ISO-1 sustained-release hydrogel repairs infected bone defects by suppressing osteoblast pyroptosis and enhancing osteogenic differentiation via the MIF-CD74/PI3K/AKT signaling pathway

Abstract

Background

Staphylococcus aureus (SA)-induced osteomyelitis causes refractory infected bone defects characterized by impaired osteogenesis and excessive inflammatory bone destruction. Pyroptosis has emerged as an important driver of infection-related tissue injury, but the upstream regulators in osteoblasts and effective local therapeutic strategies remain poorly defined. This study investigated the role of the MIF-CD74/PI3K/AKT axis in SA-mediated osteoblast dysfunction and evaluated the ISO-1 sustained-release hydrogel for repairing infected bone defects.

Methods

MIF expression was examined in human osteomyelitis sample. Bone destruction, inflammation, and osteogenesis were compared between wild-type and Mif^−/− mice using micro-CT, histology, ELISA, RT-qPCR, and immunohistochemistry. In vitro, SA-infected mouse bone marrow mesenchymal stem cells (mBMSCs) were used to assess proliferation, osteogenic differentiation, and pyroptosis, while MIF and CD74 were silenced with siRNA. RNA sequencing (RNA-seq) and co-immunoprecipitation (Co-IP) were employed to identify MIF downstream receptors and pathways, and PI3K/AKT involvement was verified using CD74 overexpression and the PI3K inhibitor LY294002. An injectable, photocrosslinkable ISO-1@β-CD/GelMA hydrogel was fabricated and characterized for structure, mechanics, and pH-responsive drug release, then implanted into rat femoral infected bone defects after debridement to evaluate bone regeneration and local inflammatory modulation.

Results

MIF was markedly upregulated in human osteomyelitis and its genetic deletion in mice significantly attenuated SA-induced bone loss, preserved trabecular microarchitecture, enhanced osteogenic marker expression, and reduced local pro-inflammatory cytokines. SA infection suppressed mBMSC proliferation and osteogenic differentiation while activating NLRP3/ASC/GSDMD-dependent pyroptosis. MIF or CD74 knockdown partially restored osteogenesis and decreased inflammatory cytokine release and pyroptotic markers. RNA-seq and Co-IP identified CD74 as a key MIF receptor, with PI3K/AKT as a critical downstream pathway regulating osteoblast fate. The ISO-1@β-CD/GelMA hydrogel showed favorable injectability, mechanical stability, and sustained ISO-1 release, and significantly improved bone volume, microarchitecture, and inflammatory microenvironment in rat infected bone defects.

Conclusion

SA-induced osteomyelitis disrupts bone homeostasis via the MIF-CD74/PI3K/AKT axis, thereby suppressing osteogenesis and promoting NLRP3/ASC/GSDMD-mediated osteoblast pyroptosis. Targeting MIF with an ISO-1 sustained-release hydrogel remodels the inflammatory and pyroptotic microenvironment, enhances osteogenic differentiation, and effectively promotes repair of infected bone defects, providing a promising translational platform for osteomyelitis-related bone regeneration.

Graphical abstract

Highlights

• •Attenuated bone destruction and reduced local pro-inflammatory cytokines in femoral and tibial osteomyelitis in Mif^−/− mice.
• •MIF-driven osteoblast pyroptosis and Inhibition of osteogenic differentiation as a key mechanism underlying osteomyelitis-related bone repair failure.
• •SA-induced osteomyelitis disrupts bone homeostasis, thereby suppressing osteogenesis and promoting osteoblast pyroptosis via the MIF-CD74/PI3K/AKT axis.
• •ISO-1@β-CD/GelMA hydrogel remodels the inflammatory and pyroptotic microenvironment, and markedly enhanced bone regeneration in infected bone defects.

1. Introduction

Osteomyelitis is an infectious disease of bone caused by bacteria, fungi, or mixed pathogens through various mechanisms such as spread from adjacent soft tissues or dissemination through the bloodstream [1]. Epidemiological studies indicate that its global annual incidence is approximately 15 to 25 cases per 100,000 population and is increasing due to population aging, diabetes, and the rising number of implant-related surgeries [2,3]. Staphylococcus aureus (SA) is the predominant pathogen in different forms of osteomyelitis [4] and can lead to persistent infection, bone nonunion, and a higher risk of amputation, with a related mortality rate reaching about 8 % [5]. In clinical treatment, chronic osteomyelitis often requires repeated and thorough debridement to control infection [6]. However, this procedure frequently results in residual infected bone defects and marked destruction of bone microstructure, which complicates subsequent bone reconstruction. Additionally, although current management often relies on combined strategies such as surgical debridement, internal fixation, dead space filling, and bone reconstruction, difficulties in infection control and unsatisfactory repair of bone defects remain common [7]. Therefore, achieving effective local infection suppression while promoting new bone formation and structural regeneration has become a key challenge that urgently needs to be addressed in the field of infected bone defect repair.

In recent years, with the in-depth exploration of the pathological mechanisms of osteomyelitis, there is significant potential in identifying therapeutic targets for osteomyelitis at the microscopic level. During normal bone remodeling, osteoblasts and their precursor cells synthesize matrix proteins such as type I collagen (COL-Ⅰ) and osteocalcin (OCN), and complete mineralization, which are crucial for maintaining bone mass and the integrity of bone microstructure [8,9]. However, in the infection microenvironment of osteomyelitis mediated by SA, bacterial toxins and persistent high-level inflammation together disrupt the dynamic balance between osteogenesis and osteoclastogenesis [10]. Multiple studies have confirmed that SA can significantly inhibit osteoblast proliferation, decrease alkaline phosphatase (ALP) activity, reduce the expression of bone matrix components and suppress mineralization [11]. Additionally, SA induces osteoblast lineage cell apoptosis through receptor-mediated signaling pathways such as TLR2 and TNFR1, as well as its virulence factors, such as protein A, PVL, and prothrombin-activating factor [12,13]. On the other hand, SA promotes the upregulation of RANKL expression, enhances osteoclastogenesis, and increases bone resorption, leading to progressive bone loss [14]. In recent years, in addition to classical apoptosis, NLRP3 inflammasome-GSDMD-mediated pyroptosis, as an inflammatory-associated programmed cell death mechanism, has also been shown to participate in the course of SA infection [15]. In human samples and mouse osteomyelitis models, elevated expression of pyroptosis-related proteins, such as NLRP3, can be detected. Further inhibition of caspase-1 or NLRP3 activation can partially alleviate bone destruction in mouse osteomyelitis and improve osteoblast function, suggesting that pyroptosis is closely related to SA-induced osteomyelitis [16]. Therefore, SA infection not only inhibits osteoblast differentiation but also exacerbates bone destruction by inducing apoptosis and pyroptosis of osteoblasts and mesenchymal stem cells in concert. However, the regulation of different cell death patterns and their upstream signaling networks in osteomyelitis-related osteogenesis dysfunction remains incompletely understood and warrants further investigation.

Macrophage migration inhibitory factor (MIF) is a highly conserved, pleiotropic pro-inflammatory cytokine that transduces downstream signaling via CD74 and chemokine receptors such as CXCR2, CXCR4, and CXCR7, and is closely implicated in a broad spectrum of acute and chronic inflammatory as well as infectious diseases [17,18]. Zheng et al. [19] reported that Mif suppresses osteoblast-mediated mineralization and the formation of bone nodules. In our previous work, Mif knockdown inhibited NF-κB activation, thereby restoring the expression of osteogenic differentiation markers and improving the inflammatory bone microenvironment [20]. Moreover, accumulating evidence in recent years has identified MIF as a pivotal upstream regulator of the NLRP3 inflammasome-GSDMD axis. In diabetic nephropathy, Mif overexpression activated the NLRP3/caspase-1/GSDMD pathway, promoted IL-1β release, and exacerbated extracellular matrix deposition. Notably, the small-molecule inhibitor CSB6B mitigated injury by downregulating these components [21]. In models of neuropathic pain, elevated MIF levels coincided with increased GSDMD-N and upregulation of multiple pro-inflammatory cytokines, whereas treatment with the MIF-specific small-molecule inhibitor ISO-1 attenuated NLRP3 activation and alleviated neuronal pyroptosis and pain-related behaviors [22]. Similarly, in sepsis-associated acute kidney injury, MIF induced caspase-1/GSDMD-dependent pyroptosis by enhancing NF-κB p65 phosphorylation and upregulating NLRP3 expression [23]. Collectively, across these disease models, MIF emerges as an important upstream driver of inflammatory pyroptosis, and its pharmacological inhibition (e.g., ISO-1 and CSB6B) may represent an effective strategy to curb inflammation-associated pyroptosis and consequent tissue destruction. However, whether MIF also contributes to osteoblast pyroptosis and impaired osteogenic differentiation within the microenvironment of SA osteomyelitis remains insufficiently investigated.

Additionally, ISO-1, as a therapeutic agent targeting MIF, may exert anti-inflammatory effects in osteomyelitis models while suppressing pyroptosis and promoting osteogenic differentiation. However, ISO-1 is administered as a freely diffusible small molecule with a short half-life and limited local residence time, resulting in substantial constraints for treating infection-associated bone defects caused by osteomyelitis [24,25]. In recent years, diverse functional biomaterials have been extensively applied for the local delivery of small-molecule drugs and the regulation of bone regeneration, demonstrating considerable potential for osteomyelitis intervention [26]. β-Cyclodextrin (β-CD), characterized by a host–guest architecture with a hydrophobic cavity and a hydrophilic exterior, can enhance the solubility and stability of hydrophobic small molecules via inclusion complexation and has therefore been widely used in drug delivery and controlled release systems [27]. GelMA hydrogels offer additional advantages, including excellent biocompatibility, a porous three-dimensional network, injectability, and in situ photo-crosslinkable gelation. These properties facilitate drug retention and sustained diffusion at defect sites while providing a scaffold conducive to cell adhesion and new bone ingrowth [28].

Accordingly, the available evidence suggests that within the microenvironment of SA-associated osteomyelitis, MIF may act as a critical upstream hub linking inflammatory amplification, impaired osteogenic differentiation, and pyroptotic cell death. However, the receptors and signaling pathways through which MIF governs osteoblast fate remain unclear, and there is also a lack of local therapeutic strategies that simultaneously target this pathway while supporting bone regeneration. To address these gaps, we comprehensively evaluated bone destruction in a Mif^−/− mouse model of osteomyelitis. In parallel, using an in vitro SA infection model, we systematically assessed the effects of MIF deficiency on osteogenic differentiation and pyroptosis, and, by integrating transcriptomic sequencing with Co-immunoprecipitation (Co-IP), we identified the MIF–CD74/PI3K/AKT signaling axis as a key regulatory pathway underlying infection-associated osteoblast pyroptosis and differentiation impairment. Furthermore, we incorporated β-CD inclusion complexation and an injectable GelMA hydrogel to construct an ISO-1@β-CD/GelMA sustained-release system, enabling localized and controllable delivery of ISO-1 (Scheme 1). Ultimately, following debridement of osteomyelitis, this approach aims to remodel the local inflammatory and pyroptotic microenvironment at sites of infected bone defects, restore osteogenic function, and promote reconstruction of bone architecture.

Scheme 1.

Schematic illustration of the preparation process of ISO-1@β-CD/GelMA hydrogels and their mechanism of action in repairing infected bone defects by targeting the MIF-CD74/PI3K/AKT signaling pathway, thereby improving the inflammatory microenvironment, inhibiting pyroptosis in BMSCs under osteogenic induction, and promoting osteogenic differentiation.

2. Results

2.1. Exacerbated bone destruction in a mouse osteomyelitis model

To evaluate bone destruction in infected bone defects, a femoral osteomyelitis model was established in C57BL/6N mice by injecting SA into the femoral medullary cavity, with PBS injection serving as the sham control. After 4 weeks, the mice were euthanized and femurs were collected for systematic assessment (Fig. 1a). Micro-CT 3D reconstructions and cross-sectional images revealed marked cortical discontinuity, trabecular rarefaction, and medullary cavity expansion in the OM group compared with the sham group, indicating extensive bone destruction (Fig. 1b). Quantitative analyses showed that bone mineral density (BMD) and bone volume/tissue volume (BV/TV) were significantly reduced in the OM group, accompanied by decreased trabecular number (Tb.N), whereas trabecular pattern factor (Tb.Pf) and trabecular separation (Tb.Sp) were markedly increased. Trabecular thickness (Tb.Th) showed no significant change (P > 0.05), suggesting severe deterioration of trabecular architecture and aggravated osteoporosis-like alterations (Fig. 1c). Histological staining further corroborated these findings (Fig. 1d). H&E staining demonstrated regular, densely arranged trabeculae and intact marrow cavity architecture in the sham group, whereas the cortical bone of the OM group was largely replaced by infiltrating inflammatory cells, and the marrow cavity was filled with granulation and necrotic tissue. Masson staining revealed uniformly blue-stained collagen fibers within the bone matrix of the sham group, while the OM group exhibited extensive collagen fiber fragmentation along with fibrosis and cavity formation, indicating severe matrix destruction and fibrotic remodeling induced by infection.

Fig. 1.

Establishment and characterization of a Staphylococcus aureus (SA)-induced femoral osteomyelitis model in mice. a) Schematic illustration of the in vivo experimental design. C57BL/6N mice received intramedullary injection of S. aureus to establish osteomyelitis or PBS for sham control, and femurs were harvested at 4 weeks for micro-CT, histological staining (H&E and Masson), RT-qPCR, and immunohistochemistry (IHC) analyses. b) Representative micro-CT reconstructed images of femurs from sham and OM groups. Regions of interest (ROIs) for quantitative analysis are indicated: G1, coronal plane; G2, trabecular bone region; G3, transverse plane. c) micro-CT-based quantitative analysis of trabecular bone parameters in the defined ROIs, including bone mineral density (BMD), bone volume fraction (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), and trabecular pattern factor (Tb.Pf). d) Representative H&E and Masson's trichrome staining of femoral sections from sham and OM groups (low- and high-magnification views, Scale bar = 200 μm and 50 μm). e) RT-qPCR analysis of osteogenesis- and osteoclastogenesis-related genes in femoral tissues, including Ocn, Col1a1, Trap, Nfatc1, and c-fos. f) Representative IHC staining of RUNX2, COL1A1, OCN, and OPN in femoral sections from sham and OM groups. Scale bar = 50 μm. g) Quantification of IHC-positive areas for RUNX2, OCN, COL1A1, and OPN. h) Levels of pro-inflammatory cytokines (IL-1β, IL-6, IL-18, and TNF-α) in samples from sham and OM groups. Data are presented as mean ± SD; n = 5. ns, not significant; ∗, P< 0.05; ∗∗, P< 0.01; ∗∗∗, P< 0.001; ∗∗∗∗, P < 0.0001.

At the molecular level, RT-qPCR results showed that the mRNA expression of Ocn and Col1a1 was significantly downregulated in the OM group compared with the sham group, while the expression of osteoclast-related genes tartrate-resistant acid phosphatase (Trap) and c-fos was upregulated. Nfatc1 expression showed no notable difference, indicating suppressed osteogenic activity and enhanced osteoclastogenic activity during osteomyelitis (Fig. 1e). Consistently, immunohistochemical (IHC) staining for COL1A1, RUNX2, OCN, and osteopontin (OPN) displayed extensive and intense brown-yellow positivity in the sham group but was markedly weakened in the OM group, with only a small number of positive cells remaining. This suggests that osteoblast differentiation and matrix synthesis are impaired within the osteomyelitis microenvironment (Fig. 1f and g). Additionally, ELISA analyses revealed substantially elevated levels of multiple pro-inflammatory cytokines in bone tissue homogenates from the OM group (Fig. 1h). Compared with the sham group, the concentrations of the typical pro-inflammatory mediators IL-1β, TNF-α, IL-6, and IL-18 were all significantly increased in OM (P < 0.001), indicating that persistent inflammation induced by bacterial infection may be a key contributor to exacerbated bone destruction and impaired osteogenesis. Collectively, these results demonstrate that intramedullary SA infection successfully establishes a murine femoral osteomyelitis model, producing pronounced local bone loss, microarchitectural deterioration, reduced osteogenic function, and elevated inflammatory cytokines. This provides a robust animal foundation and strong evidence to support subsequent mechanistic investigations.

2.2. Attenuated bone destruction in femoral and tibial osteomyelitis in Mif^−/− mice

Given previous evidence demonstrating that MIF participates in the pathogenesis and progression of various inflammatory and infectious diseases, we collected infected bone tissue samples from six patients undergoing debridement for chronic osteomyelitis, along with three non-infected bone samples obtained during trauma surgery, to examine whether MIF is differentially expressed in OM bone tissues. Western blot (WB) analysis revealed that MIF protein expression was markedly elevated in infected bone tissues from osteomyelitis patients compared with the control group (Fig. 2a and Fig. S1). To further elucidate the role of MIF in osteomyelitis-associated bone destruction, we utilized the C57BL/6N-Mif^em1C knockout mouse model. PCR identification showed that the F1/R1 primer pair consistently amplified a specific 383 bp band (Fig. S2). Based on this model, we established femoral and tibial osteomyelitis in both wild-type and Mif^−/− mice, and collected samples at 4 weeks post-infection for micro-CT imaging, histological staining, and molecular analyses of gene and protein expression (Fig. 2b). Micro-CT imaging and quantitative assessments showed that compared with the sham group, wild-type OM mice exhibited pronounced cortical discontinuity and trabecular rarefaction in both the femur and tibia (Fig. 2c–d). Correspondingly, BMD, BV/TV, Tb.N, and Tb.Th were significantly reduced, whereas Tb.Sp was markedly increased. In contrast, the Mif^−/−/OM group displayed substantial improvement in these microarchitectural parameters relative to the OM group. BMD, BV/TV, Tb.N, and Tb.Th were partially restored, Tb.Sp was reduced, and cortical and trabecular structures remained well preserved (Fig. 2e). These results indicate that Mif deficiency effectively alleviates osteomyelitis-induced bone loss in both the femur and tibia.

Fig. 2.

Mif^−/− alleviates infection-induced bone destruction and inflammatory responses in vivo. a) WB analysis of MIF expression in control and clinical osteomyelitis tissues. b) Schematic of the animal experimental design. Wild-type and Mif^−/− mice received intramedullary injection of SA (OM) or PBS (sham) into the femur/tibia. Femurs and tibias were harvested at 4 weeks for micro-CT, histology (H&E), RT-qPCR, IHC, and ELISA analyses. c) Representative micro-CT reconstructed images of femurs from sham, OM, and Mif^−/−/OM groups. d) Representative micro-CT reconstructed images of tibias from sham, OM, and Mif^−/−/OM groups. e) micro-CT–based quantitative analysis of trabecular bone parameters in femur and tibia, including BMD, BV/TV, Tb.N, Tb.Th, and Tb.Sp. f) Representative H&E staining of femoral and tibial sections from sham, OM, and Mif^−/−/OM groups. Scale bar = 200 μm and 50 μm. g) Representative IHC staining of osteogenic markers (COL1A1, RUNX2, OCN, and OPN) in femoral and tibial sections from the indicated groups. Scale bar = 50 μm. h) Quantification of IHC-positive areas for COL1A1, RUNX2, OCN, and OPN in femur and tibia. i) RT-qPCR analysis of osteogenesis- and osteoclastogenesis-related genes in femoral and tibial tissues, including Ocn, Col1a1, Trap, Nfatc1, and c-fos. j) ELISA quantification of pro-inflammatory cytokines (IL-1β, IL-6, IL-18, and TNF-α) in femoral and tibial samples. Data are presented as mean ± SD; n = 5. ns, not significant; ∗, P< 0.05; ∗∗, P< 0.01; ∗∗∗, P< 0.001; ∗∗∗∗, P < 0.0001.

H&E staining further confirmed this trend. In the wild-type OM group, the femoral and tibial cortices were heavily infiltrated by inflammatory cells, accompanied by areas of necrotic bone. In contrast, inflammatory infiltration was markedly reduced in the Mif^−/−/OM group, in which the remaining trabeculae appeared more intact, and a certain amount of newly formed bone was observed (Fig. 2f). Regarding osteogenic indicators, the protein levels of COL1A1, RUNX2, OCN, and OPN were substantially diminished in the femur and tibia of the wild-type OM group compared with the sham group, whereas these markers were markedly enhanced in the Mif^−/−/OM group (Fig. 2g–h). RT-qPCR analysis demonstrated that Ocn and Col1a1 were significantly downregulated in the femurs of the OM group, while osteoclast-related genes Trap, Nfatc1, and c-fos were significantly upregulated. In the Mif^−/−/OM group, these expression changes were largely reversed, showing significant upregulation of osteogenic genes and downregulation of osteoclastic genes. In the tibial model, Mif deletion robustly reversed the OM-induced reduction in Ocn and Col1a1 mRNA expression and suppressed the elevation of Trap mRNA levels (Fig. 2i). These findings indicate that Mif deficiency ameliorates the suppression of osteoblast differentiation under osteomyelitis conditions. ELISA analysis of inflammatory cytokines in femoral and tibial tissue homogenates revealed that levels of IL-1β, IL-6, IL-18, and TNF-α were markedly elevated in the wild-type OM group, whereas these cytokines were significantly reduced in the Mif^−/−/OM group (Fig. 2j), demonstrating that Mif deficiency effectively alleviates the highly inflammatory microenvironment associated with osteomyelitis. Taken together, Mif knockout significantly mitigates SA-induced bone destruction in the femur and tibia, partially restores osteogenic function, and suppresses inflammatory responses.

2.3. SA infection inhibits BMSC proliferation and osteogenic differentiation and promotes pyroptosis

To simulate the osteomyelitic microenvironment in vitro, BMSCs were first exposed to different concentrations of SA (0.5, 1, 10, and 50 MOI), and cell viability at various time points was assessed using the CCK-8 assay. The results showed a dose- and time-dependent decrease in cell viability with increasing SA concentration and prolonged exposure. Notably, 1.0 MOI SA induced a clear but non-lethal reduction in cell viability at approximately 48 h, followed by partial recovery at 72 h and 96 h (Fig. 3a). Based on these findings, 1.0 MOI was selected for subsequent osteogenic induction experiments to better mimic a persistent infection environment. BMSCs were then divided into three groups: Control, Osteogenic induction (Osteo), and Osteogenic induction under SA infection (Osteo + SA). EdU proliferation assay showed that the proportion of EdU-positive cells increased over time (24, 48, 72 h) in the Osteo group and remained comparable to the Control group. In contrast, the Osteo + SA group exhibited a significantly reduced number of EdU-positive cells (P < 0.05, Fig. 3b and S3).

Fig. 3.

SA impairs BMSC proliferation and osteogenic differentiation and activates inflammasome-associated pyroptosis in vitro. a) Cell viability of BMSCs after exposure to increasing concentrations of SA (0.5, 1.0, 10.0, and 50.0 MOI) for the indicated times. b) Representative EdU incorporation staining of BMSCs in the control, Osteo, and Osteo + SA groups at 24, 48, and 72 h. c) Quantification of alkaline phosphatase (ALP) activity in the indicated groups during osteogenic differentiation (Day 1, Day 3, Day 5, and Day 7). d) Representative Alizarin Red S (ARS) staining (Day 21) and corresponding quantification of mineralized nodule formation in the indicated groups. e) RT-qPCR analysis of osteogenesis-related genes (Runx2, Col1a1, Ocn, and Opn) in Control, Osteo, and Osteo + SA groups. f) Representative WB and densitometric quantification of osteogenic proteins (RUNX2, COL1A1, OCN, and OPN). g) WB analysis and quantification of pyroptosis-associated proteins (NLRP3, ASC, and cleaved GSDMD-NT) in the indicated groups. h) Representative flow cytometry plots and quantification of apoptotic cells in control, Osteo, and Osteo + SA groups. Data are presented as mean ± SD; n = 3. ns, not significant; ∗, P< 0.05; ∗∗, P< 0.01; ∗∗∗, P< 0.001; ∗∗∗∗, P < 0.0001. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

We further investigated the potential effects of SA on the osteogenic differentiation of BMSCs. RT-qPCR analysis showed that osteogenic induction markedly increased Alp mRNA expression, whereas SA infection significantly suppressed Alp mRNA at days 3, 5, and 7 (Fig. 3c), indicating that SA inhibits early-stage osteogenic differentiation. Alizarin Red S (ARS) staining demonstrated that after 21 days of osteogenic induction, the Osteo group formed abundant, densely distributed mineralized nodules, whereas the Osteo + SA group exhibited a pronounced reduction in mineralized nodule formation, with significantly decreased staining area (P = 0.0006, Fig. 3d). Further assessment of osteogenesis-related genes revealed that Runx2, Col1a1, Ocn, and Opn mRNA levels were markedly upregulated in the Osteo group, whereas SA infection significantly downregulated their expression, returning them close to control levels (Fig. 3e). Consistent with the mRNA expression patterns, WB results demonstrated robust increases in RUNX2, COL1A1, OCN, and OPN protein levels in the Osteo group. In contrast, protein levels of RUNX2, COL1A1, and OCN were significantly reduced in the Osteo + SA group, indicating that SA infection profoundly impairs the osteogenic differentiation capacity of BMSCs (Fig. 3f).

In recent years, an increasing number of programmed cell death pathways have been shown to play key roles in the pathogenesis of osteomyelitis, and SA-induced pyroptosis has drawn particular attention [29,30]. To evaluate the impact of SA on pyroptosis in BMSCs, we examined the expression of classical pyroptosis-related proteins NLRP3, ASC, and the cleaved form of gasdermin D (GSDMD-NT). Compared with the Control and Osteo groups, the Osteo + SA group exhibited significantly elevated levels of NLRP3, ASC, and GSDMD-NT (Fig. 3g), indicating SA-induced activation of the inflammasome and enhanced GSDMD-mediated membrane pore formation. Flow cytometry further revealed a substantial increase in the proportion of Annexin V/PI double-positive cells in the Osteo + SA group, with a higher cell death rate than in the Control and Osteo groups (Fig. 3h). Taken together, these findings demonstrate that SA infection in vitro inhibits BMSC proliferation and osteogenic differentiation, and promotes osteoblast pyroptosis through activation of the canonical NLRP3 inflammasome/GSDMD pathway.

2.4. siMIF improves SA-mediated inhibition of BMSC osteogenic differentiation and promotion of pyroptosis

Given the crucial role of Mif deficiency in reversing bone destruction in murine femoral and tibial osteomyelitis models, we next sought to clarify the regulatory function of MIF in SA-mediated impairment of osteogenic differentiation and induction of pyroptosis in BMSCs. To this end, MIF expression in BMSCs was knocked down using siRNA. CCK-8 assays showed that SA exposure significantly reduced cell proliferation, whereas knockdown of MIF markedly restored cell viability compared with the SA group (Fig. 4a). Consistent with this observation, EdU proliferation assays and fluorescence quantification demonstrated that siMIF partially rescued the proliferative capacity of BMSCs under SA infection at 72 h (Fig. 4b and S4).

Fig. 4.

siMIF partially restores osteogenic differentiation of BMSCs under SA stimulation and attenuates inflammasome-associated pyroptosis and inflammatory cytokine release. a) Quantification of BMSC proliferation at 72 h in the indicated groups. b) Representative EdU incorporation staining of BMSCs (72 h) in control, Osteo, Osteo + siNC + SA, and Osteo + siMIF + SA groups. c) RT-qPCR analysis of osteogenesis-related genes (Runx2, Col1a1, Ocn, and Opn) and ALP expression (Day 7) in the indicated groups. d) Representative ARS staining (Day 21) showing mineralized nodule formation after osteogenic induction under the indicated treatments. e) ELISA quantification of pro-inflammatory cytokines (TNF-α, IL-1β, and IL-18) in culture supernatants from the indicated groups. f) Representative WB and densitometric quantification of RUNX2, COL1A1, OCN, and OPN. g) Representative flow cytometry plots and quantification of apoptotic cells in the indicated groups. h) WB analysis and quantification of NLRP3, ASC, and cleaved GSDMD-NT in the indicated groups. Data are presented as mean ± SD; n = 3. ∗, P< 0.05; ∗∗, P< 0.01; ∗∗∗, P< 0.001; ∗∗∗∗, P < 0.0001.

We further examined the regulatory role of siMIF on osteogenic differentiation under SA stimulation. Analysis of osteogenesis-related gene expression revealed that, compared with the Osteo group, SA markedly downregulated Runx2, Col1a1, Ocn, and Alp mRNA levels. However, Mif knockdown significantly restored the expression of these osteogenic markers (Fig. 4c). ARS staining showed that the Osteo group formed abundant and dense mineralized nodules, whereas the SA-treated group exhibited a substantial reduction in both nodule number and staining area. Notably, the siMIF + SA group demonstrated a significant recovery of mineralized nodule formation (Fig. 4d and S5), indicating that siMIF effectively reverses SA-induced impairment of BMSC mineralization capacity. WB analysis further confirmed that SA markedly suppressed RUNX2, COL1A1, OCN, and OPN protein expression, whereas siMIF treatment significantly upregulated all four proteins (Fig. 4f), consistent with the mRNA expression patterns.

Regarding regulation of the inflammatory microenvironment, ELISA results showed that TNF-α, IL-1β, and IL-18 levels in the cell supernatant were markedly elevated in the SA group, whereas siMIF treatment significantly reduced the concentrations of these pro-inflammatory cytokines (Fig. 4e), indicating that Mif knockdown helps alleviate SA-induced inflammatory responses. To further explore the potential role of siMIF from the perspective of pyroptosis, flow cytometry analysis revealed that the proportion of Annexin V/PI-positive cells was substantially higher in the SA group than in the Control and Osteo groups, while Mif knockdown markedly reduced the cell death rate (Fig. 4g). Moreover, SA stimulation strongly increased the expression of NLRP3, ASC, and GSDMD-NT, whereas siMIF treatment significantly suppressed the upregulation of these pyroptosis-related proteins (Fig. 4h), suggesting that siMIF attenuates activation of the NLRP3 inflammasome and GSDMD-mediated membrane pore formation. Taken together, Mif knockdown not only partially restores the osteogenic differentiation capacity of BMSCs under SA infection, but also reduces osteoblast pyroptosis by inhibiting the NLRP3/ASC/GSDMD pathway. These findings further demonstrate that MIF plays a key pro-pathological role in infection-associated osteogenic dysfunction and pyroptotic cell death.

2.5. Transcriptome sequencing (RNA-seq) and Co-IP reveal the interaction between MIF and its receptor CD74

To further elucidate the downstream receptors and signaling pathways through which MIF mediates infection-related osteoblastic damage, we performed transcriptome sequencing on three groups: untreated BMSCs (BMSC), SA-treated BMSCs (BMSC + SA), and SA-treated BMSCs with Mif knockdown (siMIF/BMSC + SA). Cluster heatmaps showed tight clustering within each group and clear segregation between groups with different treatments, indicating strong reproducibility and reliability of the transcriptomic data (Fig. 5a). The volcano plots illustrated the significantly upregulated and downregulated genes between each comparison group (Fig. 5b). In the BMSC vs. BMSC + SA comparison, a total of 1016 upregulated genes and 826 downregulated genes were identified. In the BMSC + SA vs. siMIF/BMSC + SA comparison, 795 genes were upregulated and 949 genes were downregulated.

Fig. 5.

Transcriptomic profiling and Co-immunoprecipitation (Co-IP) identifies MIF-CD74 as a key axis involved in SA-induced osteogenic dysfunction in BMSCs. a) Heatmap showing hierarchical clustering of differentially expressed genes among BMSCs, BMSC + SA, and siMIF/BMSC + SA (n = 3 per group). b) Volcano plots of differentially expressed genes (DEGs) comparing BMSC vs BMSC + SA (left) and BMSC + SA vs siMIF/BMSC + SA (right), and significantly upregulated and downregulated genes are highlighted. c) Gene Ontology (GO) enrichment analysis of DEGs, showing the most enriched terms across biological process, cellular component, and molecular function categories. d) KEGG pathway enrichment analysis of DEGs comparing BMSC vs BMSC + SA (top) and BMSC + SA vs siMIF/BMSC + SA (bottom); numbers of upregulated and downregulated genes enriched in each pathway are indicated. e) Increased Mif expression in BMSC + SA compared with BMSCs (left), and reduced Cd74 mRNA level after siMIF under SA stimulation (right). f) WB analysis confirming the expression of CD74 in BMSCs transfected with siMIF compared with SA group. g) Co-IP assay showing the physical association between MIF and CD74 in BMSCs (anti-MIF IP versus IgG control). Data are presented as mean ± SD; n = 3. ∗, P< 0.05; ∗∗∗, P < 0.001.

GO enrichment analysis further showed that differentially expressed genes in the BMSC + SA group were mainly enriched in biological processes such as response to bacterium, cellular response to lipopolysaccharide, and angiogenesis; in cellular components associated with the extracellular space, plasma membrane, and extracellular region; and in molecular functions such as integrin binding and growth factor activity (Fig. 5c). These results imply that SA infection may activate membrane receptors on osteogenic cells and triggers inflammatory signaling. KEGG pathway analysis revealed that, compared with the BMSC group, differentially expressed genes in the BMSC + SA group were significantly enriched in classical inflammatory pathways, including PI3K-Akt signaling, SA infection, and cytokine-cytokine receptor interaction (Fig. 5d). Importantly, after siMIF treatment, the PI3K-Akt pathway remained significantly enriched, suggesting that MIF may mediate SA-induced osteoblast dysfunction and pyroptosis partly through regulation of downstream pathways such as PI3K/AKT.

To further identify potential downstream receptors, we focused on CD74, a receptor molecule that showed strong correlation with MIF in the transcriptomic dataset. Quantitative transcriptome analysis revealed that SA treatment markedly upregulated Mif expression in BMSCs, whereas Cd74 transcript levels were significantly reduced following siMIF intervention (P < 0.05, Fig. 5e), suggesting that MIF may positively regulate CD74 expression. WB validation further demonstrated that siMIF transfection substantially decreased intracellular CD74 protein levels compared with the SA-induced group (Fig. 5f and S6), confirming a positive relationship between the expression of MIF and CD74. Finally, to determine whether a direct physical interaction exists between MIF and CD74, we performed Co-IP. The results showed that, upon immunoprecipitation with an anti-MIF antibody, a clear CD74 band was detected in the precipitated complex, whereas no specific signal was observed in the IgG control under the same experimental conditions (Fig. 5g). This finding indicates that MIF can form a complex with CD74. Together, RNA-seq and Co-IP analyses demonstrate that CD74 serves as a key receptor for MIF in the context of SA infection and may mediate SA-induced impairment of osteogenic differentiation and pyroptosis through regulation of the PI3K/AKT signaling pathway. These findings provide a mechanistic basis for subsequent functional studies targeting CD74 using siCD74 interventions. To further substantiate that the MIF-CD74 axis modulates PI3K/AKT signaling in SA-induced osteogenic cells, WB analysis was performed to detect total and phosphorylated PI3K and AKT in the Osteo, Osteo + siNC + SA, Osteo + siMIF + SA, and Osteo + siCD74+SA groups (Fig. S7). Compared with the Osteo group, SA treatment markedly reduced p-AKT/AKT ratios. Notably, siMIF or siCD74 significantly reversed this inhibition and significantly elevated p-PI3K/PI3K and p-AKT/AKT ratios compared with the Osteo + siNC + SA group. These data indicate that SA suppresses PI3K/AKT pathway activation during osteogenic induction, whereas MIF or CD74 knockdown restores PI3K/AKT phosphorylation, supporting PI3K/AKT as a key downstream signaling node of MIF-CD74 in infection-associated osteogenic dysfunction.

2.6. siCD74 improves SA-mediated inhibition of BMSC osteogenic differentiation and promotion of pyroptosis

The transcriptomic and Co-IP findings indicated that MIF may exert its effects through its receptor CD74, we further knocked down Cd74 expression in BMSCs using siCD74. CCK-8 assays showed that SA treatment markedly reduced cell proliferation, whereas Cd74 knockdown significantly increased proliferative activity compared with the SA group (Fig. 6a). EdU proliferation assays at 72 h revealed a pronounced reduction in EdU-positive cells in the SA group, while siCD74 treatment restored the number of EdU-positive cells (P = 0.0332, Fig. 6b and S8), indicating that Cd74 knockdown helps alleviate SA-induced suppression of BMSC proliferation.

Fig. 6.

siCD74 rescues osteogenic differentiation of BMSCs under SA stimulation and mitigates inflammasome-associated pyroptosis and inflammatory cytokine release. a) Quantification of BMSC proliferation in the indicated groups. b) Representative EdU incorporation staining of BMSCs (72 h) in control, Osteo, Osteo + siNC + SA, and Osteo + siCD74+SA groups. c) Quantification of ALP expression during osteogenic differentiation (Day 1, Day 3, Day 5, and Day 7) under the indicated treatments. d) Representative ARS staining (Day 21) showing mineralized nodule formation in the indicated groups. e) RT-qPCR analysis of Runx2, Opn, Ocn, and Col1a1 in the indicated groups. f) Representative WB and densitometric quantification of RUNX2, OCN, OPN, and COL1A1. g) ELISA quantification of pro-inflammatory cytokines (TNF-α, IL-1β, and IL-18) in culture supernatants from the indicated groups. h) WB analysis and quantification of pyroptosis-associated proteins (NLRP3, ASC, and cleaved GSDMD-NT) in the indicated groups. i) Representative flow cytometry plots showing apoptotic cells in the indicated groups. Data are presented as mean ± SD; n = 3. ns, not significant; ∗, P< 0.05; ∗∗, P< 0.01; ∗∗∗, P< 0.001; ∗∗∗∗, P < 0.0001.

The assessment of osteogenic differentiation showed that Cd74 knockdown significantly increased Alp mRNA at all examined time points (days 1, 3, 5, and 7) compared with the SA group (P = 0.0332, Fig. 6c). While the Osteo group formed abundant and dense mineralized nodules, the SA group exhibited a marked reduction in both the number and area of mineralized nodules. In contrast, the siCD74 group displayed a substantial increase in mineralized nodule formation and staining area (P = 0.0056, Fig. 6d and S9). RT-qPCR analysis showed that SA infection significantly downregulated the mRNA levels of Runx2, Ocn, Opn, and Col1a1, whereas Cd74 knockdown markedly restored their expression (Fig. 6e). Consistent with the mRNA trends, WB results demonstrated that SA markedly suppressed the protein levels of RUNX2, OCN, OPN, and COL1A1, while siCD74 treatment significantly upregulated all four proteins (Fig. 6f), indicating that siCD74 effectively reverses SA-mediated inhibition of osteogenic differentiation.

To evaluate the effects of CD74 on inflammation and pyroptosis, we examined changes in pro-inflammatory cytokines in the cell supernatant as well as pyroptosis-related protein expression. ELISA results showed that SA infection markedly increased TNF-α, IL-1β, and IL-18 levels, whereas these cytokines were significantly reduced in the siCD74 group (Fig. 6g), indicating that Cd74 knockdown attenuates SA-induced pro-inflammatory cytokine secretion. Additionally, SA markedly upregulated the protein levels of NLRP3, ASC, and GSDMD-NT, while Cd74 knockdown significantly decreased their expression (Fig. 6h), confirming that CD74 participates in activating the NLRP3 inflammasome and the GSDMD-dependent pyroptotic process. Flow cytometry further demonstrated that the proportion of Annexin V/PI-positive cells was substantially increased in the SA group, whereas the siCD74 group showed a pronounced reduction in cell death rate (Fig. 6i and S10). Therefore, Cd74 knockdown under SA infection conditions restores the osteogenic differentiation capacity of BMSCs and alleviates osteoblast pyroptosis by suppressing activation of the NLRP3/ASC/GSDMD axis. These findings functionally validate the critical role of the MIF-CD74 signaling pathway in infection-associated osteogenic differentiation impairment and pyroptotic cell death.

2.7. The PI3K/AKT signaling pathway mediates MIF-CD74 regulation of osteoblast differentiation and pyroptosis under SA stimulation

Transcriptomic analysis indicated that PI3K/AKT may serve as a key downstream signaling pathway of the MIF-CD74 axis. To verify its functional role, BMSCs under SA infection were transfected with siMIF, followed by either Cd74 overexpression (oe-CD74) or treatment with the PI3K-specific inhibitor LY294002. CCK-8 assays showed that siMIF significantly enhanced cell proliferation under SA stimulation, whereas both oe-CD74 and LY294002 markedly attenuated this proliferative recovery (Fig. 7a). Consistently, EdU proliferation assays demonstrated that siMIF substantially increased the proportion of EdU-positive cells in the presence of SA, while oe-CD74 or PI3K inhibition reduced EdU positivity at 72 h (Fig. 7b and S11). These findings suggest that the PI3K/AKT pathway participates in MIF/CD74-mediated regulation of osteogenic cell proliferation under SA-induced stress.

Fig. 7.

MIF regulates SA-induced osteogenic dysfunction and inflammatory/pyroptotic responses through the CD74/PI3K/AKT axis in BMSCs. a) Quantification of BMSC proliferation under osteogenic induction with the indicated treatments: Osteo, Osteo + siNC + SA, Osteo + siMIF + SA, Osteo + siMIF + oe-CD74+SA, and Osteo + siMIF + LY294002+SA. b) Representative EdU incorporation staining of BMSCs at 72 h in the indicated groups. c) Time-course analysis of ALP expression during osteogenic differentiation (Day 1, Day 3, Day 5, and Day 7) under the indicated treatments. d) Representative ARS staining (Day 21) showing mineralized nodule formation in the indicated groups. e) ELISA quantification of pro-inflammatory cytokines (TNF-α, IL-1β, and IL-18) in culture supernatants from the indicated groups. f) RT-qPCR analysis of Runx2, Opn, Ocn, and Col1a1 in the indicated groups. g) Representative WB and densitometric quantification of RUNX2, OCN, OPN, and COL1A1. h) WB analysis and quantification of pyroptosis-associated proteins (NLRP3, ASC, and cleaved GSDMD-NT) in the indicated groups. i) Representative flow cytometry plots showing apoptotic cells in the indicated groups. Data are presented as mean ± SD; n = 3. ns, not significant; ∗, P< 0.05; ∗∗, P< 0.01; ∗∗∗, P< 0.001; ∗∗∗∗, P < 0.0001.

Regarding the regulation of osteogenic differentiation, RT-qPCR results showed that compared with the siMIF group, oe-CD74 downregulated Alp mRNA expression on days 1 and 7, and LY294002 treatment caused a pronounced reduction in Alp expression at all examined time points (Fig. 7c). ARS staining at day 21 further demonstrated that siMIF markedly increased the area of mineralized nodules, while oe-CD74 or PI3K inhibition weakened the restorative effect of siMIF on mineralization capacity (Fig. 7d and S12). Analysis of osteogenesis-related genes revealed that under SA stimulation, siMIF significantly upregulated Runx2, Ocn, Opn, and Col1a1 mRNA levels. However, oe-CD74 notably reduced Opn and Col1a1 expression, and LY294002 treatment resulted in significant downregulation of Runx2, Ocn, Opn, and Col1a1 mRNA levels (Fig. 7f). WB validation showed that oe-CD74 markedly decreased OCN, OPN, and COL1A1 protein levels, while LY294002 treatment significantly reduced RUNX2, OCN, OPN, and COL1A1 protein expression (Fig. 7g). These findings indicate that the enhancement of osteogenic differentiation induced by Mif knockdown in the SA-mediated inflammatory microenvironment depends on the integrity of the CD74-PI3K/AKT signaling pathway.

Furthermore, inflammation- and pyroptosis-related assays provided additional support for this result. ELISA measurements showed that siMIF significantly reduced SA-induced secretion of TNF-α, IL-1β, and IL-18, whereas LY294002 treatment markedly increased the levels of these pro-inflammatory cytokines, and oe-CD74 significantly enhanced IL-1β and IL-18 expression (Fig. 7e). WB analysis of pyroptosis-related proteins revealed that siMIF downregulated SA-induced NLRP3, ASC, and GSDMD-NT expression, while LY294002 treatment led to a pronounced upregulation of these pyroptosis markers. In the oe-CD74 group, NLRP3 and ASC levels were significantly elevated (Fig. 7h). Flow cytometry further demonstrated that siMIF markedly reduced the proportion of Annexin V/PI-positive cells, whereas oe-CD74 or PI3K inhibition significantly increased cell death, partially counteracting the protective effects of siMIF (Fig. 7i and S13). Taken together, these findings indicate that the PI3K/AKT signaling pathway serves as a critical downstream mediator of MIF-CD74 regulation of osteogenic differentiation and pyroptosis under SA infection. Mif knockdown suppresses CD74 expression, thereby activating PI3K/AKT signaling, enhancing osteogenic differentiation, and attenuating NLRP3-ASC/GSDMD-mediated osteoblast pyroptosis.

2.8. Preparation and physicochemical characterization of ISO-1@β-CD/GelMA hydrogel

To achieve localized sustained delivery of ISO-1, we first formed a host–guest inclusion complex between β-CD and ISO-1, which was subsequently mixed with GelMA prepolymer solution containing a photoinitiator. The mixture was crosslinked under 405 nm light irradiation to obtain the ISO-1@β-CD/GelMA hydrogel. The preparation process is illustrated in Fig. 8a. Both GelMA and ISO-1@β-CD/GelMA hydrogels underwent rapid gelation within 15 s under 405 nm light exposure (Fig. 8b). The injectability and photo-induced gelation behavior of the hydrogels were further evaluated. The precursor solution remained fluid at room temperature and rapidly formed a stable gel upon brief light exposure. Moreover, it could be smoothly extruded through a fine needle, indicating excellent injectability and in situ gelation capability. These properties make the system suitable for rapid shaping and filling of irregular infected bone defects. Scanning electron microscopy (SEM) showed that both pure GelMA and ISO-1@β-CD/GelMA hydrogels exhibited a well-defined three-dimensional porous network with uniformly distributed and highly interconnected pores, demonstrating that incorporation of ISO-1@β-CD did not disrupt the intrinsic porous architecture of GelMA (Fig. 8c). Energy-dispersive X-ray spectroscopy (EDS) mapping and elemental distribution analysis indicated homogeneous distribution of C, N, and O across the hydrogel cross-sections. In the control group, the mass fractions of C, O, and N were approximately 46.0%, 34.3%, and 19.7%, respectively. In the ISO-1@β-CD/GelMA group, the corresponding values were about 45.3%, 37.3%, and 17.4%. The overall elemental compositions were comparable, suggesting that the inclusion complex was uniformly dispersed within the GelMA network without introducing impurities. The slightly increased O/C ratio and decreased N content in the experimental group are likely attributable to the incorporation of β-CD, which lacks nitrogen (Figs. S14 and S15).

Fig. 8.

Fabrication and physicochemical characterization of the ISO-1 sustained-release hydrogel (ISO-1@β-CD/GelMA). a) Schematic illustration of hydrogel preparation. ISO-1 was encapsulated into β-cyclodextrin (β-CD) to form the inclusion complex (ISO-1@β-CD), which was subsequently incorporated into GelMA precursor and photo-crosslinked to obtain ISO-1@β-CD/GelMA hydrogel. b) Representative photographs showing the sol–gel transition of GelMA and ISO-1@β-CD/GelMA before and after 405 nm light irradiation, and the injectability of the precursor solution through a syringe/needle. c) Representative SEM and elemental mapping images of GelMA and ISO-1@β-CD/GelMA hydrogels. d) FTIR spectra of GelMA and ISO-1@β-CD/GelMA. e) XPS survey spectrum of ISO-1@β-CD/GelMA. High-resolution XPS spectra of (f) C 1s, (g) N 1s, and (h) O 1s. i) Cumulative release profile of ISO-1 from ISO-1@β-CD/GelMA under different pH conditions (PBS, pH 7.4; acetate buffer, pH 6.5). j) Strain sweep showing storage modulus (G′) and loss modulus (G″). k) Frequency sweep of G′ and G″. l) Time-dependent evolution of G′ and G″.

In terms of chemical structural characterization, FTIR spectra of both hydrogels exhibited typical GelMA features, including the broad O–H stretching band, the C-H stretching peak (2951 cm⁻¹), and characteristic amide I and amide II bands (N-H bending vibration, 1540 cm⁻¹). In the ISO-1@β-CD/GelMA hydrogel, slight changes in peak intensities were observed at the O-H and C=C stretching regions (Fig. 8d). The minor attenuation of the C=C stretching signal may result from the encapsulation of ISO-1 within the β-CD cavity, whereas the enhanced O–H broad peak is likely attributable to the introduction of β-CD polysaccharide units. To further analyze chemical states, XPS measurements were performed. The survey spectrum revealed that the material was primarily composed of C, N, and O, with no detectable impurity peaks, indicating a simple and high-purity composition (Fig. 8e). High-resolution C 1s spectra displayed two prominent peaks at 284.8 eV and 287.9 eV, corresponding to the GelMA/β-CD backbone and the aromatic structures of ISO-1, respectively (Fig. 8f). The N 1s peak around 399.5 eV was attributed to -CONH- groups, reflecting the presence of the GelMA backbone and the pyrrolidone ring of ISO-1 (Fig. 8g). Deconvolution of the O 1s spectrum revealed two signals at approximately 531.0 eV (C=O) and 532.2 eV (C-O), consistent with the abundant carbonyl, hydroxyl, and ether bonds present in GelMA, β-CD, and ISO-1 (Fig. 8h). These findings collectively indicate that the ISO-1@β-CD inclusion complex is successfully integrated into the GelMA network and remains on the hydrogel surface in the form of stable organic functional groups.

2.9. Sustained-release performance and degradation behavior of ISO-1@β-CD/GelMA hydrogel

Importantly, we employed a multilevel physical-barrier strategy to achieve controlled, sustained release of ISO-1. GelMA serves as a macroscopic slow-release matrix, while β-CD functions as a molecular-scale drug capsule, and their combination produces a synergistic release-regulating effect. ISO-1 release profiling showed that ISO-1@β-CD/GelMA exhibited a slow and continuous release pattern in both neutral PBS (pH 7.4) and mildly acidic acetate buffer (pH 6.5). Notably, under weakly acidic conditions, the cumulative release was slightly higher and the release rate marginally faster, reaching 63.7% cumulative release at 14 days (Fig. 8i). This suggests that the hydrogel can achieve more efficient drug release in infection-associated acidic microenvironments, supporting precise local therapy for osteomyelitis. We further evaluated the physicochemical properties of the hydrogels. Swelling tests revealed that both hydrogels rapidly absorbed water and expanded within a short time, reaching equilibrium after approximately 8-12 h. The equilibrium swelling ratios for the ISO-1-loaded group and the GelMA group stabilized at 408 % and 477 %, respectively (Fig. S16), indicating that ISO-1 loading did not significantly alter hydrogel hydrophilicity. In vitro degradation assays in PBS showed a gradual mass reduction in both hydrogels over time. However, GelMA degraded more rapidly, retaining only about 35% of its initial mass by day 28, whereas ISO-1@β-CD/GelMA maintained over 45% of its original mass (Fig. S17). This suggests that incorporation of the inclusion complex enhances network stability, providing prolonged mechanical integrity and structural persistence that can support tissue regeneration in bone defects.

Additionally, the rheological properties of the hydrogels were further evaluated. In the strain amplitude sweep test (Fig. 8j), the storage modulus G′ of both hydrogels was markedly higher than the loss modulus G″ across a broad strain range and remained relatively stable. The hydrogels were within the linear viscoelastic region and exhibited the characteristic behavior of elastic solids. The ISO-1@β-CD/GelMA hydrogel displayed a consistently higher G′ than GelMA, suggesting a denser and more stable network structure. The critical strain was approximately eighty-five percent, beyond which the hydrogel network ruptured and transitioned into a sol-like state. Frequency sweep results (Fig. 8k) showed that within the tested frequency range, G′ remained much higher than G″ and showed little sensitivity to frequency changes between 0.1 and 10 Hz. This demonstrates that both hydrogels possessed favorable solid-like viscoelasticity. The ISO-1@β-CD/GelMA group again maintained higher modulus values throughout the test. Time sweep analysis (Fig. 8l) further revealed that under constant strain and frequency, the G′ and G″ values of both hydrogels remained essentially stable over time. The ISO-1@β-CD/GelMA hydrogel maintained a consistently higher and stable G′, indicating superior mechanical performance and structural stability capable of maintaining scaffold strength during prolonged in vivo application. In summary, the injectable photo-crosslinked ISO-1@β-CD/GelMA hydrogel exhibited a uniform porous structure, robust mechanical stability, controlled drug release behavior, and suitable swelling and degradation characteristics. These properties establish a strong material foundation for its potential in vivo application for the repair of infected bone defects.

2.10. In vivo tissue repair effects of ISO-1@β-CD/GelMA hydrogels

To evaluate the in vivo therapeutic efficacy of the ISO-1 sustained-release hydrogel in repairing infected bone defects, a rat femoral osteomyelitis model was established by intramedullary inoculation of S. aureus, followed by complete debridement to generate a standardized infected bone defect. To further clarify the therapeutic contribution of ISO-1 itself and the added value of the sustained-release formulation, animals were randomly assigned into four groups: OM, GelMA, single ISO-1 (10 mg/kg), and ISO-1@β-CD/GelMA. Specimens were harvested at 4 and 8 weeks post-treatment for radiological and histological analyses (Fig. 9a).

Fig. 9.

ISO-1@β-CD/GelMA promotes osseointegration in infected bone defect models after debridement. a) Schematic timeline of the in vivo study. Femoral osteomyelitis was induced by SA injection. After 28 days, complete debridement was performed to create an infected bone defect (Day 0), followed by implantation of GelMA, single ISO-1 or ISO-1@β-CD/GelMA hydrogel. Samples were collected at 4 and 8 weeks post-implantation for micro-CT and histological evaluation, including inflammatory and osseointegration assessments. b) Representative micro-CT 3D reconstructions and corresponding transverse/coronal views of femurs from OM, GelMA, single ISO-1, and ISO-1@β-CD/GelMA groups at 4 and 8 weeks. Red dashed circles indicate the defect region. c) Micro-CT-based quantitative analysis of trabecular bone parameters in the defect region at 4 and 8 weeks, including BV/TV, Tb.Th, Tb.N, and relative bone defect diameter. Scale bar = 5 mm. d) Representative H&E staining of femoral sections at 4 and 8 weeks. Boxed regions indicate the areas shown at higher magnification. Scale bars = 200 μm and 50 μm. Data are presented as mean ± SD; n = 5. ∗∗, P< 0.01; ∗∗∗∗, P < 0.0001. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Micro-CT 3D reconstruction and cross-sectional/coronal images revealed distinct differences in defect repair among the four groups (Fig. 9b). At 4 weeks, the OM group exhibited an obvious residual defect with irregular margins and limited new bone formation. The GelMA group showed only mild trabecular infiltration at the defect boundary. Notably, the single ISO-1 group displayed increased new bone formation compared with the OM and GelMA groups, indicating that ISO-1 alone exerted a partial osteoprotective effect. In contrast, the ISO-1@β-CD/GelMA group demonstrated more substantial trabecular extension toward the defect center, with more continuous bridging structures and improved defect filling. At 8 weeks, these differences became more pronounced. The OM group still presented a large defect with persistent cortical discontinuity, while the GelMA group exhibited partial bone regeneration but incomplete restoration of bone continuity. The single ISO-1 group showed improved repair compared with OM and GelMA; however, the defect remained incompletely filled. Importantly, the ISO-1@β-CD/GelMA group exhibited near-complete reconstruction of cortical continuity and dense trabecular architecture, suggesting superior long-term osteogenic regeneration.

Quantitative micro-CT analysis further supported these observations (Fig. 9c). At both 4 and 8 weeks, BV/TV, Tb.Th, and Tb.N were significantly higher in the ISO-1@β-CD/GelMA group compared with the OM and GelMA groups. Although free ISO-1 significantly improved BV/TV and trabecular parameters relative to OM, the ISO-1@β-CD/GelMA group consistently achieved the highest bone regeneration indices. Additionally, compared with the single ISO-1 treatment, the ISO-1@β-CD/GelMA group exhibited a significantly reduced relative defect diameter, particularly at 8 weeks, indicating that the sustained-release ISO-1 hydrogel provided additional benefits in promoting bone mass restoration and trabecular microarchitecture reconstruction in infected bone defects.

H&E staining revealed extensive fibrous tissue hyperplasia and marked inflammatory cell infiltration in the OM group at 4 and 8 weeks. The GelMA group showed abundant fibrotic tissue and inflammatory infiltration. In the single ISO-1 group, inflammatory infiltration was partially reduced and trabecular bone formation was moderately increased. In contrast, the ISO-1@β-CD/GelMA group exhibited substantial newly formed bone filling the bone defect, with a more organized trabecular arrangement, accompanied by markedly decreased inflammatory cell infiltration (Fig. 9d). Masson's trichrome staining further confirmed enhanced matrix remodeling in the ISO-1 sustained-release group. At 4 weeks, the OM and GelMA groups displayed disorganized collagen deposition and extensive fibrotic tissue occupying the defect area. Single ISO-1 treatment promoted collagen deposition and trabecular-like tissue regeneration to a certain extent. Notably, the ISO-1@β-CD/GelMA group demonstrated more extensive collagen-rich matrix formation. At 8 weeks, the ISO-1@β-CD/GelMA group exhibited the most mature and continuous matrix deposition. The single ISO-1 group exhibited intermediate improvement, indicating that sustained ISO-1 delivery facilitates accelerated bone matrix maturation and remodeling (Fig. 10a).

Fig. 10.

ISO-1@β-CD/GelMA enhances matrix remodeling and attenuates osteoclast activity and local inflammation in infected bone defects. a) Representative Masson's trichrome staining of femoral sections at 4 and 8 weeks post-implantation in OM, GelMA, single ISO-1, and ISO-1@β-CD/GelMA groups. Scale bar s= 200 μm and 50 μm. b) Representative immunohistochemical (IHC) staining of osteogenic markers OCN and OPN in defect-site sections from OM, GelMA, single ISO-1, and ISO-1@β-CD/GelMA groups. c) Representative TRAP staining in defect-site sections from the indicated groups. d) Representative WB and densitometric quantification of p-AKT/AKT and NLRP3. e) Representative IHC staining of inflammatory-associated cytokines IL-6 and IL-10 in sections from the indicated groups. Dashed boxes indicate regions shown at higher magnification. Scale bars = 100 μm and 25 μm. Data are presented as mean ± SD; n = 5.

2.11. In vivo osteogenesis, osteoclast activity, and inflammatory modulation induced by ISO-1@β-CD/GelMA hydrogels

Consistent with these findings, immunohistochemical (IHC) staining of osteogenic markers demonstrated that, compared with the OM group, both the single ISO-1 and ISO-1@β-CD/GelMA groups exhibited a significantly increased OCN-positive staining area. Importantly, relative to single ISO-1 injection, the ISO-1@β-CD/GelMA group showed the strongest expression of OCN and OPN proteins, indicating that sustained ISO-1 delivery more effectively enhanced osteogenic differentiation in vivo (Fig. 10b and S18). In contrast, the TRAP-stained area was markedly reduced in both the single ISO-1 and ISO-1@β-CD/GelMA groups, with no significant difference between the two groups, suggesting that ISO-1 administration may suppress osteoclast activity during defect remodeling (Fig. 10c and S19). To further validate the in vivo regulatory effects of ISO-1@β-CD/GelMA on key signaling pathways and pyroptosis-related molecules, WB analysis was performed on local tissues from infected bone defects. Compared with the OM group, the ISO-1@β-CD/GelMA–treated group exhibited a significantly increased p-AKT/AKT ratio and a marked reduction in NLRP3 protein expression (Fig. 10d). These findings indicate that ISO-1@β-CD/GelMA enhances AKT pathway activation and suppresses NLRP3 inflammasome formation in vivo, consistent with the in vitro results.

Furthermore, immunohistochemical analysis of inflammation revealed that, compared with the OM group, both the single ISO-1 and ISO-1@β-CD/GelMA groups significantly reduced local IL-6 protein expression while promoting the upregulation of IL-10 protein expression (Fig. 10e and S20). Notably, although no significant difference in IL-10 protein levels was observed between the two groups, the ISO-1@β-CD/GelMA group exhibited a markedly lower IL-6 expression level than the single ISO-1 group. Collectively, the injectable sustained-release ISO-1@β-CD/GelMA hydrogel improved the early local inflammatory microenvironment in an osteomyelitis-induced infected bone defect model by downregulating IL-6 and upregulating IL-10, thereby providing favorable conditions to support subsequent osseointegration.

To further assess the systemic biosafety of the ISO-1@β-CD/GelMA hydrogel, major organs including the heart, liver, spleen, lung, and kidney were harvested from different treatment groups at the experimental endpoint and subjected to H&E staining. As shown in Fig. S21, all examined organs exhibited preserved histoarchitecture without evident inflammatory cell infiltration, tissue necrosis, hemorrhage, or structural disruption.

3. Discussion

This study systematically reveals that the SA-induced osteomyelitis microenvironment upregulates MIF and its receptor CD74, thereby driving osteoblast pyroptosis, suppressing osteogenic differentiation, and disrupting bone microarchitecture. It further demonstrates for the first time that the MIF-CD74/PI3K/AKT signaling axis exerts a central pathogenic role in infection-associated impairment of bone repair. Using both in vivo and in vitro models, we found that SA infection markedly attenuates osteoblast proliferation and mineralization capacity and activates the canonical NLRP3/ASC/GSDMD pyroptotic pathway, whereas Mif deficiency or knockdown significantly ameliorates osteogenic damage and inhibits pyroptosis. In parallel, transcriptomic profiling and co-immunoprecipitation experiments showed that MIF can physically associate with CD74 and modulate PI3K/AKT signaling activity, thereby regulating osteoblast fate. Moreover, we developed an injectable, photocrosslinkable, and sustained-release ISO-1@β-CD/GelMA hydrogel, which enables continuous local inhibition of MIF, improves the inflammatory microenvironment, and markedly promotes the repair of infected bone defects after osteomyelitis debridement. Collectively, this work not only elucidates the molecular mechanisms underlying osteomyelitis-induced osteogenic injury, but also proposes an innovative MIF-targeted, materials-based therapeutic strategy, providing a new theoretical basis and feasible approach for the precision treatment of infected bone defects.

In previous studies, rat and mouse models of osteomyelitis have most commonly been established by direct inoculation of bacteria into a bone defect or the medullary cavity, thereby recapitulating the clinical process of post-traumatic osteomyelitis [31,32]. Typically, bone defects with predefined diameters and depths are created in the tibia or femur and then injected with varying doses of SA suspension; for example, 1 × 10⁴-10⁶ CFU of SA may be introduced into a diaphyseal or metaphyseal tibial defect in rats. Alternatively, gel-based carriers with a defined bacterial load can be used to fill the defect to ensure a stable localized infection [33,34]. In mouse models, a unicortical bone defect combined with SA instillation or injection has also been widely adopted, enabling the successful reproduction of inflammatory responses and infection-driven bone destruction within smaller skeletal structures [35]. Consistent with these approaches, the present study likewise employed intramedullary SA injection. However, unlike protocols that useexogenous scaffolds to sustain infection, we induced an osteomyelitis phenotype exhibiting both acute and chronic features by directly delivering a bacterial suspension into the femoral medullary cavity of mice. This strategy minimizes confounding perturbations of the local microenvironment introduced by foreign materials and more closely mirrors the clinical course of osteomyelitis, which typically initiates in the marrow cavity and subsequently involves cortical and trabecular bone. Importantly, this model was established in C57BL/6N mice, aligning with prior reports that murine osteomyelitis models are well suited for genetic manipulation and investigations of inflammatory and immune responses [36,37]. This enabled us to conduct osteomyelitis induction under Mif knockout conditions and to directly test the contribution of MIF to infection-associated bone destruction, providing a highly consistent and reproducible infection model for mechanistic studies.

Osteomyelitis-induced destruction of bone microarchitecture is fundamentally driven by an imbalance between bone resorption and bone formation. Within the inflammatory microenvironment, RANKL expression is upregulated, promoting the differentiation of monocytes into osteoclasts and thereby enhancing bone resorption [38]. Meanwhile, OPG expression is downregulated, suppressing osteoblastic activity and leading to dysregulated bone remodeling [39]. At the microscopic level, elevated levels of inflammatory cytokines, such as IL-1β, IL-6, and TNF-α, directly inhibit the osteogenic differentiation of BMSCs and reduce the expression of osteogenic markers including RUNX2, OCN, and OPN [40]. In our mouse femur and tibia osteomyelitis model, we similarly observed hallmark features of enhanced bone resorption. Micro-CT analysis revealed significant decreases in BMD, BV/TV, and Tb.N, accompanied by a marked increase in Tb.Sp, indicating rapid deterioration of trabecular architecture and a shift toward bone-resorptive processes. Histological evaluations further confirmed that the infected regions were heavily infiltrated by inflammatory cells, accompanied by pronounced trabecular bone loss. At the molecular level, osteogenesis-related genes such as Ocn, Col1a1, and Runx2 were markedly downregulated following infection, whereas osteoclast-related genes Trap and c-fos were upregulated, consistent with the morphological manifestations of bone destruction. Additionally, ELISA results showed significantly elevated levels of the pro-inflammatory cytokines IL-1β, IL-6, and TNF-α in infected bone tissue, highlighting the pivotal role of the inflammatory microenvironment in suppressing osteogenesis and promoting bone resorption. Overall, the observations from the animal model clearly demonstrate the osteomyelitis-mediated destruction of bone microarchitecture.

MIF, as a pleiotropic pro-inflammatory cytokine, plays a pivotal regulatory role in amplifying inflammatory responses. Specifically, it promotes the expression of a variety of secondary inflammatory mediators and drives macrophage activation, chemotaxis, and secretory responses, thereby intensifying local inflammatory cascades [41,42]. Under inflammatory stimulation, including LPS and TNF-α, MIF expression and release are elevated several-fold in many cell types and tissues [43]. Consistent with previous reports, our study likewise observed markedly increased MIF protein expression in human infected bone tissue. We also found significant upregulation of pro-inflammatory cytokines in both the murine osteomyelitis model and the SA-induced cellular model. Importantly, Mif knockout or knockdown resulted in a pronounced reduction of pro-inflammatory cytokines in vivo and in vitro, accompanied by markedly alleviated inflammatory infiltration, suggesting that MIF may function as a key regulator of inflammatory responses. This regulatory role has been demonstrated in other inflammatory disease models. For example, in a mouse model of glomerulonephritis, Mif deficiency significantly reduced renal IL-6 levels and lowered serum IL-12 expression during systemic infection [44]. Another study reported that macrophages from Mif^−/− mice exhibited reduced synthesis of TNF-α and IL-1β, along with diminished lymphocyte production of IL-6, IL-17, and IL-23 [45].

Beyond its role in regulating inflammation, MIF also influences osteoblast differentiation. Previous studies have shown that MIF promotes the expression of genes related to matrix degradation and osteogenic regulation, such as Mmp-13, c-fos, and c-jun, and modulates osteoblast function through the ERK1/2 and AP-1 pathways [46]. In addition, some reports suggest that MIF inhibitors enhance osteogenic differentiation by suppressing the NF-κB pathway [47]. Our findings are consistent with these observations. In the SA-induced infectious microenvironment, elevated MIF expression was strongly associated with impaired osteogenic differentiation, whereas Mif knockdown restored the expression of RUNX2, COL1A1, and OCN and increased mineralized nodule formation. Collectively, these results suggest that MIF may induce osteogenic suppression and matrix degradation under infectious conditions, forming a critical nexus that links inflammatory responses to osteogenic impairment.

This study is the first to investigate the potential therapeutic effect of targeting MIF to suppress pyroptosis in the context of osteomyelitis. Previous research across multiple inflammation-related diseases has suggested a close association between MIF and pyroptotic cell death. In a diabetic nephropathy model, elevated MIF expression was shown to activate the NLRP3 inflammasome, whereas treatment with the MIF inhibitor CSB6B markedly downregulated NLRP3, Caspase-1, and GSDMD, reduced IL-1β secretion, and suppressed pyroptosis, thereby providing significant protection against tissue injury [21]. In sepsis-associated acute kidney injury, further studies have identified MIF as a critical structural or auxiliary activating component of the NLRP3 complex. IGF2BP1 promotes MIF expression by upregulating E2F1, subsequently inducing tubular epithelial cell pyroptosis; blocking this regulatory axis substantially alleviates organ damage [48]. Likewise, Li et al. [23] demonstrated in an LPS-induced acute kidney injury model that MIF upregulation facilitates NLRP3 inflammasome activation, while ISO-1 treatment reverses this process and inhibits pyroptosis. Our findings extend this mechanistic link into the field of osteomyelitis. We observed that SA infection markedly increased the expression of NLRP3, ASC, and GSDMD-N in osteoblasts and elevated the proportion of Annexin V/PI-positive cells, indicating that pyroptosis contributes to osteoblast injury. Upon Mif knockdown, pyroptosis-related markers were significantly reduced, accompanied by decreased release of inflammatory cytokines, suggesting that MIF serves as a key promoter of SA-induced osteoblast pyroptosis. Furthermore, Cd74 knockdown similarly attenuated the upregulation of pyroptotic signaling, implying that the MIF-CD74 receptor axis may represent an important pathway regulating osteoblast pyroptosis. Collectively, this study not only provides direct evidence for osteoblast pyroptosis in osteomyelitis but also demonstrates that targeting MIF effectively blocks the NLRP3/ASC/GSDMD-mediated pyroptotic pathway, offering a promising therapeutic strategy.

ISO-1, a specific inhibitor of the tautomerase activity of MIF both in vitro and intracellularly [49], has demonstrated strong therapeutic potential across multiple infectious and inflammatory diseases. Al-Abed et al. [50] reported that even delayed administration of ISO-1 significantly improved survival in mouse models of endotoxemia and CLP-induced sepsis, with effects comparable to anti-MIF antibodies. Moreover, the study found that the protective role of ISO-1 does not arise from inhibition of an essential enzymatic substrate; rather, it is likely mediated by altered binding interactions between MIF and downstream signaling proteins, thereby blocking its pro-inflammatory function. In chronic kidney injury models, ISO-1 likewise exerts potent organ-protective effects [51]. By inhibiting the MIF-CD74 receptor axis, ISO-1 markedly alleviates tissue damage and fibrosis, suggesting that disruption of MIF-CD74 signaling reduces inflammation-induced organ injury. In sepsis-associated acute kidney injury, ISO-1 also restores PINK1/Parkin-mediated mitophagy and attenuates apoptosis [52]. Notably, recent work has expanded the application of ISO-1 to central nervous system diseases. In an ischemic stroke model, a multifunctional dual-crosslinked hydrogel responsive to the pathological microenvironment (mild acidity and elevated ROS) was engineered to enable co-delivery of ISO-1 and nitric oxide (NO) [25]. This hydrogel scavenged ROS, reduced pro-inflammatory cytokines including MIF and IL-6, and provided sustained NO release to promote angiogenesis and neuroprotection, ultimately leading to substantial improvement in neurological recovery. This study demonstrated the therapeutic value of ISO-1 at sites of tissue injury from a material-delivery perspective, particularly its effective regulatory capacity in acute inflammation-driven pathological environments. The present study further extends ISO-1–based material-delivery strategies to the repair of infected bone defects. Within the inflammatory and acidic microenvironment that persists after osteomyelitis debridement, our ISO-1@β-CD/GelMA sustained-release hydrogel, leveraging GelMA as a slow-release matrix and β-CD as a molecular drug encapsulant, achieved multistage synergistic release of ISO-1. This effectively reduced local pro-inflammatory cytokine expression, improved osteoblast function, and promoted reconstruction of bone microarchitecture. Mechanistically, local ISO-1 administration efficiently inhibited the MIF-CD74 signaling pathway and restored PI3K/AKT activation, alleviating both osteogenic suppression and NLRP3/ASC/GSDMD-mediated osteoblast pyroptosis. Together, these findings provide robust mechanistic and materials-based support for the translational application of MIF inhibitors in treating infected bone defects and suggest future opportunities to develop targeted, locally delivered MIF-modulating strategies for the repair of osteomyelitis-induced bone loss.

Notably, macrophages may indirectly influence osteogenesis through paracrine regulation of BMSCs within the infected bone microenvironment [53]. Accumulating evidence indicates that even in the absence of direct cell-cell contact, macrophage-derived soluble mediators and extracellular vesicles can substantially modulate MSC/BMSC lineage commitment and osteogenic programs. For instance, cytokines secreted by osteal macrophages, such as oncostatin M (OSM), are considered key regulators of bone remodeling and osteogenic differentiation, with OSM-associated signaling involved in coupling immune activation to osteoblast maturation [54,55]. Moreover, macrophage-derived exosomes have emerged as an important mechanism of intercellular communication. A previous study reported that exosomes derived from M2 macrophages promoted BMSC osteogenic differentiation by downregulating miR-423-5p [56], highlighting the regulatory potential of macrophage exosomes in osteogenesis. Conversely, BMSCs themselves can also modulate immune responses via paracrine signaling and have been shown to reprogram macrophage polarization toward a reparative phenotype, thereby alleviating inflammatory osteolysis [57]. These findings collectively suggest the existence of a bidirectional osteo-immunomodulatory circuit. From the perspective of osteomyelitis, persistent bacterial stimulation often drives macrophages toward a pro-inflammatory phenotype, leading to increased secretion of cytokines and inflammatory mediators, which may impair osteogenic differentiation of bone marrow mesenchymal stem cells and prolong inflammation-associated bone destruction [58,59]. Under such conditions, macrophage-related MIF signaling may represent an additional regulatory mechanism beyond the direct osteoblastic effects mediated by MIF. Notably, a previous study has demonstrated that MIF can induce macrophage phagocytosis through autocrine and paracrine mechanisms, supporting its role as an immune amplifier within inflammatory niches [60]. Therefore, MIF inhibition may reshape the paracrine function of macrophages, thereby creating a more favorable osteoimmune microenvironment for BMSC-driven osteogenesis during the repair of infected bone defects. However, the present study did not directly investigate macrophage-BMSC paracrine crosstalk, which remains an important direction for future research. Further studies integrating macrophage-BMSC co-culture systems with secretome profiling will help elucidate how MIF inhibition modulates macrophage paracrine signaling to support osteogenic differentiation of BMSC in infected bone defects.

Although this study systematically elucidates the critical role of the MIF-CD74/PI3K/AKT signaling axis in osteomyelitis-related osteoblast dysfunction and pyroptosis, and verifies the therapeutic potential of an ISO-1 sustained-release hydrogel for the repair of infected bone defects, several limitations remain. Firstly, this work primarily relies on mouse and rat osteomyelitis models as well as in vitro cell models. Although these systems reasonably recapitulate the infectious microenvironment, they still differ from human bone tissue architecture, immune responses, and infection dynamics; therefore, further validation of the material's efficacy and safety in larger animal models is required. Secondly, the ISO-1@β-CD/GelMA hydrogel developed in this study exhibits favorable sustained-release behavior in vivo, but the precise release kinetics of the drug within a complex infectious milieu still need to be optimized and quantitatively characterized. Additionally, the long-term safety of local ISO-1 administration, its potential immunogenicity, and its effects under conditions of recurrent infection have not yet been fully clarified. Finally, the repair of infected bone defects is a multifactorial and highly complex process. Although targeting MIF alone can ameliorate inflammation and enhance osteogenic capacity, whether it can act synergistically with antibiotic therapy, immunomodulation, or other regenerative strategies remains to be determined in future studies.

4. Conclusion

In conclusion, we demonstrate that SA-induced osteomyelitis disrupts bone homeostasis through MIF-CD74-dependent dysregulation of the PI3K/AKT signaling axis, thereby inhibiting osteogenic differentiation and promoting NLRP3/ASC/GSDMD-mediated pyroptosis. MIF deficiency effectively restores osteogenic activity, reduces the release of inflammatory cytokines, and attenuates pyroptosis, indicating that MIF acts as a key regulatory factor linking infection-driven inflammation to osteogenic dysfunction. Based on this mechanism, we developed an injectable ISO-1@β-CD/GelMA hydrogel for the sustained delivery of the MIF inhibitor ISO-1, which improved the inflammatory microenvironment and significantly promoted bone regeneration and microarchitectural restoration in a rat model of infected bone defects. Therefore, a MIF-targeted hydrogel delivery system provides a promising therapeutic platform for osteomyelitis-associated infected bone defects and may offer insights for the treatment of other inflammatory diseases.

5. Materials and methods

5.1. Reagents and antibodies

The small-molecule MIF inhibitor ISO-1 and the PI3K inhibitor LY294002 were purchased from MedChemExpress (Shanghai, China). β-CD and the photoinitiator LAP were obtained from Aladdin Scientific (Shanghai, China). GelMA (EFL-GM-60) was supplied by the Suzhou Intelligent Manufacturing Research Institute (Suzhou, China). The CCK-8 assay kit, EdU cell proliferation assay kit, ALP activity assay kit, ARS staining solution, ELISA kits for IL-1β, IL-6, IL-18, and TNF-α, as well as the apoptosis detection kit were purchased from Beyotime Biotechnology (Shanghai, China). Antibodies against MIF, CD74, RUNX2, COL1A1, OCN, OPN, NLRP3, ASC, PI3K, p-PI3K, AKT, and p-AKT were obtained from Proteintech Group, Inc. (Wuhan, China). Antibodies against IL-6 and IL-10 were purchased from Abcam (Cambridge, UK). The antibody against GSDMD-NT was acquired from Thermo Fisher Scientific Inc. (Shanghai, China).

5.2. Collection of clinical osteomyelitis samples

Six osteomyelitis specimens were obtained from hospitalized patients who underwent lesion debridement and reconstruction of bone defects for chronic osteomyelitis. All patients met the following inclusion criteria [1]: chronic osteomyelitis of the long bones of the extremities confirmed based on clinical manifestations, imaging examinations, and bacterial culture results [2]; planned radical debridement with intraoperative exposure of infected bone tissue; and [3] age ≥18 years. Three control bone specimens were collected during the same period from non-infected bone tissue acquired in surgeries such as removal of internal fixation after fracture, corrective osteotomy for deformity, or traumatic osteotomy; these cases were required to have no preoperative evidence of infection, no purulent secretions intraoperatively, and negative bacterial cultures. Under sterile conditions, approximately 0.5 cm³ of bone tissue was rapidly excised at the operating table, immediately snap-frozen in liquid nitrogen, and stored at −80 °C for subsequent protein and RNA extraction. This study was approved by the Ethics Committee of 920th Hospital of Joint Logistics Support Force of PLA, and all procedures were conducted in strict accordance with the principles of the Declaration of Helsinki.

5.3. Source and genotyping of Mif^−/− mice

In this study, a constitutive whole-body Mif knockout mouse line on a C57BL/6N background (Mif^em1C) was generated by Cyagen Biosciences Inc. (Suzhou, China). This line was established using CRISPR/Cas9-mediated deletion of the critical exon region of the MIF gene to create a functionally null allele. Genotyping was performed by PCR amplification according to the primer scheme provided by the supplier, and the genotype was determined based on amplicon size by agarose gel electrophoresis. All mice used in the experiments were verified by PCR to ensure correct genotypes. The primers used for genotyping are listed in Table S1.

5.4. Establishment of femoral and tibial osteomyelitis in mice

Eight-week-old male C57BL/6N wild-type mice and Mif^−/− mice (body weight, 20–25 g) were used. Mice were randomly assigned to the Sham, OM, and Mif^−/−/OM groups (n = 10 per group) for the establishment of femoral and tibial osteomyelitis models and subsequent assessment of bone destruction. Prior to modeling, mice were anesthetized via intraperitoneal injection of sodium pentobarbital (25 mg/kg), followed by routine shaving, disinfection, and sterile draping. A 5-8 mm longitudinal incision was made at the distal right femur and the proximal one-third of the tibia. After sharp separation of the muscles, the intercondylar region of the femur or the tibial plateau was exposed. A 0.5 mm micro-drill was used to create an entry hole into the medullary cavity via the femoral intercondylar fossa or the tibial tuberosity, and a defined volume of SA suspension was slowly injected (1 × 10⁶ CFU/mL, 10 μL). The Sham group received an equal volume of sterile PBS (Solarbio, Beijing, China). After injection, the bone hole was sealed with bone wax or a gelatin sponge, and the muscles and skin were sutured sequentially. Detailed procedures followed our previous publication [20]. Mice were euthanized at 4 weeks postoperatively, and bilateral femora and tibiae were harvested for micro-CT scanning, histological staining, RT-qPCR, and immunohistochemical analyses to evaluate osteomyelitis-induced bone destruction and the impact of MIF deficiency on bone injury. The experiment was approved by the Ethics Committee of 920th Hospital of Joint Logistics Support Force of PLA (Grant No. 2024-051-01).

5.5. Micro-CT scanning and histological staining

After euthanasia, bilateral femora and tibiae were harvested, and soft tissues were removed. Specimens were fixed in 4% paraformaldehyde for 24-48 h and subsequently subjected to micro-CT scanning (Bruker, Germany). Scanning parameters were standardized as follows: tube voltage, 70 kV; tube current, 200 μA; exposure time, 350 ms; rotation angle, 180°. The spatial resolution was 6.5 μm for all samples, and a calibration phantom scanned in the same batch was used for density calibration. Raw data were reconstructed into three-dimensional images using NRecon software (Version 1.7.4.2, Bruker, Germany) with the following settings: smoothing = 5, beam hardening = 8, and ring artifact correction = 25 %, to achieve optimal image quality. Quantitative analyses were then performed within predefined ROIs using CTAn software (Version 1.20.3.0, Bruker, Germany) to determine bone microstructural parameters, including BMD, BV/TV, Tb.Th, Tb.N, and Tb.Sp. Bone defect filling and reconstruction were also assessed on 3D reconstructions and cross-sectional/coronal images.

Following micro-CT scanning, specimens were further fixed in 4% paraformaldehyde and decalcified until the bone became pliable, then routinely dehydrated and embedded in paraffin. Serial sections (4-5 μm) were prepared through the center of the defect. Some sections were stained with H&E to evaluate trabecular architecture, inflammatory cell infiltration, and marrow cavity reconstruction. Masson's trichrome staining was used to assess collagen fibers and bone matrix deposition, and TRAP staining was performed to visualize osteoclasts in bone tissue. Additional sections were used for IHC staining to examine the protein expression of RUNX2, COL1A1, OCN, OPN, IL-6, and IL-10. Standard IHC procedures were followed, including overnight incubation with primary antibodies at 4 °C, incubation with HRP-conjugated secondary antibodies at room temperature, DAB development, hematoxylin counterstaining, dehydration, and mounting. All sections were imaged under a light microscope, and image analysis software was used for semi-quantitative assessment of collagen deposition and the area of positive staining.

5.6. Osteogenic induction and establishment of an SA infection model

Murine BMSCs (mBMSCs; Cyagen Biosciences, Suzhou, China) were cultured in α-MEM complete medium supplemented with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin (50 μg/mL; Sigma-Aldrich, MO, USA) at 37 °C in a humidified atmosphere containing 5% CO₂. Cells between passages 3 and 6 were used for subsequent experiments. For osteogenic induction, mBMSCs were seeded into 96-well plates and, after stable attachment, the medium was replaced with osteogenic induction medium consisting of the basal medium supplemented with 10 mM β-glycerophosphate, 50 μg/mL ascorbic acid, and 10 nM dexamethasone; the medium was refreshed every 2 days. SA was cultured overnight in tryptic soy broth (TSB) at 37 °C with shaking and then adjusted to an appropriate MOI (1.0 MOI) based on preliminary CCK-8 assays. To establish an SA-infected BMSC model, after cell attachment, 1.0 MOI SA was added to the corresponding groups. Cells and supernatants were collected after treatment for different durations for subsequent assays and analyses. In addition, the BMSC, BMSC + SA, and siMIF/BMSC + SA groups were included for transcriptome sequencing. siRNA and plasmid transfection was performed when cell confluency reached 50%–70%. Small interfering RNAs targeting Mif and Cd74, as well as negative-control siRNA, were purchased from Guangzhou RiboBio Co., Ltd. (Guangzhou, China). According to the manufacturer's instructions, the transfection medium was replaced with complete medium containing 10% FBS after 4 h, and cells were further cultured. At 48 h post-transfection, SA infection or osteogenic induction was performed. The specific target sequences are detailed in Table S2.

5.7. Cell proliferation assays

mBMSC proliferation was evaluated using the CCK-8 assay and the EdU incorporation assay. Log-phase mBMSCs were seeded into 96-well plates at an appropriate density. After cell attachment and completion of the designated treatments (osteogenic induction, SA infection, and/or siRNA transfection), CCK-8 reagent was added at the predefined time points and incubated at 37 °C for 2 h. Absorbance was then measured at 450 nm (OD450) to reflect cell viability. For the EdU assay, treated BMSCs were further incubated in EdU-containing medium for a specified period. EdU incorporation was detected via click-chemistry labeling according to the manufacturer's protocol, followed by DAPI counterstaining of nuclei. Fluorescence images were acquired using a fluorescence microscope, and ImageJ (National Institutes of Health, USA) was used to calculate the proportion of EdU-positive cells relative to total cells in each group to assess the proliferation level of BMSCs.

5.8. Assessment of osteogenic differentiation

mBMSCs were seeded into 6-well plates at a predetermined density. After attachment, the medium was replaced with osteogenic induction medium, and cells were continuously induced with medium changes every 2 days. On days 1, 3, 5, and 7 of osteogenic induction, cells were harvested, and Alp mRNA expression was quantified by RT-qPCR. Late-stage mineralization capacity was evaluated by ARS staining. On day 21 of osteogenic induction, the medium was removed, cells were washed with PBS, fixed with 4% paraformaldehyde, rinsed with PBS, and then stained with 0.2% Alizarin Red solution (Solarbio, Beijing, China) for 15 min. The distribution of mineralized nodules was observed and recorded under a light microscope.

5.9. RNA extraction and RT-qPCR

Gene expression levels were quantified by RT-qPCR. Total RNA was extracted from cultured mouse BMSCs and bone tissues using TRIzol Reagent (Ambion Inc., Austin, TX, USA). Cells were washed with PBS and lysed directly in TRIzol, whereas bone tissues were ground in liquid nitrogen and then homogenized in TRIzol. For cDNA synthesis, 1 μg of total RNA was reverse-transcribed using the SweScript RT I First Strand cDNA Synthesis Kit (Servicebio, Wuhan, China) following the manufacturer's instructions. RT-qPCR was performed using Universal Blue SYBR Green qPCR Master Mix (Servicebio, Wuhan, China) on a real-time PCR system with the following cycling conditions: initial denaturation at 95 °C for 1 min, followed by 40 cycles of 95 °C for 20 s (denaturation), 55 °C for 20 s (annealing), and 72 °C for 30 s (extension). Melt-curve analysis was carried out at the end of amplification to confirm specificity. Primers were designed to amplify osteogenic-related genes (Runx2, Col1a1, Alp, Ocn, and Opn), with Gapdh used as the internal control. Relative mRNA expression levels were calculated using the 2^−ΔΔCt method, normalized to the housekeeping gene and expressed relative to the control or Sham group. The primer sequences are shown in Table S3.

5.10. Western blot

RIPA lysis buffer (Servicebio, Wuhan, China) supplemented with protease and phosphatase inhibitors (Servicebio, Wuhan, China) was used to extract total cellular proteins from cultured BMSCs and bone tissue samples. Lysates were centrifuged at 15,000 rpm for 20 min at 4 °C to remove debris, and protein concentration in the supernatant was determined using a BCA Protein Assay Kit (Beyotime, Shanghai, China). Equal amounts of protein were separated on 10% SDS-polyacrylamide gels (Solarbio, Beijing, China) and transferred onto PVDF membranes (Millipore, MA, USA). Membranes were blocked with 5% (w/v) skimmed milk in TBST for 1 h at room temperature and then incubated overnight at 4 °C with primary antibodies against MIF, CD74, RUNX2, COL1A1, OCN, OPN, NLRP3, ASC, GSDMD-NT, p-AKT, AKT, and GAPDH diluted in 1% skimmed milk/TBST. After three washes with TBST, membranes were incubated with HRP-conjugated secondary antibodies for 1 h at room temperature, and signals were detected using an ECL chemiluminescent substrate (Millipore) and imaged on a gel documentation system. Band intensities were quantified densitometrically using ImageJ software, with target proteins normalized to GAPDH.

5.11. ELISA measurement of inflammatory cytokines

For in vitro experiments, cell culture supernatants were collected at predetermined time points. For in vivo experiments, bone tissue from the defect/infection site was harvested to prepare tissue homogenates; the supernatants were clarified by centrifugation and used to quantify murine IL-1β, IL-6, IL-18, TNF-α, and IL-10 levels. All procedures were performed strictly according to the ELISA kit instructions. After stopping the reaction, absorbance was measured at 450 nm, and cytokine concentrations were calculated from the standard curves.

5.12. Flow cytometry assay

BMSC apoptosis was assessed by flow cytometry using Annexin V-APC/PI double staining. Briefly, log-phase BMSCs were seeded into 6-well plates and subjected to osteogenic induction, SA infection, and siRNA treatment as described above. At the designated time points, the culture medium was removed, and cells were washed twice with ice-cold PBS. Cells were then gently detached using EDTA-free trypsin, collected, and centrifuged at 1000 rpm for 5 min. After discarding the supernatant, cells were resuspended at approximately 1 × 10⁶ cells/mL. Annexin V-APC and PI were added according to the manufacturer's protocol and incubated for 10 min at room temperature in the dark. Stained cells were analyzed on a flow cytometer, with Annexin V-APC-A on the x-axis and PI-A on the y-axis. Data were processed using FlowJo or similar software, and cells were classified into Annexin V⁻/PI⁻ (viable), Annexin V⁺/PI⁻ (early apoptosis), Annexin V⁺/PI⁺ (late apoptosis/pyroptosis), and Annexin V⁻/PI⁺ (necrosis) subpopulations to evaluate the extent of SA-induced inflammatory cell death under different treatment conditions.

5.13. Co-IP

To verify the interaction between MIF and CD74, co-immunoprecipitation was performed. Briefly, treated BMSCs were lysed on ice using IP lysis buffer supplemented with protease and phosphatase inhibitors. After centrifugation to remove debris, the supernatant was collected as total protein. Samples were first precleared with Protein A/G agarose beads to reduce nonspecific binding, followed by incubation with an anti-MIF antibody or isotype-matched IgG control at 4 °C with rotation overnight. Protein A/G agarose beads were then added and incubated for an additional 2 h to allow formation of immune complexes. After repeated washes, bound proteins were eluted by boiling the beads in SDS loading buffer. The precipitated proteins were subjected to Western blotting, and CD74 was detected in the MIF immunoprecipitates using an anti-CD74 antibody.

5.14. RNA sequencing and bioinformatics analysis

To investigate the effects of SA infection and MIF intervention on the BMSC transcriptome, log-phase mBMSCs were assigned to three groups: BMSC, BMSC + SA, and siMIF/BMSC + SA, with three biological replicates per group. Total RNA was extracted using TRIzol reagent. RNA concentration, purity, and integrity were assessed using a NanoDrop spectrophotometer and an Agilent 2100 Bioanalyzer, and samples meeting quality requirements were submitted to Qinglianbio (Beijing, China) for library preparation and sequencing. Differential expression analysis was performed in R (Version 4.3.2) using the “limma” package. Differentially expressed genes and transcripts were identified based on the thresholds of |log₂ (fold change)| > 1 and FDR ≤0.05, and expression patterns were visualized using volcano plots and clustered heatmaps. Functional annotation and enrichment analyses of differentially expressed genes were conducted using the GO and KEGG databases, and enrichment bar plots were generated for visualization. In addition, a protein–protein interaction (PPI) network was constructed using the STRING database, and hub genes were visualized and screened in Cytoscape (Version 3.9.1) to explore MIF-associated signaling pathways and their potential roles in SA-induced impairment of osteogenic differentiation and pyroptosis in BMSCs.

5.15. Preparation of the ISO-1@β-CD/GelMA hydrogel

ISO-1/β-CD inclusion complexes were first prepared. Briefly, an appropriate amount of β-CD was dissolved in deionized water or PBS and magnetically stirred at 37 °C until completely dissolved. A pre-dissolved ISO-1 solution (ethanol or a small amount of DMSO used as a cosolvent, with the final concentration maintained within a cell-safe range) was then slowly added. The mixture was combined at an approximate molar ratio of 1:1 and stirred continuously for 24 h under light-protected conditions to form host–guest inclusion complexes. The resulting solution was subsequently lyophilized to obtain solid ISO-1@β-CD powder. GelMA hydrogel was dissolved in sterile PBS. Specifically, GelMA (1 g) was mixed with 10 mL PBS containing 0.25% (w/v) LAP photoinitiator and heated in a 60 °C water bath to prepare a 10% (w/v) solution. After complete dissolution, 1 g of ISO-1@β-CD powder was added at a predetermined mass ratio and gently agitated until uniformly dispersed, yielding the ISO-1@β-CD/GelMA precursor solution (1 wt% ISO-1@β-CD). A GelMA-only solution served as the control. Prior to use, the precursor solution was maintained at 37 °C to preserve fluidity. For in vitro experiments, the precursor was directly added to cell culture wells and photocrosslinked under 405 nm light for several seconds to form a self-supporting hydrogel. For in vivo studies, the appropriate volume of precursor solution was injected into the bone defect cavity after thorough debridement and photocrosslinked in situ under 405 nm illumination.

5.16. Characterization of hydrogel morphology and chemical structure

The micro-morphology of the hydrogels was examined by SEM (Gemini, Zeiss, Oberkochen, Germany). Elemental composition and spatial distribution were analyzed by EDS. Under the same field of view, elemental mapping and corresponding spectra for C, N, and O were acquired to compare differences in surface elemental content and distribution between the two hydrogel groups. The chemical structure of the hydrogels was characterized by FTIR (Nicolet, Thermo Fisher Scientific, USA) and XPS (Thermo ESCALAB 250XI, USA) to confirm the successful incorporation of ISO-1@β-CD and the formation of a stable organic functional group environment within the GelMA network.

5.17. Hydrogel property evaluation and ISO-1 release assay

GelMA and ISO-1@β-CD/GelMA hydrogels were fabricated into cylindrical specimens (6-8 mm in diameter and 3 mm in thickness), and their dry weight ($W_d$) was recorded. The samples were then immersed in PBS (pH 7.4) at 37 °C under static conditions. At predetermined time points (1, 2, 4, 6, 12, 24, 48, and 72 h), specimens were removed, gently blotted dry with filter paper, and immediately weighed to obtain the swollen weight ($W_s$). The swelling ratio was calculated as $\left(W_s-W_d\right)/W_d\times 100\%$ to compare the water uptake and volumetric expansion capacities between the two hydrogels. To assess degradation behavior, the initial dry weight of lyophilized hydrogels ($W_0$) was measured, and samples were incubated in PBS at 37 °C for an extended period. At predefined time points [1,3,5,7,14,21,28], specimens were retrieved, gently rinsed with PBS, and dried at 37 °C or re-lyophilized to constant weight to obtain the remaining dry weight ($W_t$). The percentage of remaining mass was calculated as $W_t/W_0\times 100\%$, and a remaining mass–time curve was plotted to evaluate the degradation profiles of the different hydrogels.

To investigate the sustained-release behavior of the hydrogel toward ISO-1, a defined amount of ISO-1@β-CD/GelMA hydrogel was placed in centrifuge tubes and subjected to release studies in PBS (pH 7.4) and acetate buffer (pH 6.5) at 37 °C. At the predetermined time points [1,3,5,7,9,11,14], all supernatants were collected for measurement and immediately replaced with an equal volume of fresh release medium. The absorbance of each sample was measured at the maximum absorption wavelength of ISO-1 using a UV-vis spectrophotometer. ISO-1 concentration and cumulative release were calculated based on a previously established ISO-1 standard curve.

The viscoelastic properties of the hydrogels were measured using a rotational rheometer (Kinexus Pro, Malvern, UK), with the testing temperature maintained at 25-37 °C. Crosslinked GelMA and ISO-1@β-CD/GelMA hydrogels were placed between parallel-plate geometries. Strain sweep tests were first performed at a fixed frequency (1.0 Hz) to determine the linear viscoelastic region (LVR). Subsequently, a small strain within the LVR (5%) was selected for frequency sweep measurements from 0.1 to 10 Hz, and the frequency-dependent changes in storage modulus (G′) and loss modulus (G″) were recorded. In addition, time sweep tests were conducted under fixed frequency and strain conditions to assess the temporal stability of G′ and G″. By comparing the magnitudes and trends of G′ and G″ between the two hydrogel groups, the effects of ISO-1 loading on the mechanical performance and network stability of the hydrogels were evaluated.

5.18. Establishment of rat osteomyelitis-associated infected femoral bone defect model and hydrogel implantation

Briefly, healthy male SD rats (6-8 weeks old, 250-300 g) were anesthetized, and the right femur was exposed under sterile conditions. An entry hole was created to access the medullary cavity, followed by intramedullary inoculation with a standardized SA suspension to induce osteomyelitis. The inoculum was freshly prepared for each operation, and the bacterial concentration was standardized by colony-forming unit (CFU) quantification (serial dilution plating) before administration. A fixed inoculation volume (1 × 10⁶ CFU/mL, 10 μL) was used for all animals to minimize inter-animal variability. After inoculation, the entry channel was sealed using bone wax to prevent leakage and ensure consistent establishment of intramedullary infection. The incision was closed in layers. Postoperatively, rats were monitored daily for general condition and local signs of infection. The osteomyelitis model was considered successfully established when local infection signs were present and osteomyelitis was confirmed by at least one objective assessment, including positive bacterial culture from the infected femur and/or typical imaging features consistent with osteomyelitis (osteolysis, inflammatory infiltration, and marrow destruction). After 28 days of infection induction, rats underwent complete debridement to remove necrotic/infected tissues and to create a standardized infected bone defect. The defect cavity was thoroughly irrigated, and animals were randomly assigned to receive OM control, GelMA hydrogel, single ISO-1 (10 mg/kg), or ISO-1@β-CD/GelMA hydrogel treatment, according to the grouping design. The hydrogel precursor was injected into the defect cavity and photocrosslinked in situ to form a stable implant. Rats were sacrificed at 4 and 8 weeks post-implantation, and femurs were harvested for micro-CT evaluation and histological/IHC analyses.

5.19. Statistical analysis

All data are presented as the mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism (Version 8.0.1; GraphPad Software, CA, USA). Student's t-test was used for comparisons between two groups. For comparisons among multiple groups, one-way ANOVA followed by Tukey's post hoc test was applied as appropriate. For all in vitro experiments, each condition was tested in at least three independent biological replicates, and for animal experiments the number of animals per group is indicated in the figure legends. A P value < 0.05 was considered statistically significant.

CRediT authorship contribution statement

Xiangwen Shi: Conceptualization, Data curation, Formal analysis, Funding acquisition, Validation, Visualization, Writing – original draft. Mingjie Wei: Formal analysis, Investigation, Methodology, Resources. Zhe Yin: Data curation, Validation, Visualization. Mingjun Li: Software, Validation, Visualization. Jianjun Wang: Data curation, Formal analysis. Pengcheng Fu: Validation, Visualization. Libo Yuan: Investigation, Methodology, Software. Yi Li: Conceptualization, Writing – review & editing. Yipeng Wu: Conceptualization, Funding acquisition, Writing – original draft, Writing – review & editing. Yongqing Xu: Conceptualization, Funding acquisition, Project administration, Supervision, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Supporting Information

Supporting Information is available.

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Data availability

Data will be made available on request.