Intelligence-responsive wettability switch coating on magnesium implants for treating osteoporotic fracture

Abstract

Osteoporotic fracture healing is frequently complicated by elevated levels of reactive oxygen species and disrupted bone homeostasis. Biodegradable magnesium alloys are promising orthopedic implants, but rapid degradation in oxidative, acidic osteoporotic microenvironments limits their clinical application. In this study, we develop a reactive oxygen species-responsive hydrogel coating of tannic acid and gelatin methacryloyl on magnesium implants via a metal-phenolic network. We show that this firmly adhering coating significantly decelerates magnesium degradation while scavenging reactive oxygen species on demand. In osteoporotic rat models, we demonstrate that coated implants effectively reduce oxidative stress and facilitate bone healing. Mechanistically, single-cell transcriptomics reveals that the coating activates the nuclear factor erythroid 2-related factor 2 signaling pathway in bone marrow mesenchymal stromal cells and bone marrow-derived macrophages, enhancing osteogenesis while inhibiting osteoclastogenesis. Consequently, this multifunctional coating provides corrosion protection and a targeted therapeutic approach to enhance osteoporotic fracture healing under oxidative stress.

Osteoporotic fractures often heal poorly due to high oxidative stress. Here, the authors develop an intelligent responsive hydrogel coating for magnesium implants that decelerates implant degradation, alleviates oxidative stress, and promotes bone regeneration.

Introduction

Osteoporosis is identified as a systemic skeletal disorder, leading to decreased bone mass, bone structure deterioration, and increased fracture susceptibility. Fractures associated with osteoporosis are a primary reason for disability and mortality in elderly individuals, thereby imposing a substantial burden on both affected individuals and society¹. In contrast to standard fractures, osteoporotic fractures present the considerable challenge of an unfavorable healing environment. This is partly due to the osteoporotic condition, which is marked by impaired osteoblast function and hyperactive osteoclast activity, resulting in a pronounced decline in bone regeneration capacity². Additionally, the persistent inflammatory response at the fracture site results in an overproduction of reactive oxygen species (ROS), creating an oxidative stress microenvironment^3–5. Increased ROS levels cause direct damage to surrounding cells, including bone marrow mesenchymal stromal cells (BMSCs) and osteoblasts, impairing their capacity for osteogenic differentiation⁶, while also continuously activating pro-inflammatory signaling pathways^7–9. This exacerbates local inflammation and establishes a vicious cycle that hinders bone healing. Consequently, the precise regulation of the oxidative stress microenvironment at the fracture site, alongside the enhancement of osteogenic capacity, represents a significant scientific challenge in advancing the healing process of osteoporotic fractures.

Surgical reduction and fixation are widely employed treatments for osteoporotic fractures. Recently, biodegradable magnesium (Mg) alloys have shown significant potential in the realm of orthopedic implants, attributed to their superior biocompatibility, elastic modulus comparable to human bone (~45 GPa), and biodegradable characteristics^10–12. The study conducted by Qin et al. demonstrated that Mg implants promoted bone fracture healing by upregulating neuronal calcitonin gene-related peptide¹³. Nevertheless, the rapid degradation rate and non-uniform degradation pattern of Mg alloys present substantial challenges to their clinical application, resulting in the premature loss of mechanical integrity of the implant, localized generation of excessive hydrogen gas, and an elevation in the pH of bodily fluids^14,15. The situation is particularly aggravated when it comes to osteoporosis. Elevated ROS levels and the localized acidic microenvironment may further exacerbate the non-uniform corrosion property of Mg implants¹⁶. The extended healing duration of osteoporotic fractures necessitates stringent control over the degradation behavior of the implant. Consequently, enhancing the uniform degradation of Mg-based bone implants within high-ROS inflammatory microenvironments is a critical challenge that must be resolved prior to their clinical application.

Recently, surface modification techniques are considered an efficient method to manage the degradation rate and pattern of Mg implants^17,18. Previous studies have explored methods such as anodizing, plasma electrolytic oxidation, and electrodeposition for preparing coatings on Mg alloy surfaces to retard matrix degradation and impart specific biological properties on the implant surface¹⁹. Nonetheless, challenges persist, including single functionalization, non-uniformity of coating, and lack of environmental response characteristics¹⁷. Compared to previous coatings, hydrogel coating has garnered significant attention due to its distinctive hydrophilic three-dimensional network structure, excellent biocompatibility, and capacity for functionalization^20,21. Hydrogel coatings can serve as a physical barrier, effectively isolating corrosive media and delaying the degradation of Mg alloys by design. More importantly, their flexible modifiability enables researchers to incorporate various functional molecules, such as antibacterial agents, growth factors, and anti-inflammatory drugs, into the coating²¹. This approach constructs a multifunctional platform that not only provides active biological functions but also protects the substrate. Nevertheless, hydrogel coatings frequently exhibit inadequate adhesion strength to Mg, which undermines their effectiveness for long-term stable protection^22,23. Consequently, specific coating design strategies are imperative to improve the adhesion between the hydrogel and Mg. Tannic acid (TA), a natural polyphenolic compound, has attracted significant attention in the biomedical field due to its potent antioxidant and anti-inflammatory properties^24,25, indicating its potential for application in the treatment of osteoporotic fractures. However, the direct application of TA is constrained by limitations such as poor stability and susceptibility to oxidation^26,27. Actually, its functionality and application value can be enhanced by forming Metal-Phenolic Networks (MPN) through coordination with metal ions²⁸ and employing hydrogel-based controlled drug delivery platforms²⁹. Compared with other common methods for enhancing coating adhesion, MPNs exhibit superior biocompatibility, a simple preparation process, and strong binding ability with both metal substrates and hydrogel layers. The MPN framework significantly improves the adhesion strength and stability of the coating while imparting the material with superior antioxidant barrier properties and controlled release capabilities. Concurrently, TA within the hydrogel can undergo oxidation to form quinone structures in environments with elevated ROS^30,31. This oxidation process not only augments the hydrophobicity of materials but also increases the cross-linking density and compactness of the coating, thereby effectively impeding the infiltration of corrosive agents and retarding the degradation of Mg alloys. This oxidation-induced structural transformation offers an approach for achieving prolonged corrosion protection of Mg alloys in the high-ROS microenvironment characteristic of osteoporosis.

Hence, this study has developed a ROS-responsive hydrogel coating, designated as TA and gelatin methacryloyl (TA-GelMA), for application on Mg surfaces. The objective is to synergistically address the dual challenges of corrosion protection and fracture healing in Mg-based implants under osteoporotic conditions. The hydrogel coating, using the MPN created mainly by the interaction of polyphenols and Mg ions, demonstrates superior adhesion strength to Mg-based implants and effectively retards the initial degradation of the Mg alloy. TA, serving as the principal functional component of the hydrogel coating, is responsive to the elevated ROS levels at fracture sites, facilitating the prompt scavenging of excess ROS and promoting the healing of osteoporotic fractures. Mechanistically, single-cell sequencing analysis has revealed that the TA-GelMA-coated Mg implants (Mg@MPN@TA-GelMA) enhance osteogenesis and inhibit osteoclast activity by activating the nuclear factor erythroid 2-related factor 2 (Nrf2) signaling pathway. As a result, the Mg@MPN@TA-GelMA exhibits potential as a promising therapeutic strategy for the treatment of osteoporotic fractures, as illustrated in Fig. 1.

Fig. 1. Schematic illustration for the application of TA-GelMA-coated Mg implants in enhancing osteoporotic fracture healing.

a The TA-GelMA-coated Mg implant was placed in the femoral region of osteoporotic fracture rats. b The TA-GelMA coating scavenged excess reactive oxygen species (ROS) and reduced Mg degradation via a quinone reaction that improved coating compactness and hydrophobicity. c TA-GelMA facilitated the upregulation of antioxidant enzyme expression via the Nrf2 signaling pathway, thereby promoting osteogenesis and inhibiting bone resorption, ultimately promoting osteoporotic fracture healing.

Results

Design, synthesis, and characterization of Mg@MPN@TA-GelMA

The methodology for preparing the TA-GelMA coating is illustrated in Fig. 2a. Initially, high-purity Mg was treated with hydrofluoric acid (HF), resulting in the formation of a black, dense MgF₂ coating. Subsequently, the MgF₂ surface was treated with TA to develop a golden-yellow MPN coating (Mg@MPN). Scanning electron microscopy (SEM) results of the hydrogel group (GelMA) displayed uniformly dense pores with an approximate diameter of 25 μm, which is favorable for in vivo osseointegration and provides a robust foundation for fracture healing. Additionally, SEM images of the TA-GelMA group revealed the presence of numerous dense drug particles (Fig. 2b). Atomic Force Microscopy (AFM) analysis revealed that the untreated Mg exhibited an average surface roughness (R_a) of 100.15 nm, which increased to 199.64 nm following HF treatment (MgF₂-coated Mg). In comparison, the R_a of the MPN coating was measured at 99.55 nm, indicating that the MPN formed by the interaction of TA and Mg ions effectively reduced the overall surface roughness of the acid-etched Mg (Fig. 2c). The R_a of the MPN was comparable to that of untreated Mg, and the roughness value remained within an acceptable range. Besides, the element mapping results (the composition of both the surface and cross-section of TA-GelMA) reveal a robust bond between the hydrogel coating and Mg surface (Fig. 2d, and Supplementary Fig. 1b). These observations suggest the successful fabrication of TA-GelMA coating and its satisfactory uniformity. The chemical bonds in MPN, GelMA, and TA-GelMA were analyzed using Fourier Transform Infrared Spectroscopy (FTIR). Both MPN and TA-GelMA showed an ether bond absorption peak at 1070 cm⁻¹, which is likely attributed to TA. Whereas this peak was absent in GelMA, indicating complete coverage of the Mg and MPN coating by GelMA (Fig. 2e). Subsequently, measurements of static water contact angles were performed to assess the hydrophilicity of different surfaces. In comparison to the Mg group (69.57 ± 1.67°), the TA-GelMA surface demonstrated a significantly reduced water contact angle (34.54 ± 3.12°), indicating enhanced surface hydrophilicity (Fig. 2f). The protective efficacy of the coating is contingent upon its strong adhesion to the metal substrate. We quantitatively evaluated the adhesion strength of GelMA and TA-GelMA coatings to Mg using lap shear testing. The adhesion strength of TA-GelMA was notably stronger than that of the GelMA group, both in the material-to-material group (For material, the adhesion strength of TA-GelMA group was 150.02 ± 5.24 kPa, while that of GelMA group was 29.58 ± 4.08 kPa.) and the material-to-tissue group (For tissue, the adhesion strength of TA-GelMA group was 48.89 ± 3.17 kPa, while that of GelMA group was 10.24 ± 1.89 kPa.), as illustrated in Fig. 2g.

Fig. 2. Design, synthesis and characterization of Mg@MPN@TA-GelMA.

a A schematic representation of the synthesis process for Mg@MPN@TA-GelMA. b SEM images of Mg, MgF₂, Mg@MPN, Mg@MPN@GelMA, and Mg@MPN@TA-GelMA, with red arrows highlighting TA powder. c AFM images of Mg, MgF₂, and Mg@MPN groups. d SEM images and elemental mapping results of Mg@MPN@TA-GelMA. e FTIR spectra of Mg@MPN, Mg@MPN@GelMA, and Mg@MPN@TA-GelMA, with a red arrow indicating the ether bond absorption peak at 1070 cm⁻¹. f Images and statistical analysis of static water contact angles (n = 6 biologically independent replicates). g A schematic diagram of the lap shear test and statistical analysis of coating adhesion strength (n = 5 biologically independent replicates). The numerical data are presented as the mean ± SD (f, g). Data were analyzed using one-way ANOVA with Tukey’s post hoc test (f) or two-sided unpaired Student’s t-test (g), and ^***P < 0.001 denoting statistical significance. Source data and exact p values are provided as a Source Data file.

In vitro degradation behavior and ROS scavenging capacity of Mg@MPN@TA-GelMA under simulated oxidative stress microenvironments

To examine the degradation of Mg implants and their interaction with ROS in osteoporotic microenvironments, in vitro experiments were conducted utilizing hydrogen peroxide (H₂O₂) to simulate oxidative stress. SEM images showed that the pore size of GelMA increased progressively after five cycles of H₂O₂ immersion. In contrast, TA-GelMA demonstrated enhanced cross-linking post-immersion, resulting in a reduced pore size and a denser coating structure (Fig. 3a). The polyphenols in TA-GelMA underwent oxidation, leading to the formation of quinones. This transformation was confirmed by FTIR, which exhibited characteristic absorption peaks in the range of 1650–1750 cm⁻¹ (Fig. 3b)³². Static water contact angle measurements revealed that H₂O₂-treated TA-GelMA exhibited significantly increased hydrophobicity (69.13 ± 2.98°) compared to the untreated control (34.54 ± 3.12°), indicating oxidation-induced surface modification (Fig. 3c). Besides, in vitro ROS scavenging experiments demonstrated the rapid antioxidant capacity of TA-GelMA, achieving 86% 1,1-diphenyl-2-picrylhydrazyl (DPPH) radical clearance within 10 min and reaching 97% by 30 min. Subsequent cyclic performance testing revealed maintained efficacy, with TA-GelMA sustaining significant radical scavenging activity through multiple H₂O₂ immersion cycles. Remarkably, long-term stability tests confirmed persistent antioxidant functionality, as the material retained robust ROS neutralization capability even after prolonged air exposure (Fig. 3d). These results demonstrate that TA-GelMA coatings on Mg implants formed a denser, more hydrophobic structure under oxidative conditions. This transformation endowed the TA-GelMA coating with potent, rapid, and remarkably durable antioxidant properties (Fig. 3e). Subsequently, immersion corrosion tests were conducted to examine the protective impact of TA-GelMA coating on Mg after 24-h immersion with 0.3 mM H₂O₂. By evaluating the concentration of Mg ions released and the volume of hydrogen gas generated during in vitro corrosion, it was verified that the TA-GelMA coating notably decreased the initial rapid corrosion rate of Mg substrate and offered long-term stable protection for the Mg implant (Fig. 3f, g). To compare the protective effect of TA-GelMA coating in normal and high ROS environments, we performed immersion experiments with and without H₂O₂, and confirmed that the protective effect was activated in high ROS environments by detecting the concentration of released Mg ions (Fig. 3h, and Supplementary Fig. 2).

Fig. 3. In vitro degradation behavior and ROS scavenging capacity of Mg@MPN@TA-GelMA under simulated oxidative stress microenvironments.

a SEM images of TA-GelMA following H₂O₂ immersion. b The FTIR spectra of TA-GelMA post-H₂O₂ immersion, with a red arrow indicating the absorption peak at 1650–1750 cm⁻¹. c Representative images and statistical analysis of the static water contact angles of TA-GelMA post-H₂O₂ immersion (n = 6 biologically independent replicates). d A statistical analysis of the DPPH radical scavenging rate in ROS clearance assays (n = 3 biologically independent replicates). e A schematic diagram illustrating the ROS scavenging mechanism of TA-GelMA under in vitro conditions. f The released concentration of Mg ion during immersion tests. g The volume of hydrogen evolution during immersion tests. h The released concentration of Mg ion during immersion tests. The numerical data are presented as the mean ± SD (c, d, f, g). Data analysis was conducted using two-sided unpaired Student’s t-test (c), or one-way ANOVA with Tukey’s post hoc test (d), and ^***P < 0.001 denoting statistical significance. Source data and exact p values are provided as a Source Data file.

In vitro osteogenic and osteoclastogenic activities

To verify the role of TA-GelMA in scavenging ROS at the cellular level in a high ROS environment, BMSCs (Fig. 4a) and bone marrow-derived macrophages (BMMs) (Fig. 4i) were treated with 50% diluted material extracts to assess the efficiency of intracellular ROS clearance (Supplementary Fig. 3). The concentration of Mg ions in Mg, MPN, GelMA, and TA-GelMA extracts was approximately 1.76 mM, 1.71 mM, 1.02 mM, and 0.71 mM, and the pH value was 6.8, 6.7, 7.2, and 7.1, respectively. Analysis of cell viability indicated that exposure to H₂O₂ diminished the viability of both BMSCs and BMMs, whereas treatment with TA-GelMA mitigated the cell damage induced by H₂O₂ (Fig. 4b, j). Fluorescence imaging employing the 2’,7’-Dichlorodihydrofluorescein diacetate (DCFH-DA) probe revealed a substantial increase in fluorescence intensity in BMSCs and BMMs following H₂O₂ treatment (Supplementary Fig. 4). However, TA-GelMA treatment resulted in a reduction of fluorescence intensity by approximately 80% in BMSCs and approximately 90% in BMMs (Fig. 4c, k). Collectively, these findings demonstrate that the TA-GelMA coating exhibits substantial antioxidant properties.

Fig. 4. In vitro osteogenic and osteoclastogenic activities of Mg@MPN@TA-GelMA.

a Detection of intracellular ROS in BMSCs is performed after 24-h culture with Mg, MPN, GelMA, or TA-GelMA in a medium containing 0.3 mM H₂O₂ to simulate a high ROS environment. b The viability of BMSCs (n = 5 biologically independent replicates). c Statistical evaluation of ROS levels within BMSCs (n = 3 biologically independent replicates). d Representative images of ALP staining following a 7-day culture period and ARS staining after a 21-day culture period of BMSCs with extracts from various material groups. e Statistical evaluation of ALP staining (n = 3 biologically independent replicates). f Statistical evaluation of ARS staining (n = 3 biologically independent replicates). g Western blot analysis of Runx2, ALP, and OCN protein expression levels. h Quantitative evaluation of Western blot data (n = 3 biologically independent replicates). i Intracellular ROS detection in BMMs is carried out after 24-h culture with Mg, MPN, GelMA, or TA-GelMA in a medium containing 0.3 mM H₂O₂ to simulate a high ROS environment. j The viability of BMMs (n = 5 biologically independent replicates). k Statistical evaluation of ROS levels within BMMs (n = 3 biologically independent replicates). l Representative images of TRAP staining and bone slice resorption assays of BMMs cultured with extracts from various material groups, with red arrows indicating osteoclasts and resorption pits. m Quantitative evaluation of osteoclast numbers (n = 6 biologically independent replicates). n Quantitative evaluation of bone resorption pit areas (n = 3 biologically independent replicates). o Western blot analysis of NFATc1 and CTSK protein expression levels. p Quantitative evaluation of Western blot data (n = 3 biologically independent replicates). The numerical data are presented as the mean ± SD (b, c, e, f, h, j, k, m, n, p). Data were analyzed using one-way ANOVA with Tukey’s post hoc test (b, c, e, f, h, j, k, m, n, p), and ^***P < 0.001 denoting statistical significance. Source data and exact p values are provided as a Source Data file.

Subsequently, the impact of material extracts on the osteogenic differentiation of BMSCs was investigated. No significant cytotoxic effects of the various material extracts were observed on BMSCs, as indicated by the cell viability assay. Alkaline phosphatase (ALP) staining analysis demonstrated more intense staining in the TA-GelMA group. Furthermore, alizarin red S (ARS) staining revealed numerous aggregated red mineralized calcium nodules in the TA-GelMA group. These findings from ALP and ARS staining confirmed that TA-GelMA enhances the transformation of BMSCs into osteogenic cells (Fig. 4d–f, and Supplementary Fig. 5a). Additionally, the levels of genes related to osteogenesis were assessed using Western blot analysis, including Runt-related transcription factor 2 (Runx2), ALP, and Osteocalcin (OCN). During the process of osteogenic differentiation, Runx2 is initially expressed, then ALP expression increases from the early to mid-stages, and OCN is highly expressed at the late stage³³. Quantitative analysis via Western blot and reverse transcription quantitative polymerase chain reaction (RT-qPCR) showed that the expression levels of Runx2, ALP, and OCN proteins and mRNA in BMSCs cultured with TA-GelMA were notably elevated compared to the control groups (Fig. 4g, h, and Supplementary Fig. 5b). The findings strongly support the superior capability of TA-GelMA to induce osteogenic differentiation in vitro. We also supplemented in vitro experiments using BMSCs isolated from rats, and the results were consistent with those obtained using mouse cells (Supplementary Fig. 6).

The excessive activation of osteoclasts serves as the primary pathological alteration in osteoporosis, leading to increased bone resorption. Consequently, we proceeded to explore the impact of TA-GelMA on BMM differentiation into osteoclasts in vitro. The cell viability results indicated that the various material extracts did not exhibit significant cytotoxic effects on BMMs. Tartrate-resistant acid phosphatase (TRAP) staining revealed that TA-GelMA prevented the development of mature osteoclasts, resulting in a 93% reduction in osteoclast count compared to the control group. Concurrently, TA-GelMA significantly suppressed the bone resorption activity of osteoclasts, decreasing the resorption area on bone slices by 80% relative to the control group (Fig. 4l–n, and Supplementary Fig. 5c). Furthermore, the expression levels of genes related to osteoclasts were evaluated through Western blot analysis. Nuclear factor of activated T cells c1 (NFATc1) is predominantly expressed during early differentiation, whereas cathepsin K (CTSK) is expressed during mid-to-late differentiation and serves as a characteristic marker of osteoclasts³⁴. Quantitative analysis via Western blot and RT-qPCR revealed that NFATc1 and CTSK expression levels in BMMs cultured with TA-GelMA were notably lower than the control groups (Fig. 4o, p, and Supplementary Fig. 5d). These findings confirm the excellent ability of TA-GelMA to inhibit osteoclast differentiation in vitro. We also supplemented in vitro experiments using BMMs isolated from rats, and the results were consistent with those obtained using mouse cells (Supplementary Fig. 6).

In vivo ROS scavenging capacity and osteogenic activity for fracture healing

To explore the protective impact of the TA-GelMA coating on Mg and its ROS-scavenging properties in vivo, an osteoporotic fracture model was successfully developed using Sprague-Dawley (SD) rats, followed by intramedullary nailing treatment (Fig. 5a, b, and Supplementary Fig. 7). Notably, significant gas production was detected around the Mg implant as early as 24 h post treatment. By 12 weeks post operation, the Mg implant exhibited clear signs of fracture (Supplementary Fig. 8). Micro-computed tomography (micro-CT) analysis of intramedullary nail degradation in vivo revealed non-uniform degradation of the implant, with pitting corrosion evident on the surface of bare scaffold, indicating that the fracture of the Mg implant may result from localized corrosion (Fig. 5c). Statistical analysis of changes in implant volume showed that the TA-GelMA coating significantly mitigated the corrosion of the intramedullary nail compared to the control group. Specifically, the corrosion volume was reduced to 25% of the control at 4 weeks, 43% at 8 weeks, and 50% at 12 weeks, thereby preventing fracture failure of the intramedullary nail (Fig. 5d). Photoacoustic (PA) imaging of local ROS at the fracture site revealed that the osteoporotic fracture group had significantly elevated ROS levels compared to the control group (Supplementary Fig. 9). Treatment with TA-GelMA resulted in a substantial reduction of ROS levels at the fracture site post-operation, particularly at 2 and 4 weeks post operation, where the PA signal intensity in the TA-GelMA group was 75% and 58% of that observed in the control group, respectively (Fig. 5e, f). Micro-CT analysis of rat femur demonstrated that TA-GelMA significantly enhanced the healing of osteoporotic fractures, with fractures in the TA-GelMA group being nearly healed by 12 weeks (Fig. 5g). Further examination of trabecular bone parameters revealed that the TA-GelMA group outperformed the control group (Fig. 5h–m). These findings confirm the efficacy of TA-GelMA in enhancing bone healing in osteoporotic fracture models.

Fig. 5. In vivo ROS scavenging capacity and osteogenic activity for fracture healing.

a A schematic diagram illustrating the development of the Sprague-Dawley rat model for osteoporotic fractures and the subsequent in vivo implantation procedure (Some elements in this figure were created by figdraw.com, under authorization ID: WAPATa4c4a). b Macroscopic images of implants from each group. c Micro-CT reconstruction images depicting in vivo implant degradation. d Statistical analysis of implant volume (n = 3 biologically independent replicates). e Representative images depicting in vivo ROS detection, with blue coloration indicating ROS levels. f Statistical analysis of in vivo ROS levels (n = 5 biologically independent replicates). g 3D reconstruction images of rat femurs. h–m Quantitative analysis of trabecular bone parameters bone mineral density (BMD), bone volume per tissue volume (BV/TV), bone surface per tissue volume (BS/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), and trabecular separation (Tb.Sp) (n = 4 biologically independent replicates). The numerical data are presented as the mean ± SD (d, f, h–m). Data were analyzed using one-way ANOVA with Tukey’s post hoc test (d, f, h–m), and ^*P < 0.05, ^**P < 0.01 and ^***P < 0.001 denoting statistical significance. Source data and exact p values are provided as a Source Data file.

In addition, histological and immunohistochemical analyzes were conducted on samples collected from each group at 4, 8 and 12 weeks following implant treatment. The results, including images of Hematoxylin and Eosin (HE), Safranin O/Fast Green (S-O), ALP, OCN, and TRAP staining of the fracture sites after various treatments, are presented in Fig. 6a. HE staining facilitates the visualization of the general structure and cellular composition of bone tissue, whereas S-O staining highlights newly formed cartilage tissue, which appears red³⁵. At all postoperative time points, the cartilage fraction in the TA-GelMA group exceeded that of other groups, with a notable 82% increase at 4 weeks post operation (Fig. 6b). Regarding the osteogenesis and osteoclastogenesis evaluation, TA-GelMA-treated osteoporotic fractures exhibited the highest expression levels of ALP and OCN across different time points (Fig. 6c, d), while osteoclast formation was reduced compared to other groups (Fig. 6e). Actually, bone regeneration is inherently linked to mineral accumulation³⁶. Consequently, to further elucidate the spatiotemporal dynamics of bone formation following TA-GelMA implantation, alizarin red and calcein were employed as fluorescent markers to monitor new bone formation and mineralization (Fig. 6f). Following the implantation of TA-GelMA in rats, a pronounced red and green fluorescence was detected surrounding the newly formed bone, suggesting that TA-GelMA significantly enhances osteogenesis during the healing of osteoporotic fractures (Fig. 6g). Quantitative analysis revealed that TA-GelMA demonstrated the highest mineral apposition rate (MAR), with a 27% increase compared to the control, thereby facilitating mineral deposition at the fracture site (Fig. 6h). Furthermore, the mechanical strength of the healed tissue was evaluated. At 8 and 12 weeks post operation, the TA-GelMA group exhibited a notably higher maximum load than the control group, with a 13% increase, indicating that TA-GelMA not only augmented the bone volume of the callus but also enhanced its mechanical strength (Fig. 6i). To further demonstrate the clinical advantages of TA-GelMA-coated Mg implants, we also supplemented micro-CT and histological staining results of the titanium (Ti) group, and found that the fracture healing effect of the TA-GelMA group was superior to that of the Ti group, confirming its potential clinical application value (Supplementary Fig. 10). Beyond that, the biosafety of the implants was assessed across all groups, and no significant organ toxicity or adverse effects on liver and kidney function were observed (Supplementary Figs. 11 and 12).

Fig. 6. Histological, immunohistochemical, and immunofluorescence analyses of in vivo osteogenic and osteoclastogenic activity.

a Representative images of HE, TRAP, S-O, ALP, and OCN staining. b Quantitative analysis of cartilage fraction based on S-O staining (n = 6 biologically independent replicates). c Quantitative analysis of ALP expression levels derived from ALP staining (n = 6 biologically independent replicates). d Quantitative analysis of OCN expression levels from OCN staining (n = 6 biologically independent replicates). e Quantitative analysis of osteoclast presence using TRAP staining (n = 6 biologically independent replicates). f Schematic illustration of the procedure for the triple fluorescent labeling experiment. g Representative images from the triple fluorescent labeling experiment, with red coloration indicating alizarin red S and green coloration indicating calcein (Some elements in this figure were created by figdraw.com, under authorization ID: WAPATa4c4a). h Quantitative analysis of the mineral apposition rate (MAR) (n = 5 biologically independent replicates). i Quantitative assessment of healing quality through four-point bending biomechanical testing, with maximum load as the primary evaluation metric (n = 3 biologically independent replicates). The numerical data are presented as the mean ± SD (b–e, h, i). Data were statistically analyzed using one-way ANOVA with Tukey’s post hoc test (b–e, h, i), and ^*P < 0.05 and ^***P < 0.001 denoting statistical significance. Source data and exact p values are provided as a Source Data file.

Single-cell sequencing analysis of TA-GelMA in regulating bone metabolism

To elucidate the mechanism by which TA-GelMA facilitates fracture healing, single-cell RNA sequencing (scRNA-seq) was carried out on bone marrow samples at 3 weeks post operation (Fig. 7a). At this stage, ROS levels at the fracture site remain elevated, osteogenesis is highly active, and osteoclasts begin to emerge within the callus, initiating the remodeling process³⁷. Consequently, this time point was selected to evaluate the equilibrium between osteogenesis and osteoclast activity during the healing of osteoporotic fractures. Cell populations were predominantly categorized into B cells, neutrophils, monocytes, NK cells, stem cells, T cells, and basophils based on the expression profiles of various marker genes (Fig. 7b). The proportion of stem cells in the TA-GelMA group (11.98%) was significantly higher than the control group (5.44%), whereas the proportion of monocytes (1.22%) was lower than the control group (4.09%) (Fig. 7c). The differentially expressed genes (DEGs) were analyzed using Gene Set Enrichment Analysis (GSEA), indicating that metabolic pathways associated with the response to oxidative stress were activated in the TA-GelMA group (Fig. 7d).

Fig. 7. Single-cell sequencing analysis on the mechanism of Mg@MPN@TA-GelMA in regulating bone metabolism.

a Flowchart illustrating the process of single-cell sequencing (Some elements in this figure were created by figdraw.com, under authorization ID: WAPATa4c4a). b t-distributed stochastic neighbor embedding (t-SNE) visualization and categorization of bone marrow cells into eight distinct subsets based on marker genes of various cell types. c Proportional composition of the different cell subsets. d Results of GSEA. e Visualized t-SNE plots of BMSCs in Mg group and TA-GelMA group. f Volcano map of differentially expressed genes (DEGs). g Heatmap of gene expression profile. h GO biological process enrichment analysis of DEGs. i KEGG pathway enrichment analysis of DEGs. j Feature plot and violin plot show Nrf2, CAT, and SOD1 relative expression level in Mg group and TA-GelMA group within the BMSCs subset. k Visualized t-SNE plots of BMMs in Mg group and TA-GelMA group. l Volcano map of DEGs. m Heatmap of gene expression profile. n GO biological process enrichment analysis of DEGs. o KEGG pathway enrichment analysis of DEGs. p Feature plot and violin plot show Nrf2, CAT, and SOD1 relative expression level in Mg group and TA-GelMA group within the BMMs subset. Source data and exact p values are provided as a Source Data file.

Following the identification and isolation of BMSCs and BMMs using marker genes, the expression of DEGs was analyzed. In both BMSCs (Fig. 7e–j) and BMMs (Fig. 7k–p) subsets, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis showed that these DEGs were primarily associated with the response to oxidative stress and Nrf2 signaling pathway. The expression of genes linked to oxidative stress was also examined, revealing an increase in Nrf2 and antioxidant genes like catalase (CAT) and superoxide dismutase 1 (SOD) (Fig. 7j, p). As a key transcription factor, Nrf2 is essential in the cellular defense against oxidative stress, enhancing osteogenic differentiation and suppressing osteoclast activity³⁸. Overall, TA-GelMA was found to activate oxidative stress-related pathways during the healing of osteoporotic fractures, thereby reducing intracellular ROS levels, promoting osteogenesis, and inhibiting osteoclastogenesis. These alterations are probably directly linked to the modulation of the Nrf2 signaling pathway.

Regulation of the Nrf2 signaling pathway by TA-GelMA

To examine the impact of TA-GelMA on the Nrf2 signaling pathway, we evaluated the expression levels of Nrf2, heme oxygenase-1 (HO-1), and NAD(P)H quinone dehydrogenase 1 (NQO1) proteins, alongside the antioxidant enzymes CAT and SOD1, using Western blot analysis. In both BMSCs and BMMs, TA-GelMA treatment resulted in a notable increase in Nrf2 nuclear translocation and an elevation in the expression of HO-1, NQO1 and downstream antioxidant enzymes CAT and SOD (Fig. 8a–d, and Supplementary Figs. 13 and 14). Moreover, component-specific comparative experiments were designed to verify the effects of each component (Mg ions, TA, GelMA) on the Nrf2 signaling pathway, and the results demonstrated that the regulation of Nrf2 was mainly driven by the synergistic effect of TA-mediated ROS scavenging and Mg ion release (Supplementary Fig. 15). We next examined Nrf2 expression at the fracture site in a rat model through immunohistochemical staining. At 4, 8, and 12 weeks post operation, TA-GelMA was observed to enhance Nrf2 expression levels (Fig. 8e, f). These results demonstrate that TA-GelMA activates the Nrf2 signaling pathway in both BMSCs and BMMs, upregulating antioxidant enzymes to counteract oxidative stress in osteoporotic fractures. This dual regulation promotes osteogenesis while suppressing osteoclastogenesis, thereby accelerating fracture healing (Fig. 8g).

Fig. 8. Regulation of the Nrf2 signaling pathway by TA-GelMA in vitro and in vivo.

a Western blot analysis of Nrf2, HO-1, NQO1, CAT and SOD proteins in BMSCs. b Statistical analysis of Western blot results in BMSCs (n = 3 biologically independent replicates). c Western blot analysis of the same proteins in BMMs. d Statistical analysis of Western blot results in BMMs (n = 3 biologically independent replicates). e Representative images of Nrf2 immunohistochemical staining. f Quantitative analysis of the Nrf2 positive ratio (n = 6 biologically independent replicates). g A schematic diagram illustrating the impact of Mg@MPN@TA-GelMA on osteogenesis and osteoclastogenesis. The numerical data are presented as the mean ± SD (b, d, f). Data analysis was performed using one-way ANOVA with Tukey’s post hoc test (b, d, f), and ^**P < 0.01 and ^***P < 0.001 denoting statistical significance. Source data and exact p values are provided as a Source Data file.

Additional experiments were conducted using cell models to clarify if TA-GelMA indeed stimulated the Nrf2 signaling pathway to promote osteogenesis and suppress osteoclastogenesis. We inhibited Nrf2 signaling pathway using Nrf2-targeting siRNA and the Nrf2 inhibitor ML385 with an optimal dose of 5 µM. TA-GelMA could no longer strongly promote osteogenesis and suppress osteoclastogenesis in the presence of siRNA or ML385 (Fig. 9, and Supplementary Fig. 16). The attenuation of the effect of TA-GelMA upon Nrf2 inhibition substantiated the conclusion that TA-GelMA promoted osteogenesis and suppressed osteoclastogenesis via the Nrf2 signaling pathway.

Fig. 9. Inhibition of Nrf2 signaling pathway blocks TA-GelMA-mediated regulation of osteogenesis and osteoclastogenesis.

a Representative images of ALP staining following a 7-day culture period and ARS staining after a 21-day culture period of BMSCs. b Statistical evaluation of ALP staining (n = 3 biologically independent replicates). c Statistical evaluation of ARS staining (n = 3 biologically independent replicates). d Western blot analysis of Runx2, ALP, and OCN protein expression levels. e Quantitative evaluation of Western blot data (n = 3 biologically independent replicates). f Representative images of TRAP staining and bone slice resorption assays of BMMs cultured with extracts from various material groups. g Quantitative evaluation of osteoclast numbers (n = 3 biologically independent replicates). h Quantitative evaluation of bone resorption pit areas (n = 3 biologically independent replicates). i Western blot analysis of NFATc1 and CTSK protein expression levels. j Quantitative evaluation of Western blot data (n = 3 biologically independent replicates). The numerical data are presented as the mean ± SD (b, c, e, g, h, j). Data were analyzed using one-way ANOVA with Tukey’s post hoc test (b, c, e, g, h, j), and ^***P < 0.001 denoting statistical significance. Source data and exact p values are provided as a Source Data file.

Discussion

Mg alloys represent an alternative class of biodegradable metal implant materials, offering the distinct advantage of eliminating the need for secondary removal surgery while enhancing osteogenesis through the release of Mg ions as main degradation products³⁹. Nonetheless, their rapid degradation rate poses a significant challenge to clinical application. This challenge is particularly acute in the treatment of osteoporotic fractures, where the slow healing process necessitates prolonged mechanical support from the implant, while the elevated levels of ROS and acidic pH in the local microenvironment may exacerbate matrix corrosion^40,41. Actually, an optimal biodegradable implant should degrade substantially only after fulfilling its mechanical role during the fracture healing period, necessitating a degradation rate that aligns with the bone healing process^42–44. This study effectively modulated the degradation of Mg alloys by developing a TA-GelMA composite coating. The TA-GelMA coating displayed uniformly dense pores with dense TA particles, and the MPN-mediated coating methodology employed in this research augments interfacial adhesion strength through the coordination between TA and Mg ions. Immersion tests in vitro and microstructural analyses in vivo demonstrated that this coating delayed the corrosion of Mg implants, which was confirmed by quantitative analysis of the released concentration of Mg ions, the volume of hydrogen evolution during immersion tests, and implant volume obtained from micro-CT results. Notably, the TA-GelMA coating demonstrates distinctive dynamic responsive characteristics. Utilizing FTIR spectroscopy, this study identified a characteristic absorption peak at 1650–1750 cm⁻¹, indicative of the oxidation of phenolic groups in TA molecules to quinone structures. This chemical transformation precipitated significant alterations in the physical properties of the coating: contact angle measurements revealed that the water contact angle of the oxidized coating increased from 35° to 68°, suggesting enhanced hydrophobicity; SEM images illustrated that the coating became denser and more uniform. This structural modification is critical for corrosion protection^45,46, as increased hydrophobicity reduces the attachment of water molecules to the coating surface, and the densified network structure effectively blocks the penetration of corrosive ions like Cl⁻. This oxidation-responsive behavior notably aligns with the fracture healing process. When ROS levels are elevated in the initial healing stages, the coating enhances protection through oxidative densification. As healing progresses and ROS levels diminish, the degradation rate of the implants increases gradually, releasing Mg ions to promote osteogenesis⁴⁷. In contrast to traditional static protective coatings, this dynamic responsive characteristic achieves an optimal balance between protective efficacy and bioactivity. Additionally, quinone structures are capable of undergoing Michael addition reactions with protein thiols or amino groups, thereby enhancing the integration between the coating and the surrounding tissue⁴⁸.

One of the primary factors contributing to impaired healing in osteoporotic fractures is the persistent oxidative stress microenvironment at the local site³. This study has verified the presence of abnormally increased ROS levels during the healing process of osteoporotic fractures using in vivo ROS PA imaging. Elevated ROS concentrations not only directly impair mitochondrial function in BMSCs, thereby inhibiting their osteogenic differentiation potential⁴⁹, but also facilitate osteoclastogenesis by activating NF-κB signaling pathways⁵⁰. The TA-GelMA coating addresses this pathological condition through a dual antioxidant mechanism: firstly, the phenolic groups in TA molecules can directly donate electrons to neutralize free radicals, with in vitro ROS clearance experiments demonstrating its ability to rapidly scavenge approximately 85% of ROS; secondly, TA can enhance the intracellular endogenous antioxidant enzyme system by activating the Nrf2 signaling pathway, as evidenced by Western blot analyzes that reveal a significant increase in HO-1, NQO1, CAT, and SOD protein expression levels. Furthermore, the localized sustained-release antioxidant strategy facilitated by TA-GelMA mitigates the adverse effects commonly associated with systemic administration and ensures a more prolonged and stable therapeutic action⁵¹.

Recent discoveries have highlighted the pivotal role of the Nrf2 signaling pathway, a central regulatory system of cellular antioxidant defense, in bone metabolism^52,53. This study utilized various methodologies, including Western blot, immunohistochemistry, and scRNA-seq, to demonstrate that TA-GelMA effectively promotes the Nrf2 nuclear translocation, thereby initiating the expression of downstream genes regulated by the antioxidant response element. In the context of osteogenesis, Nrf2 activation not only shields BMSCs from oxidative damage by reducing ROS levels but also directly influences the expression of osteogenesis-related genes. Experimental results revealed a significant upregulation of Runx2, ALP, and OCN expression, with ALP and ARS staining indicating enhanced osteogenic differentiation. Mechanistically, Nrf2 may alleviate the suppression of osteogenic differentiation by inflammatory factors through the inhibition of NF-κB activation³⁸, while concurrently promoting the expression of osteogenic genes by regulating the Wnt/β-catenin signaling pathway⁵⁴. In the context of osteoclastogenesis, the activation of Nrf2 inhibits osteoclast formation through several mechanisms. Firstly, the reduction of ROS levels attenuates the RANKL-induced activation of NF-κB and MAPK signaling pathways⁵⁵. Secondly, Nrf2 activation directly suppresses the expression of critical osteoclast transcription factors, including c-Fos and NFATc1⁵⁶, as evidenced by Western blot analysis showing downregulated expression of NFATc1 and CTSK. Thirdly, it protects osteoclast precursor cells from apoptosis induced by excessive oxidative stress by upregulating antioxidant enzymes, thereby maintaining a proper balance in osteoclast function⁵⁷. This bidirectional regulatory effect is crucial for the healing of osteoporotic fractures, as it not only facilitates bone formation but also inhibits abnormal bone resorption, thus rectifying the bone metabolic imbalance associated with osteoporosis. Compared to the systemic administration of Nrf2 agonists, the localized activation strategy provided by TA-GelMA circumvents potential side effects related to systemic Nrf2 activation^58,59, thereby enhancing clinical application safety.

Beyond the magnesium substrates investigated in this study, the TA-GelMA coating exhibits broad translational potential for modifying other clinically relevant metals, including Ti and zinc (Zn) alloys. For Ti implants, this coating confers dynamic ROS-scavenging capacity and Nrf2 activation, addressing their inherent limitation of insufficient osseointegration in oxidative stress-imbalanced microenvironments. For emerging biodegradable Zn alloys, the MPN interlayer enhances adhesion, while the TA-GelMA hydrogel modulates ion release and mitigates oxidative tissue damage. Extending beyond osteoporotic fractures, this system is well-suited for challenging physiological contexts. In dental applications, the TA component simultaneously alleviates inflammation via ROS scavenging and inhibits oral pathogens⁶⁰, reducing infection risk while promoting bone regeneration. Similarly, in infected bone defects, the ROS-scavenging ability synergizes with the antibacterial properties of TA and the physical barrier function of hydrogel to suppress biofilm formation and foster a pro-healing microenvironment. Collectively, the modular design of TA-GelMA—integrating robust adhesion, ROS regulation, and osteogenesis promotion—establishes it as a promising surface modification strategy adaptable to diverse implant materials and pathological conditions.

In conclusion, this study developed a ROS-responsive hydrogel coating (TA-GelMA) for Mg alloy implants. TA-GelMA demonstrated superior biocompatibility, adhesion, corrosion resistance, and ROS scavenging capability. In vitro and in vivo studies both proved that TA-GelMA effectively scavenges local ROS at the fracture site, increases the levels of intracellular antioxidant enzymes via the activation of the Nrf2 signaling pathway, promotes osteogenesis, and inhibits osteoclastogenesis, thereby expediting fracture healing. This innovative coating demonstrates substantial promise for applications in bone tissue engineering and holds significant potential for clinical application.

Methods

Preparation and characterization of TA-GelMA coating

High-purity Mg (>99.99%) was provided by the research group of Professor Yufeng Zheng (Peking University, Beijing, China). MgF₂ was obtained by immersing the Mg in 0.1 M HF for 24 h. A metal-phenolic network coating (Mg@MPN) was prepared on the Mg surface by immersing the acid-pickled Mg in a 2 mg/ml TA solution (Sigma Aldrich) for 3 h. GelMA hydrogel was purchased from Engineering For Life Co., Ltd (Suzhou, China). GelMA hydrogel coating (Mg@MPN@GelMA) and GelMA hydrogel coating containing 5 mg/ml TA (Mg@MPN@TA-GelMA) were prepared on the MPN surface.

The roughness of Mg, MgF₂, and MPN groups was detected by AFM (Bruker Dimension Icon, Germany). After the hydrogel-containing samples were freeze-dried, the surface morphology was examined using SEM (ZEISS Sigma 300, Germany), and EDS mapping was utilized to assess the elemental composition of the sample surface and cross-section. Besides, FTIR (Thermo Fisher Scientific Nicolet iS20, USA) was performed to assess structural changes in the samples. The hydrophilicity of different surfaces was examined using static water contact angle measurements. Afterwards, lap shear tests were utilized to determine the adhesion of the hydrogel coating to both Mg material and skin tissue.

ROS scavenging assay

Prepared samples were immersed in a 1 mM H₂O₂ solution, with each 24-h period constituting one cycle, and the H₂O₂ solution was replaced with fresh solution for each cycle. SEM and FTIR were utilized to examine the morphological characteristics and structural transformations of the coating after oxidation. The antioxidant capacity of the materials was assessed through the DPPH radical scavenging assay. The absorption peak of DPPH is notably at 517 nm; when it reacts with radical scavengers, the color transitions from purple to yellow, and the absorbance at this peak diminishes. To each sample, 200 μL of DPPH solution in anhydrous ethanol was added. After reacting in the dark for 10, 20, and 30 min, the absorbance of the supernatant at 517 nm was measured. For the cyclic antioxidant performance test, the surface of each sample was treated with 200 μL of DPPH solution, and the supernatant was aspirated to measure absorbance after reacting in the dark for 30 min. The samples underwent washing with deionized water prior to each subsequent cycle, and this process was repeated for 5 cycles. To test the long-term antioxidant performance of the materials, samples from each group were exposed to air for 7, 15, and 30 days, and then the supernatant absorbance was measured following the steps described above.

To calculate the DPPH scavenging rate, the following formula was applied:

$$\text{Scavenging Rate}=\left(1-\frac{{\text{OD}}_{\text{sample}}-{\text{OD}}_{\text{blank}}}{{\text{OD}}_{\text{control}}}\right)\times 100\%.$$ 1

OD_sample is the absorbance of the DPPH solution incubated with samples, OD_blank is the absorbance of the sample incubated with anhydrous ethanol, and OD_control is the absorbance of the unreacted DPPH ethanol solution.

Immersion test

To simulate Mg ion release in the physiological environment, samples were sterilized using 29 kGy of ⁶⁰Co γ-ray irradiation and then aseptically immersed in α-MEM medium maintained at 37 °C within an incubator supplemented with 5% CO₂. The Mg ion concentration was determined using ICP-AES measurements. To assess the corrosion rate of the materials, samples from each group were immersed in Hank’s solution at 37 °C, with a solution volume to sample surface area ratio of 60 ml/cm². The solution was changed every 48 h, and daily measurements were taken of the hydrogen gas volume released.

Preparation of sample extracts

Samples were sterilized using 29 kGy of ⁶⁰Co γ-ray irradiation and then immersed in α-MEM medium, incubated at 37 °C for 72 h, with a sample surface area to immersion medium volume ratio of 1.25 cm²/mL. The collected extract medium was stored at 4 °C for later applications in cell culture.

Cell culture

BMSCs and BMMs were extracted from C57BL/6 mice. Briefly, under sterile conditions, bilateral femurs were isolated from mice, thoroughly removing attached soft tissues. The ends of the femurs were cut with scissors, and the medullary cavity was repeatedly flushed with culture medium using a syringe. The gathered mixture was subsequently filtered through a mesh and cultured in culture dishes. After 16–20 h, non-adherent cells were gathered from the supernatant and grown in α-MEM medium with 50 ng/ml M-CSF to obtain BMMs. Adherent cells continued to be cultured in α-MEM medium to obtain BMSCs. A high ROS environment was simulated in vitro using medium supplemented with 0.3 mM H₂O₂, with an intervention time of 24 h.

Cell viability assay

Cell viability was evaluated through the CCK-8 method (Beyotime, China). BMSCs and BMMs were placed in 96-well plates with a concentration of 5 × 10³ cells per well. After treatment with medium containing 0.3 mM H₂O₂, ML385 (S8790, Selleck, China) or various material extracts, the cells underwent further incubation with a 10% CCK-8 solution for a duration of 1–2 h. The absorbance at 450 nm (OD₄₅₀) was measured to assess cell activity.

Intracellular ROS level measurement

Intracellular ROS levels were detected using DCFH-DA (Beyotime, China). BMSCs and BMMs were placed in 24-well plates. After treatment for 24 h with medium containing 0.3 mM H₂O₂ or various material extracts, the culture medium was removed, and DCFH-DA working solution diluted 1:1000 was added and incubated for 30 min. Cells were examined and imaged using a laser scanning confocal microscope (Leica, Germany).

Nrf2 siRNA cell transfection

Cells were seeded in 6-well plates at a density of 1 × 10⁵ cells per well. When cellular confluence reached 50–70%, transfection with negative control siRNA (si-Ctrl) and Nrf2-targeting siRNA (Sangon Biotech, Shanghai, China) was performed using Lipofectamine® 3000 (L3000015; Invitrogen, Carlsbad, CA, USA). Subsequent to transfection, the cells were subjected to osteogenesis and osteoclastogenesis.

ALP and ARS staining

BMSCs were isolated and cultured as described above. Cells were planted in 6-well plates. When the cells reached 70% confluence, the medium was changed to an osteogenic induction differentiation medium. After 7 days of culture, BMSCs were fixed with 4% paraformaldehyde and stained using an ALP staining kit (Beyotime, China). After 21 days of osteogenic induction differentiation, BMSCs were fixed with 4% paraformaldehyde and stained using an Alizarin Red S staining kit (Solarbio, China). Microscopic evaluation and image acquisition were subsequently performed, followed by quantitative analysis via ImageJ (NIH, Maryland, USA).

TRAP staining

BMMs were placed in 96-well plates, and the medium was replaced with osteoclast induction medium after 24 h of culture. After 5–7 days of induction, multinucleated mature osteoclasts could be observed under the microscope. Cells were stained with a TRAP staining kit (Takara, Japan) and then examined under a microscope.

Bone resorption pit assay

BMMs were cultured on bone slices and induced for 5–7 days. When osteoclast formation was observed in the wells, the medium was switched to medium containing material extract, and the culture proceeded for an additional 2–3 days. Following the complete removal of cells, the slices were rinsed and dried. The size of the bone resorption pits was detected using SEM.

Western blot

After various treatments, total protein or nuclear protein was extracted from cells. Protein lysates were prepared with loading dye, resolved via SDS-PAGE, and blotted onto PVDF membranes. After blocking, the membranes were incubated overnight at 4 °C with primary antibodies targeting Runx2 (CST, 12556S, 1:1000), ALP (Abcam, AB229126, 1:1000), OCN (proteintech, 16157-1-AP, 1:1000), NFATc1 (abclonal, A19597, 1:1000), CTSK (proteintech, 11239-1-AP, 1:1000), Nrf2 (Abclonal, A0674, 1:1000), HO-1 (proteintech, 10701-1-AP, 1:1000), NQO1 (proteintech, 67240-1-Ig, 1:5000), CAT (proteintech, 21260-1-AP, 1:2000), SOD1 (proteintech, 10269-1-AP, 1:5000), Lamin B1 (proteintech, 12987-1-AP, 1:5000), and β-actin (CST, 4970, 1:1000). After TBST washing, the membranes were exposed to secondary antibodies (Beyotime, China; 1:1000) at room temperature for 1 h, followed by detection of protein bands.

RT-qPCR

We isolated total RNA with the EZBioscience extraction kit (USA) and synthesized cDNA utilizing Takara’s PrimeScript RT Master Mix. RT-qPCR was carried out with a SYBR Premix ExTaq kit (Takara) on a QuantStudio 5 Real-Time PCR System. The primer sequences used for RT-qPCR are detailed in Supplementary Table 1.

Establishment of animal models

All animal surgical procedures were performed in accordance with protocols approved by the Animal Ethics Committee of Renji Hospital, School of Medicine, Shanghai Jiao Tong University, with the corresponding approval number: 2025-819. We purchased 8-week-old female Sprague-Dawley (SD) rats and 6-week-old male C57BL/6 mice from Charles River (Beijing, China) and housed all animals in the animal facility of Shanghai Jiao Tong University (Shanghai, China) (25 ± 2 °C, 60 ± 10% humidity, continuous air circulation, and a standard 12-h light/dark schedule). Rat osteoporosis models were established by ovariectomy, and osteoporotic fracture models were established by femoral shaft fracture and intramedullary nailing implantation, according to methods reported in previous studies⁶¹. Briefly, female SD rats were anesthetized, and bilateral ovaries were removed via dorsal incision. After rearing for 3 months, the success of the osteoporosis model was assessed by micro-CT and section staining. The osteoporotic rats were then anesthetized, a fracture gap was created in the mid-shaft of the femur using a 0.3 mm diameter wire saw, and implants (Mg, MPN-coated Mg, GelMA-coated Mg, and TA-GelMA-coated Mg) were inserted into the medullary cavity via the patellar groove. All experimental procedures were performed under sterile conditions. At designated time points post-operation, rats were euthanized, and femurs, blood, visceral organs, and other tissues were collected for subsequent evaluation.

Micro-CT analysis

After fixing rat femurs with 4% paraformaldehyde, micro-CT scanning was performed using a Bruker SkyScan1276 system (Belgium) with the following parameters: scan thickness of 18 μm, voltage of 79 kV, and current of 125 μA. Data were analyzed using CTAn software to measure bone mineral density (BMD, g/cm³), bone volume per tissue volume (BV/TV, %), bone surface per tissue volume (BS/TV, 1/mm), trabecular number (Tb.N, 1/mm), trabecular thickness (Tb.Th, mm), and trabecular separation (Tb.Sp, mm). Femurs and implants were reconstructed in three dimensions to visually assess implant degradation and fracture healing.

In vivo ROS detection

This study used PA imaging technology with a liposomal nanoprobe to monitor ROS at the fracture site. The probe was provided by Soochow University (Suzhou, China) and was prepared according to methods established in previous studies⁶². Briefly, a liposomal nanocarrier containing both HRP and its substrate was designed for detecting ROS in vivo through PA imaging. At different time points post-fracture surgery, 50 μL of liposomal nanoprobe was injected into the fracture site. After 15 min, the area of the injection was covered with a layer of ultrasound gel, and ROS levels were detected at the fracture site using an ultrasound detector (FUJIFILM VisualSonics Inc.).

Histological analysis

After fixing rat femurs with 4% paraformaldehyde, they were decalcified in 10% ethylenediaminetetraacetic acid (EDTA) for 4 weeks. Following paraffin embedding, the femoral samples were cut into sections of 4 μm thickness and subjected to HE, S-O, and TRAP staining. Subsequently, immunohistochemical staining was conducted with the aid of primary antibodies for ALP (Abcam, ab224335, 1:200), OCN (Abcam, ab198228, 1:200), Nrf2 (Abclonal, A0674, 1:100) and corresponding secondary antibodies (Beyotime, 1:50).

Triple fluorescent labeling

A triple sequential fluorescent labeling technique was employed to identify newly formed bone tissue and mineralization processes. Oxytetracycline hydrochloride (25 mg/kg), alizarin red S (30 mg/kg), and calcein (20 mg/kg) were injected intraperitoneally into rats at 1, 4, and 8 weeks post operation. One week after the calcein injection (week 9), animals were euthanized, and samples were collected. After preparing hard tissue sections, observations and photography were performed under a fluorescence microscope.

Biomechanical testing

To assess fracture healing quality, a four-point bending test was performed using a Zwick Z020 tester (Zwick/Roell, Ulm, Germany). The ends of the femurs were cut, and the implants were removed. The femurs were positioned on two supporting bars set 26 mm apart, while two upper loading bars were arranged 10 mm apart, with the aim of maintaining the fracture callus at the central position. The compression test was conducted at a rate of 5 mm/min. The maximum load was determined from the load-deformation curve.

Specimen harvest for bulk RNA-seq and single cell RNA-seq

The osteoporotic fracture model in mice was established as described above. Bone marrow cells were harvested as described previously⁶³. Briefly, both femurs were excised from euthanized mice, and implants were carefully extracted using sterile forceps. The bone marrow cavity and implant surfaces were gently lavaged with phosphate-buffered saline (PBS) to collect the cell suspension, and the suspension was passed through a 40 μm cell strainer to obtain a single-cell suspension. To eliminate red blood cells, the cell pellet was resuspended in 1× red blood cell lysis buffer and incubated on ice for 5 min. Cell viability was determined by Trypan blue exclusion assay and counted using a hemocytometer.

Single cells from the bone marrow surrounding the implants were processed using the 10X Genomics Chromium Single-Cell Platform according to the manufacturer’s instructions, encapsulated into gel bead-in-emulsions (GEMs) along with gel beads coated with unique molecular barcodes. For single-cell RNA sequencing (scRNA-seq) library preparation, reverse transcription was primed using a mix of Illumina R1 sequence, a 16-nucleotide (nt) 10X Barcode, a 10-nt Unique Molecular Identifier (UMI), and a poly-dT primer; incubation of the GEMs generated barcoded, full-length cDNA from polyadenylated mRNA, which was then amplified by polymerase chain reaction (PCR) to obtain sufficient material for library construction.

To produce gene-cell count matrices, we routed the raw sequencing outputs through the 10X Genomics Cell Ranger analytical pipeline, with low-quality cells (e.g., those with fewer than 200 detected genes or a high proportion of mitochondrial reads) filtered out. Normalization of gene expression measurements for each cell was performed using the “LogNormalize” method, which normalizes by total expression, multiplies by a default scale factor of 10,000, and applies a log transformation; further data scaling was conducted using the ScaleData function in Seurat (v4.0) to standardize the expression of each gene across cells (mean = 0, variance = 1) and regress out unwanted variation (e.g., number of detected genes, mitochondrial gene expression). Differential expression analysis was performed on raw count data using the edgeR package (R version 4.1) with thresholds of false discovery rate (FDR) < 0.05 and log2 fold change (logFC) > 0.25 to identify DEGs.

Functional enrichment analysis of DEGs was conducted using R packages: GO enrichment analysis with clusterProfiler mapped DEGs to terms in the GO.db database (version 3.8.2), calculating gene counts per term and identifying significantly enriched terms via hypergeometric test (FDR < 0.05). KEGG pathway enrichment analysis utilized the KEGG database (Release 94) to screen metabolic or signal transduction pathways significantly enriched in DEGs compared to the whole-genome background, with pathways meeting FDR < 0.05 considered significant.

Statistical analysis

Data are expressed as means accompanied by their standard deviations. All experimental data were analyzed utilizing GraphPad Prism software version 9.5.0 (San Diego, USA). For statistical evaluations, we applied either a one-way ANOVA (followed by Tukey’s post hoc) or an unpaired, two-tailed Student’s t-test, depending on the experimental layout. All experiments were biologically replicated (n ≥ 3).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

G.Q., T.M., and Y.W. were instrumental in conceptualizing the experimental designs, executing the majority of the experiments, and drafting the manuscript. Y.Y. contributed to the preparation of histology sections and performed animal surgeries. Z.L. helped with specimen preparation and analysis. N.D. and C.T. provided suggestions for the project. B.L. helped with experimental designs and helped with manuscript revisions. K.C., Y.Z., and Z.W. played a pivotal role in developing the overarching concept, supervising the project, designing the experiments, and authoring the majority of the manuscript.

Peer review

Peer review information

Nature Communications thanks Carolyn Shasha, Jiacan Su and the other anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data are presented within the main text or the supplementary materials. The RNA-seq data generated in this study have been deposited in the NCBI Sequence Read Archive database under accession number “PRJNA1378706 [https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1378706]”. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-72683-3.