation into protein and facilitating cellular osteogenesis.

To further confirm that the entire effect is mediated by METTL3-dependent m6A modification of RhoA, we conducted a crucial rescue experiment. We demonstrated that reintroducing wild-type rather than m6A site mutant RhoA transcripts in the presence of Mg²⁺ could rescue the osteogenic defect caused by METTL3 knockdown (Fig. S6A–C).

3.6. Magnesium facilitates the fusion of rat caudal vertebrae via the regulation of METTL3 and RhoA

To investigate the impact of the magnesium environment on intervertebral fusion, a model of osteogenic fusion at the 5–6 segments of rat caudal vertebrae, combined with magnesium plate implantation, was established (Fig. 6A). Based on the anatomical features of rat caudal vertebrae, a surgical technique was developed for bone fusion of the rat caudal vertebrae. High-purity magnesium metal plates with thicknesses of 0.2 mm or 0.5 mm were locally implanted in the bone fusion tissue. Two weeks after surgery, samples of the caudal vertebrae were harvested. The results showed that all rats in the preliminary experiment survived the surgery, exhibited good tail mobility, and experienced satisfactory wound healing. Although the bone tissue had not fully healed, the fused segments remained relatively stable under the fixation of the caudal ligaments. Partial degradation of the magnesium metal plates was observed (Fig. S7A–D), indicating the successful establishment of the rat caudal vertebrae model.

Fig. 6.

Magnesium facilitates the fusion of rat caudal vertebrae via the regulation of METTL3 and RhoA. (A) Flowchart of animal experiments. (B) Magnesium ion quantification assays were performed on newly formed bone tissues from different groups using a magnesium ion quantitative kit. (C) Determination of magnesium ion concentration in tissues at different time points after surgery. Based on the in vitro experimental findings of this study, the effective concentration range is 2–6 mmol, with the safe concentration range being <16 mmol. (D) MeRIP-qPCR was used to detect the m6A modification abundance of RhoA mRNA in the newly formed bone tissue. (E) qPCR experiments were conducted on the new bone tissues of different groups to determine the expression levels of different mRNAs. (F) Western Blotting experiments were conducted on bone tissues of different groups to determine the expression levels of different proteins. (Data are presented as mean ± SD from three independent experiments (n = 3). Statistical differences were analyzed using one-way ANOVA. Post-hoc pairwise comparisons were conducted using the LSD test.).

Local tissue was harvested to determine the magnesium ion concentration and screen the thickness of the implanted magnesium plates. Newly formed bone tissue samples were homogenized, and the magnesium ion concentration was measured using a magnesium ion assay kit. The results showed that the local magnesium ion concentration in the tissues of rats without magnesium plate implantation was approximately 2 mmol/L, while in rats with magnesium plate implantation, the concentration was around 6 mmol/L. These results suggest that the implanted magnesium plates effectively regulate and release the magnesium ion concentration in the intervertebral region of rats within an appropriate range. Additionally, the variation in local magnesium ion concentration was relatively minor between rats implanted with magnesium plates of different thicknesses and widths (Fig. S7E). A magnesium plate with dimensions of 0.5 × 3 × 8 mm was selected for subsequent experimental implantation.

Rats were divided into four groups: the control group (Ctrl group), which underwent bone graft fusion without magnesium plate implantation; the magnesium ion controlled-release group (Mg group), which received magnesium plate implantation; the METTL3 inhibitor group (METTL3-IN group), which received magnesium plate implantation and METTL3 inhibitor administration; and the RHOA inhibitor group (RHOA-IN group), which received magnesium plate implantation and RHOA inhibitor administration. Two weeks after surgery, samples of new bone tissue in the intervertebral space from each group were collected. Magnesium ion concentration measurement revealed that the Ctrl group tissue had a magnesium ion concentration of approximately 2 mmol/L. The magnesium ion concentrations in the Mg, METTL3-IN, and RHOA-IN groups were around 6 mmol/L, with no significant differences among the groups, indicating that the sustained-release effect of Mg²⁺ was optimal (Fig. 6B). We then collected tissue samples around the fusion site at 2-, 4-, 6- and 8-weeks post-surgery and measured the local magnesium ion concentration. The local Mg²⁺ concentration peaked at Week 2 (reaching approximately 5.66 mM) due to the initial rapid degradation, followed by a gradual decrease to 4.43 mM at Week 4 and 2.45 mM at Week 8 (Fig. 6C). In addition, we also detected the m6A modification abundance of RhoA mRNA in the newly formed bone tissue of rats in different groups two weeks after the operation. The results showed that the enrichment degree of m6A on RhoA mRNA in the Mg group was significantly higher than that in the control group (Fig. 6D).

RT-qPCR analysis of METTL3 and RhoA, along with Western blotting analysis of METTL3 and ROCK1, showed that magnesium plate implantation upregulated the expression of both genes, while the application of inhibitors effectively attenuated this upregulation. Specifically, RT-qPCR results indicated that the osteogenic phenotype gene Runx2 was significantly upregulated in the Mg group compared to the Ctrl group. In contrast, in both the RHOA-IN and METTL3-IN groups, the expression of Runx2 was downregulated following the administration of inhibitors (Fig. 6E). Western blotting further confirmed these findings, showing a similar trend in the expression of the RUNX2 protein (Fig. 6F). These results suggest that magnesium plate implantation promoted a more favorable early osteogenic phenotype in rats. However, the use of METTL3 and RHOA inhibitors partially reversed this beneficial effect.

Firm fusion was defined as continuous bony bridging across the intervertebral space in three consecutive slices. Bone CT scan results indicated that the Mg group achieved relatively firm fusion at 6 weeks postoperatively, while the RHOA-IN and METTL3-IN groups reached firm fusion at 8 weeks. Over time, the intervertebral tissue was gradually replaced by newly formed trabecular bone from the two end vertebrae (Fig. 7A). Inter-group comparisons revealed that the bone volume (BV) and the ratio of bone volume to tissue volume (BV/TV) in the Mg group were higher than in the other groups (Fig. S7F–H), indicating that the amount of newly formed bone in the Mg group was greater. The bone surface area (BS) and the ratio of bone surface area to bone volume (BS/BV) in the Mg group also increased significantly, suggesting a higher level of bone formation activity in the tissue (Fig. S7I–J). Both BV and BS were higher in the Mg, METTL3-IN, and RHOA-IN groups compared to the Ctrl group, suggesting that magnesium plate implantation promoted osteogenesis during intervertebral fusion. Further trabecular bone analysis showed that rats in the Mg group outperformed those in other groups in terms of trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), trabecular number (Tb.N), and bone mineral density (BMD). In contrast, rats in the Ctrl group exhibited poorer performance in all these metrics compared to the other groups (Fig. S7K–N). These results suggest that magnesium plate implantation facilitated bone formation during intervertebral fusion, leading to a more compact trabecular structure, improved morphology, and enhanced load-bearing capacity. However, inhibition of METTL3 or RhoA partially counteracted this beneficial effect. A more detailed comparison between the RHOA-IN and METTL3-IN groups revealed that RhoA inhibition primarily increased Tb.Sp, which likely resulted in suboptimal morphology of the newly formed trabeculae. In contrast, METTL3 inhibition led to reduced Tb.Th and lower bone mineralization, potentially causing insufficient calcium salt deposition in the newly formed trabeculae and compromising their compressive strength. These results indicate that each component of this pathway plays a unique but complementary role in the regulation of bone quality.

Fig. 7.

Imaging and histological detection of intervertebral fusion of the tail vertebrae in rats. (A) Micro-CT scan images of bones at different times in different groups. (B) H&E-stained sections of bone tissues from each group at 2 weeks post-surgery. (C) Goldner - stained sections of bone tissues from each group at 2 weeks post - operation. (D) Masson - stained sections of bone tissues from each group at 2 weeks post - operation. (E) Immunofluorescent staining sections of bone tissues from each group at 2 weeks post - surgery (DAPI stained blue, RhoA stained red, METTL3 stained yellow, and Runx2 stained green).

3.7. Histological staining evidenced that the implantation of magnesium plates facilitated the intervertebral fusion of caudal vertebrae in rats

H&E staining across different time points revealed that bone tissue progressively filled the intervertebral space between 2 and 8 weeks. In the Mg group (Fig. S9A), at 2 weeks, inflammatory and fibrotic tissue occupied the gap, with bone formation beginning at the adjacent vertebrae. By 4 weeks, significant new bone formation was observed, and by 6 weeks, trabeculae largely bridged the vertebrae. At 8 weeks, complete fusion with mature cancellous architecture was achieved.

The Mg group exhibited a significantly faster fusion rate compared to the other groups, with visible fusion as early as 2 weeks, while the Ctrl group showed the slowest progress (Fig. S8A). At 2 weeks, the Mg group had well-mineralized new bone trabeculae, whereas the Ctrl group exhibited sparse bone formation and inflammatory infiltration. The METTL3-IN group presented osteoid with low mineralization, and the RhoA-IN group showed irregular trabeculae with a chondrogenic tendency (Fig. 7B). By 6 weeks, the Mg group showed complete fusion with minimal connective tissue remnants. The Ctrl group lagged behind, resembling the Mg group at 4 weeks. The RhoA-IN group showed poor trabecular merging, and the METTL3-IN group had low mineralization despite adequate trabecular growth (Fig. S9B).

Goldner staining highlighted mineralization dynamics (Fig. S9C), with the Mg group displaying the largest mineralized osteoid area and highest maturity at 2 weeks (Fig. 7C, Fig. S8B). The METTL3-IN group produced abundant osteoid but poor mineralization, while the RhoA-IN group exhibited congested trabeculae with inflammatory residues. By 6 weeks (Fig. S9D–E), the Mg group achieved full trabecular apposition, while the RhoA-IN group showed improved but ununited trabeculae, and the METTL3-IN group had immature trabeculae despite rapid growth.

Masson staining revealed the progression from fibrovascular tissue to mature trabecular bone with collagen deposition over 2-8 weeks (Fig. S9F). At 2 weeks, the Mg group exhibited significantly larger areas of new trabeculae and collagen coverage compared to the other groups. The RhoA-IN group showed inflammatory infiltration and irregular trabeculae, while the METTL3-IN group produced numerous but poorly mineralized trabeculae with minimal collagen attachment (Fig. 7D, Fig. S8C). By 6 weeks, the Mg group displayed mature, densely collagen-coated trabeculae, while the Ctrl group lagged in maturation. The RhoA-IN group exhibited slow ossification despite collagen presence, and METTL3-IN showed low collagen adhesion and immaturity. These structural findings were consistent with Goldner staining results (Fig. S9G–H).

Immunofluorescence at 2 weeks revealed significantly elevated expression of Runx2, RhoA, and METTL3 in the Mg group, particularly along nascent trabeculae (Fig. 7E). The Ctrl group showed weak signals overall. The RhoA-IN group displayed attenuated RhoA expression with inflammatory localization of Runx2 and METTL3, along with suboptimal trabecular morphology. The METTL3-IN group showed reduced expression of all three markers, particularly RhoA. These results suggest that Mg²⁺ enhances osteogenesis via coordinated upregulation of METTL3-RhoA-Runx2 signaling, and inhibition of either RhoA or METTL3 partially reverses this pro-osteogenic effect.

Overall, Mg²⁺ implantation significantly accelerated intervertebral fusion through enhanced osteogenesis and mineralization, while inhibition of METTL3 or RhoA impaired maturation and integration, delaying fusion.

4. Discussion

Mg²⁺, an essential macromineral in the human body, has been the focus of extensive research due to its pivotal role in bone metabolism. Advances in materials science have facilitated the development of magnesium-based materials capable of regulating Mg²⁺ release [27]. Simultaneously, research efforts have sought to clarify the role of Mg²⁺ in promoting bone function, with significant progress reported in recent years [28]. The present study revealed that the addition of 4 mmol/L Mg²⁺ to conventional culture media can mimic the physiological magnesium concentration that promotes bone formation, evidenced by upregulation of key osteogenic transcription factors, such as Runx2, OCN, Col1a1, and BMP4. However, when Mg²⁺ concentration exceeds a certain threshold, this positive osteogenic regulatory effect is reversed, and cytotoxic effects begin to manifest. Specifically, excessive magnesium concentrations lead to downregulation of osteogenic marker genes, reduced osteoblast differentiation capacity, and decreased mineralization of the extracellular matrix. Although this study focuses primarily on osteoblasts, previous research has demonstrated that Mg²⁺ also inhibits osteoclast formation [29]. Thus, Mg²⁺ may exert a dual regulatory role by promoting osteoblast activity and inhibiting osteoclast formation, creating a favorable microenvironment for bone fusion.

The threshold effect observed may stem from the multi-pathway regulatory effects of Mg²⁺ on cells. The osteogenic effects of magnesium are primarily mediated by the activation of multiple signaling pathways that promote osteoblast proliferation and differentiation. These pathways are associated with mechanotransduction regulation and the role of Mg²⁺ in mediating extracellular matrix mineralization. When Mg²⁺ concentrations exceed physiological levels, their inhibitory effects on osteogenesis may involve the suppression of the Wnt/β-Catenin signaling pathway and dysregulation of oxidative stress responses [[30], [31], [32]].

Given the critical role of Mg²⁺ in numerous enzyme-catalyzed reactions, it was hypothesized that Mg²⁺ may also regulate m6A methylation modification enzymes in cellular environments. To explore this, RNA was extracted from MC3T3-E1 cells exposed to a magnesium-rich environment, and RT-qPCR was performed to assess the expression levels of potential m6A-regulated genes. The results indicated that Mg²⁺ induced a significant upregulation of METTL3 transcript levels in cells, while no significant changes were observed in other m6A regulatory genes. METTL3, the core enzyme responsible for m6A modification, showed a positive correlation between its increased expression and the previously observed rise in cellular m6A modification levels. The lack of substantial changes in other m6A-regulated genes further supports the hypothesis that Mg²⁺ modulates cellular m6A modification levels by specifically regulating METTL3.

Our findings align with Wu et al.'s pioneering work [33], which identified METTL3 as a positive regulator of osteogenesis. However, our study extends this knowledge to regenerative biomaterials and reveals a distinct mechanistic axis. While Wu et al. focused on intrinsic genetic regulation and hormonal signaling (e.g., parathyroid hormone, PTH), our study discovered that Mg²⁺—a novel extrinsic chemical signal—can upregulate METTL3 expression. This suggests that m6A modification is not only a preprogrammed developmental process but also a dynamic response to the local ionic microenvironment created by degrading implants. Additionally, while Wu et al. identified the Pth1r pathway as the primary target in their developmental model, our multi-omics screening specifically pinpointed RhoA—a master regulator of the cytoskeleton—as the key m6A target under Mg²⁺ stimulation. This distinction is significant because Mg²⁺ is known to promote osteogenesis by modulating cell stiffness and mechanotransduction. By defining the METTL3-RhoA axis, this study clarifies how chemical signals (Mg²⁺) translate into physical cellular responses (cytoskeletal reorganization) to drive intervertebral fusion.

Given the critical role of Mg²⁺ in cellular environments, it is plausible that these ions may activate multiple downstream signaling pathways, potentially confounding results by inducing m6A modifications not directly associated with osteogenesis. Furthermore, magnesium is essential for cellular metabolism, and even trace amounts in the medium of control groups have been shown to regulate various cellular processes, resulting in inconsistencies between groups. Consequently, employing the MeRIP-Seq approach to examine the m6A epigenomics of MC3T3-E1 cells under magnesium conditions may encounter challenges such as substantial intra-group variability, insufficient detection efficiency for low-abundance RNAs, and difficulty accurately quantifying m6A modification percentages at the transcript level [34]. To overcome these issues, the present study utilized m6A epigenomic transcriptomics chips for MeRIP microarray analysis, followed by bioinformatics analysis, which identified RhoA as a gene with significant m6A modification differences under magnesium deficiency.

RhoA, a key GTPase, plays a pivotal role in regulating cytoskeletal dynamics and signal transduction. It is involved in various cellular processes, including cell migration, adhesion, and bone formation [[35], [36], [37]]. Previous studies have highlighted RhoA's pivotal role in mechanical transduction during osteoblast differentiation by modulating transcription factors, such as Runx2 and Osterix, which are essential regulators of osteogenic gene expression [38]. In the present study, RhoA overexpression significantly enhanced the activity of the RhoA/ROCK pathway, leading to stronger osteogenic responses, including increased ALP activity, enhanced matrix mineralization, and upregulation of osteogenesis-related gene expression. However, the osteogenic effects induced by Mg²⁺ and METTL3 were blocked by ROCK inhibition. m6A modification-mediated RhoA expression enhances ROCK activation and downstream signaling events, which may involve the phosphorylation of MLC and LIM kinases—key mediators of cytoskeletal remodeling [39]. Furthermore, our research demonstrated that RhoA/ROCK signaling plays a critical role in regulating Runx2 expression, thereby promoting pre-osteoblast differentiation. These findings are consistent with prior research, highlighting the role of RhoA/ROCK in regulating bone formation and remodeling [35,40].

An intriguing question raised by our findings is the precise upstream mechanism by which extracellular Mg²⁺ upregulates METTL3 expression. While this study focused on downstream epitranscriptomic targets, it is hypothesized that this regulation may involve specific magnesium transporters and cofactor-dependent stability. One potential candidate is the transient receptor potential melastatin 7 (TRPM7) channel. Previous studies have shown that Mg²⁺ influx via TRPM7 can activate intracellular signaling pathways (e.g., PI3K/Akt or ERK) that, in turn, activate osteogenic transcription factors [41,42]. These signaling pathways may then bind to the promoter region of Mettl3, initiating its transcription. Additionally, as an essential divalent cation, Mg²⁺ serves as a cofactor for numerous nuclear enzymes. An increase in intracellular Mg²⁺ concentration could enhance the stability of the METTL3 protein complex or facilitate the activity of RNA polymerases responsible for its synthesis. While defining the exact Mg²⁺-TRPM7-Transcription Factor-METTL3 axis is beyond the scope of this study, it presents an exciting avenue for future research.

YTHDF1, a widely studied m6A reader protein, typically enhances the translation efficiency of m6A-modified mRNA [14]. By interacting with the translation initiation complex, YTHDF1 recruits translation factors such as eukaryotic initiation factor 3 (eIF3), promoting protein synthesis [43]. In the present study, the interaction between RhoA and YTHDF1 was site-specific, likely due to the sequence at the 3′UTR of RhoA and the spatial structure of YTHDF1. Using SRAMP, potential m6A modification sites were predicted on RhoA, which were experimentally validated. Our results demonstrated that YTHDF1 recognizes and binds to m6A-modified RhoA mRNA. This interaction promotes the efficient translation of RhoA mRNA, thereby activating the downstream RhoA/ROCK signaling pathway. Although our in vitro functional assays (knockdown and rescue experiments) firmly established YTHDF1 as the critical ‘reader’ protein bridging METTL3-mediated methylation to RhoA translation, its specific role in the animal model was inferred primarily from the upregulation of its downstream target, RhoA. We did not perform YTHDF1-specific genetic interventions (such as conditional knockout or AAV-mediated silencing) in the rat fusion model. Therefore, the conclusion regarding the necessity of YTHDF1 for Mg-induced spinal fusion is largely extrapolated from our cell culture findings. Future studies utilizing YTHDF1-deficient animal models are needed to definitively verify its in vivo contribution to this epigenetic axis.

Previous studies on magnesium-based materials that promote bone formation have predominantly utilized cranial defect and femoral nail models to validate their degradation performance and osteogenic effects [2]. However, a notable disparity in the morphology of peripheral bones and vertebrae has been observed, which limits the efficacy of Mg²⁺ in facilitating intervertebral fusion. The rat tail vertebral fusion model offers distinct advantages: the rat tail ligament surrounding the vertebrae attaches to the vertebral segments, providing the necessary stability for intervertebral fusion [44]. The movement of the rat tail relies on the traction of the tail ligament, with the fifth and sixth vertebral segments located at the base of the tail exhibiting limited mobility. To maintain a stable magnesium environment at the modeling site, it is essential to avoid fluctuations in local magnesium ion concentrations induced by postoperative magnesium ion injections and mitigate the risk of hypermagnesemia, as well as the concentration effects that magnesium-based fusion devices may experience. In line with clinical applications of magnesium-based materials, high-purity magnesium plates were selected for sustained-release implantation to provide a magnesium-rich environment at the fusion site. During model construction, a magnesium plate measuring 0.5 × 3 × 8 mm was implanted between the rat vertebral body and caudal ligament. This plate gradually degraded over two weeks while maintaining an appropriate magnesium ion concentration. Compared to the Ctrl group, implantation of magnesium plates in the Mg group led to significant induction of bone fusion and promotion of osteogenic repair. Micro-CT and tissue staining techniques revealed that the Mg group exhibited earlier formation of more mature, abundant, and highly mineralized trabeculae. These findings suggest that osteoblasts play a critical role in trabecular bone formation, consistent with the osteogenic differentiation and mineralization effects observed and validated in vitro.

This study has several limitations. First, besides osteoblasts, other cell types—including osteoclasts, immune cells, vascular cells, and neural tissues—also contribute to intervertebral fusion, and their roles need further exploration. Additionally, the potential involvement of other m6A readers, such as YTHDF2 or YTHDF3, was not investigated. Finally, while validated in a rat model, differences in biomechanics between rat and human vertebrae limit the clinical applicability of these findings, highlighting the need for studies in large animal models and human trials.

5. Conclusion

In conclusion, this study demonstrates that Mg²⁺ creates an osteogenic microenvironment by upregulating METTL3. Mechanistically, METTL3 installs m6A modifications on the 3′UTR of RhoA mRNA, which are then recognized by YTHDF1 to enhance translation efficiency, thereby activating the RhoA/ROCK signaling axis to promote cytoskeletal reorganization and osteoblastic differentiation. In vivo, Mg-based implants significantly accelerate intervertebral fusion, an effect that depends on this epigenetic axis (Fig. 8). These findings reveal a novel connection between biomaterial degradation products and epitranscriptomics. Future research should focus on elucidating the precise upstream mechanism by which intracellular Mg²⁺ triggers METTL3 expression and validating this therapeutic strategy in large weight-bearing animal models.

Fig. 8.

Magnesium ions upregulate METTL3 expression, enhancing m6A modification on RhoA mRNA. The m6A reader YTHDF1 recognizes and binds to the modified sites, promoting RhoA translation. This activates the RhoA/ROCK signaling pathway, ultimately driving osteogenic differentiation, bone remodeling, and intervertebral fusion.

Author contributions

Haocheng Xu, Linli Li, Fan Zhang and Minghao Shao: Methodology, Software, Writing - original draft. Chenyan Li, Yitong Xue, Dachuan Li, Zhidi Lin, Zhaoyang Gong and Jiongdong Wu: Methodology, Software. Zhiwen Luo, Zhicai Shi, Xinlei Xia, Hongli Wang, Xiaosheng Ma and Jianyuan Jiang: Supervision; Validation; Visualization. Xiaochuan Gu, Yang Liu, Xiao Lu and Feizhou Lyu: Funding acquisition, Investigation, Project administration. All authors have read and approved the final manuscript.

Data availability statement

The data that support the findings of this study are available in the main text or the Supplementary Materials or from the corresponding author upon reasonable request.

Declaration of generative AI in scientific writing

The authors declare that no generative AI or AI-assisted technologies were used in the writing of this manuscript.

Funding

This work was supported by National Natural Science Foundation of China, China (82502884 and 82272518); Shanghai Sailing Program, China (22YF1405100); The Youth Talent Cultivation Program for Basic Research at Huashan Hospital (2025JC012); Shanghai Education Development Foundation and Shanghai Municipal Education Commission Research and Innovation Program Major project (2023ZKZD45); Shanghai Municipal Commission of Health and Family Planning Program Excellent academic leader project (2022XD009).

Declaration of competing interest

The authors declare that they have no conflict of interest.

Abbreviations

OM

osteogenic induction medium

si

small interfering

OE

overexpression

ALP

alkaline phosphatase

ARS

alizarin red s

RT-qPCR

reverse transcription quantitative polymerase chain reaction

WB

western blotting

BSA

bovine serum albumin

ELISA

enzyme-linked immunosorbent assay

CHX

cycloheximide

CT

computed tomography

BIC

bone-implant contact

H&E

haematoxylin & eosin

DAPI

4,6-diamidino-2-phenylindole

ANOVA

one-way analysis of variance

GO

gene ontology

KEGG

Kyoto encyclopedia of genes and genomes

BV

bone volume

TV

tissue volume

BS

bone surface area

Tb.Th

trabecular thickness

Tb.Sp

trabecular separation

Tb.N

trabecular number

BMD

bone mineral density

Appendix A. Supplementary data

The following is the Supplementary data to this article.