Abstract

   The specific mechanisms underlying bacteria‐triggered cell death and
   osteogenic dysfunction in host bone marrow mesenchymal stem cells
   (BMSCs) remain unclear, posing a significant challenge to the repair of
   infected bone defects. This study identifies ferroptosis as the
   predominant cause of BMSCs death in the infected bone microenvironment.
   Mechanistically, the bacteria‐induced activation of the innate immune
   response in BMSCs leads to upregulation and phosphorylation of
   interferon regulatory factor 7 (IRF7), thus facilitating IRF7‐dependent
   ferroptosis of BMSCs through the transcriptional upregulation of
   acyl‐coenzyme A synthetase long‐chain family member 4 (ACSL4).
   Moreover, it is found that intervening in ferroptosis can partially
   rescue cell injuries and osteogenic dysfunction. Based on these
   findings, a hydrogel composite 3D‐printed scaffold is designed with
   reactive oxygen species (ROS)‐responsive release of antibacterial
   quaternized chitosan and sustained delivery of the ferroptosis
   inhibitor Ferrostatin‐1 (Fer‐1), capable of eradicating pathogens and
   promoting bone regeneration in a rat model of infected bone defects.
   Together, this study suggests that ferroptosis of BMSCs is a promising
   therapeutic target for infected bone defect repair.

   Keywords: 3D‐printed scaffold, BMSC, bone infections, ferroptosis,
   infected bone defects
     __________________________________________________________________

   Ferroptosis significantly impacts the viability and osteogenic
   potential of bone marrow mesenchymal cells (BMSCs) in infected bone
   environments. S. aureus and E. coli activate Toll‐like receptor
   signaling and initiate IRF7‐dependent ferroptosis in BMSCs. This study
   highlights the potential of targeting ferroptosis to enhance bone
   regeneration in infected bone defects using a ROS‐responsive
   antibacterial/anti‐ferroptosis 3D‐printed scaffolds.

   graphic file with name ADVS-11-2404453-g009.jpg

1. Introduction

   Infected bone defects represent one of the most persistent and
   devastating consequences in orthopedic clinics. Patients with
   persistent infected bone disease suffer a high rate of pathological
   fractures, nonunion, limb deformities, and even death.^[ [60]^1 ^]
   Gram‐positive Staphylococcus aureus (S. aureus) is the most notorious
   and predominant pathogen in bone and joint infections (BJI), and
   Gram‐negative bacteria such as Pseudomonas aeruginosa (P. aeruginosa)
   and Escherichia coli (E. coli) also participate in BJI.^[ [61]^2 ^] The
   residual bacteria and chronic antimicrobial inflammation result in
   tissue injury and impede the normal osteogenic regenerative process,
   while stimulating osteoclast‐mediated bone resorption. Consequently,
   infected bone defects may ultimately develop into sequestrum formation,
   fibrous pseudo‐union, nonunion, or osteolytic bone lesions.^[ [62]^3 ^]
   Although antimicrobial treatments, such as debridement, antibiotics,
   and antibiotic‐PMMA beads, have greatly enhanced the effectiveness of
   treatment, the management of infected bone defects still confronts a
   high failure rate due to reasons such as incomplete eradication of
   bacteria and dysfunction of osteogenic regeneration.^[ [63]^4 ^] The
   underlying mechanisms of osteogenic dysfunction caused by bacteria
   remain unclear and developing an effective therapy for repairing
   infected bone defects is still an ongoing challenge that needs to be
   addressed.

   Optimal bone defect regeneration and successful union rely on the
   abundant availability and highly orchestrated osteogenic
   differentiation of bone progenitor cells, including BMSCs and
   periosteum‐derived stem cells (PSCs).^[ [64]^5 ^] Previous studies
   showed that BMSCs take the main role in the repair of drilled bone
   injuries and PSCs become the primary cell type responsible for bone
   repair in bicortical fractures.^[ [65]^6 ^] Over the past decades,
   extensive research on bone infections revealed that toxins and
   metabolites secreted by pathological bacteria, along with various
   inflammatory factors and cytokines present in the infection
   microenvironment, collectively contribute to the injury of BMSCs and
   osteoblasts, leading to impairment of their osteogenic potential.^[
   [66]^7 ^] For example, Staphylococcal protein A (SPA) and
   phenol‐soluble modulins (PSMs) could induce the death of osteoblast
   lineage cells.^[ [67]^8 ^] Immune activation by bacterial components or
   inflammatory factors also causes injury to BMSCs and hampers normal
   osteogenic differentiation.^[ [68]^7 , [69]^9 ^] However, these studies
   roughly attributed the death of infected BMSCs or osteoblasts to
   apoptosis or cell lysis, which lacks specific morphological and genetic
   evidence. In addition, recent studies emphasized the importance of
   intracellular survival of S. aureus, the predominant pathogen in BJI,
   in BMSCs and osteoblasts.^[ [70]^10 ^] Intracellular bacteria could
   adjust the expression and release of virulence factors to avoid
   cytolysis, which helps the bacteria escape the bactericidal effects of
   antibiotics.^[ [71]^11 ^] This type of intracellular infection makes
   the mechanisms of BMSC death more elusive and intriguing. The exact
   mechanisms of direct injury and cell death of BMSCs caused by live
   bacteria remain unclear.

   Ferroptosis is a newly identified form of programmed cell death that is
   characterized by iron‐dependent lipid peroxidation.^[ [72]^12 ^] Recent
   evidence suggests that ferroptosis plays an important role in infection
   and bacteria‐related cell death. The ferroptosis of macrophages and
   bronchial epithelium cells could be induced by Mycobacterium
   tuberculosis (M. tuberculosis) and P. aeruginosa.^[ [73]^13 ^]
   Meanwhile, ferroptosis also served as a potent therapeutic target for
   infectious diseases. Pharmacological inhibition of ferroptosis by Fer‐1
   treatment effectively alleviates pathological damage and bacteria
   burden in the lung of the M. tuberculosis‐infected model.^[ [74]^13a ^]
   Therefore, targeting ferroptosis may also be an effective strategy to
   rescue BMSC function and promote infected bone defect repair.

   Treatment of critical‐sized infected bone defects usually requires the
   application of bioactive bone repair materials in a two‐stage operation
   after complete clearance of residual pathogens.^[ [75]^14 ^] Recent
   advances shed light on the promising application of 3D‐printed
   scaffolds and antimicrobial hydrogel in infected bone defect repair.^[
   [76]^15 ^] However, most materials release antimicrobial and bioactive
   factors at a constant velocity and lack the capability of
   infection‐responsive releasing, which could not adapt to the dynamic
   bacterial burden and lead to failure of early pathogen eradication and
   alleviation of tissue damage. There is still a lack of an effective
   strategy for protecting bone progenitor cells and reserving tissue
   osteogenic potential in early infection, owing to a lack of
   understanding of the mechanism of bacteria‐induced BMSC death.
   Long‐lasting release of bioactive factors in repair materials is needed
   to activate tissue regenerative potential and restore innate
   physiological homeostasis of regeneration in defect sites.

   In this study, we established acute osteomyelitis models and
   implant‐associated bone infection models in vivo, as well as a
   BMSC‐bacteria co‐culture model in vitro. Ferroptosis was identified to
   play a vital role in the death and osteogenic dysfunction of BMSCs
   induced by S. aureus and E. coli infections. Additionally, we designed
   and developed a dual‐functional antimicrobial/anti‐ferroptosis
   3D‐printed bone repair scaffold with ROS‐responsive bioactive
   factors‐releasing activities. This scaffold represents a novel strategy
   for repairing critical‐sized infected bone defects. These results
   provide insights into the underlying mechanisms of infected bone
   defects and offer a promising therapeutic approach for infected bone
   defects repair.

2. Results

2.1. S. aureus and E. coli Induced Lipid Peroxides Accumulation in Bone
Marrow and BMSCs

   To elucidate the microenvironmental changes within infected bone
   tissues, we established murine models of acute osteomyelitis and
   implant‐associated bone infections using S. aureus and E. coli, primary
   causative pathogens in clinical bone infections. Lipid peroxidation
   levels in bone marrow cells were determined by measuring protein
   modifications of lipid peroxidation products, such as 4‐Hydroxynonenal
   (4‐HNE) and malondialdehyde (MDA) protein modifications. Initially, we
   assessed these protein modifications in bone marrow within the acute
   osteomyelitis model. As illustrated in Figure [77] 1A, in comparison to
   the uninfected mice, both 4‐HNE and MDA protein modifications were
   markedly elevated following S. aureus and E. coli exposure. This
   signifies a heightened lipid peroxidation within the infected bone
   marrow.

Figure 1.

   Figure 1
   [78]Open in a new tab

   S. aureus and E. coli infections induced lipid peroxides accumulation
   in infected bone microenvironment and BMSCs. A) Lipid peroxidation
   levels in bone marrow on Day 3 post‐infection in murine osteomyelitis
   model induced by S. aureus and E. coli were determined by WB analysis
   of 4‐HNE and MDA protein modifications. Each group contained four mice.
   B) Representative tissue immunofluorescence images of LepR, MDA, 4‐HNE,
   ACSL4, GPX4 in the uninfected murine femur and implant‐associated bone
   infection model induced by S. aureus and E. coli. Scale bar = 50 µm. C)
   The schematic route of animal experiments and gating strategy were
   depicted. D) Lipid peroxide regulation by S. aureus and E. coli
   infections and Fer‐1 in CD45^−CD31^−Ter119^−LepR^+ BMSCs were analyzed
   in murine osteomyelitis model by Liperfluo staining and analyzed by
   flow cytometry. E,F) Quantification of median fluorescent intensity in
   flow cytometry analysis in (D). Values are means ± SDs. Each group
   contained six mice. Multiple comparison was performed by one‐way
   analysis of variance (ANOVA) with Tukey's post‐hoc analysis.

   Given the tight association between lipid peroxidation and ferroptosis,
   we further examined protein‐lipid peroxidation modifications and
   markers of ferroptosis in the implant‐associated bone infection model.
   Tissue immunofluorescence staining (Figure [79]1B) revealed pronounced
   fluorescence signals for 4‐HNE and MDA protein modifications in the
   peri‐implant bone marrow cells 7 days post‐infection. These signals
   overlapped with leptin receptor positive (LepR^+) cells, hinting at
   possible BMSCs involvement. Concurrently, we observed an increase of
   acyl‐CoA synthetase long‐chain family member 4 (ACSL4), an important
   contributor to ferroptosis, and a decrease of glutathione peroxidase 4
   (GPX4), a key enzyme to prevent ferroptosis, within LepR^+ cells.

   To pinpoint lipid peroxide in BMSCs and ascertain the role of BMSCs
   ferroptosis in vivo within an infected bone environment, we used flow
   cytometry to analyze the bacterial induced lipid peroxide and the
   potential mitigative effects of the Fer‐1. Leptin receptor was
   recognized to be a reliable marker for BMSCs in adult mice, so
   CD45^−CD31^−Ter119^−LepR^+ cells in bone marrow were gated as BMSCs
   (Figure [80]1C) according to previous studies.^[ [81]^16 ^]
   Figure [82]1D–F depict a considerable increase of lipid peroxide in
   BMSCs subjected to S. aureus and E. coli intratibial injections
   (pre‐treated with phosphate buffered saline (PBS)) by 1.92 ± 0.20‐fold
   and 1.71 ± 0.16‐fold, respectively, compared to the PBS pre‐treated
   uninfected group (1.00 ± 0.16‐fold). Treatment with daily 10 mg kg^−1
   intraperitoneal Fer‐1 administration effectively alleviated the
   accumulation of lipid peroxides of BMSCs induced by S. aureus
   (1.50 ± 0.20‐fold) and E. coli (1.38 ± 0.16‐fold). These findings
   demonstrated that S. aureus and E. coli induced lipid peroxidation
   within the bone marrow and triggered ferroptosis in BMSCs in vivo.

2.2. S. aureus and E. coli Induced Ferroptosis of BMSCs In Vitro

   While the accumulation of lipid peroxides in the infected bone
   microenvironment and ferroptosis in BMSCs has been identified, the
   primary mechanism driving cell injury and subsequent death in infected
   BMSCs remains ambiguous. To elucidate this question, we designed an in
   vitro BMSCs‐bacteria co‐culture model. We determined the optimal
   multiplicity of infection (MOI) to be 10 based on Live/Dead staining
   results (Figure [83]S1, Supporting Information), ensuring an adequate
   number of cells for subsequent analysis.

   Utilizing flow cytometry (Figure [84] 2A–C), we found that among the
   array of death inhibitors, including inhibitors targeting ferroptosis
   (Fer‐1 and Liproxstatin‐1 (Lpx‐1)), necroptosis (Nec‐1S), apoptosis,
   and pyroptosis (Z‐VAD‐FMK), only ferroptosis inhibitors considerably
   reduced the mortality rate of BMSCs infected with S. aureus and E.
   coli. To rule out any antimicrobial activity of these inhibitors, we
   conducted antimicrobial assays. The results confirmed no significant
   antimicrobial properties of these inhibitors at specified
   concentrations (Figure [85]S2, Supporting Information), emphasizing the
   pivotal role of ferroptosis in bacterial‐induced BMSC injury.

Figure 2.

   Figure 2
   [86]Open in a new tab

   S. aureus and E. coli infections trigger ferroptosis in BMSCs in vitro.
   A) The rescuing effects of inhibitors array on S. aureus and E.
   coli‐induced BMSCs death was analyzed by Live/Dead staining assay and
   flow cytometry analysis. BMSCs were pretreated with various death
   inhibitors for 12 h and then infected with S. aureus (MOI: 10) and E.
   coli (MOI: 10) for 12 h before analysis. B,C) Quantification of the
   rate of dead cells in flow cytometry analysis in (A). Values are
   means ± SDs (n = 3). Multiple comparison was performed by one‐way
   analysis of variance (ANOVA) with Tukey's post‐hoc analysis. D) The
   levels of lipid peroxides and ferrous iron in uninfected or S. aureus‐
   or E. coli‐infected BMSCs were determined by fluorescent staining of
   C11‐BODIPY, Liperfluo and FerroOrange. Representative fluorescent
   images of C11‐BODIPY staining (Scale bar = 40 µm), Liperfluo staining
   (Scale bar = 40 µm), and FerroOrange staining (Scale bar = 40 µm) were
   presented. E) WB analysis of the expression of GPX4, HO‐1, PTGS2, and
   ACSL4 in BMSCs. Quantification of fluorescent intensity of oxidized
   lipid F,G), lipid peroxides H,I), and ferrous iron J,K). Values are
   means ± SDs (n = 3). Multiple comparison was performed by one‐way
   analysis of variance (ANOVA) with Tukey's post‐hoc analysis. L)
   Representative TEM images of BMSCs in the uninfected group, S.
   aureus‐infected, and E. coli‐infected groups with different
   magnification degrees. Scale bar = 5 µm for low magnification, 750 nm
   for middle magnification, and 250 nm for high magnification.
   Mitochondria was abbreviated as Mit, and bacteria was abbreviated as
   Bac.

   Ferroptosis is typically characterized by augmented levels of lipid
   peroxides and a surge in intracellular ferrous ion accumulation. To
   quantify these markers, we utilized specific fluorescent probes:
   C11‐BODIPY and Liperfluo for lipid peroxides, and FerroOrange for
   ferrous ions. Intriguingly, Both E. coli and S. aureus infections
   robustly enhanced these markers, with Fer‐1 treatment effectively
   mitigating these increments (Figure [87]2D,F–K). Quantitative analysis
   of fluorescence intensity (Figure [88]2F,G) showed that levels of
   oxidized lipid stained by C11‐BODIPY in BMSCs increased to
   4.15 ± 0.18‐fold (E. coli‐infected) and 6.64 ± 0.28‐fold (S.
   aureus‐infected) compared with that of uninfected BMSCs
   (1.00 ± 0.21‐fold). And accumulation of ferrous ions in BMSCs increased
   to 3.39 ± 0.14‐fold in the E. coli‐infected group and 2.89 ± 0.37‐fold
   in the S. aureus‐infected group, which was mitigated by Fer‐1
   administration (Figure [89]2J,K).

   Further investigation into the expression patterns of
   ferroptosis‐related proteins revealed that while S. aureus and E. coli
   infections downregulated the ferroptosis‐suppressing protein GPX4, yet
   simultaneously elevated the levels of ferroptosis‐promoting proteins
   such as ACSL4, heme oxygenase (HO‐1), and Prostaglandin‐endoperoxide
   synthase 2 (PTGS2) (Figure [90]2E). Transmission electron microscopy
   (TEM) (Figure [91]2L) provided morphological evidence corroborating
   ferroptosis induction in infected BMSCs. Typical ferroptosis features
   such as the smaller size of the mitochondrion and the destroyed
   structure of mitochondrial cristae were observed. Additionally,
   intracellular bacteria were observed in S. aureus‐infected BMSCs, but
   not in E. coli‐infected BMSCs, hinting at different signaling pathway
   underlying the pathogen‐induced ferroptosis.

   Combining the findings from the mortality rescue trials, lipid peroxide
   and ferrous ion fluorescence analysis, and molecular and morphological
   characterizations, we proposed that ferroptosis is the principal cell
   death mechanism in S. aureus‐ and E. coli‐infected BMSCs, potentially
   contributing to osteogenic dysfunction.

2.3. RNA‐Sequencing of BMSCs Infected by S. aureus and E. coli

   To unveil the underlying mechanisms of infection‐triggered ferroptosis
   in BMSCs, RNA‐sequencing analysis was performed. As depicted in
   Figure [92] 3A, uninfected BMSCs and BMSCs infected by S. aureus
   (MOI:10 for 12 h) and E. coli (MOI:10 for 12 h) were collected and
   subjected to RNA‐sequencing. The principal component analysis (PCA)
   (Figure [93]S3A, Supporting Information) underscored the consistency in
   gene expression patterns within individual treatment groups. Moreover,
   it highlighted distinct differences in the gene expression patterns
   between the three treatment groups. Indeed, E. coli and S. aureus
   triggered intense transcriptional changes. As depicted in volcano plots
   (Figure [94]3B,C), compared with uninfected BMSCs, a total of 3624
   differentially expressed genes (DEGs) were identified in E.
   coli‐infected BMSCs, with 1149 DEGs upregulated and 2475 DEGs
   downregulated. As for S. aureus‐infected BMSCs, 866 upregulated DEGs
   and 1029 downregulated were identified compared with uninfected BMSCs.
   Among DEGs (Figure [95]3F,G), we observed significant downregulation of
   pivotal genes such as Runx2, Sp7, Wnt4, Fzd2, Dmp1, and
   ferroptosis‐suppressing such as Acsl3 in E. coli‐infected and S.
   aureus‐infected BMSCs, which correlated with osteogenic dysfunction in
   infected BMSCs. Meanwhile, DEGs indicate that pathways correlated with
   ferroptosis (Acsl4, Hmox1, Fth1, Cybb, Lcn2, slc7a11, Ptgs2),
   osteoclast differentiation (Tnfsf11, Csf1), toll‐like receptor (TLR)
   signaling (Myd88, Irf7, Fos, Ikbke), necroptosis and apoptosis were
   upregulated in E. coli‐infected and S. aureus‐infected BMSCs. Part of
   DEGs related to ferroptosis were validated by qRT‐PCR as shown in
   Figure [96]S3B (Supporting Information).

Figure 3.

   Figure 3
   [97]Open in a new tab

   S. aureus and E. coli infections activated ferroptosis and toll‐like
   signaling pathways in BMSCs. A) Schematic illustration of
   BMSCs‐bacteria co‐culture model. RNA samples of uninfected BMSCs and
   BMSCs infected with S. aureus (MOI: 10 for 12 h) and E. coli (MOI: 10
   for 12 h) were collected for RNA‐sequencing analysis. Volcano plot of
   differentially expressed genes (DEGs) in E. coli ‐infected BMSCs B) and
   S. aureus‐infected BMSCs C) compared with uninfected BMSCs. A fold
   change of protein (FC) ≥ 1.2‐fold and Q value < 0.05 were considered as
   DEPs. Upregulated DEGs are labeled with red dots, and downregulated
   DEGs are labeled with blue dots. Gray dots indicate nonsignificant
   differentially changed proteins. Some important DEGs related to
   ferroptosis and osteogenic differentiation were marked with gene names.
   KEGG pathway enrichment analysis of DEGs in E. coli‐infected BMSCs D)
   and S. aureus‐infected BMSCs E). Some DEGs related to different
   biological functions in E. coli‐infected BMSCs F) and S.
   aureus‐infected BMSCs G) were visualized by heatmap. H) GSEA enrichment
   of toll‐like receptor signaling pathway in E. coli‐infected BMSCs
   (Enrichment score = 0.659 and FDR Q value < 0.001) and S.
   aureus‐infected BMSCs (Enrichment score = 0.647 and FDR Q value <
   0.001). I) Top 5 upregulated transcription factors in E. coli‐infected
   BMSCs and S. aureus‐infected BMSCs.

   The results of the KEGG pathway enrichment analysis (Figure [98]3D,E)
   demonstrated that ferroptosis displayed the highest enrichment factor
   among all death‐related signaling pathways in either E. coli‐infected
   and S. aureus‐infected BMSCs, which is consistent with the former
   results in vitro and in vivo and consolidated the predominant role of
   ferroptosis in bacteria‐induced death. In addition, our analysis also
   spotlighted significant alterations in the toll‐like receptor signaling
   pathway in BMSCs infected by either bacterial strain. The significant
   change in the TLR signaling pathway was subsequently confirmed through
   the GSEA enrichment analysis, as visualized in Figure [99]3H. These
   results consolidated the occurrence of ferroptosis and osteogenic
   dysfunction in S. aureus‐infected and E. coli‐infected BMSCs at the
   molecular level and unveiled the activation of innate immune responses
   of infected BMSCs.

2.4. TLR Signaling and IRF7 Regulate Infection‐Induced BMSCs Ferroptosis

   Mounting evidence supported a strong connection between innate immune
   response and ferroptosis.^[ [100]^17 ^] TLR and their downstream
   signaling take a vital role in recognizing pathogen invasion and
   initiating antibacterial inflammation responses.^[ [101]^18 ^] In the
   results of RNA‐Seq, the toll‐like receptor signaling pathway was
   significantly altered in both S. aureus‐infected and E. coli‐infected
   BMSCs. In addition, interferon regulatory factor 7 (IRF7) was the
   highest upregulated transcriptional factor in BMSCs either infected
   with S. aureus or E. coli (Figure [102]3I). This evidence leads us to
   the hypothesis postulating a potential interlinkage between the
   activation of the TLR signaling pathway and the onset of ferroptosis.
   Expression levels of downstream signaling molecules of toll‐like
   receptor signaling were determined by western blot. As shown in
   Figure [103] 4A, E. coli‐infection induced upregulation of IRF7 and
   phosphorylation of IRF7 via activating MyD88‐IRAK4‐IRAK1/TRAF6 axis.
   Different from E. coli infection, S. aureus infection induced IRF7 and
   phosphorylation of IRF7 through activating TRAF3‐IKKε/TBK1 axis,
   bypassing MyD88. This differential modulation by the two bacterial
   strains may be attributed to their unique molecular patterns
   interacting with specific TLR on BMSCs. Subsequent in vivo validation
   was performed by evaluating IRF7 expression in LepR^+ cells within an
   implant‐associated bone infection model. The immunofluorescence
   staining reinforced our in vitro observations, demonstrating a
   pronounced upregulation of IRF7 (Figure [104]4B).

Figure 4.

   Figure 4
   [105]Open in a new tab

   Toll‐like receptor signaling and IRF7 regulate infection‐induced BMSC
   ferroptosis. A) The regulating effects of S. aureus and E. coli
   infections on toll‐like receptor signaling proteins including MyD88,
   IRAK4, IRAK1, TRAF6, TRAF3, IKKε, p‐TBK1/TBK1, and p‐IRF7/IRF7. B)
   Representative tissue immunofluorescence images of LepR, IRF7 in the
   uninfected murine femur and implant‐associated bone infection model
   induced by S. aureus and E. coli. Scale bar = 50 µm. C) IRF7 knockdown
   efficiency was evaluated by qRT‐PCR. The Hsp90b1 gene was used as an
   internal reference gene. Values are means ± SDs (n = 4). Multiple
   comparison was performed by one‐way analysis of variance (ANOVA) with
   Tukey's post‐hoc analysis. D) WB analysis of IRF7 knockdown efficiency.
   E,F) Representative fluorescent images of C11‐BODIPY staining (Scale
   bar = 40 µm) and FerroOrange staining (Scale bar = 40 µm). BMSCs were
   transfected with indicated siRNA for 48 h and then infected with S.
   aureus (MOI: 10) and E. coli (MOI: 10) for 12 h before observation. G)
   The effects of IRF7 on S. aureus and E. coli‐induced BMSC death were
   analyzed by IRF7 knockdown and Live/Dead staining assay and flow
   cytometry analysis. BMSCs were transfected with indicated siRNA for
   48 h and then infected with S. aureus (MOI: 10) and E. coli (MOI: 10)
   for 12 h before analysis. H,I) Quantification of the rate of dead cells
   in flow cytometry analysis in (G). Values are means ± SDs (n = 3).
   Multiple comparison was performed by one‐way analysis of variance
   (ANOVA) with Tukey's post‐hoc analysis. Quantification of fluorescent
   intensity of oxidized lipid J,K), ferrous iron L,M). Values are
   means ± SDs (n = 3). Multiple comparison was performed by one‐way
   analysis of variance (ANOVA) with Tukey's post‐hoc analysis. N) The
   schematic route of animal experiments of in vivo si‐RNA transfection in
   murine osteomyelitis model. Each group contained six mice. O)
   Representative flow cytometry images of PE‐IRF7 in
   CD45^−CD31^−Ter119^−LepR^+ BMSCs in mice transfected with si‐RNA. P)
   Quantitative analysis of the median fluorescent intensity of PE‐IRF7 in
   (O), Values are means ± SDs (n = 6). Multiple comparison was performed
   by one‐way analysis of variance (ANOVA) with Tukey's post‐hoc analysis.
   Q) Representative flow cytometry images of lipid peroxides stained by
   Liperfluo in CD45^−CD31^−Ter119^−LepR^+ BMSCs were analyzed in a murine
   osteomyelitis model transfected with si‐RNA. R) Quantitative analysis
   of the median fluorescent intensity of Lipid peroxides in (O), Values
   are means ± SDs (n = 6). Comparison between two groups was performed by
   a two‐tailed Student's t‐test.

   To precisely delineate the role of IRF7 in mediating bacteria‐induced
   ferroptosis in BMSCs, si‐RNAs targeting IRF7 were designed to knock
   down IRF7. As evidenced in Figure [106]4C,D, all three si‐RNAs
   targeting Irf7 (namely, si‐Irf7‐1, si‐Irf7‐2, and si‐Irf7‐3)
   proficiently reduced the expression of IRF7 both at the mRNA and
   protein levels. Then si‐Irf7‐1 and si‐Irf7‐2 were selected for the
   following experiments.

   In the bacteria‐BMSCs co‐culture model, as shown in Figure [107]4E,F,
   IRF7 silence in BMSCs significantly alleviated the upsurge of lipid
   peroxides and ferrous ions triggered by S. aureus and E. coli
   infections. Quantitative analysis of fluorescence intensity
   consolidated these findings (Figure [108]4J–M). Based on these results,
   the effects of Irf7 knockdown or combined with Fer‐1 on
   bacteria‐triggered ferroptosis were investigated with Live/Dead flow
   cytometry analysis. The results (Figure [109]4G–I) demonstrated that
   knockdown of IRF7 mitigated the death rate of BMSCs infected by S.
   aureus and E. coli, and a combined regimen of Fer‐1 and si‐Irf7
   manifested pronounced synergistic anti‐ferroptosis effects in the
   context of both bacterial infections.

   To validate the modulatory influence of IRF7 on ferroptosis in vivo,
   IRF7 was silenced through intratibial injections of si‐Irf7 in a murine
   osteomyelitis model induced by S. aureus infection (Figure [110]4N).
   Figure [111]4O,P showed that in vivo intratibial si‐RNA transfection
   effectively downregulated the expression of IRF7 in
   CD45^−CD31^−Ter119^−LepR^+ BMSCs in uninfected mice or osteomyelitis
   mice with S. aureus injection. Subsequent results confirmed that IRF7
   knockdown alleviated lipid peroxide levels in
   CD45^−CD31^−Ter119^−LepR^+ BMSCs in the bone marrow, as depicted
   demonstrated in Figure [112]4Q,R.

2.5. Infection‐Induced Upregulation of ACSL4 was Transcriptionally Activated
by IRF7

   To elucidate the mechanistic underpinnings by which IRF7 modulates
   ferroptosis, we executed an IRF7 knockdown in BMSCs and subsequently
   probed its regulatory impact on the expression of
   ferroptosis‐associated genes, as identified through RNA‐seq and qRT‐PCR
   analysis. Figure [113] 5A–C distinctly illustrated that the suppression
   of IRF7 leads to a marked downregulation of ACSL4 both at mRNA and
   protein levels. IRF7 knockdown moderately upregulated mRNA expression
   of Hmox1, but did not have an observable effect on HO‐1 expression at
   the protein level. Furthermore, both immunofluorescence and western
   blot assays underscored that IRF7 silencing mitigated the upregulated
   expression of ACSL4 in BMSCs infected with S. aureus and E. coli
   (Figure [114]5D,E). Functionally, ACSL4 catalyzes the synthesis
   reaction of polyunsaturated fatty acids‐containing phospholipids
   (PUFA‐PLs) and directly contributes to the occurrence of ferroptosis.^[
   [115]^12 , [116]^19 ^] Ferroptosis could be blocked by inactivation or
   knockout of ACSL4.^[ [117]^20 ^] Collectively, these findings
   substantiate the pivotal role of IRF7 in modulating ACSL4 expression,
   thereby shedding light on the potential mechanism through which IRF7
   exerts control over ferroptosis. Given the canonical role of IRF7 as a
   transcription factor, primarily influencing biological outcomes by
   modulating downstream gene expression, it is plausible to surmise that
   its regulatory influence on ACSL4 operates through transcriptional
   modulation.

Figure 5.

   Figure 5
   [118]Open in a new tab

   Infection‐induced upregulation of ACSL4 was transcriptionally activated
   by IRF7. A–C) Regulating effects of IRF7 silencing on Acsl4 and Hmox‐1
   expression at mRNA level and protein level. For qRT‐PCR, Values are
   means ± SDs (n = 4). Multiple comparison was performed by one‐way
   analysis of variance (ANOVA) with Tukey's post‐hoc analysis. D)
   Representative images of cell immunofluorescence of BMSCs stained with
   ACSL4 (green), Phalloidin (red) and DAPI (blue). BMSCs were transfected
   with indicated siRNA for 48 h and then infected with S. aureus (MOI:
   10) and E. coli (MOI: 10) for 12 h before observation. Scale
   bar = 40 µm. E) The regulation effects of IRF7 silencing on bacterial
   infection‐induced ACSL4 upregulation were analyzed by WB analysis. F)
   Illustration of the binding motif of IRF7. G) Schematic diagrams of
   potential IRF7 binding site on Acsl4 promoter covering 2‐kb nucleotide
   sequences upstream from the transcription initiation site. Binding site
   prediction was performed using the JASPAR database. H) Promoter
   occupancy analysis of IRF7 in Acsl4 promoter using ChIP‐PCR assay in
   BMSCs cells. Values are means ± SDs (n = 3). Multiple comparison was
   performed by one‐way analysis of variance (ANOVA) with Tukey's post‐hoc
   analysis. I) Representative images of agarose gel electrophoresis of
   ChIP‐PCR assay. The regulating effects of IRF7‐overexpression J) and
   IRF7 knockdown K) on Acsl4 promotor activity were determined by
   dual‐luciferase reporter assay. Values are means ± SDs (n = 4).
   Multiple comparison was performed by one‐way analysis of variance
   (ANOVA) with Tukey's post‐hoc analysis.

   To validate our hypothesis that IRF7 directly regulates ACSL4
   transcription, transcription factor binding site predictions, chromatin
   immunoprecipitation (ChIP) and dual‐luciferase reporter gene assay were
   performed. First, we used the JASPAR database to analyze potential IRF7
   binding sites within the 2000 bp upstream promoter sequence of the
   Acsl4 gene. Based on the binding motif of IRF7 (Figure [119]5F), the
   computational analysis (Figure [120]5G) demonstrated that there existed
   three potential binding sites, from −1563 to −1550 bp (Binding site 1),
   from −987 to −974 bp (Binding site 2), and from −389 to −376 bp
   (Binding site 3). ChIP‐PCR assay in BMSCs was performed with IRF7
   antibody to verify whether IRF7 could directly bind to these sites. The
   results of ChIP‐PCR (Figure [121]5H,I) showed that IRF7 bound only to
   binding site 1. Moreover, to validate the regulating effects of IRF7 on
   Acsl4 promotor activity, the Acsl4 promotor fragment was fused to a
   luciferase reporter plasmid (pGL4.10). And dual luciferase reporter
   assay was performed in BMSCs. The results (Figure [122]5J,K) indicated
   that overexpression of IRF7 with pcDNA‐IRF7 plasmid increased Acsl4
   promotor activity in a dose‐dependent manner, while silencing IRF7 by
   siRNA knockdown significantly inhibited Acsl4 promoter activity. In
   summary, these results demonstrated that IRF7 acts as a transcription
   stimulator of Acsl4 in BMSCs, which further contributed to
   bacteria‐induced ferroptosis in BMSCs.

2.6. Targeting‐Ferroptosis Rescued Osteogenic Differentiation Potential of
Infected BMSCs

   Our preliminary investigations highlighted the dominant role of
   ferroptosis in the damage and subsequent bacteria‐induced mortality of
   BMSCs. Based on these findings, we hypothesized that obstructing
   ferroptosis with specific inhibitors could rescue cell death and
   restore osteogenic differentiation potential in BMSCs. To validate this
   hypothesis, we designed a consecutive experiment involving ferroptosis
   inhibition and osteogenic induction in a bacteria‐BMSCs co‐culture
   model (Figure [123] 6A).

Figure 6.

   Figure 6
   [124]Open in a new tab

   Targeting ferroptosis rescued osteogenic differentiation potential of
   infected BMSCs. A) Schematic diagram of osteogenic differentiation of
   BMSCs after S. aureus and E. coli infections. The single rescuing
   effects of ferroptosis inhibitors Fer‐1 (10 µm) or Lpx‐1 (1 µm) on
   osteogenic differentiation of infected BMSCs were measured by ALP
   staining, Alizarin Red S staining, qRT‐PCR and WB at the indicated
   time. B) Representative images of ALP staining on Day 7 post osteogenic
   induction. C) Representative images of Alizarin Red S staining on Day
   21 post osteogenic induction. D–G) Gene expression of Alpl, Col1,
   Runx2, Sp7 was determined by qRT‐PCR on Day 7 post osteogenic
   induction. The Hsp90b1 gene was used as an internal reference gene.
   Values are means ± SDs (n = 4). Multiple comparison was performed by
   one‐way analysis of variance (ANOVA) with Tukey's post‐hoc analysis. H)
   WB analysis of the expression of ALP, COL1, RUNX2, OSX in BMSCs on Day
   14 post osteogenic induction.

   We first evaluate osteogenic differentiation by Alkaline Phosphatase
   (ALP) staining and Alizarin Red S (ARS) staining on Day 7 and Day 21
   post‐osteogenic induction, respectively. Interestingly, the findings
   (Figure [125]6B,C) revealed that the application of Fer‐1 (10 µm) and
   Lpx‐1 (1 µm) to uninfected BMSCs had no discernible impact on standard
   osteogenic differentiation when compared with the vehicle group.
   However, a significant inhibitive effect in osteogenic differentiation
   emerged in BMSCs exposed to S. aureus (MOI: 10) and E. coli (MOI: 10).
   Fer‐1 and Lpx‐1 notably enhanced ALP levels and calcified node
   formation relative to their respective vehicle cohorts (S. aureus and
   E. coli post‐infection).

   This restorative effect of ferroptosis inhibitors on osteogenic
   capability was further corroborated by qRT‐PCR and western blot
   analysis. As delineated in Figure [126]6D–G, the expression profiles of
   key osteogenic markers in BMSCs on Day 7—namely Alpl, Col1, Runx2, and
   Sp7—were markedly suppressed upon S. aureus and E. coli exposure.
   Strikingly, this suppression was effectively alleviated by the
   ferroptosis inhibitors Fer‐1 and Lpx‐1. Expression of these genes on
   Day 14 at the protein level also showed a similar trend
   (Figure [127]6H).

   Collectively, our results underscore the potential of Fer‐1 and Lpx‐1
   in ameliorating the osteogenic dysfunction in BMSCs infected by S.
   aureus and E. coli infections. These findings suggest that
   ferroptosis‐targeted intervention may serve as a promising therapeutic
   strategy for the regeneration of infected bone defects.

2.7. Synthesis of ROS‐Responsive Dual‐Functional
Antimicrobial/Anti‐Ferroptosis 3D‐Printed Scaffold

   Ideal bone repair materials for infected bone defects are proposed to
   release antimicrobial and tissue‐protecting factors in an
   infection‐responsive manner in the early stage of bone infection, which
   could adapt to the dynamic pathogen burden and inflammation load. In
   the middle and late stages of bone repair, sustained release of
   tissue‐protecting factors is required to maintain a homeostatic repair
   environment. Based on this concept, we sought to develop a bone repair
   scaffold that integrates a ROS‐responsive hydrogel with a 3D‐printed
   scaffold, offering adjustable release of antimicrobial and
   anti‐ferroptosis agents in early stage and sustained release of
   anti‐ferroptosis agents in the middle and late stage of bone repair
   (Figure [128] 7B).

Figure 7.

   Figure 7
   [129]Open in a new tab

   Synthesis of ROS‐responsive HACC‐DP/Fer‐1 hydrogel and construction of
   antibacterial/anti‐ferroptosis 3D printed scaffold. A) Schematic
   illustration showing synthesis route of ROS‐responsive HACC‐DP/Fer‐1
   hydrogel and construction of antibacterial/anti‐ferroptosis 3D printed
   scaffold. B) Schematic illustration showing surgical procedures of
   animal model and two‐phase release of HACC and Fer‐1 from the
   antibacterial/anti‐ferroptosis 3D printed scaffold. C) Representative
   images of shape adaptability of HACC‐DP/Fer‐1 hydrogel containing 1%
   HACC‐PBA in different patterns. Detailed contour indicates flexible
   in‐situ shape adaptability. D) Images of PCL/MBG scaffold and
   Fer‐1@HACC‐DP‐PCL/MBG scaffold (Scale bar = 1 mm). Cumulative release
   FITC in HACC‐DP/FITC hydrogel containing 0.5% HACC‐PBA E), 1.0%
   HACC‐PBA F), 1.5% HACC‐PBA G) and PCL/MBG/FITC scaffold H) in SBF and
   SBF+H[2]O[2] (100 µm) at 37 °C were measured at predetermined time
   points. A significantly higher release pattern in SBF+H[2]O[2] compared
   with SBF indicated the ROS‐responsive release capability of HACC‐DP
   hydrogel. Values are means ± SDs (n = 3). Two‐way analysis of variance
   (ANOVA) was performed. I) Representative images of CFUs counting in S.
   aureus treated by Fer‐1@HACC‐DP‐PCL/MBG scaffolds with different
   HACC‐PBA content for 3, 6, and 24 h. J) Quantitative analysis of CFU
   number in (I). Values are means ± SDs (n = 3). Multiple comparison was
   performed by one‐way analysis of variance (ANOVA) with Tukey's post‐hoc
   analysis.

   A notable characteristic of infected sites is the heightened production
   of ROS due to inflammation,^[ [130]^21 ^] which can be harnessed for
   controlled release. Our prior research spotlighted the effectiveness of
   a particular quaternized chitosan (hydroxypropyltrimethyl ammonium
   chloride chitosan, HACC) with a 26% substitution degree. It showed
   notable antimicrobial activity while simultaneously facilitating bone
   regeneration.^[ [131]^22 ^] This led to our proposal of synthesizing a
   photo‐cross‐linking hydrogel that would exhibit ROS‐responsive release
   dynamics of both HACC and Fer‐1.

   Illustrated in Figure [132]7A, we modified HACC with phenylboronic
   acid, obtaining HACC‐PBA. In a parallel process, Poly (ethylene glycol)
   diacrylate (PEGDA) and DL‐Dithiothreitol (DTT) were combined to create
   a photo‐cross‐linking polymer named PEGDA‐DTT, which could form
   ROS‐responsive boronic esters in alliance with the PBA group from
   HACC‐PBA. Next, biodegradable scaffolds, formulated from a blend of
   polycaprolactone, and mesoporous bioactive glass loaded with Fer‐1,
   were constructed using 3D printing. This scaffold was designed to
   provide sustained tissue protection and osteoinduction. Before inducing
   photo‐cross‐linking within the 3D‐printed scaffold, we added a
   calculated quantity of HACC‐PBA and Fer‐1 into the PEGDA‐DTT solution,
   and stirred for 2 h, yielding HACC‐DP/Fer‐1. Subsequently, this
   mixture, complemented with the photo‐initiator Lithium Phenyl
   (2,4,6‐trimethyl benzoyl) phosphinate (LAP), was added into the
   PCL/MBG/Fer‐1 scaffold. In‐situ photo‐cross‐linking was triggered by
   blue light exposure, yielding the composite Fer‐1@HACC‐DP‐PCL/MBG
   scaffold.

   HACC‐PBA, the pivotal component of the HACC‐DP hydrogel, was
   synthesized successfully, as evidenced in Figure [133]S4A,B (Supporting
   Information). The synthesis was authenticated by both Fourier‐transform
   infrared spectroscopy (FTIR) and ^1H‐nuclear magnetic resonance (^1H
   NMR). In the FTIR spectrum, a characteristic peak at 1344 cm^−1
   corresponding to B‐O was observed. Similarly, the ^1H NMR spectrum
   displayed characteristic proton signals: one at 7.11 ppm (‐BOH) and
   another at 7.47 ppm (‐C6H4‐).^[ [134]^23 ^] Based on these findings, we
   initiated the photo‐cross‐linking of the HACC‐DP hydrogel. Observations
   from Figure [135]S4D (Supporting Information) indicated that the
   hydrogel efficiently formed under blue light irradiation when the
   HACC‐PBA content ranged between 0% and 1.5%. The formation of HACC‐DP
   hydrogel was further validated by a rheological test (Figure [136]S4E,
   Supporting Information), which demonstrated that the storage modulus
   (G’) value was higher than the loss modulus (G’’) value in all HACC‐DP
   hydrogel groups after blue light irradiation. The results (Figure
   [137]S4G, Supporting Information) of shear‐thinning effect and
   injectability tests indicated that HACC‐DP hydrogels had relative good
   injectability and suitable for filling in scaffolds. The shape
   adaptability of the HACC/DP hydrogels, depicted in Figure [138]7C,
   emphasizes its compatibility with the geometry of the PCL/MBG scaffold.
   As illustrated in Figure [139]7D, both the PCL/MBG/Fer‐1 scaffold and
   the integrated Fer‐1@HACC‐DP‐PCL/MBG scaffold were effectively
   constructed. The compression modulus (Figure [140]S4C, Supporting
   Information) of both scaffolds reached over 20 MPa, which were relative
   higher in 3D printed scaffold and suitable for bone repair.

2.8. In Vitro and In Vivo Degradation Test and Evaluation of Antibacterial
Effects

   We sought to investigate the release behavior of the
   Fer‐1@HACC‐DP‐PCL/MBG scaffold. For this, fluorescein isothiocyanate
   (FITC), a small fluorescent molecule, was employed to simulate the
   Fer‐1 release in both HACC‐DP/Fer‐1 hydrogel and PCL/MBG/Fer‐1 scaffold
   within simulated body fluid (SBF) in vitro. In our observations
   (Figure [141]7E–G), HACC‐DP/FITC hydrogel with HACC‐PBA concentrations
   of 1% and 1.5% demonstrated a consistent FITC release over 6 days in
   SBF. In contrast, hydrogel containing 0.5% HACC‐PBA almost exhibited
   complete release within the initial 24 h. Interestingly, when the SBF
   was supplemented with 100 µm H[2]O[2], FITC release from the
   HACC‐DP/FITC hydrogel was notably accelerated, underlining the
   hydrogel's distinct ROS‐responsive release capabilities. In the case of
   the PCL/MBG/FITC scaffold (Figure [142]7H), we noted an initial burst
   of FITC release within the first 24 h, which then transitioned into a
   more controlled and steady release phase. Different from ROS‐responsive
   hydrogel, the presence or absence of H[2]O[2] in the SBF did not
   manifest in significant differences in FITC release patterns. By the
   end of the observation period at 28 days, only ≈19.5% of the total FITC
   content (equivalent to 800 µg) in the scaffold had been released,
   aligning with the scaffold's gradual biodegradation, as further
   illustrated in Figure [143]S4I–L (Supporting Information).

   The antimicrobial efficacy of the Fer‐1@HACC‐DP‐PCL/MBG scaffold
   containing different HACC‐DP hydrogel was subsequently evaluated. As
   shown in Figure [144]7I,J, the scaffolds with HACC‐DP‐1, HACC‐DP‐2 and
   HACC‐DP‐3 hydrogels exhibited an inhibition rate of above 90% against
   planktonic S. aureus in tryptic soy broth (TSB) at 3 h. The bacteria
   level decreased by four orders of magnitude at 24 h compared with those
   in Ctrl group and scaffold with HACC‐DP‐0. This result demonstrated
   that the scaffold with HACC‐DP‐1, HACC‐DP‐2 and HACC‐DP‐3 hydrogels
   could effectively inhibit the growth of planktonic bacteria. The
   scanning electronic microscopy (SEM) was used to observe the structure
   change of S. aureus and biofilm formation on scaffolds in this study.
   As shown in Figure [145]S4H (Supporting Information), Bacteria on
   PCL/MBG scaffold and Fer‐1@HACC‐DP‐0‐PCL/MBG scaffold (without HACC‐PBA
   in hydrogel) showed intact round shape and mature biofilm on scaffolds.
   In contrast, no biofilm was observed on Fer‐1@HACC‐DP‐2‐PCL/MBG
   scaffold (containing 1.0% HACC‐PBA in hydrogel), with few dead bacteria
   with pore formed on bacteria surface. This indicated that bacterial
   cell wall was damaged by HACC‐DP‐2 hydrogel. We also evaluated the
   biocompatibility of the Fer‐1@HACC‐DP‐PCL/MBG scaffold. As illustrated
   in Figure [146]S4M (Supporting Information), BMSCs could proliferate on
   Fer‐1@HACC‐DP‐PCL/MBG scaffolds and cell viability of BMSCs grown on
   Fer‐1@HACC‐DP‐PCL/MBG scaffold was above 80% compared with BMSCs seeded
   in culture disk (control), which showed good biocompatibility of
   Fer‐1@HACC‐DP‐PCL/MBG scaffolds. Given these findings, we opted to
   utilize the HACC‐DP hydrogel with 1% HACC‐PBA for the subsequent
   fabrication of the composite scaffold and the ensuing in vivo
   evaluations.

   To clarify the relationship of in vivo degradation of hydrogel and
   antibacterial effects, we conducted a subcutaneous degradation test of
   HACC‐DP‐2/Fer‐1 hydrogel. As indicated in Figure [147]S5A (Supporting
   Information), the results of H&E staining demonstrated that the
   HACC‐DP‐2/Fer‐1 hydrogel (indicated by blue arrow) implanted in the
   uninfected subcutaneous cavity gradually degraded from Day 1 to Day 7
   post‐operation and fully degraded by Day 14 post‐operation.
   Correspondingly, the subcutaneous cavity gradually closed, and healing
   tissue formed on Day 14. In contrast, the degradation rate of
   HACC‐DP‐2/Fer‐1 hydrogel was markedly accelerated by subcutaneous
   infection microenvironment. The HACC‐DP‐0/Fer‐1 hydrogel followed a
   similar degradation pattern as the HACC‐DP‐2/Fer‐1 hydrogel.

   To clarify the antibacterial effects of these hydrogels, we conducted
   Giemsa staining to label residual bacterial in tissue. As shown in
   Figure [148]S5B (Supporting Information), bacteria could be observed in
   subcutaneous tissue around the implanted hydrogels in HACC‐DP‐2/Fer‐1
   and HACC‐DP‐0/Fer‐1 with infection on Day 1. As the release and
   degradation of HACC‐DP‐2/Fer‐1 hydrogel occurred, bacterial colonies
   were undetectable from Day 3 to Day 14. However, dense bacterial
   colonies could be observed in tissue in HACC‐DP‐0/Fer‐1 group from Day
   3 to Day 14. This result demonstrated that the release and degradation
   of HACC‐DP‐2/Fer‐1 hydrogel could exert effective antibacterial effects
   on subcutaneous S. aureus compared with HACC‐DP‐0 hydrogel. The
   antibacterial effects could also be reflected by the level of
   tissue‐infiltrated inflammatory cells from histological evaluation. As
   shown in high magnification images in H&E staining (Figure [149]S5A,
   Supporting Information), dense inflammatory cell infiltration could be
   observed in tissue around hydrogel in HACC‐DP‐0/Fer‐1 hydrogel group
   from Day 1 to Day 7, and abscess tissue formed on Day 14, indicating
   persistent bacterial activity. In contrast, the level of inflammatory
   cell infiltration in HACC‐DP‐2/Fer‐1 group was moderate compared to
   that in HACC‐DP‐0/Fer‐1 group from Day 1 to Day 7 and healing tissue
   successfully formed on Day 14, indicating effective elimination of
   tissue bacteria.

   The overall results demonstrated that the HACC‐DP‐2/Fer‐1 hydrogel
   could degraded in an infection‐responsive pattern from Day 1 to Day 7
   in vivo and exert potent antibacterial effect on S. aureus infection,
   leading to lower tissue inflammatory level and tissue healing.

2.9. Intervention in Ferroptosis Facilitates Osteogenic Regeneration of
Infected Bone Defect in a Rat Model

   To assess the therapeutic effectiveness of the
   antimicrobial/anti‐ferroptosis scaffold in the repair of infected bone
   defects, we created critical‐sized infected bone defect models in the
   distal femur of rats. The study comprised four groups based on the
   implanted materials: Bone wax (blank control), PCL/MBG scaffold
   (lacking antimicrobial/anti‐ferroptosis functions), HACC‐DP‐PCL/MBG
   scaffold (antimicrobial properties but without anti‐ferroptosis
   features), and Fer‐1@HACC‐DP‐PCL/MBG scaffold (possessing both
   antimicrobial and anti‐ferroptosis characteristics). The procedural
   timeline is illustrated in Figure [150] 8A.

Figure 8.

   Figure 8
   [151]Open in a new tab

   Intervention in ferroptosis facilitates regeneration of infected bone
   defect in a rat infected distal‐femur defect model. A) Schematic
   diagram of animal experiment. A critical‐sized bone defect with a
   diameter of 3 mm and a depth of 4 mm was made on the distal femur of
   rats. 50 µL of 2 × 10^8 CFUs mL^−1 S. aureus (Xen 29) suspension was
   injected into the defect sites. Bone wax or different scaffolds were
   then implanted into defect sites. Each group contained 12 rats, with 6
   rats sacrificed at Week 6 and 6 rats sacrificed at Week 12
   post‐operation. B) Representative of in vivo bioluminescence images
   from rats in each group. Bioluminescence reflected bacterial burden at
   the operation sites. C) Quantitative analysis of total photon flux at
   operation sites. Values are means ± SDs (n = 12 for all groups from Day
   3 to Week 6, n = 6 for all groups at Week 12). Multiple comparison was
   performed by two‐way analysis of variance (ANOVA) with Tukey's post‐hoc
   analysis. * indicates P < 0.05, ** indicates P < 0.01. D)
   Representative of micro‐CT images from rats in each group at Week 6 and
   Week 12. The overall 3D reconstruction view was on the left (Scale
   bar = 1 mm). The coronal plane view was on the top right corner (Scale
   bar = 0.5 mm). The 3D reconstruction image of bone regeneration at
   defect sites was on the bottom right corner (Scale bar = 0.5 mm). E)
   Quantitative analysis of BV/TV at defect sites. Values are means ± SDs
   (n = 6). Multiple comparison was performed by one‐way analysis of
   variance (ANOVA) with Tukey's post‐hoc analysis. F) Representative
   images of Masson trichrome staining from rats in each group at Week 6
   and Week 12. Scale bar = 1 mm (low magnification). Scale bar = 50 µm
   (high magnification). G) Representative images of Giemsa staining from
   rats in each group at Week 6 and Week 12. Scale bar = 25 µm. H).
   Representative immune fluorescence images of OCN (Green) and LepR (Red)
   from rats in each group at Week 6 and Week 12. Scale bar = 100 µm.

   The in vivo imaging system (IVIS) was utilized for continuous
   monitoring of bacterial loads. Figure [152]8B,C displayed those
   bacterial signals persisted in the surgical sites of the Bone wax and
   PCL/MBG scaffold groups, detectable from Day 3 up to Week 12. In
   contrast, the groups with HACC‐DP‐PCL/MBG and Fer‐1@HACC‐DP‐PCL/MBG
   scaffold exhibited a marked reduction in bacterial presence on Day 3
   post operation, with undetectable levels from Week 1 to Week 12. This
   suggested that the HACC‐DP hydrogel within the scaffold effectively
   eradicated bacteria at the defect sites.

   At the predetermined time points of Week 6 and Week 12 post‐surgery,
   rats were sacrificed, and their femurs were analyzed using micro‐CT,
   X‐ray and histological evaluations. The micro‐CT results
   (Figure [153]8D) revealed significant osteolysis and pathological
   fractures in the Bone wax and PCL/MBG scaffold groups on Week 6. By 12
   weeks post‐operation, the pathology deteriorated further, with
   substantial fractures, pronounced joint damage, and deformities. In
   contrast, bone regeneration and reactive new bone formation were
   observed in the defect periphery in groups with HACC‐DP‐PCL/MBG
   scaffold and Fer‐1@HACC‐DP‐PCL/MBG scaffold on Week 6. The central
   defect region exhibited signs of bone formation, especially evident in
   the Fer‐1@HACC‐DP‐PCL/MBG scaffold group on Week 12. Quantitative
   micro‐CT analysis (Figure [154]8E) revealed that new bone formation
   (BV/TV) in the group with Fer‐1@HACC‐DP‐PCL/MBG scaffold surpassed
   those in the group with HACC‐DP‐PCL/MBG at both the 6 and 12‐week time
   points. Furthermore, BV/TV in the Fer‐1@HACC‐DP‐PCL/MBG and
   HACC‐DP‐PCL/MBG scaffold groups significantly exceeded those in the
   Bone wax and PCL/MBG scaffold groups. The X‐ray images (Figure [155]S6,
   Supporting Information) were also consistent with the results of
   micro‐CT. Conclusively, while scaffolds possessing only antimicrobial
   function moderately prevent osteolysis, joint destruction, and support
   bone regeneration in critical‐sized infected bone defects, targeting
   ferroptosis to mitigate cell damage substantially augments bone repair
   in these defects.

   We further carried out histological analysis to evaluate the
   therapeutic efficacy of these scaffolds. As shown in Masson trichrome
   staining (Figure [156]8F), a large amount of granular tissue and
   fibrous tissue filled the defect in the bone wax group on Week 6 and
   Week 12, which showed that the critical‐sized infected bone defect
   could not be repaired by self‐healing. In the PCL/MBG scaffold group on
   Week 6, an apparent fibrous capsule formed around the scaffold and
   fibrous tissue filled the defect. This situation worsened on Week 12,
   with the scaffold being completely squeezed out of the defect area by
   fibrous tissue. Meanwhile, in the HACC‐DP‐PCL/MBG group, the new bone
   tissue could be observed around the scaffold on Week 6 and more
   appeared on Week 12. Fer‐1@HACC‐DP‐PCL/MBG group showed the most
   apparent trabecular bone tissue in the central defect area on Week 6
   and Week 12, which validates the results of micro‐CT analysis. The
   Giemsa staining (Figure [157]8G) indicates the bacteria presence in the
   groups with bone wax and PCL/MBG scaffold, but not in the groups with
   HACC‐DP‐PCL/MBG scaffold and Fer‐1@HACC‐DP‐PCL/MBG scaffold, which
   demonstrated that HACC containing scaffolds could effectively eradicate
   pathogen in infected bone defects. The tissue immunofluorescence
   (Figure [158]8H) showed that LepR and osteocalcin (OCN) signals in the
   HACC‐DP‐PCL/MBG scaffold were significantly higher than other three
   groups, which reflected more BMSCs and more mature osteogenic
   differentiation induced by Fer‐1 releasing scaffold.

   In summary, the above results demonstrate that ROS‐responsive releasing
   of HACC and sustained delivery of Fer‐1 could effectively eradicate
   bacteria and protect the cell viability of BMSCs, thus promoting bone
   regeneration in infected bone defect models.

3. Discussion

   During the pathological progression of infected bone defects, the
   persistence of viable bacterial pathogens and their associated
   metabolites induce reprogramming of cells within the infection
   microenvironment. While several virulence factors from S. aureus and E.
   coli have been identified for their direct cytotoxicity,^[ [159]^8 ,
   [160]^24 ^] the direct regulatory impacts of living bacteria and their
   metabolites on bone progenitor cells remain elusive. The mechanisms by
   which living bacteria induce cell death may diverge from those induced
   by individual bacterial components. Numerous studies support the notion
   of intracellular S. aureus behaving as a pathogen within osteoblasts,
   osteocytes, and macrophages, further complexifying the cellular injury
   and death mechanisms.^[ [161]^10 , [162]^25 ^] In this study, we
   identified the predominant cell death pathways in BMSCs upon infection
   with S. aureus and E. coli, using an array of cell death inhibitors.
   Our findings demonstrated a pronounced ferroptosis phenotype in
   infected BMSCs, characterized by increased intracellular lipid
   peroxides, overload of ferrous iron, distinctive mitochondrial
   morphology, and the upregulation of ferroptosis‐related genes. In vivo
   evidence, including the heightened protein modifications of lipid
   peroxidation products in the infected bone marrow and amplified lipid
   peroxides in CD45^−CD31^−Ter119^−LepR^+ BMSCs, corroborated the
   manifestation of ferroptosis in a bone infection environment.

   Recently, the role of ferroptosis in bacterial‐induced cell death has
   been thrust into the spotlight. For instance, both M. tuberculosis and
   P. aeruginosa have been reported to induce ferroptosis within
   macrophages and bronchial epithelium cells, respectively.^[ [163]^13a,c
   ^] Certain bacterial components, such as Tyrosine phosphatase A (PtpA)
   and lipoxygenase (pLoxA), have been pinpointed as ferroptosis triggers
   in infections by these bacteria. The virulence factors of S. aureus and
   E. coli involved in inducing ferroptosis remain to be explored and
   elucidated. We also observed significant alterations in the expression
   of genes related to apoptosis, pyroptosis, and necroptosis pathways, as
   evidenced by the elevated expression levels of Casp7, Casp1, Nlrp3, and
   Mlkl. This suggests that a multitude of death signaling pathways might
   converge to determine the injury mechanisms in bacteria‐induced BMSC
   death. Emerging evidence indicates potential interactions between
   different death signaling pathways. For instance, Caspase 7 (Casp7) has
   been noted to activate acid sphingomyelinase (ASM) to mend gasdermin
   and perforin pores in infected hepatocytes.^[ [164]^26 ^] This
   reparative mechanism potentially defers pore‐driven lysis, steering the
   cell toward a moderated death program. In light of this, it is
   plausible that multiple death signaling pathways in BMSCs are
   co‐activated upon infection with S. aureus and E. coli. These pathways
   might interplay and culminate in a predominant ferroptosis death
   phenotype.

   Ferroptosis is a newly recognized form of programmed cell death. The
   activation of the innate immune response can initiate ferroptosis in
   various types of cells. For instance, the activation of IFNR by binding
   with IFN‐γ leads to the activation of IRF1, which upregulates ACSL4
   expression. When combined with arachidonic acid, IFN‐γ induces
   pronounced ferroptosis in various tumor cell lines.^[ [165]^20a ^] The
   inhibition of toll‐like receptor 4 (TLR4) by a TLR4 inhibitor
   ameliorates ferroptosis in the hypoxic‐ischemic injury of hippocampal
   neurons.^[ [166]^17a ^] IRF7 belongs to the transcription factors
   family of IRFs. Previous studies identified the role of IRF7 in
   transcriptionally activating IFN‐α/β expression, which resists
   infection by pathogens.^[ [167]^27 ^] In our study, we first found that
   TLR signaling and downstream IRF7 are involved in ferroptosis in BMSCs
   infected with S. aureus and E. coli by regulating ACSL4 expression, and
   IRF7 was the most upregulated transcriptional factor. However, the
   downstream pathways of TLR leading to IRF7 phosphorylation in S.
   aureus‐infected BMSCs (TRAF3‐IKKε/TBK1‐IRF7 axis) differ from those in
   E. coli‐infected BMSCs (MyD88‐IRAK4‐IRAK1/TRAF6‐IRF7 axis). This
   distinction might be due to the different bacterial components
   recognized by various TLRs in BMSCs. E. coli contains the classical
   TLR4 agonist, lipopolysaccharides (LPS), which can be recognized by
   TLR4 on the cell membrane, leading to MyD88‐dependent activation of
   Toll‐like receptor signaling.^[ [168]^28 ^] However, we also noticed
   that some downstream signal transducers of TLR3, including TRAF3, IKKε,
   and TBK1 could be activated by E. coli. This phenomenon was possibly
   caused by biological crosstalk in downstream of toll‐like signaling
   pathway, as the TLR4 downstream TRAM could activate TRAF3‐IKKε/TBK1
   axis through TRIF.^[ [169]^29 ^] And the TRAF3‐IKKε/TBK1 axis may act
   as a signal bypass of E. coli‐activated IRF7. The classical upstream
   TLR for TRAF3 activation is TLR3, an intracellular receptor that
   recognizes double‐stranded RNA (dsRNA) from pathogens.^[ [170]^30 ^]
   Given that S. aureus secretes immunomodulatory RNA via membrane
   vesicles,^[ [171]^31 ^] we hypothesize that intracellular S. aureus may
   release dsRNA, triggering ferroptosis by activating TLR3 and downstream
   IRF7. However, this hypothesis remains to be further investigated and
   validated.

   The importance of IRF7 in infection‐related immune response and tissue
   injury has recently garnered more attention. A recent study showed that
   self‐limited innate immune responses depend on the balance between IRF7
   and IRF3. IRF3 inactivation leads to the overexpression of IRF7,
   exacerbating kidney pathology in E. coli infections. Conversely, IRF7
   deletion protects mice from infection‐induced renal damage
   significantly.^[ [172]^32 ^] In our study, we observed a similar change
   in the IRF7/IRF3 ratio in RNA‐seq, with IRF7 upregulation and IRF3
   downregulation. Our findings further elucidate the role of IRF7 in
   regulating ferroptosis by transcriptionally regulating ACSL4
   expression, suggesting that IRF7 could be a promising therapeutic
   target to protect BMSCs from bacterial infections (as illustrated in
   the schematic diagram). Additionally, a notable upregulation HO‐1,
   encoded by Hmox1, was observed in both S. aureus‐infected and E.
   coli‐infected BMSCs. This phenomenon was consistent with the
   upregulated HO‐1 in macrophages infected by M. tuberculosis, which
   triggers ferroptosis in macrophages.^[ [173]^33 ^] Recent evidence
   supported that over‐activation of HO‐1 enhance intracellular Fenton
   reaction by increasing intracellular ferrous iron level thus
   facilitating ferroptosis.^[ [174]^34 ^] In this study, apparent upsurge
   of ferric iron level was found in S. aureus‐infected and E.
   coli‐infected BMSCs, which were supposed to be closely associated with
   upregulation of HO‐1. However, we found that upregulation of HO‐1 in
   BMSCs was not transcriptionally regulated by activating IRF7. Previous
   studies showed HO‐1 was mainly regulated by mitogen‐activated protein
   kinases (MAPK) signaling, PI3K‐Akt signaling, and Keap1/Nrf2
   signaling.^[ [175]^35 ^] The exact mechanism underlying upregulation of
   HO‐1 in S. aureus‐infected and E. coli‐infected BMSCs remained elusive
   and need further investigation.

   The limited availability of BMSCs and osteogenic dysfunction pose a
   challenge to infected bone defect repair. Past studies have intensively
   investigated the role of cytokines and inflammatory factors produced in
   the infection environment. The appropriate level of orchestrated
   inflammation was essential for neovascularization, BMSCs migration and
   osteogenic differentiation, thus facilitating bone regeneration in
   defect sites.^[ [176]^1a ^] However, persistent activity of residual
   bacteria leads to altered cellular composition and cytokine profile,
   which exert obvious inhibitive effects on osteogenic differentiation.^[
   [177]^7d ^] Although these findings laid a strong foundation for
   understanding pathological progression in infected bone defects, the
   role of direct cytotoxicity of bacteria toward BMSCs in osteogenic
   dysfunction and bone repair was unclear due to unclarified death
   mechanisms of infected BMSCs. We first identified ferroptosis as the
   predominant death type of BMSCs infected with S. aureus and E. coli.
   Based on these findings, we used Fer‐1 to tentatively rescue osteogenic
   dysfunction caused by S. aureus and E. coli infections in vitro. The
   results demonstrated partially osteogenic‐rescuing effects of Fer‐1,
   with improved ALP expression, cell calcification and expression of
   important osteogenic transcription factors Runx2 and Osterix, which
   shed light on anti‐ferroptosis therapy for infected bone defects.

   The repair of critical‐sized bone defects often necessitates the use of
   bone repair materials. Composite artificial bone scaffolds loaded with
   antibacterial components and bioactive factors have become a hotspot in
   the advances of orthopedic materials. For instance, a wood‐derived
   scaffold infused with chitosan quaternary ammonium salt (for
   antibacterial action) and dimethyloxalylglycine (pro‐osteogenic and
   pro‐angiogenic) has been designed for infected bone defect repair.^[
   [178]^36 ^] Another innovation includes the 3D‐printed PLGA/Cu(I)@ZIF‐8
   scaffold designed for infected bone repair.^[ [179]^37 ^] The
   3D‐printed scaffold has shown huge potential in bone repair due to its
   adjustable internal structure and outstanding anatomical adaptability.
   However, most of these recently developed scaffolds tend to
   mechanically combine antimicrobial components with osteoinductive
   factors, often overlooking the injury‐mitigating effects on BMSCs in
   the infected sites. Additionally, these scaffolds frequently lack
   mechanisms for the responsive and sustained release of bioactive
   factors, hindering their adaptability to varying bacterial loads and
   sustained bone regeneration. Recently, phenylborate ester‐based
   ROS‐responsive biomaterials attracted researchers’ attention. The
   phenylborate ester bond is a type of ROS‐responsive covalent bond,^[
   [180]^38 ^] and was used as a bridging segment connecting HACC and
   PEGDA‐DTT, the two polymer backbones in HACC‐DP/Fer‐1 hydrogel. When
   ROS levels increase in the infection environment in vivo or in the
   presence of H[2]O[2] in vitro, the phenylborate ester bond was cleaved,
   leading to accelerated backbone breakage of hydrogel. Correspondingly,
   the release rate of encapsulated Fer‐1 in hydrogel was accelerated by
   ROS.

   In this study, we constructed an antibacterial/anti‐ferroptosis 3D
   printed scaffold with ROS‐responsive releasing of HACC and Fer‐1 in the
   acute phase and long‐term releasing of Fer‐1 in the bone regeneration
   process. ROS‐responsive releasing of HACC and Fer‐1 from HACC‐DP
   hydrogel could adjust the antibacterial and ferroptosis‐rescuing effect
   according to the intensity of inflammation. And long‐term releasing of
   Fer‐1 from PCL/MBG/Fer‐1 scaffold could sustainably favor osteogenic
   differentiation. In vivo evaluations indicated that, while traditional
   treatments like bone wax and standard PCL/MBG scaffolds failed to
   effectively repair S. aureus‐infected bone defects, leading to
   pronounced osteolysis and pathological fractures, our newly developed
   scaffold demonstrated superior results. Specifically, the
   Fer‐1@HACC‐DP‐PCL/MBG scaffold, endowed with both antibacterial and
   anti‐ferroptosis properties, exhibited the most promising bone
   regeneration outcomes in infected defects. These findings suggest that
   targeting bacteria‐induced ferroptosis could significantly enhance bone
   regeneration in infected bone defects.

4. Conclusion

   In conclusion, we identified ferroptosis as the predominant cell death
   of BMSCs infected with S. aureus and E. coli both in vitro and in vivo.
   Additionally, we ascertained that the osteogenic dysfunction of
   infected BMSCs can be effectively alleviated by inhibiting ferroptosis
   using Fer‐1. Stemming from these findings, we designed and developed a
   ROS‐responsive antibacterial/anti‐ferroptosis 3D‐printed bone repair
   scaffold. This scaffold is intended for the treatment of critical‐sized
   infected bone defects and has demonstrated robust antimicrobial
   properties and high efficiency in promoting bone regeneration in vivo.
   Moreover, we found that the TLR signaling pathway, along with its
   downstream component IRF7, plays a pivotal role in the initiation and
   execution of lipid peroxidation in BMSC ferroptosis by
   transcriptionally regulating ACSL4 expression.

5. Experimental Section

Bacteria Culture

   The S. aureus strain (ATCC 25923) and E. coli strain (ATCC 25922) were
   acquired from the American Type Culture Collection (ATCC). The Xen 29
   bioluminescent strain (IVISbrite 119240) was obtained from PerkinElmer.
   For the preparation of the bacterial inoculum, a fresh single bacterial
   colony was suspended in TSB and then incubated in a shaker at 37 °C and
   200 rpm for 6 h.

Murine Models of Acute Osteomyelitis and Implant‐Associated Infection

   The murine acute osteomyelitis model and implant‐associated infection
   model were established according to previous studies.^[ [181]^39 ^] All
   operation procedures in the animal experiment had been reviewed and
   approved by the Animal Ethical Committee of the Shanghai Institute of
   Immunity and Infection, Chinese Academy of Sciences, with approval
   number No. A2023035. Adult male C57BL/6J mice (14‐week‐old) were
   purchased from Shanghai Jihui Laboratory Animal Company and were
   acclimated to the facilities for 3 days before the operation. For Fer‐1
   pretreatment, mice were given a 10 mg Kg^−1 administration (i.p. q.d.)
   for 3 days before the operation. To establish the acute osteomyelitis
   model, mice were first anesthetized with isoflurane. Then the skin near
   the knee joint was prepared and sterilized with a 75% ethanol solution.
   Next, 2 µL PBS or 10^6 CFUs bacteria in 2 µl PBS were injected into the
   bone marrow cavity with a microsyringe. Compression hemostasis was
   performed, and all mice were carefully monitored until they awoke from
   anesthesia. On Day 3 post‐operation, the mice were sacrificed, and the
   bone marrow cells from the tibia were collected for further western
   blot (n = 4 per group) or flow cytometry (n = 6 per group) analysis. In
   addition, intratibial si‐Irf7 transfection was performed using an in
   vivo‐jetPEI siRNA delivery reagent (101000040, Polyplus, France)
   according to the instructions from the manufacturer. Ten micrograms
   si‐RNA (si‐Irf7) dissolved in the delivery reagent was intratibially
   injected 3 days before the bacteria injection. For the establishment of
   the murine implant‐associated infection model, procedures of animal
   preparation and anesthesia were the same as the above‐mentioned steps.
   Then an incision beside the femur was made with a scalpel. Soft tissues
   around the femur were carefully separated, and a small hole was made by
   a 29G needle at the middle of the femoral cortex. Next, a sterilized
   steel wire with a diameter of 0.3 and 4 mm length was immersed in the
   bacterial inoculum (10^8 CFUs mL^−1) and implanted in the bone marrow
   cavity of the femur by squashing the bacterial‐loaded wire into the
   hole on the femur. After this, the wound was sutured layer by layer,
   and animals were monitored until they awoke from anesthesia. On Day 7
   post‐operation, mice were sacrificed, and the femur was isolated for
   further histological immunofluorescence staining (n = 3 per group).

In Vivo Flowcytometry

   After the mice were sacrificed, the tibias were isolated using sterile
   scissors and forceps. The bone marrow cells were rinsed with precooled
   FBS‐free RPMI 1640 culture medium and then filtered through a 70 µm
   cell filter (352350, BD Falcon, USA). After that, cell counting was
   performed and 2 × 10^6 cells were harvested for further treatment and
   analysis. Next, cells were blocked with an anti‐CD16/32 antibody
   (1:300, 14‐0161‐86, eBioscience, USA) and then stained with an antibody
   cocktail containing PerCP/Cy5.5‐CD45 antibody (1: 300, 103131,
   BioLegend, USA), PerCP/Cy5.5‐CD31 antibody (1: 300, 102419, BioLegend,
   USA), PerCP/Cy5.5‐Ter119 antibody (1: 300, 116227, BioLegend, USA), and
   AF647‐LepR antibody (1 µg mL^−1, 20966‐1‐AP, Proteintech, China,
   conjugated with AF647 using the FlexAble CoraLite Plus 647 Antibody
   Labeling Kit, Proteintech, China). The cells were then washed with
   FBS‐free RPMI 1640 medium and stained with the Liperfluo probe (1 µm,
   L248, DOJINDO, Japan). After staining, the cells were washed and
   subjected to flow cytometry analysis. At least 5 × 10^5 cells were
   collected in each sample, and CD45^−CD31^−Ter119^−LepR^+ cells were
   gated as BMSCs. The median fluorescence intensity (MFI) of the FITC
   channel (Liperfluo) was analyzed. For intracellular IRF7 staining, the
   cells that were stained with CD45/CD31/Ter119/LepR were fixed with the
   Foxp3/Transcription Factor Staining Buffer Set (00‐5523‐00,
   eBioscience, USA) and subsequently stained with PE‐IRF7 antibody
   (1:300, 12‐5829‐82, eBioscience, USA). The cells were then washed and
   subjected to flow cytometry analysis.

Cell Culture

   Mouse BMSCs were isolated as previously described and expanded using
   the MesenCult™ Expansion Kit (05513, STEMCELL TECHNOLOGIES, Canada)
   according to the instructions from the manufacturer.^[ [182]^40 ^] In
   brief, bone marrow cells harvested from the femur and tibia of adult
   male C57BL/6J mice (14‐week‐old) were rinsed with MesenCult™ Expansion
   culture medium and then cultured in a culture plate at 37 °C in a 5%
   CO[2] environment for 48 h. Then, the suspended cells were removed, and
   the culture medium was replaced every 2 days until passage. Purified P2
   generation BMSCs were used for subsequent experiments.

Bacteria‐BMSCs Co‐Culture Model

   BMSCs (P2 generation) were seeded in 6‐well culture plates at a density
   of 5 × 10^5 cells per well and cultured with α‐MEM culture medium
   (supplemented with 10% FBS without antibiotics) overnight at 37 °C, in
   a 5% CO[2] incubator. For the pretreatment with death inhibitors, BMSCs
   were pretreated with the indicated concentrations of the inhibitors:
   Fer‐1 (10 µm), Lpx‐1 (1 µm), Nec‐1s (20 µm), and Z‐VAD‐FMK (50 µm) when
   the cells were seeded for 12 h. BMSCs were then treated with α‐MEM
   culture medium containing freshly activated S. aureus and E. coli with
   an MOI of 10 for the indicated time. After this, the bacteria were
   washed off, and the bacteria‐infected BMSCs were used for further
   analysis. The inhibitors Fer‐1 (S7243), Lpx‐1 (S7699), Nec‐1s (S8641),
   and Z‐VAD‐FMK (S7023) were purchased from Selleck (Selleck, China).

C11‐BODIPY, Liperfluo, and FerroOrange Staining

   C11‐BODIPY, Liperfluo, and FerroOrange staining were performed
   according to the instructions from the manufacturer. In brief, cells
   following the indicated stimulations were washed with Hanks' Balanced
   Salt Solution (HBSS, C0218, Beyotime, China) and subsequently incubated
   in working solutions of C11‐BODIPY (2.5 µm, D3861, Invitrogen, USA),
   Liperfluo (1 µm, L248, DOJINDO, Japan), and FerroOrange (1 µm, F374,
   DOJINDO, Japan). Finally, cells were washed and then observed using a
   confocal laser scanning microscope (CLSM, Leica, Germany). The
   intracellular lipid peroxides were determined by quantification of the
   fluorescence of C11‐BODIPY and Liperfluo, while the intracellular
   ferrous ions were determined by quantification of the fluorescence of
   FerroOrange.

Live/Dead Staining Assay

   The Live/Dead staining assay was performed using Live/Dead Fixable Dead
   Cell Stain Kits ([183]L34970, Invitrogen, USA) and flow cytometry
   analysis. BMSCs following the indicated stimulation were stained with
   the reconstituted fluorescent reactive dye and fixed with 3.7%
   formaldehyde. The stained cells were then subjected to flow cytometry
   analysis. The dead cell rate was calculated and analyzed.

Transmission Electron Microscopy

   Transmission electron microscopy was performed as previously
   described.^[ [184]^41 ^] BMSCs after infection with S. aureus (MOI: 10)
   and E. coli (MOI: 10) for 12 h were fixed with 2.5% glutaraldehyde. The
   fixed cells were then subjected to gradient dehydration, followed by
   resin embedding and slicing. Then, samples were observed with
   transmission electron microscopy (TEM, FEI Talos L120C, Thermo
   Scientific, USA).

Western Blot

   Total cellular proteins were obtained by lysing cells with RIPA lysis
   (P0013B, Beyotime, China) containing a protease and phosphatase
   inhibitor cocktail (78445, Thermo Scientific, USA). Protein solutions
   were treated with an ultrasonic cell disruptor, and the protein
   concentration was quantified with a BCA assay. Each protein sample
   (20 µg) was separated by electrophoresis and transferred to a
   polyvinylidene fluoride membrane (0.22‐µm, Millipore, USA). The
   membrane was then blocked with 5% bovine serum albumin (BSA) solution
   and incubated with the primary antibody at 4 °C for 12 h. The next day,
   the membrane was washed with TBST and incubated in a secondary antibody
   at 4 °C for 2 h. The membrane was then imaged with a chemiluminescence
   imager (Touch Imager, e‐BLOT, China). The primary antibodies were as
   follows: 4‐HNE antibody (MAB3249, R&D SYSTEMS, USA), MDA antibody
   (NBP2‐59367, Novus, USA), ACTIN antibody (14395‐1‐AP, Proteintech,
   China), GPX4 antibody (ab125066, Abcam, UK), PTGS2 antibody
   (66351‐1‐Ig, Proteintech, China), HO‐1 antibody (10701‐1‐AP,
   Proteintech, China), ACSL4 antibody (ab155282, Abcam, UK), MyD88
   antibody (23230‐1‐AP, Proteintech, China), IRAK1 antibody (D51G7, Cell
   Signaling Technology, USA), IRAK4 antibody (4363, Cell Signaling
   Technology, USA), TRAF6 antibody (ab 40675, Abcam, UK), TRAF3 antibody
   (33640, Cell Signaling Technology, USA), IKKε antibody (3416, Cell
   Signaling Technology, USA), TBK1 antibody (38066, Cell Signaling
   Technology, USA), Phospho‐TBK1 antibody (5483, Cell Signaling
   Technology, USA), IRF‐7 antibody (77 260, Cell Signaling Technology,
   USA), Phospho‐IRF‐7 (24129, Cell Signaling Technology, USA), ALPL
   Antibody (MA5‐24845, Invitrogen, USA), Collagen I antibody (66761‐1‐Ig,
   Proteintech, China), RUNX2 antibody (12556, Cell Signaling Technology,
   USA), Osterix antibody (ab209484, Abcam, UK).

Cell Transfection

   For IRF7 knockdown, BMSCs were transfected with siRNA targeting IRF7
   using Lipofectamine RNAiMAX reagent (13778500, Thermo Scientific, USA)
   according to the instructions from the manufacturer. The siRNA
   sequences are listed in Table [185]S1 (Supporting Information). For
   IRF7 overexpression, BMSCs were transfected with a mouse
   IRF7‐overexpression plasmid (54123) purchased from HanBio Co. Ltd
   (China) using Lipofectamine 3000 reagent (L3000075, Thermo Scientific,
   USA) according to the instructions from the manufacturer.

RNA‐Sequencing

   BMSCs were co‐cultured with S. aureus (MOI:10) and E. coli (MOI:10) for
   12 h. The total RNA from both infected and uninfected BMSCs was
   extracted using a TRIzol reagent. The RNA samples were then sent to
   Sinotech Genomics Co. Ltd (Shanghai, China) for RNA‐sequencing
   analysis. A gene expression fold change (FC) ≥ 2‐fold and a Q value <
   0.05 were considered indicative of differentially expressed genes.

Immunofluorescence

   BMSCs, after the indicated stimulation, were fixed with 4%
   paraformaldehyde and then permeabilized with 0.3% Triton X‐100 in PBS.
   The samples were then blocked with goat serum solution and incubated
   with primary ACSL4 antibody (ab155282, Abcam, UK) for 12 h. The next
   day, samples were washed with PBST and incubated with the secondary
   Anti‐rabbit IgG‐AF488 antibody (4412, Cell Signaling Technology, USA)
   for 2 h. Subsequently, cells were washed and stained with phalloidin
   for 40 min and observed using confocal laser scanning microscopy.

ChIP Assay

   ChIP assays were performed as previously described using a SimpleChIP
   Plus Enzymatic Chromatin IP Kit (9005, Cell Signaling Technology,
   USA).^[ [186]^42 ^] Chromatin from BMSCs was cross‐linked with 1%
   formaldehyde and then subjected to sonic treatment to produce ≈400 bp
   fragments. These fragments were immunoprecipitated with an anti‐IRF7
   antibody (77260, Cell Signaling Technology, USA), normal IgG antibody
   (2729, Cell Signaling Technology, USA), and Histone H3 antibody (4620,
   Cell Signaling Technology, USA), and then pulled down by magnetic
   beads. The obtained DNA content was used for qRT‐PCR and agarose gel
   electrophoresis.

Dual‐Luciferase Reporter Assay

   The dual‐luciferase reporter assay was performed according to what was
   previously described.^[ [187]^34b ^] In brief, the Acsl4‐WT‐Fluc
   reporter plasmid was constructed using a pGL4.10 vector. A pRLuc‐SV40‐N
   reporter plasmid (D2762) was purchased from Beyotime (China). For
   plasmid transfection in BMSCs, the Lipofectamine 3000 reagent was used
   (L3000075, Thermo Scientific, USA) according to the instructions from
   the manufacturer. For siRNA and plasmid co‐transfection, the
   Lipofectamine 2000 reagent (11668027, Thermo Scientific, USA) was used
   according to the instructions from the manufacturer. The BMSCs were
   subjected to bioluminescence detection 48 h post transfection.

Osteogenic Induction Assay

   BMSCs treated by the indicated stimulation were induced to osteogenic
   differentiation by osteogenic medium containing 10% FBS, 5 mm
   β‐Glycerophosphate disodium salt hydrate (G9422, Sigma‐Aldrich, USA),
   50 µg mL^−1 ascorbic‐acid (1043003, Sigma‐Aldrich, USA), 1 × 10^−7 m
   dexamethasone (D4902, Sigma‐Aldrich, USA), 20 µg mL^−1 Gentamicin
   (S4030, Selleck, China) and 10 µg mL^−1 Oxacillin (S4974, Selleck,
   China). The osteogenic medium was replaced every 3 days. On Day 7 and
   Day 21 post‐induction, alkaline phosphatase staining and Alizarin Red S
   staining were performed, respectively.

qRT‐PCR

   qRT‐PCR was performed as previously described.^[ [188]^43 ^] In brief,
   total RNA from BMSCs was extracted with a TRIzol reagent. cDNA was
   reverse‐transcribed from the extracted RNA using a Script Reverse
   Transcription Kit (A2800, Promega, USA). qRT‐PCR was performed with
   SyBR Green qPCR Master Mix ([189]B21202, Selleck, USA) on a Real‐Time
   PCR platform (QuantStudio 6, Thermo Scientific, USA). The 2^−ΔΔCt
   method was used to analyze the relative expression level of mRNA.
   Primer sequences are listed in Table [190]S2 (Supporting Information).

HACC‐PBA Synthesis

   HACC with a substitution degree of 26% was synthesized according to the
   previous studies.^[ [191]^44 ^] In brief, 4 g chitosan (C105799,
   Aladdin, China) was dissolved in 200 mL 1% acetic acid solution and
   stirred for 3 h for complete dissolution. Then 7.5 g
   3‐Glycidoxypropyldimethoxymethylsilane (G298766, Aladdin, China) was
   dissolved in 40 mL 1% acetic acid and dropwisely added to the chitosan
   solution. The solution was then stirred for reaction at 80 °C for 24 h.
   The reaction product was packed in a dialysis bag (molecular weight
   cut‐off of 3500) and subjected to dialysis in ddH[2]O for 72 h. After
   dialysis, the product was frozen at −80 °C and went through
   lyophilization to obtain the final HACC product. For HACC‐PBA
   synthesis, a specified amount of 4‐formylphenylboronic acid was added
   to a solution of HACC (200 mg mL^−1) dissolved in deionized water. The
   mixture was stirred for 1 h at room temperature. Subsequently, an
   amount of NaBH3CN equivalent to three moles was added to the solution
   and the mixture was stirred for an additional 20 h. Finally, the
   product, PBA‐modified HACC, underwent dialysis and freeze‐drying. The
   resulting product was collected for further hydrogel synthesis.

Preparation of ROS‐Responsive Antimicrobial/Anti‐Ferroptosis Hydrogel

   The antimicrobial/anti‐ferroptosis hydrogel was formulated utilizing a
   dual‐cross‐linking system that combines the photo‐cross‐linking of
   PEGDA‐DTT polymers with the covalent cross‐linking between HACC‐PBA and
   PEGDA‐DTT. Initially, PEGDA‐DTT was prepared by mixing DTT
   (100 mg mL^−1) with PEGDA (150 mg mL^−1) at a specified ratio for 2 h,
   as detailed in the prior study.^[ [192]^45 ^] Subsequently, HACC‐PBA
   (ranging from 0% to 1.5%), Fer‐1 (to achieve a final concentration of
   20 mg mL^−1), and 0.5% LAP were introduced into the PEGDA‐DTT solution.
   This mixture was stirred for 20 min. Once homogenously mixed, the
   solution was poured into molds or applied onto 3D‐printed PCL/MBG/Fer‐1
   scaffolds. The final step involved photo‐cross‐linking the
   HACC‐DP/Fer‐1 hydrogel by exposing it to blue light (405 nm) for 20 s.

Characterization

   The synthesis of HACC and HACC‐PBA was characterized using both FTIR
   and NMR spectroscopy. FTIR spectra were obtained using an FT‐IR
   Spectrometer (FT/IR‐4600, JASCO, Japan). Meanwhile, NMR spectra were
   recorded using an NMR spectrometer (Bruker‐400 MHz, Bruker, USA). To
   assess the rheological properties of the HACC‐DP hydrogel, measurements
   were taken using a HAAKE rheometer (Thermo Scientific, USA) with 1%
   strain and 37 °C. To assess the shear‐thinning property of HACC‐DP
   hydrogel, the uncross‐linked hydrogel was subjected to a rheological
   test with a shear rate from 0.1–100 (1/s) at 37 °C. The compression
   modulus of the scaffold was measured using an electronic universal
   testing machine (5982, INSTRON, USA).

Scanning Electron Microscopy

   The morphology of bacteria treated with different scaffolds was
   observed with scanning electron microscopy. In brief, S. aureus
   inoculum was freshly activated by overnight shaking and diluted to 10^7
   CFUs mL^−1 in TSB. Then, different scaffolds were incubated in 100 µL
   bacteria inoculum at 37 °C for 6 h. The scaffolds were then fixed with
   2.5% glutaraldehyde and subjected to gradient dehydration. The
   dehydrated scaffolds were coated with gold for observation under a
   scanning electron microscope (S4800, HITACHI, Japan).

Construction of Antimicrobial/Anti‐Ferroptosis 3D‐Printed Scaffold for Bone
Repair

   PCL/MBG/Fer‐1 scaffolds were fabricated following the previously
   established protocols.^[ [193]^46 ^] Initially, a mixture powder of MBG
   and Fer‐1 was prepared. MBG and Fer‐1 were combined in a 7:1 mass ratio
   and then lyophilized to form the composite powder. PCL was dissolved in
   methylene chloride, and either the MBG/Fer‐1 composite or just MBG was
   introduced at a 4:9 mass ratio (either PCL: MBG/Fer‐1 or PCL:MBG). This
   blend was stirred thoroughly for 2 h. After allowing the organic
   solvent to partially evaporate, the uniform mixture was loaded into a
   3D printer (Biobuild‐S, Regenovo, China). 3D‐printing was conducted
   using the following parameters: temperature = 25 °C, the diameter of
   nozzle = 0.25 mm, speed = 1.8 mm s^−1, pressure = 3.0 bar, the distance
   between strips = 500 µm, the diameter of layer = 3 mm, scaffold
   height = 4 mm. After printing, the scaffold was allowed to fully dry to
   ensure complete evaporation of the organic solvent. Once dried, the
   scaffold was incorporated with the HACC‐DP/Fer‐1 hydrogel. This
   hydrogel was then photo‐cross‐linked under the illumination of blue
   light (405 nm) for 20 s.

Releasing and Degradation of HACC‐DP Hydrogel and PCL/MBG 3D‐Printed Scaffold

   Release and degradation patterns of Fer‐1 in HACC‐DP hydrogel and
   PCL/MBG scaffold were simulated by measuring the release of FITC in
   these two systems. In brief, HACC‐DP/FITC hydrogel and PCL/MBG/FITC
   scaffold were prepared with the same content of FITC as Fer‐1.
   HACC‐DP/FITC hydrogel was immersed in SBF or SBF containing 100 µm
   H[2]O[2] (volume ratio of HACC‐DP/FITC:SBF = 1:500). PCL/MBG/FITC
   scaffold was immersed in SBF or SBF containing 100 µm H[2]O[2] with a
   ratio of 1 g:100 mL. At predetermined time points, the supernatants
   were collected and stored at −20 °C in the dark until further analysis.
   Hydrogel and scaffold were washed with ddH[2]O and dried by
   lyophilization, and the mass was determined by an electronic balance.
   The release amount of FITC was calculated using a standard
   fluorescence‐concentration curve of FITC (Figure [194]S4F, Supporting
   Information).

Antimicrobial Assay

   The antimicrobial assay was performed as previously described.^[
   [195]^47 ^] In brief, S. aureus (ATCC 25923) was freshly activated by
   suspending a single bacteria colony in TSB broth. After incubation at
   37 °C for 3, 6, and 24 h with shaking, the bacterial concentration was
   counted and diluted to 10^6 CFUs mL^−1. Then,
   antimicrobial/anti‐ferroptosis 3D‐printed scaffolds containing
   different concentrations of HACC‐PBA in loaded HACC‐DP hydrogels were
   immersed in 100 µL of 1 × 10^6 CFU mL^−1 bacterial suspension and
   incubated at 37 °C for indicated periods. After that, 1 mL of sterile
   PBS was added into the tube and it was subjected to ultrasonic
   oscillation to sufficiently resuspend the bacteria. Serial dilution and
   drop plate methods were performed and the number of colonies was
   counted. All groups were normalized to the Ctrl group (TSB) in the
   loaded hydrogel.

CCK‐8 Assay

   The cell viability of BMSCs seeded on Fer‐1@HACC‐DP‐PCL/MBG scaffolds
   was measured by CCK‐8 assay (Beyotime Biotechnology, China) according
   to the manufacturer's instructions. In brief, 2 × 10^4 BMSCs were
   seeded on Fer‐1@HACC‐DP‐PCL/MBG scaffolds in 48‐well culture plates.
   The cell viability of BMSCs was measured by OD[450] on Day 1, 4, and 7.

Rat Model of Distal Femur Critical‐Sized Infected Bone Defect

   All operation procedures in the animal experiment had been reviewed and
   approved by the Animal Ethical Committee of the Shanghai Institute of
   Immunity and Infection, Chinese Academy of Sciences, with approval
   number No. A2023035. Adult male SD rats (250–300 g per rat) were
   obtained from the Charles River Animal Company. The rats were
   accustomed to breeding facilities for 4 days. On the day of the
   operation, rats were anesthetized with isoflurane. The skin near the
   knee joint was prepared and sterilized using a 75% ethanol solution.
   Next, a longitudinal incision was made on the inner side of the joint
   midline. The joint capsule was opened and the medial condyle of the
   femur was exposed. Then, a round defect with a diameter of 3 mm and a
   depth of 4 mm was made on the lateral side of the medial condyle using
   a trephine. After that, 50 µL of 2 × 10^8 CFUs mL^−1 S. aureus (Xen 29)
   suspension was injected into the defect sites. Bone wax or scaffolds
   were then implanted into defect sites. The wound was sutured layer by
   layer, and animals were monitored until they awakened from anesthesia.
   The animals were randomly designated to four groups: Bone wax group
   (control group), PCL/MBG scaffold group (without
   antibacterial/anti‐ferroptosis properties), HACC‐DP‐PCL/MBG scaffold
   (antibacterial scaffold), and Fer‐1@HACC‐DP‐PCL/MBG scaffold
   (antibacterial/anti‐ferroptosis scaffold). Each group contained 12
   rats, with 6 rats sacrificed at Week 6 and 6 rats sacrificed at Week 12
   post‐operation. At predetermined time points, IVIS was performed to
   monitor bacterial burden. Rats were separately sacrificed at Week 6 and
   Week 12. After that, the isolated femur was fixed with 4%
   paraformaldehyde solution and stored at 4 °C for further analysis.

Rat Model of Subcutaneous Infection

   All operation procedures in the animal experiment had been reviewed and
   approved by the Animal Ethical Committee of the Shanghai Institute of
   Immunity and Infection, Chinese Academy of Sciences, with approval
   number No. A2023035. Adult male SD rats (200 – 300 g per rat) were
   obtained from the Charles River Animal Company. The rats were
   accustomed to breeding facilities for 4 days. The HACC‐DP‐2/Fer‐1
   hydrogel and HACC‐DP‐0/Fer‐1 hydrogel were photo‐cross‐linked in a
   48‐well plate with a volume of 150 µl. On the day of the operation,
   rats were anesthetized with isoflurane. The skin on the back of the rat
   was prepared and sterilized with 75% ethanol. Then a longitudinal
   incision was made with sterile ophthalmic scissors, and a subcutaneous
   cavity was made by blunt dissection. In the uninfected HACC‐DP‐2/Fer‐1
   hydrogel group, HACC‐DP‐2/Fer‐1 hydrogel was implanted in the
   subcutaneous cavity and 50 µL PBS was injected into the cavity. In
   infected groups, HACC‐DP‐2/Fer‐1 hydrogel or HACC‐DP‐0/Fer‐1 hydrogel
   was implanted in the subcutaneous cavity, and 50 µL 2 × 10^8 CFUs mL^−1
   S. aureus (Xen 29) was injected into the cavity. The wound was sutured
   layer by layer, and animals were monitored until they awakened from
   anesthesia. Each group contained 3 rats. Rats were sacrificed on Day 14
   post‐operation and then the skin tissue on the back was separated and
   fixed in 4% paraformaldehyde. The tissue was then subjected to
   histological evaluation.

X‐Ray and Micro‐CT Analysis

   X‐ray radiography of femur from rats was performed using a small animal
   X‐ray imager (MX20, Faxitron Bioptics, USA). Micro‐CT scanning and 3D
   image reconstruction were performed using a Micro‐CT system (µCT80,
   Scanco Medical, Switzerland).

Histological Analysis

   Histological analysis was performed as previously described.^[ [196]^48
   ^] In brief, rat specimens fixed with 4% paraformaldehyde were
   decalcified for 2 months and then underwent paraffin embedding and
   sectioning. Tissue sections, 5 µm thick, were then stained with Masson
   trichrome, Giemsa, and immunofluorescence (IF) staining. The primary
   antibodies used for IF were listed as follows: OCN (23418‐1‐AP,
   Proteintech, China), LepR (20966‐1‐AP, Proteintech, China).

Statistical Analysis

   Independent experiments or repeated measurements were conducted for all
   data (n = 3, 4, or 6, respectively). All data were presented as
   mean ± standard deviation (SD). A P value less than 0.05 was considered
   statistically significant. Comparisons between the two groups were
   conducted using a two‐tailed Student's t‐test. Multiple comparisons
   were conducted using one‐way ANOVA followed by Tukey's post hoc test.
   For multivariate parametric data, two‐way ANOVA followed by Tukey's
   post hoc test was used. Analysis was conducted using Graphpad Prism
   Software (Version 8.0, Boston, USA).

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   K.Y., Y.Y., and Y.L. contributed equally to this work. K.Y., Y.Y., and
   Y.L. designed and performed the experiments and analyzed the data.
   K.Y., S.Y., H.L., Y.L., and T.T. conceived and designed the overall
   study as principal investigators. F.Z., Y.L., K.H., W.K., F.L., T.K.,
   Y.W., C.C., H.A., and Z.Y. performed the experiments and analyzed the
   data. H.C. and Y.L. contributed to the transmission electron microscopy
   imaging and analysis. J.H. contributed to the analysis of flow
   cytometry. All authors contributed to the writing and editing of the
   manuscript and approved the final manuscript.

Supporting information

   Supporting Information
   [197]ADVS-11-2404453-s001.docx^ (2.4MB, docx)

Acknowledgements