Dual-action bisphosphonate hydrogel preserves cortical bone integrity and restores healing cascade in osteoporotic fracture-related infection

Abstract

In osteoporotic fracture-related infection (FRI), the coordinated bone healing is compromised by cortical bone lysis and persistent bacterial colonization. Current treatments primarily focus on infection control, but fail to preserve bone microarchitecture or restore disrupted healing cascade, particularly in fragile osteoporotic bone. Here, we develop a bioactive nanocomposite hydrogel based on bisphosphonate-modified hyaluronan (HABP) and either free bisphosphonate (BP) or pyrophosphate (TSPP), carrying antibacterial Ag⁺ ions and Vancomycin. Triggered by the pathological pH of biofilms, this hydrogel enables site-specific release of antimicrobials. In a rat osteoporotic FRI model, prominent cortical bone lysis and osteocyte apoptosis impair healing. HABP hydrogel effectively eradicates the Methicillin-resistant Staphylococcus aureus infection, preserves cortical bone integrity, and rescues osteocyte viability by inhibiting osteoclast activation and osteocyte apoptosis. The TSPP-based formulation, delivering a lower BP dose, enhances callus remodeling. Our study demonstrates the promising potential of the dual-action hydrogel for treating osteoporotic FRI via preventing bone lysis and rejuvenating coordinated healing cascades.

In osteoporotic fracture-related infection (FRI), coordinated bone healing is compromised by cortical bone lysis and persistent bacterial colonization. Here, Li et al. report a dual-action nanocomposite bisphosphonates hydrogel for treatment of osteoporotic FRI that preserves cortical bone integrity and restores fracture healing.

Introduction

Bone fractures caused by musculoskeletal trauma can be life-threatening events. While modern advancements in orthopedic fixation techniques and medical devices have significantly facilitated bone repair, one persisting complication, fracture-related infection (FRI), remains unresolved¹. Normally, the healing of long bone fracture is a well-coordinated biological process, including the inflammatory phase, endochondral bone formation, and callus remodeling². However, bacterial colonization at the fracture site generates a chronic inflammatory environment, triggering prominent bone lysis that destabilizes the fracture site and disrupts the natural healing process³. The clinical outcomes of FRI are further worsened in elderly patients with comorbidities such as osteoporosis, in which the cortical microarchitecture is more vulnerable, and the bone’s ability to regenerate is severely diminished^3,4. This intersection of infection and osteoporosis not only worsens fracture instability and delays healing but also increases the risk of recurrent infection and mortality⁵. Despite these risks, the majority of current therapies for FRI are designed for otherwise healthy bone and fail to address the unique pathological synergy between infection and osteoporosis^6,7.

A central pathological feature of FRI, particularly in osteoporotic bone, is cortical bone lysis, driven by excessive osteoclast activation and apoptosis of osteocytes⁸. It is known that the overactivated osteoclasts (OCs) are directly responsible for bone lysis and microarchitecture damage⁹, while osteocytes, the most abundant and major resident cells in cortical bone, serve as key regulators of bone remodeling¹⁰. Recent evidence suggests that osteocyte apoptosis, especially in inflammatory conditions like FRI, not only compromises bone integrity but also exacerbates inflammation and facilitates bacterial invasion through empty lacunae¹¹. Yet, few therapeutic approaches have directly targeted both cell types in the context of osteoporotic FRI, underscoring a major gap in current treatment strategies⁸.

The current standard of care for FRI is to maintain stability and allow for bone healing through debridement, antibiotics, and implant retention (DAIR). However, systemic antibiotics often fail to eradicate bacteria completely, causing a high DAIR failure rate ranging from 21.4% to 43%^12,13. Moreover, treatments aimed at enhancing bone formation (e.g., administration of BMP-2) lack specificity for infected and osteoporotic environments and do not prevent pathological bone lysis^14,15. Therefore, an integrated therapy that addresses both microbial eradication and bone catabolism is urgently needed, particularly in osteoporotic FRI where bone fragility and immune dysregulation coexist.

Bisphosphonates (BP) are carbon-substituted analogs of pyrophosphates widely used to treat osteoporosis, bone metastasis, and Paget’s disease^16,17. However, systemic application of BP is limited by poor bioavailability and gastrointestinal irritation¹⁸. BP inhibits OC maturation and induces OC apoptosis, thereby leading to a reduction in bone resorption. Notably, BP has been shown to reduce the porosity of cortical bone and promote bone mineralization in osteoporotic models¹⁹, and can also inhibit osteolysis and protect osteocytes from TNF-α-induced apoptosis²⁰. These combined functions suggest the potential of BP in treating osteoporotic FRI. Besides, given the unique chemical structure of BP, comprising two phosphate groups bound to a central carbon atom, they can form dynamic coordination bonds with various bioactive cations, enabling the creation of dynamic hydrogels²¹. Our previous work has demonstrated a nanocomposite hydrogel that is capable of releasing BP, inhibiting osteoclast maturation, and facilitating bone regeneration²². In addition, BP-based hydrogel can also be engineered to load drugs and antimicrobial agents, offering the dual benefit of infection control and bone healing augmentation²³. These features highlight the potential of BP-based hydrogels as a targeted therapy for osteoporotic FRI.

Here, we develop a BP-based bioactive nanocomposite hydrogel (HABP-BP) co-loaded with Vancomycin and Ag⁺ for the treatment of osteoporotic FRI. This hydrogel exploits the acidic microenvironment at infection sites to specifically trigger the release of antimicrobials, while the released BPs inhibit osteoclast activity and prevent osteocyte apoptosis, thereby preserving cortical bone integrity. Furthermore, using a rat model with osteoporotic FRI, we identify that the HBAP-BP-V-Ag⁺ hydrogel eradicates infection, prevents cortical bone lysis and osteocyte apoptosis, and consequently re-establishes the bone healing course through callus formation and mineralization. Interestingly, we demonstrate that the hydrogel with a high dosage of BP may delay fracture callus remodeling, which is ameliorated by using a modified formulation (HABP-TSPP-V-Ag⁺) with reduced BP concentration while maintaining therapeutic efficacy (Fig. 1). This refined system offers a tunable platform that balances infection control and bone regeneration, tailored for complex osteoporotic environments. Our findings suggest that BP-based bioactive nanocomposite hydrogels provide a disease-specific, dual-action therapy that directly addresses the overlooked intersection of osteoporosis and FRI. By targeting both pathogenic bacteria and bone-destructive mechanisms, it presents a comprehensive and translationally relevant therapy for infected fractures in osteoporotic patients.

Fig. 1. An illustration of the effects of nanocomposite bisphosphonate-based antibacterial hydrogel in ameliorating cortical bone lysis and facilitating bone healing in osteoporotic FRI.

A The osteoporotic FRI presents with cortical bone lysis and bacteria invasion of osteocyte-lacuno-canalicular network. The obtained injectable hydrogel was applied to the fracture site. B The acidic infection environment triggered the disassembly of the BP-Ca²⁺-Ag⁺ nanoparticles and the release of Ag⁺. Cortical bone integrity was preserved through inhibiting osteoclast activation and osteocyte apoptosis, and cortical bone bacterial invasion was prevented. C The HABP hydrogel treatment restores the cortical bone integrity and enables coordinated healing cascades.

Results

The BP-based bioactive hydrogels mediate pathological pH-specific release of antimicrobials

Leveraging the dynamic coordination between BP and bioactive cations, we synthesized HABP macromers (Supplementary Fig. 1) and fabricated the self-assembled nanocomposite hydrogel by simply mixing HABP, free BP, and Ca²⁺ (denoted as “HABP-BP”). To enhance its antibacterial activity, we further incorporated Ag⁺ into the hydrogel using the same coordination strategy, yielding the “HABP-BP-Ag⁺” formulation (Fig. 2A). To tailor the localized BP concentration, pyrophosphate (TSPP), a BP analog that shares a similar chemical structure but lacks anti-resorptive bioactivity, was used as a substitute for free BP, generating the “HABP-TSPP” and “HABP-TSPP-Ag⁺” hydrogels. Rheological time sweep tests showed that all hydrogels exhibited similar storage and loss moduli, ranging from 600 to 700 Pa, indicating that the inclusion of different components did not compromise their structural integrity (Fig. 2B and Supplementary Fig. 2). Alternating high (400%) and low (0.1%) shear strain rheological experiments demonstrated the self-healing properties of the HABP and HABP-BP-Ag⁺ hydrogels, which quickly reformed their structure after shear disruption. In contrast, the HABP-TSPP and HABP-TSPP-Ag⁺ groups showed slightly diminished self-healing capacity (Fig. 2C and Supplementary Fig. 3), reflecting the superior coordination dynamics of BP compared to TSPP, which enhances the remodeling and self-healing capacity of the hydrogels. Therefore, the HABP-BP-Ag⁺ hydrogel demonstrated robust injectability, successfully passing through a needle with an inner diameter of 0.45 μm and forming stable, moldable structures upon extrusion, making it suitable to fit the irregular bone defects (Supplementary Movie 1). The hydrogel fabrication process entails a straightforward mixing of precursor solutions and vortexing, resulting in a stable hydrogel formation within 10 seconds. Subsequent syringe-based injection enables direct delivery to the injury site, offering substantial advantages for clinical translation (Supplementary Fig. 4). SEM images revealed uniform porous structures and well-distributed BP/TSPP-Ag⁺/Ca²⁺ nanoparticles in all hydrogel formulations, confirming their stability and consistent morphology (Fig. 2D). Next, to assess the release profiles of antimicrobial agents, Vancomycin was incorporated into the hydrogels (“HABP-BP-V-Ag⁺” and “HABP-TSPP-V-Ag⁺”). Both hydrogels exhibited an initial burst release of Vancomycin and Ag⁺ over the first 3 days, followed by a sustained release over 14 days. ¹H NMR analysis showed chemical shift changes in vancomycin’s α-carbon hydrogens (amide, hydroxyl, carboxyl), suggesting hydrogen bonding with BP phosphates, which may account for the sustained release of Vancomycin (Supplementary Fig. 6). Notably, the release of both Vancomycin and Ag⁺ can be significantly accelerated under acidic environment (pH ~5.5), simulating the infected bone microenvironment (Fig. 2E–G and Supplementary Fig. 5). This pH-responsive release can be attributed to the protonation of BPs, which weakens the coordination bonds between BP and cations, leading to the disassembly of BP/TSPP-Ag⁺/Ca²⁺ nanoparticles and the subsequent release of antimicrobial agents. Besides, the HABP-BP-V-Ag⁺ hydrogel also showed a significantly higher cumulative release of BP compared to the HABP-TSPP-V-Ag⁺ group, as the latter formulation lacks free BP and can only provide BP through the degradation of the HABP macromers.

Fig. 2. Pathological pH-specific release of the antimicrobials from the bisphosphonate-based bioactive hydrogels.

A Schematic illustration of the hydrogel formulation based on the coordination between HABP and self-assembled BP-Ag⁺/Ca²⁺ nanoparticles. B Rheological analysis of the HABP nanocomposite hydrogels. C Rheological data showing the shear-thinning and self-healing behaviors of the HABP-BP-Ag⁺ hydrogel under alternating high (400%) and low (0.1%) shear strains (representative flow curves of the rheological behavior following 3 independent replicates). D SEM images of the HABP-BP, HABP-BP-Ag⁺, HABP-TSPP, and HABP-TSPP-Ag⁺ hydrogels. E–G Cumulative release profiles of Ag⁺, Vancomycin, and BP from the HABP-BP-V-Ag⁺ hydrogel at physiological (pH 7.4) and pathological (pH 5.5) conditions. Data are presented as mean ± standard deviations (SD). (n = 3 biologically independent samples/group), **p < 0.0001, two-way ANOVA, group effects.

The BP-based bioactive hydrogels inhibit osteoclast differentiation and osteocyte apoptosis in vitro

To investigate the biological effects of the BP-based hydrogels on bone cellular functions, we first assessed their influence on the osteogenic differentiation of bone mesenchymal stem cells (BMSCs) in vitro. BMSCs were cultured with an osteogenic induction medium supplemented with 10% hydrogel extracts, in which the concentration of BP (9.8 μM) exhibited no significant cytotoxicity (Supplementary Fig. 7). The alkaline phosphatase (ALP) staining and ALP activity assays revealed comparable ALP activity in HABP-BP-V-Ag⁺ or HABP-TSPP-V-Ag⁺ hydrogel-treated groups. (Fig. 3A, B). Additionally, HABP-BP, HABP-TSPP, HABP-BP-V-Ag⁺, and HABP-TSPP-V-Ag⁺ hydrogels all exhibit higher mineralization capacity compared to the control group, which was accompanied by upregulated expression of Col1a1 and Bglap (Fig. S8). In contrast, TSPP showed no detectable effect on the mineralization of BMSCs, as confirmed by Alizarin Red S staining (Supplementary Fig. 9). Next, we evaluated the impact of hydrogel extracts on OC differentiation in primary bone marrow monocytes (BMM). The tartrate-resistant acid phosphatase (TRAP) staining showed that the cells exposed to the HABP-BP or HABP-BP-V-Ag⁺ hydrogel extracts had a significantly reduced number of multinuclear (>3) OCs compared to untreated controls. Consistently, phalloidin and DAPI staining revealed a significant reduction in actin ring formation in the HABP-BP and HABP-BP-V-Ag⁺ groups (Fig. 3A, C). These reductions was accompanied by the reduced expression of osteoclast-specific genes, including Ctsk, Nfatc1, and C-fos, particularly in the HABP-BP-V-Ag⁺ group, where the addition of Ag⁺ further inhibited OC activity (Fig. 3D)²⁴. In contrast, the HABP-TSPP-based hydrogels, which lack free BP, demonstrated less pronounced inhibition of OC differentiation, suggesting that BP release plays a critical role in this effect. However, when Ag⁺ was incorporated into the HABP-TSPP-based hydrogels (HABP-TSPP-V-Ag⁺), the inhibitory effects on OC differentiation were enhanced, underscoring the combined anti-resorptive effects of BP and Ag⁺.

Fig. 3. The BP-based hydrogels allow for osteogenesis while inhibiting osteoclastogenesis and osteocyte apoptosis.

A Representative images of ALP staining (BMSCs), TRAP staining (OCs), and F-actin staining (OCs) after hydrogel extract treatment. B Quantitative analysis of ALP activity in BMSCs treated with hydrogel extracts. Data are presented as mean ± SD (n = 3 biologically independent samples/group). C Numbers of the multinucleated OCs and the F-actin rings in OCs treated with different hydrogel extracts. Data are presented as mean ± SD (n = 4 biologically independent samples/group). D The gene expression levels of osteoclast differentiation markers (C-fos, Ctsk, and Nfatc1). Data are presented as mean ± SD (n = 3 biologically independent samples/group). E, F Flow cytometry analysis and quantification of TNF-α-induced apoptosis in MLO-Y4 osteocytes treated with hydrogel extracts. Data are presented as mean ± SD (n = 3 biologically independent samples/group). *p < 0.05, **p < 0.0001, one-way ANOVA, multiple comparisons with Tukey correction.

As the dominant cells in cortical bone, osteocytes play a key role in orchestrating bone homeostasis and maintaining the integrity of the cortical structure²⁵. Osteoporotic FRI is often associated with elevated inflammatory cytokines such as TNF-α⁸, which can induce osteocyte apoptosis and compromise cortical bone integrity²⁶. To simulate this condition in vitro, we treated MLO-Y4 osteocytes with TNF-α and evaluated apoptosis levels in the presence of hydrogel extracts. Flow cytometry analysis showed significantly elevated apoptosis induced by TNF-α, which was attenuated by treatment with HABP-based hydrogel extracts (Fig. 3E, F). The MLO-Y4 cells were further treated with BP or TSPP alone, confirming that this protective effect was mediated by BP rather than TSPP, as only BP treatment reversed TNF-α-induced changes in apoptotic gene expression, including upregulated Caspase-3 and downregulated Bcl-2 (Supplementary Fig. 7B, C). This protective mechanism is consistent with previous reports indicating that BP can inhibit osteoblast and osteocyte apoptosis through the opening of CX43 hemichannels and downstream Src/ERK signaling pathways²⁷. Collectively, these findings demonstrate that both HABP-BP-V-Ag⁺ and HABP-TSPP-V-Ag⁺ can suppress OC maturation and resorptive activity while protecting osteocytes from TNF-α-induced apoptosis. Importantly, these hydrogels support osteogenic differentiation of BMSCs, highlighting their potential to mitigate cortical bone lysis and restore bone homeostasis in osteoporotic FRI.

The BP-based bioactive hydrogels exhibit bactericidal effects and inhibit biofilm formation

To examine the antimicrobial efficacy of the BP-based hydrogels in vitro, we performed a disc diffusion test against methicillin-resistant Staphylococcus aureus (MRSA). The HABP-BP-Ag⁺ hydrogel exhibited a significantly larger inhibition zone compared to the HABP-BP-V hydrogel, confirming the enhanced antimicrobial capacity with the addition of Ag⁺(Fig. 4A, D). Given the propensity of bacteria to form biofilms on implants during FRI²⁸, we next investigated the ability of the hydrogels to disrupt biofilms using live/dead staining. After 12-hour treatment, no visible reduction of SYTO-9 fluorescence was observed in the HABP-BP group, indicating limited intrinsic antimicrobial ability. In contrast, the HABP-BP-V-Ag⁺ and HABP-TSPP-V-Ag⁺ hydrogels effectively disrupted biofilms, as evidenced by the scattered propidium iodide (PI) signals interspersed with reduced SYTO-9 fluorescence (Fig. 4B). The confocal images further confirmed the enhanced biofilm reduction capacity of the HBAP-BP-V-Ag⁺ hydrogel compared to the HBAP-BP-V group (Fig. 4C). Furthermore, the bacterial colony-forming unit (CFU) assay was conducted to quantify the bactericidal capacity of the hydrogels. The HABP-BP-V hydrogel achieved an over 2-log reduction in bacterial load (99.9% reduction, from an average 8.7 to 5.9 log₁₀ CFU/ml) after 24 h of co-culture compared to Ctr group without intervention. This effect was further enhanced in the presence of Ag⁺, achieving a >5-log reduction (from an average 8.7 to 3.2 log₁₀ CFU/ml) in bacterial load with the HABP-BP-V-Ag⁺ hydrogel compared to control group. (Fig. 4E). Notably, the bactericidal efficacy of the HABP-BP-V-Ag⁺ hydrogel was comparable to that of the HABP-TSPP-V-Ag⁺ one, confirming the robust antimicrobial effects of both systems. Further, we found that the co-culture with the HABP-BP-V-Ag⁺ and HABP-TSPP-V-Ag⁺ hydrogel significantly elevated the production of ROS in MRSA biofilms than the HABP-BP-V group, confirming that Ag⁺ release has the potential to trigger ROS generation that may synergize with vancomycin in biofilm eradication (Supplementary Fig. 10). Ag⁺ can disrupt bacterial cell walls and trigger bacterial apoptosis while simultaneously generating ROS. The ROS penetrates the biofilm matrix, compromising its structural integrity and facilitating Vancomycin access to embedded bacteria, thereby enhancing bacteria eradication (Fig. 4F)²⁹. Collectively, these findings demonstrate that both HABP-BP-V-Ag⁺ and HABP-TSPP-V-Ag⁺ hydrogels achieve a clinically significant > 5-log reduction in bacterial load, effectively disrupting biofilms and eliminating embedded bacteria. This highlights their potential as powerful therapeutic agents for managing FRI in clinical settings³⁰.

Fig. 4. The BP-based bioactive hydrogels have potent antibacterial effects and disrupt MRSA biofilms in vitro.

A The disk diffusion assay of BP-based bioactive hydrogels shows inhibition zones against MRSA. B Live/dead staining of bacterial biofilm treated with hydrogels. PI staining indicates bacterial death, while SYTO-9 shows viable cells. C Confocal images of the biofilm treated with BP-based bioactive hydrogels. D, E Quantification of disk diffusion test and CFU reduction in MRSA co-cultured with different hydrogels. Data are presented as mean ± SD (n = 3 biologically independent samples/group). F Schematic illustration of the proposed bactericidal mechanisms of the hydrogel. The images are generated using PowerPoint, provided by Microsoft, and 3DMax software, provided by Autodesk. **p < 0.0001, one-way ANOVA, multiple comparisons with Tukey correction.

BP-based bioactive hydrogels prevent cortical bone lysis and promote fracture healing in osteoporotic bone with FRI

The therapeutic potential of BP-based bioactive hydrogels was evaluated using an osteoporotic femur fracture rat model inoculated with MRSA (“OVX-M”) to simulate FRI (Fig. 5A)⁸. Ovariectomy surgery was performed on 6-month-old SD rats to induce osteoporosis over 3 months, followed by femur fracture surgery with bacteria inoculation. At week 4 post-surgery, the X-ray images showed extensive bone lysis and periosteal bone reaction in the OVX-M group, characterized by new bone formation with inflammatory irritation of the periosteal membrane. Quantitative analysis indicated significantly increased callus width and callus area in OVX-M group than those in the OVX control. Micro-CT analysis further confirmed substantial loss of high-density bone tissue at the fracture site, particularly in the axial view, indicating severe cortical bone resorption. This progressive bone loss at the fracture site likely compromised fracture stability, impairing callus bridging and creating dead space that exacerbated bacterial infection (Fig. 5B, C). Three-dimensional reconstructed images using different thresholds to distinguish newly formed low-density bone (blue) from the high-density bone tissue (gray), further revealed prominent cortical bone loss in the OVX-M group. BVh (threshold from 350 to 1000) represents mature cortical bone that is progressively eroded by osteoclast overactivation, while BVl (threshold from 165 to 350) is characterized by disorganized, poorly mineralized bone formation around lytic areas. By week 8, the cortical bone lysis becomes more prominent, with a dramatic reduction of high-density cortical bone in the OVX-M group compared to the OVX group (−88.3% in BV, p < 0.0001). The coexistence of BVh loss and BVl reveals the uncoupling of normal remodeling where destructive processes outpace effective repair. Despite extensive formation of low-density bone tissue, no bridging of the fracture gap was observed in any of the rats in the OVX-M group (Fig. 5D, E), reflecting the impaired healing phenotype characteristic of osteoporotic FRI, which mimics the radiographic features observed in clinical scenarios¹.

Fig. 5. BP-based bioactive hydrogels prevent cortical bone lysis and promote fracture healing in osteoporotic bone with FRI.

A The schematic illustration of the animal experiment. OVX, ovariectomy. B, C The representative X-ray and micro-CT images, and quantification of the fracture callus and bone volume at week 4. In the 3D reconstructed view, the blue color indicates newly formed bone with low density, whereas the gray color indicates high-density cortical bone. BVh, bone volume with high density; BVl, bone volume with low density. Data are presented as mean ± SD (n = 5 animals per group). D, E The representative X-ray and micro-CT images, and quantitative analysis at week 8. Data are presented as mean ± SD (n = 5 animals per group). *p < 0.05, **p < 0.0001, one-way ANOVA, multiple comparisons with Tukey correction.

Notably, treatment with both HABP-BP-V-Ag⁺ and HABP-TSPP-V-Ag⁺ significantly increased bone volume (BVh) compared to untreated OVX-M controls by week 4, indicating ameliorated osteolysis. In contrast, no such protective effect was observed in the OVX-M-V group, underscoring the limitation of systemic antibiotic treatment. Besides, the unresolved bone lysis and non-union in the OVX-M-HABP-BP group further confirmed that the HABP-BP hydrogel alone was insufficient to support bone healing without effective infection control. By week 4, both OVX-M-HABP-BP-V-Ag⁺ and OVX-M-HABP-TSPP-V-Ag⁺ group showed reduced callus area compared to the OVX-M group (Fig. 5B, C). Notably, the HABP-BP-V-Ag⁺ group exhibited a substantial volume of high-density bone tissue in the fracture callus by the study endpoint (Fig. 5D, E). The increased callus area relative to OVX controls suggests delayed callus remodeling, consistent with reports that high systemic BP concentrations can delay the transition from woven bone to lamellar bone and impair callus remodeling³¹. To address this, the HABP-TSPP-V-Ag⁺ hydrogel was designed with a reduced BP concentration (20 mM), resulting in significantly lower callus area and improved callus remodeling compared to the HABP-BP-V-Ag⁺ group.

BP-based bioactive hydrogels restore disturbed formation of fracture callus in FRI and rejuvenate coordinated healing cascades

The integrity of cortical bone is the prerequisite for stabilizing the fracture site and facilitating the formation of a cartilage callus to bridge the fracture gap³². However, in osteoporotic FRI with severe bone lysis, the disturbance of callus formation has not been fully characterized. In the OVX-M group, H&E staining demonstrated extensive necrotic tissue and inflammatory cell infiltration at the fracture gap (Fig. 6A), while Safranin O staining showed no visible cartilage formation (Fig. 6B). Despite increased periosteal bone formation surrounding the cortical bone, the fracture gap remained unbridged. These pathological changes suggest that the loss of fracture site stability, compounded by ongoing infection, impaired the formation of a cartilage callus. Next, we performed IHC staining of osteocalcin (OCN) and collagen-1 (COL-1) to assess bone regeneration. In the non-infected control group, a brown-stained area adjacent to the fracture site indicated active new bone formation. Comparatively, the expression OCN and COL-1 was significantly decreased in the OVX-M group compared to that in the OVX group (Fig. 7A, C, D and Supplementary Fig. 11), highlighting impaired bone regeneration. In FRI, the overshoot of inflammatory cytokines, particularly TNF-α, is known to drive bone cell death and osteolysis⁹. Immunofluorescence staining demonstrated significantly increased TNF-α expression at the fracture site in the OVX-M group compared to the OVX control (Fig. 7B, E). At week 8, high-magnification H&E staining images demonstrated aggravated cortical bone erosion and empty osteocyte lacunae (Supplementary Fig. 12), accompanied by persistently elevated TNF-α levels (Fig. 7B, F). These findings elucidate that the normal healing process was diverted toward an infection-associated inflammatory phenotype in untreated osteoporotic FRI.

Fig. 6. Histological analysis reveals restored callus formation and coordinated bone healing with BP-based hydrogel treatment.

A The H&E staining of harvested fractured femur at week 4 and week 8. B Safranin O staining of the cartilage callus at the fracture site. C, D Semi-quantitative scoring of bone healing and callus area at week 4 and week 8, respectively. Data are presented as mean ± SD (n = 5 animals per group). E, F Quantification of cartilage formation at week 4 and week 8, respectively. Data are presented as mean ± SD (n = 5 animals per group). G Schematic illustration of the non-union occurred in the untreated osteoporotic FRI. A coordinated healing cascade was restored following HABP-TSPP-V-Ag⁺ hydrogel treatment. The images are generated using PowerPoint, provided by Microsoft. *p < 0.05, **p < 0.0001, one-way ANOVA, multiple comparisons with Tukey correction.

Fig. 7. The BP-based hydrogel treatment enhances osteogenesis and reduces local inflammation at the fracture site.

A IHC staining of OCN at week 4 and week 8, respectively. B Immunofluorescent staining of TNF-α at the fracture site. C–F The quantitative analysis of the expression of OCN and TNF-α at the fracture site. Data are presented as mean ± SD (n = 5 animals per group) **p < 0.0001, one-way ANOVA, multiple comparisons with Tukey correction.

In the OVX-M-V group, systemic vancomycin treatment did not mitigate bone lysis or improve healing outcomes, consistent with previous findings that systemic antibiotics have limited effects against FRI due to poor local drug penetration²⁸. Similarly, the HABP-BP hydrogel, which lacks antimicrobial agents, demonstrated no therapeutic benefit either, as infection remained uncontrolled. While in the HABP-BP-V-Ag⁺ treated group, H&E staining demonstrated an intact cortical bone structure, in which a bony callus was formed surrounding the fractured bone. IHC staining showed increased expression of OCN and COL-1 compared to OVX-M. These observations suggest that HABP-BP-V-Ag⁺ hydrogels successfully mitigated inflammatory necrosis and preserved the cortical bone structure. However, Safranin O staining demonstrated a reduced area of cartilage callus compared to the OVX control (Fig. 6B,E). At week 8, the fracture gap in the HABP-BP-V-Ag⁺ group was connected by fibrotic tissue, around which a large volume of highly mineralized callus was formed (Fig. 6A). This suggests that while the hydrogel promoted fracture healing, high BP concentrations may have hindered callus remodeling, delaying the transition from woven bone to lamellar bone.

In contrast, the HABP-TSPP-V-Ag⁺ hydrogel demonstrated superior fracture healing outcomes. At week 4, Safranin O staining showed cartilage formation at the center of the fracture gap, which progressively mineralized into bone, leading to complete bridging of the fracture by week 8 (Fig. 6B, E, F). Further, we assess the histological bone healing with tissue quality score, which classifies callus from 0 (bony union, best) to 4 (mostly inflammatory cells, worst), with higher scores indicating poorer callus quality and impaired healing. The semi-quantitative score confirmed the restored healing in both OVX-M-HABP-BP-V-Ag⁺ and OVX-M-HABP-TSPP-V-Ag⁺ group at week 4, whilst by week 8, a more mature bony callus was observed in OVX-M-HABP-TSPP-V-Ag⁺ group compared to the fibrotic callus in OVX-M-HABP-BP-V-Ag⁺ group (Fig. 6C, D). Next, we performed 4-point bending test to assess the mechanical strength of the fracture callus. Compared to the OVX-M group, both hydrogel-treated groups showed enhanced ultimate load and stiffness. Notably, HABP-TSPP-V-Ag⁺ group exhibited even better ultimate load and stiffness compared to HABP-BP-V-Ag⁺ (Supplementary Fig. 12). These findings suggest that while a high dosage of topical BP in HABP-BP-V-Ag⁺ hydrogel does not impair bony callus formation, it may hinder callus remolding. Comparatively, a lower dosage of BP concentration in HABP-TSPP-V-Ag⁺ hydrogel yields a more coordinated healing process, promoting optimal callus remodeling and functional bone regeneration (Fig. 6G).

BP-based hydrogels eradicate the infection and prevent bacterial colonization of cortical bone

Eradicating bacteria at the fracture site is essential for improving bone healing outcomes in FRI. The biofilm-associated bacteria residing in the osteocyte-lacuno-canalicular network (OLCN) are highly recalcitrant and beyond surgical margins, causing persistent infection and treatment failure³³. Here, we first evaluated bacterial contamination of bone tissue using Gram staining. Notably, the bacteria colonies were identified in the cortical bone tissue and OLCN in the OVX-M group at week 4, coinciding with the onset of cortical bone lysis. Reportedly, the dispersing of bacteria and colonization of OLCN is more prominent in osteoporotic bone with FRI due to the high cortical porosity⁸. In the OVX-M-V group, the presence of bacteria in bone tissue revealed limited therapeutic effect of systemic antibiotics against deep-seated infections (Fig. 8A). In contrast, negative CFU cultures and the absence of Gram-positive bacteria in bone tissue demonstrated that bacteria were eradicated in both the OVX-M-HABP-BP-V-Ag⁺ and OVX-M-HABP-TSPP-V-Ag⁺ groups, confirming the effect of pH-triggered release of Ag⁺ and antibiotics in vivo. In the OVX-M-V group, the bacteria load on bone tissue was significantly reduced, but such effects were not observed on the k-wire. (Fig. 8B, D, E).

Fig. 8. The BP-based hydrogels effectively eliminate in vivo MRSA infection and prevent biofilm formation on cortical bone and implants.

A Schematic illustration of the bacteria invasion of the cortical bone in osteoporotic FRI. B Gram staining showing Gram-positive bacteria within the osteocyte lacunar in the cortical bone tissue. CB cortical bone. C SEM images showing the biofilm formation on untreated K-wire, whilst absent in hydrogel-treated groups. D, E Quantitative analysis of the bacteria CFU from k-wire and bone tissue. Data are presented as mean ± SD (n = 3 animals per group). *p < 0.05, **p < 0.0001. One-way ANOVA, multiple comparisons with Dunnett correction.

Additionally, the bacterial biofilm formed on the implant is a major challenge in FRI management, serving as a persistent reservoir for reinfection^34,35. SEM images of the K-wire surface in the OVX-M and OVX-M-V groups reveal extensive bacterial biofilm and necrotic tissue adherence (Fig. 8C). High-magnification images confirmed bacterial cells enclosed within the biofilm matrix. Notably, both HABP-BP-V-Ag⁺ and HABP-TSPP-V-Ag⁺ treatments prevented biofilm formation, and CFU plating confirmed the successful eradication of bacteria on the implant. These results collectively demonstrated the in vivo antimicrobial capacity of the HABP-BP-V-Ag⁺ and HABP-TSPP-V-Ag⁺ hydrogels.

BP-based hydrogels alleviate osteoclast activation and osteocyte apoptosis in vivo

As the key contributors to cortical bone damage, activated osteoclasts are major therapeutic targets in FRI⁹. By assessing osteoclast activity through immunohistochemistry (IHC) for cathepsin K (CTSK) (Fig. 9A), a large number of positively-stained osteoclasts was identified surrounding the cortical bone in the OVX-M group. Both HABP-BP-V-Ag⁺ and HABP-TSPP-V-Ag⁺ treatments significantly reduced CTSK expression, indicating inhibited osteoclast activation (Fig. 9C). The therapeutic effects were absent in the HABP-BP hydrogel-treated group. Notably, a similar reduction of osteoclast activation in HABP-TSPP-V-Ag⁺ groups indicates that a low dosage of BP can effectively prevent excessive bone loss in osteoporotic FRI. Additionally, elevated serum levels of CTSK in the OVX-M group also indicate bone catabolism, which was ameliorated by both HABP-BP-V-Ag⁺ and HBAP-TSPP-V-Ag⁺ treatments (Supplementary Fig. 13). Importantly, in the late stage of fracture repair, activated osteoclasts are indispensable for reducing the callus size and facilitating callus remodeling³⁶. At week 8, we identified that the number of osteoclasts in the bone tissue around fracture site was significantly decreased in HABP-BP-V-Ag⁺ compared to HABP-TSPP-V-Ag⁺. (Fig. 9A, D). Consistently, in the OVX-M-HABP-BP-V-Ag+ group, the expression of bone matrix-deposited cytokines, including TGF-β1 and IGF1, was markedly reduced compared to the control group, whilst the expression of RANKL was remained comparable. In contrast, the OVX-M-HABP-TSPP-V-Ag⁺ group exhibited TGF-β1 and IGF1 levels similar to those of the OVX control, demonstrating preserved osteoclast-osteoblast coupling (Supplementary Fig. 14). These findings suggest that the enhanced callus remodeling observed in the HABP-TSPP-V-Ag⁺ group is primarily attributable to the reduced BP dosage, which prevents excessive osteoclast suppression and sustains the coupled remodeling process. In particular, osteoclasts play a critical role by first resorbing mineralized tissue and cartilage, then working in tandem with osteoblasts in coordinated cycles to rebuild strong, functional bone². If this delicate balance is disrupted by excessive BP, it can hinder the timely restoration of proper bone structure and strength.

Fig. 9. The BP-based hydrogels suppress osteoclast activation and protect osteocytes from apoptosis at the fracture site.

A Immunohistochemistry staining of CTSK in cortical bone at week 4 and week 8. B TUNEL staining of apoptotic osteocytes embedded in the cortical bone tissue. C, D Quantitative analysis of the CTSK expression. Data are presented as mean ± SD (n = 5 animals per group). E Quantitative analysis of the apoptotic osteocytes. Data are presented as mean ± SD (n = 5 animals per group). F Schematic illustration of the restored callus formation and bone healing following hydrogel treatments. The images are generated using PowerPoint, provided by Microsoft. *p < 0.01, **p < 0.0001. One-way ANOVA, multiple comparisons with Tukey correction.

In cases of FRI, inflammatory cytokines such as TNF-α, and IL-1β are major initiators for inflammatory response⁹. Osteocyte apoptosis in inflammatory microenvironment is found to cause bone resorption and tissue damage^26,37. Consequently, the increase of osteocyte-mediated cortical matrix degradation may exacerbate the spreading of the bacteria³⁸, and the invasion of deeply embedded osteocytes and forming an escalated infection²⁸. In our study, TUNEL staining showed significantly increased osteocyte apoptosis at the fracture site in the OVX-M group compared to the control group (Fig. 9B). The proportion of apoptotic osteocytes in cortical bone tissue was significantly reduced in the HABP-BP-V-Ag⁺ and HABP-TSPP-V-Ag⁺ groups (Fig. 9E), which can be attributed to the anti-apoptotic effects of BP on TNF-α-induced osteocyte death. Here, we found histological evidence of OLCN colonies in osteoporotic FRI, which was prevented by both HABP-BP-V-Ag⁺ and HABP-TSPP-V-Ag⁺ treatment. Taken together, we proposed that HABP hydrogel therapy can reduce osteoclast activation and osteocyte apoptosis, thereby strengthening cortical bone integrity, and preventing bacterial invasion of the cortical bone. With an appropriate dosage of BP, these effects will jointly provide a stable fracture site, promote the callus formation cascades, and facilitate healing in osteoporotic bone (Fig. 9F). Further we evaluated the presence of pro-inflammatory M1 macrophages in the bone tissue with IHC staining for iNOS. At week 8, neither the HABP-BP-V-Ag⁺ nor HABP-TSPP-V-Ag⁺ groups showed significant increases in CD68 or iNOS levels in the bone tissue compared to control group (Supplementary Fig. 15). ELISA test shows no significant elevation of TNF-α and IL-6 in both hydrogel treated groups, confirming that the HBAP hydrogel does not trigger prominent immune response (Supplementary Fig. 15). Moreover, H&E staining of the heart, kidney, and liver showed no signs of systemic toxicity following hydrogel treatment, confirming the biocompatibility and localized action of BP-based hydrogels (Supplementary Fig. 16). This underscores the safety and therapeutic potential of BP-based hydrogels as a targeted treatment for osteoporotic FRI.

Discussion

In osteoporotic bone with FRI, increased bone lysis and bacterial invasion of the cortical bone are key pathological features that cause nonunion. The limited efficacy of systemic antibiotic treatment poses an urgent need for developing new therapeutic interventions tailored to the unique pathological changes in osteoporotic bone. In this study, we first developed a HABP-BP/TSPP hydrogel loaded with Vancomycin and Ag⁺, which not only effectively eradicates the infection but also augments bone healing. Through simultaneously inhibiting osteoclast activation and osteocyte apoptosis, the hydrogel therapy successfully prevents excessive osteolysis and preserves cortical bone integrity. This restoration of a stable, sterile environment facilitates coordinated callus formation and healing cascades, addressing the critical deficiencies observed in osteoporotic FRI.

The cortical bone lysis is the detrimental event that impairs the stability and bone healing in FRI. In osteoporotic FRI, the well-coordinated callus formation and mineralization course has been altered⁸, and the limited effect of systemic antibiotics warrants innovative treatment. At cellular level, osteoclast is the major cause of bone lysis in bone infection. Here, we observed a large number of osteoclasts accumulated on the cortical bone surface, which was ameliorated by the HABP-based hydrogel. Additionally, as the major resident cells, osteocytes have been found to be involved in the bone catabolism of FRI. The presence of empty lacunae and increased cortical bone porosity may facilitate bacterial invasion. Yoshimoto et al. demonstrated that osteocytes predominantly drive bacterial-induced osteolysis via RANKL production³⁹. Hence, we developed the HABP-based bioactive hydrogels to inhibit osteocyte apoptosis. Through preventing OC activation and osteocyte apoptosis, the cortical bone integrity was protected. Therefore, normal callus formation and bone healing are restored, and cortical bone invasion and persistent infection are prevented.

BP-based bioactive hydrogels have been rapidly developed in recent years for bone regeneration. Reportedly, the nano-formulation of BP delivered by HA-MC/CaCO3/ZOL@PBAE-SA is found to preferentially inhibit bone resorption in osteoporotic mice without affecting bone anabolism⁴⁰. The BP-functionalized Gly-Phe-Phe hydrogel has been shown to enhance periodontal bone regeneration by inhibiting osteoclasts while facilitating the attachment and migration of BMSCs⁴¹. Nevertheless, the suggested potential of application in periodontitis has not been validated in a real infection model. In osteomyelitis, BP-conjugated antibiotics have been shown to enhance delivery efficiency due to their high affinity for bone tissue⁴². Leveraging the coordination between BP and multivalent metal ions, aqueous solutions of BP, hyaluronic acid-grafted BP can form highly tunable nanocomposite hydrogels^43,44. Unlike previous studies utilizing HABP with ex situ nanoparticles including calcium phosphate or magnesium silicate^45,46, our work employs in situ self-assembled Ag⁺-Ca²⁺ -BP nanoparticles within the hydrogel network, yielding superior mechanical properties and stability. Compared with the previously reported HA-BP/Ag⁺ hydrogels synthesized via BP-Ag⁺ coordination for infected wound healing²³, our design further incorporates BP’s intrinsic osteoprotective functions and uses TSPP to enable tunable BP dosage. This allows for pH-responsive, targeted antibiotic/Ag⁺ release, and more importantly, positions BP as both a structural and bioactive agent tailored for osteoporotic FRI. The acidic biofilm microenvironment-triggered release of antimicrobials from the hydrogel is capable of eliminating the infection in the fracture site, while the BP release at the fracture site inhibits osteoclast activation and osteocyte apoptosis. Through restoring a stable and sterile fracture site, bone union and fracture healing are achieved through the mineralization of the cartilage callus.

Supplementation with BP has been found to accelerate fracture healing by inhibiting catabolism during the early stages of callus formation⁴⁷ and increasing the abundance of trabecular-like bone throughout the callus⁴⁸. However, the potential negative effect of topical BP on callus remodeling has been documented. In this study, we observed significantly delayed callus remodeling in rats treated with HABP-BP-V-Ag⁺ hydrogel. In this group, both the free BP and grafted BP may exert biological functions in vivo, and excessive BP may cause prolonged inhibition of osteoclasts, delaying callus maturation. To address this, we formulated the HABP-TSPP hydrogel, in which BP concentration was reduced to 20 mM while maintaining comparable rheological properties and drug release profiles. TSPP is used in scaffolds or bioceramics loaded with enhanced mineralization⁴⁹. Our study demonstrated that even a low dosage of BP in the TSPP-based formulation is sufficient in reducing excessive osteoclast activation and promoting bone healing, suggesting that the grafted BP is sufficient to achieve desirable therapeutic effects in vivo.

From a translational perspective, this study is conducted in the context of FRI in osteoporotic bone. This application extends the indications for topical BP application in FRI, serving as proof of concept on simultaneous anti-osteoporosis and anti-infection therapy in managing these cases. Unlike prosthetic joint infections, the instability of the fracture site in osteoporotic FRI presents a unique challenge for clinical treatment³. Current debridement, antibiotics, and implant retention (DAIR) treatment is considered the standard option, but the dilemma is how to effectively eradicate infection. The action mode of existing therapies is killing the bacteria and biofilm but not targeting the pathological bone loss in FRI³. In light of this, our therapy offers a promising treatment option for osteoporotic bone with FRI. Here, effective bacterial eradication is an acidic biofilm microenvironment-triggered release of antimicrobials. The combination of vancomycin and Ag⁺ was used to enhance antibacterial efficacy (achieving a 5 log₁₀ CFU reduction) and biofilm disruption, as vancomycin alone was insufficient for complete infection clearance in osteoporotic FRI due to compromised immunity and biofilm resistance. Reportedly, a low dosage of Ag⁺ coated on the implant surface is safe and does not affect osteogenic differentiation⁵⁰. The clearance of bacteria and the preservation of cortical bone integrity will jointly lead to the re-establishment of coordinated healing cascades. Our proposed adjunctive therapy can synergize with DAIR to improve infection control and bone healing. This combined strategy may mitigate catastrophic bone lysis and non-union, ultimately enabling implant removal and functional bone regeneration. However, real-world application faces practical challenges including the optimal dosage and timing of intervention that integrates with DAIR, which demands precise surgical and pharmacological coordination. Further preclinical and translational validation will be essential before advancing this strategy to clinical trials.

Our study has several limitations. First, the molecular mechanisms underlying the cellular response of osteocytes to FRI, particularly in cases of osteoporosis, are not fully understood. Second, we primarily focus on the role of BP-based hydrogel in treating osteoporotic bone, while its role in normal bone remains to be elucidated. Third, our anti-infection therapy addresses the early phase of infection, but whether it can be applied to late-stage FRI with mature bacterial infection remains to be validated. Further studies will examine the hydrogel’s stability in clinically relevant environments, such as wound exudate, potential contamination, and mechanical stress, to further validate its applicability in open fracture scenarios. Additionally, while our femur fracture model effectively recapitulates key FRI features, it may not fully represent the higher-risk clinical scenario of open tibial fractures commonly seen in young males, where the insufficient soft tissue coverage may further increase the risk of infection and non-union. Future investigations are warranted to validate the therapeutic efficacy of our hydrogel in such challenging clinical settings.

In summary, we developed a BP-based hydrogel that integrates anti-infective and osteoprotective functions for treating osteoporotic bone with FRI. We demonstrated that the pH-triggered release of Ag⁺ and Vancomycin effectively eradicates bacterial infection, restoring a sterile fracture environment essential for healing. Simultaneously, BP release at the fracture site prevents cortical lysis and enhances bone integrity by inhibiting osteoclast activation and osteocyte apoptosis, thereby re-establishing coordinated healing cascades and preventing further bacterial invasion into the cortical bone. Taken together, our findings provide strong preclinical evidence supporting the dual-functionality of BP-based hydrogels as an integrated anti-infection and bone-regenerative therapy for osteoporotic FRI.

Methods

Ethics statement

The animal study protocols were approved by the Institutional Animal Care and Use Committee of Nanjing Drum Tower Hospital (Permit number 2021AE01027).

Synthesis of the HA-BP

Referring to the previously reported synthesis method, slight modifications were made. One gram of hyaluronic acid (HA) was dissolved in 100 ml of water, and stirred until dissolved, and then 2.15 ml of methacrylic anhydride solution was added in an ice bath environment. The pH of the reaction system was maintained at 8.5–9 by dropwise addition of NaOH solution while stirring rapidly, and the reaction proceeded for three hours. The mixture was dialyzed against sodium chloride solution and deionized water for 3 days, then frozen at −80 °C, and freeze-dried to obtain MeHA. 2.3 g of N-hydroxysuccinimide (NHS), 3.8 g of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC), and 0.92 g of mercaptoacetic acid were dissolved in 2 ml of N, N-dimethylformamide (DMF), and after overnight reaction, 40 ml of pyrophosphate buffer (1.12 g) was added to maintain pH at 8.5, then reacted in the dark for 24 h. After washing with anhydrous ethanol and drying, thiolated BP was obtained. Thiolated BP, MeHA, and tris(2-carboxyethyl) phosphine (TCEP) were dissolved in TEOA buffer with pH 8.5, reacted in the dark for 48 h, dialyzed for three days, and then freeze-dried to obtain HABP. The grafting rate of BP onto HA was calculated as 21% by subtracting the double-bond grafting rate of HA-BP (6%) from that of MeHA (27%), representing the quantity of double bonds consumed by thiolated BP during the thiol-ene coupling reaction.

Fabrication of the HABP-BP-V-Ag⁺ hydrogel

HABP (4% w/v), free BP (200 mM), and Ca (NO3)₂ (286 mM) were dissolved in PBS solution and vortexed to form the HABP-BP hydrogel. The HABP-BP-V-Ag⁺ hydrogel was prepared by adding AgNO₃ (20 mM) and Vancomycin (5 mg/ml) to the mixture. The other two hydrogel groups, HABP-TSPP and HABP-TSPP-V-Ag⁺, were obtained by replacing free BP with TSPP (200 mM). The total volume of the HABP-BP-Ag⁺ hydrogel was 50 μL for in vivo application.

Rheological measurement

Using a hybrid rheometer, rheological tests were conducted on hydrogel samples with a volume of 100 μL and a thickness of 2 mm. For oscillatory time sweep experiments, the constant strain and frequency were fixed at 1% and 1 Hz, respectively. In the shear strain scan tests, the shear frequency was fixed at 1 Hz, and the shear strain ranged from 0.1 to 500%. For the high and low shear strain scan tests, three cycles of high shear and low shear were performed, with shear strains of 400% and 1%, respectively.

Scanning electron microscope

The hydrogel samples were rapidly frozen in liquid nitrogen and then placed in a freeze dryer for 6 h. Afterward, they were rapidly frozen again in liquid nitrogen and sliced with a blade. The freeze-dried sample sections were mounted face-up on copper pillars and coated with a gold/palladium sputter for 60 s. Subsequently, scanning electron microscopy (SEM) images were obtained using a field emission scanning electron microscope (Zeiss Merlin) equipped with an X-MaxN20 EDS system.

Injectability test

Inject 200 μL of hydrogel into a luer-lock syringe with a needle inner diameter of 0.45 mm. Inject the hydrogel into a polytetrafluoroethylene mold in the shape of a leaf, then let it sit for 1 minute before demolding the hydrogel.

Quantification of the release of BP and Vancomycin from the hydrogel

At 37 °C, 50 μL of hydrogel was immersed with 2 ml PBS buffer. At designated time intervals, 1200 μL of the supernatant was replaced with a new release medium. The concentration of the BP in the release medium was determined through the formation with a Fe (III) ion and measured by UV/Vis spectrophotometer at 293 nm wavelength⁵¹. The Vancomycin was detected by UV/Vis spectrophotometer at the wavelength of 236 nm. The concentration of Ag⁺ was determined by inductively coupled plasma mass spectrometry (ICP-MS).

Preparation of hydrogel extracts

Briefly, the hydrogels were sterilized by submerging the samples in 70% v/v ethanol and then washed with PBS. Subsequently, 50 µL of the hydrogels was transferred to a 6-well culture plate and incubated with 2.5 mL of specific cell culture medium for BMSC, osteoblasts, and BMM-induced osteoclast for 24 h (37 °C, 5% CO₂, and 95% air).

Primary bone marrow stem cell and osteogenic differentiation

The bone marrow stem cells from both tibiae of 1-month-old SD rats were flushed out and cultured in LDMEM containing 10% FBS and 1% PSN. The BMSC in the second passage was used in these experiments. Then the cytotoxicity was measured by CCK-8⁵². Osteogenic differentiation was induced with dexamethasone (0.1 μM), ascorbate acid (50 μg/ml), and β-glycerophosphate (10 mM). ALP staining was performed on day 7, and ARS staining was performed on day 14. The ALP activity was assessed with the enzyme activity kit (Beyotime, P0321S).

Primary macrophage osteoclastogenesis, TRAP staining, and FITC-labeled phalloidin staining

The bone marrow cells from 6 weeks C57BL/6 mice were filtered through a cell strainer, and the cells were seeded in a T 25-cm² flask for proliferation. After 24 h, the cells were collected and resuspended in 5 ml α-MEM with 30 ng/ml macrophage colony-stimulating factor (M-CSF; R&D System), and the adherent cells were classified as osteoclast. After 48 h of culture, the cells were induced for osteoclast differentiation in α-MEM with 30 ng/ml and 50 ng/ml NF-κB receptor activators. After 5 days of culture, the cells were fixed with 4% paraformaldehyde and stained with TRAP working solution for 60 min. Then the TRAP-positive cells were observed and imaged by an optical microscope. The differentiated osteoclast was stained with actin labeled with DAPI, and fluorescein isothiocyanate to evaluate the formation of F-actin ring (G1248, Servicebio).

The in vitro antimicrobial effect of the HABP-BP-V-Ag⁺ hydrogel

The antibacterial was tested with MRSA (ST239). The bacteria strains were recovered and cultured in the TSB medium at 37 °C for 24 h. Then the bacteria suspension (OD600 = 0.25) was incubated with 50 μL hydrogel. Then the CFU in the medium was quantified by serial dilution and plate counting. Additionally, to observe the effect of hydrogel on the bacterial biofilm formation, the attached bacteria were cocultured with hydrogel in the confocal dishes for 24 h and stained with LIVE/DEAD bacteria viability kit (L13152, Invitrogen). ROS accumulation was measured in 24-h mature MRSA biofilms treated with PBS or hydrogels. After washing and CFU quantification, biofilms were stained with 10 μM DCFHDA (1 h, 37 °C), and fluorescence intensity normalized to CFU was analyzed via spectrometry.

Quantitative polymerase chain reaction

The total RNA was extracted from the cultured cells with the Trizol reagent, and cDNA was synthesized according to the instructions from reverse transcription. RT-qPCR was performed with the ChamQ SYBR Color qPCR master Mix kit, and the premier sequence was supplemented in Table S1.

Flow cytometry

The MLO-Y4 cells were harvested according to the instructions with Annexin V FITC apoptosis Kit (Beyotime, C1062M). Briefly, the cells were digested and resuspended in a staining buffer, and 1 × 10⁵ cells were incubated with Annexin V FITC in the dark for 15 min, propidium iodide was added and the apoptosis was detected by flow cytometry. (Accuri C6, BD Biosciences, Franklin Lakes, NJ, USA).

Animal study

Ovariectomy surgery was performed at month 6 and the rats were kept for another 3 months to induce osteoporosis. A 9-month-old Sprague-Dawley rat model with FRI was established with published protocol⁸. The animals were randomized into OVX, OVX-M (with MRSA), OVX-M-V (treated with systemic Vancomycin), or with hydrogel treatment, including OVX-M-HABP-BP, OVX-M-HABP-BP-V-Ag⁺, or OVX-M-HABP-TSPP-V-Ag⁺. The hydrogel treatment was applied 24 h after surgery to allow for the infection development. Briefly, 50 μL of the hydrogel was injected with a pipette, and the treatment was applied at 24 h post-surgery when the infection was established. Additionally, intraperitoneal injection of vancomycin was started until sacrifice (50 mg/kg body weight twice daily)⁵³.

X-ray radiograph

The anterior-posterior X-rays were taken at week 4 and week 8 post-surgery. The callus morphology, including callus width and callus area was measured in the Image J software according to the established protocol⁸.

micro-CT scanning

The femur was harvested and scanned with micro-CT (μCT-40, Scanco Medical, Switzerland). The region of interest was defined as 422 slices at the upper or lower than the fracture line. The image was visualized in the axial plane and the 3D reconstructed views. A separate reconstruction was used to differentiate the old cortical bone from the newly formed low-density bone, and the low-density bone volume (BVl) and high-density bone volume (BVh) were calculated accordingly⁸.

Histological and histomorphometric analysis

At weeks 4 and 8, femur samples were harvested, fixed in 10% neutral buffered formalin buffer, decalcified in 10% EDTA solution, and embedded in paraffin. A 5 μm slice was obtained for histological staining. A scoring system based on H&E staining was adopted to assess the histological component of the fracture callus⁵⁴. Briefly, the bone union was defined as 5 grades. Bony union: 0; 1 the majority was new bone:1; the majority was fibroblast with less new bone: 2; the majority was spindle with less inflammatory cells: 3; and the majority was inflammatory cells: 4. The modified gram staining was used to visualize the bacterial colonies in the bone tissue. The Safranin O-Fast green (SOFG) to quantify the proportion of cartilage area in the fracture callus.

Immunohistochemical staining and immunofluorescence staining

For the immunohistochemical staining, the prepared slides were rehydrated, heated in sodium citrate for antigen retrieval, and blocked with BSA. The primary antibodies of CTSK (1:200, DF6614, Affinity), Collagen I (1:200, AF7001, Affinity), Osteocalcin (1:200, DF12303, Affinity), and TNF-α (1:300, ab6671, Abcam), CD68 (1:200, DF7518, Affinity), iNOS (1:200, AF0199, Affinity), IGF1(1:200, DF6096, Affinity), TGF-β1(1:200, BF8012, Affinity), RANKL (1:200, AF0313, Affinity) were applied and incubated overnight, followed by the incubation of the Horseradish peroxidase (HRP)-conjuncted secondary antibody and visualized following reaction with 3,3’-diaminobenzidine (DAB). The positive expression of the protein at the target region of interest was quantified by Image J (NIH, USA). For immunofluorescent staining, the sections were incubated with corresponding secondary antibodies and stained with DAPI. The fluorescent images were captured and the fluorescence was quantified with Image-Pro Plus 6.0 software.

TUNEL staining

For TUENL staining, the slides were processed with terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay kit (G1507; Servicebio), followed by diaminobenzidine staining, and were then counterstained with haematoxylin.

Bacteria observation and CFU quantification

Under the sterilized condition, the infected femur was harvested, weighed, and homogenized in sterilized PBS solution. The bacteria load in bone samples was calculated by plating in serial dilution and CFU counting. The bacteria load attached to the K-wire was sonicated, and the CFU load was determined in a similar method⁸. Additionally, the K-wire was carefully removed from the harvested femur and fixed with glutaraldehyde solution, air-dried, mounted, and subject to microstructure observation under SEM.

ELISA

Enzyme-linked immunosorbent assay test was performed to quantify bone turnover markers in the serum, including Cathepsin K, Trap5b, Balp, and Osteocalcin, TNF-α and IL-6.

Mechanical test

At eight weeks post-surgery, the femur was carefully dissected free of muscle. A four-point bending test was used to test the stiffness of the fracture callus. The ultimate load (N), stiffness(N/mm), were calculated based on the stress-displacement curves.

Statistical analysis

The data was presented as mean ± standard deviation. Statistical analysis was performed with GraphPad Prism 8.0 software. An unpaired independent t-test was used for data comparison between the two groups. Data with more than two groups was analyzed by one-way ANOVA analysis with Tukey correction. A two-sided p-value less than 0.05 was considered as statistically significant.

Reporting summary

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

Author contributions

J.L., K.Z., Z.Z., L.B., Z.L. conceived and designed the study. J.L., Y.Z., Yi.L., Y.X., H.X., Q.L., Z.T., C.L., ZY.T., K.A., Ya.L., Z.H., Y.Q., performed the experiment and analyzed the data. Y.Q., K.Z., Z.Z., L.B., and Z.L. supervised the project. J.L. Y.Z. Yi.L. K.Z. Z.Z., L.B., and Z.L. drafted the manuscript; All authors participated in interpreting the data and producing the final manuscript.

Peer review

Peer review information

Nature Communications thanks Bin Li, Francis Lee, and Yuan Yuan for their contribution to the peer review of this work. A peer review file is available.

Data availability

All relevant data supporting the key findings of this study are available within the article and its Supplementary Information files. 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/s41467-025-68036-1.