Mitochondrial micro-nano reactors via enhancing mitochondrial transfer and mitophagy for alleviating metabolic crisis Authors: Jiayi Mao, Wenzheng Xia, Minxiong Li, Xin Huang, Yun Zhao, Zheyuan Hu, Yinghong Su, Juan Wang, Wenguo Cui, Tao Zan Published: 2026 ABSTRACT AbstractEnergy metabolic dysfunction is a major cause of impaired chronic wounds healing, in which disrupted mitochondrial transfer and autophagy imbalance further aggravate the cellular energy crisis. In this study, single-cell RNA sequencing (scRNA-seq) of clinical diabetic wound samples first identified a pivotal role for macrophage-to-fibroblast mitochondrial transfer in wound healing. This finding was further validated using diabetic wound models and histological analyses, highlighting these processes as potential therapeutic targets for alleviating energy metabolic stress. Based on these findings, we innovatively developed a mitochondrial micro-nano reactor (MtNR) that alleviates the energy metabolic crisis by concurrently enhancing mitochondrial transfer and autophagy. First, hypoxic-preconditioning combined with gene-edited techniques was used to generate M2 macrophage-derived mitochondria-trained apoptotic bodies (mABs). Subsequently, mABs were conjugated with piezoelectric short fibers (PSFS) via copper-free strain-promoted azide–alkyne cycloaddition (SPAAC) click chemistry to self-assemble into MtNR. This system promotes intercellular mitochondrial transport through the Miro1–mitochondria–dynein–microtubule complex. It also generates bionic electrical signals via mechano-electrical conversion, thereby restoring Pink1–Parkin–P62/SQSTM1–LC3-mediated mitophagy and mitochondrial homeostasis. In a diabetic mouse wound model, MtNR restored mitochondrial morphology, enhanced cellular energy biogenesis, reduced p62 accumulation, increased LC3 expression, and significantly promoted tissue repair, providing a promising therapeutic strategy for addressing the energy deficit in diabetic wounds.Keywords: Energy metabolism, Apoptotic bodies, Piezoelectricity, Mitochondrial transfer, Mitophagy Introduction Wound healing is an energy-dependent process, and disrupted energy metabolism is a key factor in chronic wound formation and impaired repair [1]. As the central organelle of cellular energy metabolism, mitochondria continuously generate ATP and various metabolic intermediates through oxidative phosphorylation, playing a pivotal role in maintaining cellular homeostasis and facilitating tissue repair. However, in the diabetic microenvironment, persistent glucose dysregulation, oxidative stress, and chronic inflammation impair mitochondrial structure and function. As a result, cellular metabolism shifts from oxidative phosphorylation to less efficient anaerobic glycolysis [2,3]. This metabolic reprogramming results in insufficient ATP production, lactate accumulation, and intracellular acidification, which markedly delay key reparative processes such as keratinocyte migration, fibroblast proliferation, and angiogenesis, thereby severely hindering wound healing [4,5]. Under conditions of elevated glucose and oxidative stress, mitochondrial DNA (mtDNA) tends to accumulate mutations and deletions, largely due to its lack of histone shielding and insufficient repair capacity. These alterations further compromise mitochondrial function and exacerbate energy metabolic disturbances, forming a vicious cycle of persistent dysfunction [6,7]. Therefore, restoring mitochondrial function and reestablishing its energy metabolic capacity has emerged as a potential therapeutic target for alleviating the energy crisis in diabetic wounds [8]. Currently, strategies to repair impaired mitochondrial function in diabetic wounds mainly involve promoting mitochondrial biogenesis, alleviating oxidative stress, or supplementing exogenous mitochondria [9,10]. However, these approaches primarily focus on ameliorating mitochondrial damage and passively restoring function, with relatively limited and singular intervention targets. Their therapeutic efficacy is markedly constrained, particularly in cases of mitochondrial depletion or severe mtDNA damage [11]. Recently, intercellular mitochondrial transfer has drawn considerable interest as an emerging mechanism characterized by greater proactivity and broader dimensions of intervention. In this process, healthy cells transfer functionally intact mitochondria to damaged cells through structures such as tunneling nanotubes (TNTs), extracellular vesicles (EVs), or gap junctions, thereby rapidly restoring their energy metabolic capacity, modulating immune status, and reshaping metabolic phenotype and epigenetic features [[12], [13], [14], [15], [16], [17], [18]]. This mechanism has been demonstrated to be crucial for tissue repair in various diseases [14,19]. However, in tissue injury—particularly within the diabetic wound microenvironment—factors such as calcium overload, excessive reactive oxygen species (ROS), and increased pro-inflammatory cytokines can markedly suppress mitochondrial activity and disrupt the process of intercellular mitochondrial transfer [[20], [21], [22]]. These damaging factors not only impair the activity of donor cell mitochondria but also disrupt the formation of transfer structures such as TNTs, thereby limiting the efficiency of mitochondrial delivery. A recent Nature study revealed that exogenous mitochondria, upon entering recipient cells, do not directly integrate into the existing mitochondrial pool; instead, they tend to colocalize with autophagosomes and activate the PINK1–Parkin–dependent mitophagy pathway [23]. This process is considered an important mechanism by which cells perform quality screening and functional evaluation of heterologous mitochondria. If this pathway is blocked, the therapeutic efficacy of mitochondrial transfer will be hindered by the persistent dysfunction of mitochondria, which not only compromises mitochondrial stability but also impairs their sustained support for the metabolic functions of recipient cells [24]. Notably, high glucose levels and metabolic toxins in the diabetic wound microenvironment have been shown to severely impair mitophagy, resulting in the failure to promptly clear exogenous mitochondria within cells and the consequent difficulty in maintaining their functional integrity [25]. Therefore, promoting mitochondrial transfer alone is insufficient. Restoring autophagic homeostasis is equally critical to ensure the metabolic integration and functional performance of transplanted mitochondria. This represents a key breakthrough for overcoming therapeutic bottlenecks in diabetic wound healing. Mitochondrial autophagy-mediated clearance of dysfunctional or redundant mitochondria is crucial for regulating inflammatory responses, metabolic reprogramming, and cellular remodeling [26]. Cells have evolved multiple tightly regulated pathways to ensure timely and precise activation of mitophagy in response to stress [27]. However, when mitophagy is dysregulated, mitochondrial dysfunction and reduced cellular viability ensue, necessitating exogenous interventions to restore tissue regenerative capacity [28]. Current strategies to restore mitophagy—including mitochondrial uncouplers, membrane stabilizers, and gene editing techniques—often carry toxicity risks, poor controllability, and stability limitations [29,30]. More critically, existing approaches primarily promote mitochondrial clearance without addressing mitochondrial transfer or replenishment. Simply activating mitophagy neither enhances mitochondrial transfer nor restores functional mitochondrial content in energy-depleted tissues. Under specific pathological conditions, excessive mitophagy may even exacerbate mitochondrial depletion, leading to energy insufficiency and reduced cell survival. Therefore, an ideal therapeutic strategy should not only eliminate damaged mitochondria but also promote mitochondrial replacement and stabilize post-transfer quality control. Researches indicate that piezoelectric materials can restore mitophagy function by reconstructing bioelectric fields [31]. As a form of physical stimulation, bioelectricity-mediated autophagy offers advantages such as non-invasiveness, minimal damage, and precise controllability, making it more physiologically relevant. This approach not only facilitates the clearance of damaged organelles and the improvement of energy metabolism by correcting the PINK1–Parkin–mediated mitophagy pathway, but also helps modulate inflammatory responses, restore cellular homeostasis, and promote tissue regeneration in damaged or metabolically impaired environments [31,32]. Based on this, we proposed a cascading mitochondrial micro/nanoreactor strategy. This approach synergistically enhanced mitochondrial transfer while stabilizing post-transfer mitochondrial autophagy, thereby alleviating energy metabolism crises and promoting tissue repair. In this study, we integrated scRNA-seq data from clinically healed and non-healed diabetic wounds to investigate alterations in mitochondrial transfer and autophagy pathways, clarifying their link to the energy metabolic crisis. Differentially expressed gene (DEG) analysis and Gene Set Enrichment Analysis (GSEA) indicated that impaired healing is associated with reduced mitochondrial transfer and defective mitophagy. A diabetic wound model was then established, and histological analyses confirmed diminished mitochondrial transfer (primarily due to decreased Rho family genes (Miro1) expression) and autophagy status at the wound site, highlighting these pathways as critical targets for alleviating the metabolic crisis (Fig. 1A). Based on these observations, we conceptualized a mitochondrial micro-nano reactor (MtNR) and fabricated mitochondria-functionalized, preconditioned apoptotic bodies (mABs) through hypoxic training and gene-editing techniques. The prepared mABs were then grafted onto the piezoelectric short-fiber scaffold (PSFS) via copper-free click chemistry mediated by strain-promoted azide–alkyne cycloaddition (SPAAC) to construct the MtNR (Fig. 1B). The PSFS not only serves as a structural support for mAB immobilization but also provides piezoelectric stimulation to improve the wound microenvironment and facilitate mitophagy and cellular energy recovery. In vitro, the impact of the MtNR on rescuing the energy metabolic crisis was assessed by evaluating mitochondrial transfer efficiency, mitochondrial respiratory function and membrane potential, cellular mitochondrial ROS production, and mitophagy activity. In vivo, a full-thickness skin defect model in diabetic mice was used to evaluate the tissue regenerative capacity. Finally, we combined transcriptomic analysis with in vivo electrophysiological measurements to investigate mitochondrial energy rescue. These analyses elucidated the regulatory mechanisms by which the MtNR alleviates the energy metabolic crisis, providing new theoretical insights and therapeutic strategies for diabetic wounds (Fig. 1C). Illustration of the mitochondrial micro-nano reactor facilitating diabetic wound healing via synergistic regulation of mitochondrial transfer and mitophagy. A) Elucidation of mitochondrial transfer and mitophagy mechanisms in clinical diabetic wounds through scRNA-seq. B) Mitochondrial transfer capacity and biogenesis were enhanced via lentiviral-mediated Miro1 overexpression combined with hypoxic preconditioning, leading to the formation of mitochondria-trained apoptotic bodies (mABs). Piezoelectric short-fiber scaffolds (PSFS) were fabricated using electrospinning and short-fiber techniques. Through glycoengineering modifications on the membranes of donor cells, the mABs were non-destructively conjugated onto the PSFS via click chemistry. C) The mechanism of MtNR in treating diabetic wounds, including the promotion of intercellular mitochondrial transfer, mitophagy, and mitochondrial protection. Results and discussion Mitochondrial transfer between macrophages and fibroblasts and impaired mitophagy exacerbate delayed healing in diabetic wounds We analyzed scRNA-seq data from six healing and non-healing diabetic wound samples (GSE165816) to explore key cell types and potential therapeutic targets (Fig. 2A, Supplementary Fig. 1A). We first assessed sample demographics (age, sex; Table S1–2), and scRNA-seq quality (Supplementary Fig. 1A–G). From 137,655 skin cells, we identified 22 clusters corresponding to nine major cell types, including fibroblasts, macrophages, keratinocytes, endothelial cells (ECs), CD4+ T cells, monocytes, B cells, melanocytes, and NK cells (Fig. 2B, Supplementary Fig. 1H and Supplementary Fig. 2A–E). CellChat analysis revealed that macrophage-fibroblast interactions were markedly reduced in non-healing wounds compared to healing ones (Fig. 2C and Supplementary Fig. 3A). CellPhoneDB validation confirmed that ligand-receptor pairs between macrophages and fibroblasts dropped from 107 to 58, indicating suppressed intercellular crosstalk in non-healing wounds (Fig. 2D and E). Macrophages are key innate immune cells that regulate inflammation and tissue regeneration during wound healing [33,34]. Their metabolic plasticity enables adaptation to the dynamic wound microenvironment. Fibroblasts, meanwhile, are central to extracellular matrix (ECM) synthesis; stimulated by soluble factors and metabolites from M2 macrophages, they can differentiate into myofibroblasts to drive wound contraction and collagen deposition in early healing [35]. Thus, we hypothesize that impaired functional crosstalk between fibroblasts and macrophages may be critical to the defective healing of diabetic wounds. Dysfunction of mitophagy and mitochondrial transfer in diabetic wounds. A) schematic diagram of single-cell RNA sequencing study on mitochondrial transfer and autophagy mechanisms in clinical diabetic wounds. B) UMAP plot showing cell clusters in diabetic wounds. C) CellChat analysis showing the intercellular interaction strength in healing and non-healing diabetic wounds. D-E) CellPhoneDB validation and quantification of intercellular interactions in healing and non-healing diabetic wounds. F) Chord diagram showing KEGG pathway enrichment and differentially expressed genes in macrophages from both groups. G) Volcano plot analysis highlighting differentially expressed genes. H) GSEA of the mitochondrial transfer–related gene set among the differentially expressed genes. GO analysis of I) macrophages and J) fibroblasts in diabetic wounds under different healing conditions. We further performed Differentially Expressed Gene (DEG) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses on these two key cell types. In wounds with distinct healing outcomes, macrophage-associated DEGs were mainly enriched in mitochondrial oxidative phosphorylation, cytoskeleton regulation, and endocytosis pathways (Fig. 2F). Volcano plots of DEGs revealed that, compared with the healing group, non-healing group macrophages had 1519 downregulated and 370 upregulated genes, while fibroblasts had 693 downregulated and 629 upregulated genes—indicating profound shifts in their metabolic activity and function. Using the mitochondrial transfer gene signature by Baldwin et al. [11], we found several genes differentially expressed between healing and non-healing wounds. ACTB, MYL6, CDC42SE2, and RHOC were downregulated, whereas KLF2 and ZEB2 were upregulated, suggesting impaired mitochondrial transfer in non-healing wounds (Fig. 2G). GSEA revealed significant enrichment of mitochondrial transfer–related genes in macrophages and fibroblasts, indicating suppressed mitochondrial transfer function of these two cell types in non-healing diabetic wounds. Notably, key genes involved in mitochondrial transport and TNT formation were markedly downregulated in non-healing wounds. These included cytoskeletal protein ACTB, myosin-II complex members MYL6 and MYL12A, small GTPases RHOC and RAB5C, and CDC42 downstream effector CDC42EP1 (Fig. 2H, Supplementary Fig. S3B and Supplementary Fig. 4A–B). These genes regulate actin assembly and cell morphology to support TNT formation and function. Gene Ontology (GO) enrichment analysis further indicated that organelle membrane association, metabolic functions and protein binding were the main co-regulated biological processes in both groups (Fig. 2I and J). Given ECs' role in angiogenesis and skin regeneration in the metabolically impaired diabetic wound microenvironment, their scRNA-seq data were analyzed. Mitochondrial transfer–related genes and GO pathways in ECs were also significantly downregulated. In addition, differential proportions of macrophages and ECs between healing states suggested that ECs are important for diabetic wound healing (Supplementary Fig. 4C–E). To verify single-cell transcriptomic findings from clinically healing/non-healing diabetic wounds, a diabetic rat model was established via intraperitoneal streptozotocin (STZ) injection. Skin tissues from successfully induced rats were subjected to histological analysis and compared with healthy controls. H&E and Masson staining confirmed diabetic skin metabolic dysfunction, with decreased hair follicles and collagen fibers. Transmission electron microscopy (TEM) revealed marked mitochondrial ultrastructural abnormalities in diabetic wound tissues, including swelling, disrupted cristae, and vacuolization (Fig. 3A–F). Western blotting results showed marked downregulation of Miro1 in diabetic skin. Miro1 is a small GTPase on the outer mitochondrial membrane. It acts as an adaptor linking mitochondria to microtubule motors (dynein), mediating intercellular mitochondrial movement. Reduced Miro1 levels suggest impaired mitochondrial trafficking [[36], [37], [38]]. Validation of mitochondrial transfer and autophagy in the skin of diabetic rats. Representative images of skin tissues from normal and diabetic rats: A) H&E staining; B) Masson trichrome staining; C) TEM images, with D–F) quantitative analyses (n = 5 biologically independent samples). G) Western blot of mitochondrial transfer marker (Miro1) and mitophagy markers (PINK1, P62, Parkin, and LC3) in skin tissues from normal and diabetic rats. H) Representative immunofluorescence staining of mitochondrial transfer marker (Miro1) and autophagy markers (LC3 and P62) in skin tissues from normal and diabetic rats, with I) corresponding quantitative analysis (n = 5 biologically independent samples). J) Bubble plot illustrating CD163 and FAP gene expression across various cell types. K) Feature gene plots validating cluster assignments using representative cell-specific marker genes. L) Immunofluorescence staining of fibroblast marker COL-1 (red) and FAP (green), and M2 macrophage markers CD163 (red) and CD68 (green) in diabetic rat skin tissues. The significance between the two groups was calculated using a two-tailed Student's t-test. Data are presented as mean ± SD (n = 5 independent animal samples for D, E, F, I). Meanwhile, decreased Pink1 and Parkin, upregulated P62, and reduced LC3-II/LC3-I ratio indicated significant mitophagy suppression (Fig. 3G). Further scRNA-seq analysis of autophagic function in healing vs. non-healing diabetic wounds revealed cell-type-specific mitophagy trends. Specifically, macrophages—key phagocytic cells—showed elevated mitophagy in non-healing wounds, potentially related to enhanced clearance of damaged mitochondria (Supplementary Fig. 5A and C). In contrast, mitophagy was significantly inhibited in fibroblasts, which lack primary phagocytic function (Supplementary Fig. 5B and D). Combined with Western blot results, these findings indicate that while mitophagy is generally inhibited in diabetic wounds, it exhibits an imbalance across different cell types. Thus, we aimed to achieve a certain degree of mitophagy homogenization between the two cell types via mitochondrial transfer. Immunofluorescent analysis showed decreased Miro1 and LC3 levels accompanied by P62 accumulation, indicative of autophagy suppression in diabetic wounds (Fig. 3H and I). Based on the above results, we hypothesize that impaired mitochondrial transfer and mitophagy are key mechanisms underlying dysregulated immunometabolism in diabetic wounds. Selectively optimizing mitochondrial transfer and mitophagy between macrophages and fibroblasts may help maintain a balanced pool of functional and dysfunctional mitochondria in diabetic wounds, thereby synergistically improving cellular functional status and representing a potential therapeutic strategy. To develop precise targeted therapies, we systematically evaluated surface expression profiles and characteristic gene UMAP maps of pathological cells within diabetic lesion tissues. high expression of macrophage-specific marker CD163 and fibroblast-specific marker FAP in rat skin tissues (Fig. 3J and K). Immunofluorescence validation confirmed CD163 colocalization with macrophage marker CD68 and FAP colocalization with fibroblast marker Col1, verifying elevated CD163 and FAP expression in diabetic-associated macrophages and fibroblasts, respectively (Fig. 3L). These findings indicate CD163 and FAP are promising markers for selective identification and specific intervention of cells in diabetic skin lesions. Construction and validation of mitochondria-function-trained and glycoengineered macrophages Apoptosis, programmed cell death that eliminates senescent or dysfunctional cells without releasing danger signals, is critical for development, growth, and immune homeostasis [39]. During apoptosis, cells undergo morphological and functional changes to form apoptotic bodies (ABs) [40]. Initially, ABs were considered biologically inert cellular fragments. ABs are now known to carry DNA, RNA, proteins, and organelles like mitochondria, enhancing recipient cell function upon transplantation [19,41]. ABs also mediate intercellular communication and contribute to mesenchymal stem cell–driven immunomodulation [39,42,43]. Notably, mitochondria within ABs retain respiratory function despite structural changes [44], and specialized donor cell treatments can functionalize ABs by transferring bioinformation into them [45]. In contrast to exosomes or microvesicles, apoptotic bodies exhibit larger dimensions (500–5000 nm) and a greater capacity for encapsulating intact organelles [46]. Their formation mechanism, involving membrane budding and rupture, inherently results in a rich organelle content [47]. Moreover, apoptotic bodies themselves emit signals such as the chemokine CX3CL1 (‘find me’) and phosphatidylserine (‘eat me’) to attract macrophages for phagocytosis [48]. Consequently, apoptotic bodies represent a superior carrier choice in studies targeting the delivery of functional mitochondria. To validate that bioengineered parent cells can transfer functionality to ABs, we generated mitochondria-trained parent cells and induced their apoptosis to form mitochondria-trained apoptotic bodies (mABs). Based on parental homology, M2 macrophages were selected as parent cells to construct M2-mABs, aiming for specific targeting of M2 macrophages and metabolic support [49,50]. We hypothesize these mABs enhance mitochondrial transfer and mitophagy in recipient M2 macrophages, thereby modulating metabolism and function of surrounding recipient cells [51,52]. The mAB production process involved three steps (Fig. 4A): First, Raw264.7 cells were lentivirally transduced to overexpress Miro1, selected with puromycin for stable clones, and polarized into M2 macrophages using IL-4. Second, Miro1-overexpressing M2 macrophages underwent hypoxic training with 1% cobalt chloride (CoCl2) to enhance mitochondrial biogenesis. Third, Mitochondria-trained M2 macrophages were induced to apoptosis using staurosporine (STS), and mABs were harvested by differential centrifugation and stored at −80 °C for further characterization. Production of mABs and their metabolic glyco-engineering. A) Schematic illustration of the generation of mABs, including Miro1 overexpression, hypoxic preconditioning, glycan metabolic engineering, and STS-induced ABs production. B) Representative western blots showing Miro1 and TOM20 levels in untreated ABs versus mitochondria-trained ABs (Miro1 OE and hypoxia), and C) their quantitative analysis (n = 3 independent samples). D) Representative TEM images and corresponding statistical analysis comparing untreated and hypoxia-induced ABs (n = 5 independent samples). E) Representative immunofluorescence staining of TOM20 in untreated versus hypoxia-induced ABs, and the quantitative analysis (n = 5 independent samples). F) Schematic illustration of the glycan metabolic engineering of macrophages using Ac4ManNAz-DP. G) Molecular dynamics simulation showing the spatial presentation and surface accessibility of azide-functionalized sialic acids. H) Representative fluorescence images of macrophages treated with different Ac4ManNAz-DP concentrations and stained with DBCO–PEG–rhodamine110, with quantification of mean fluorescence intensity (n = 5 independent samples). I) Flow cytometry histograms showing azide incorporation efficiency under different doses of Ac4ManNAz-DP, and quantification of mean fluorescence intensity (MFI) normalized relative to the group receiving the highest dose. J) Representative TEM images (upper) and fluorescence microscopy images (lower) of azide-modified ABs (N3-ABs). K) DLS analysis of particle size distribution of ABs and N3-ABs. L) Zeta potential of ABs and N3-ABs (n = 5 independent samples). M) Representative dual-immunofluorescence images of Miro1 and DiO in ABs derived from M2 macrophages under Miro1 OE (upper) and hypoxia induction (lower), and N) colocalization analysis. O) Representative immunofluorescence images showing tunneling nanotubes between Raw 264.7 and NIH3T3 cells with mitochondrial transfer (yellow dashed box). The significance between the two groups was calculated utilizing the two-tailed Student's t-test. Data are presented as mean ± SD (n = 3 independent samples for C; n = 5 independent samples for D, E, H, L). Studies have shown that recipient cells can improve viability upon acquiring exogenous mitochondria through natural intercellular mitochondrial transfer or artificial transplantation [53]. However, isolated donor mitochondria are prone to irreversible damage when exposed to high Ca2+ concentrations in culture medium. In contrast, mitochondria encapsulated within extracellular vesicles (EVs) maintain a certain level of respiratory activity in vitro [31,54]. We thus used TEM to examine mitochondria in isolated naked macrophage mitochondria and in ABs. Results revealed that naked mitochondria exhibited pronounced swelling and severe cristae disruption, whereas mitochondria within mABs showed substantially less damage (Supplementary Fig. 6). Miro1, a Rho GTPase family protein, is anchored to the outer mitochondrial membrane via its C-terminal transmembrane domain, while its cytoplasmic GTPase and calcium-binding domains mediate mitochondrial distribution and microtubule-dependent transport [11]. In this study, Raw264.7 cells were transduced with the lentiviral vector CV186 (Ubc-MCS-3FLAG-SV40-Cherry-IRES-puromycin) to overexpress Miro1 (Miro1 OE). Miro1 protein levels were markedly upregulated in the Miro1 OE group versus controls (Fig. 4B and C). Notably, mammalian cells sense oxygen levels and coordinate multiple adaptive responses under hypoxia [55]. Moderate hypoxia induces PGC-1α expression via the AMPK pathway to promote mitochondrial biogenesis [13,56], reduce intracellular superoxide in mesenchymal stem cells (MSCs), enhance mitochondrial membrane potential, and remodel metabolism in recipient cells [10,14]. Here, we applied hypoxic training to donor macrophages to enhance mitochondrial reserve and function. Western blot revealed significantly higher Tom20 expression in CoCl2-treated macrophages vs. normoxic controls (Fig. 4B and C). TEM revealed normoxic mitochondria were a mix of elongated tubules and short fragments, whereas 24 h hypoxia increased mitochondrial fragmentation, reduced elongation, decreased individual mitochondrial area, but increased total mitochondrial area (normoxia vs. hypoxia: 4.10 ± 1.50 vs. 7.01 ± 0.89 μm2) and number (9.00 ± 2.55 vs. 22.80 ± 4.08 per field) (Fig. 4D). Consistently, Tom20 immunofluorescence showed stronger signal under hypoxia (Fig. 4E), indicating hypoxia-induced mitochondrial biogenesis. For mABs assembly design, we first focused on chemically accessible and modifiable cell surface regions. Numerous membrane-embedded glycans form a dense, mesh-like glycocalyx on the cell surface, mediating direct environmental interactions. This glycocalyx landscape is dynamically reshaped by endogenous glycosylation, non-natural chemical groups from sugar derivatives to be incorporated [57]. Based on this, we employed an azide-modified derivative (Ac4ManNAz-DSPE-PEG, Ac4ManNAz-DP) to functionalize terminal sialic acid residues on surface-exposed glycoproteins, enabling their use as crosslinkers (Fig. 4F). Molecular dynamics (MD) simulations were conducted to examine the interactions between N3 groups on the mABs surface and DBCO moieties in PSFS during SPAAC click chemistry (Fig. 4G and Supplementary Fig. 7A–E). N3 on mABs surface and DBCO on the PSFS surface clustered via non-covalent hydrogen bonding, ultimately forming stable covalent linkages through click chemistry. Representative MD trajectories showed N3 was rapidly attracted to, contacted, and covalently bound to DBCO within 60 ns (Supplementary Fig. 8A–H, Movie S1). After geometric optimization, the van der Waals interaction energy between DBCO and N3 was −165.27 kcal/mol, indicating spontaneous binding (Supplementary Fig. 9A). Radial distribution function (RDF) analysis revealed preferred interaction distances. The O-H pair exhibited a sharp peak at ∼0.97 Å, consistent with typical O-H covalent bond length, indicating strong local ordering. The N-H pair peaked at ∼1.31 Å, reflecting hydrogen bonding or direct coordination. While the P-H RDF displayed a broader peak at ∼2.71 Å, indicating weaker and more diffuse interactions. These simulations support the spontaneous assembly of N3-functionalized mABs onto DBCO-containing materials, establishing an interconnected and robust micro/nano structural network (Supplementary Fig. 9B–D). As shown in Supplementary Fig. 10, electrostatic, van der Waals, and total energies fluctuated initially and gradually stabilized over time. Electrostatic energy showed larger fluctuations, reflecting dynamic rearrangements of charged groups, while van der Waals energy converged rapidly, indicating stabilization of non-bonded interactions. Total potential energy decreased and reached a plateau, demonstrating that the system achieved equilibrium. For surface functionalization, macrophages were treated with 0–100 μM Ac4ManNAz under standard culture conditions for 24 h. Conjugating with fluorescent DBCO-PEG-Rho119 confirmed glycometabolic engineering incorporated azide-modified sialic acids into the macrophage glycocalyx, providing reactive sites for SPAAC-mediated linkage (Fig. 4H). Flow cytometry validated this process: Ac4ManNAz treatment reduced unmodified macrophages and increased surface azide incorporation in a dose-dependent manner (Fig. 4I). Preparation and characterization of mitochondria-trained apoptotic bodies (mABs) Metabolic glyco-engineering offers a rapid, one-step passive strategy whereby cells autonomously functionalize surfaces through endogenous biosynthetic pathways under routine culture conditions, without external coupling reagents or specialized reaction environments. We characterized vesicle morphology and surface chemistry pre- and post-modification using TEM, immunofluorescence staining (IF), dynamic light scattering (DLS), and Zeta potential analysis. TEM showed both ABs and N3-ABs had vesicular morphology (average diameter ∼500 nm), with N3-ABs slightly larger than unmodified ABs (Fig. 4J, upper). DBCO-PEG-Rho110 fluorescence colabeling confirmed successful surface azide functionalization (Fig. 4J, lower). Consistently, DLS revealed N3-ABs (from metabolically engineered macrophages) had a slightly increased hydrodynamic diameter (≈596.76 nm) vs. unmodified ABs (≈577.19 nm) (Fig. 4K). Zeta potential analysis showed both ABs and N3-ABs carried negative surface charges; although N3-ABs had mildly reduced negative potential, there was no significant difference (Fig. 4L). Since azide functionalization only introduces terminal –N3 group onto the glycocalyx without altering the vesicle core, overall vesicle physical size remains largely unchanged. Therefore, the minor diameter increase may be attributed to surface charge distribution changes and associated solvation layer alterations post-chemical modification. Given that metabolically engineered macrophages exhibited enhanced mitochondrial training—evidenced by elevated Tom20 and Miro1 expression in previous experiments—we next investigated whether this parental mitochondrial training effect could be transferred to mABs. Extracted mAB membranes were labeled with DiO and immunostained for Miro1 and Tom20. Confocal imaging revealed significantly increased fluorescence intensity in mABs derived from Miro1-overexpressing cells than control ABs, while hypoxic preconditioning further upregulated Tom20 expression. These findings indicate that mitochondrial training of parental macrophages is efficiently inherited by their derived vesicles (Fig. 4M and N). To evaluate targeting specificity, immunofluorescence staining was used to examine selective mABs uptake in a macrophage-fibroblast co-culture system. Results showed Dil-labeled mABs exhibited significantly higher colocalization with CD163+ macrophages than FAP+ fibroblasts, demonstrating robust macrophage-targeting capability of the engineered mABs (Supplementary Fig. 11). Based on these findings, we hypothesized that mAB-treated macrophages would have enhanced mitochondrial transfer ability to fibroblasts. To test this, we co-cultured Mito-GFP (+) macrophages with DsRed (+) fibroblasts. After 16 h of incubation, macrophages and fibroblasts were labeled with F4/80 and FAP, respectively, and cellular interactions were visualized using confocal microscopy. Mitochondrial colocalization with TNTs demonstrated bidirectional mitochondrial transfer between macrophages and fibroblasts (macrophages to fibroblasts and vice versa). However, the green fluorescence signal within TNTs was markedly stronger than the orange signal, indicating predominant mitochondrial transfer from macrophages to fibroblasts. Bidirectional mitochondrial exchange occurred both before and after mAB treatment. However, green fluorescence intensity in TNTs increased markedly after mAB exposure, indicating enhanced mitochondrial transfer from macrophages to fibroblasts. This effect is likely due to increased functional mitochondria in macrophages following mAB engulfment (Fig. 4O and Supplementary Fig. 12). An electric field of 100 mV/mm represents the optimal stimulation intensity for enhancing mitochondrial respiratory chain function in macrophages Accumulation of excessive dysfunctional mitochondria in cells not only aggravates local inflammation and impairs energy metabolism but also hinders the transplantation and functional integration of exogenous mitochondria. Accordingly, inducing selective clearance of abnormal mitochondria and mitigating oxidative stress is considered an effective strategy to overcome diabetes-associated wound repair obstacles. In recent years, piezoelectric nanomaterials have garnered increasing attention, as they can generate bioelectrical signals under physiological conditions and directly interact with biological tissues [58]. Recent evidence indicates that bioelectrical cues from piezoelectric materials effectively activate mitophagy, thereby enhancing mitochondrial quality control, restoring cellular function, and promoting tissue regeneration [31,59]. To assess how electrical stimulation modulates mitochondrial respiratory chain activity in macrophages, controlled stimulation was delivered to cultured cells via an electrochemical workstation (Fig. 5A). Macrophages were stimulated at 50 or 100 mV/mm for 5–60 min. CCK-8 analysis after 24 h showed that 30-min stimulation (50 or 100 mV/mm) and 60-min stimulation (100 mV/mm) significantly increased proliferation. Evaluation of optimal electrical stimulation parameters and underlying mechanisms in vitro. A) Schematic illustration of the electrical stimulation culture system. B) CCK-8 assay showing the effects of different stimulation durations and intensities on macrophage proliferation (n = 5 independent samples). C) Intracellular ATP levels after stimulation at different intensities (n = 5 independent samples). D) Representative macrophage migration images at distinct stimulation levels, and E) their quantitative analysis (n = 5 independent samples). F) JC-1 staining profiles of macrophages assessed by flow cytometry under different stimulation intensities and G) statistical analysis (n = 5 independent samples). H) Analysis of mitochondrial respiratory activity via OCR before and after stimulation (n = 3 independent samples). I) Quantification of maximal respiration, basal respiration, proton leak, ATP production, and non-mitochondrial respiration (n = 3 independent samples). J) Chord plot of differentially expressed metabolites (DEMs) and KEGG pathway enrichment (n = 3). K) Volcano plot of DEGs in macrophages before and after electrical stimulation (n = 3). L) Heatmap of DEGs (n = 3). M) Enriched KEGG pathways identified from DEGs (n = 3). N) Western blot assessment of Miro1, Pink1, P62, Parkin, and LC3 in macrophages under different stimulation intensities. O) Schematic diagram of how the reduction in Ref. [60] content maintains cristae integrity, regulates mitochondrial/vascular function, and inhibits the release of mitochondrial DNA. The significance between the two groups was calculated utilizing the two-tailed Student's t-test. The significance between groups was calculated utilizing the One-way ANOVA followed by Tukey's test. Data are presented as mean ± SD (n = 5 independent samples for B, E; n = 4 independent samples for C; n = 3 independent samples for G, I). The maximal effect was observed with 30-min stimulation (Fig. 5B). Macrophages were subsequently divided into four groups: control (untreated), 50 mV/mm for 30 min (50 mV/mm), 100 mV/mm for 30 min (100 mV/mm), and 100 mV/mm for 60 min (100 mV/mm+). Cellular activity was assessed by ATP quantification and scratch assays. No significant differences in ATP generation or migration velocity were observed between the control and 50 mV/mm groups, indicating 50 mV/mm was insufficient to activate macrophages. In contrast, the 100 mV/mm group exhibited significantly increased ATP levels and enhanced migration relative to control. Further comparison showed no significant difference in ATP production or migration speed between the 100 mV/mm and 100 mV/mm+ groups, suggesting prolonged stimulation did not further enhance cellular activity (Fig. 5C–E). Collectively, these data indicate 100 mV/mm (30 min) represents the optimal stimulation condition for activating the mitochondrial respiratory chain in macrophages, markedly improving cellular activity. Oxidative stress-induced excessive ROS disrupts mitochondrial membrane potential (ΔΨm), causing depolarization and dysfunction. JC-1 staining (higher red/green ratio = preserved ΔΨm) showed both stimulation intensities increased the ratio vs. control, with 100 mV/mm more effective (Fig. 5F and G). These suggests electrical stimulation may maintain mitochondrial function or clear damaged mitochondria. Mitochondria's core function is ATP production via aerobic respiration. To further evaluate electrical stimulation's impact on mitochondrial respiration, we measured macrophage respiratory chain activity in stimulated and control groups (Fig. 5H), including basal respiration, ATP production, maximal respiration, spare respiratory capacity, proton leak, and non-mitochondrial respiration. Electrical stimulation markedly enhanced mitochondrial function: basal respiration increased from 55.31 ± 2.89 to 91.71 ± 6.03 pmol/min, ATP production from 24.45 ± 9.51 to 56.32 ± 4.83 pmol/min, and maximal respiration from 152.1 ± 6.18 to 205.0 ± 4.27 pmol/min. Collectively, these data demonstrate improved mitochondrial respiratory efficiency, indicating that electrical stimulation effectively activates the macrophage respiratory chain and augments cellular energy supply (Fig. 5I). Transcriptomic and metabolomic analyses reveal the mechanisms of electrical stimulation Previous studies have demonstrated that metabolic intermediates such as succinate and lipids regulate eukaryotic gene expression and inflammatory responses [61]. To clarify the metabolic mechanisms driving electrical stimulation's effect on mitochondrial respiration, we performed metabolomic profiling to identify and quantify differentially expressed metabolites (DEMs) between control macrophages and those stimulated at 100 mV/mm. After data normalization, Principal component analysis (PCA) analysis revealed clear separation among control, stimulated, and QC samples (Supplementary Fig. 13A). In total, 512 DEMs were identified following stimulation, including 163 up-regulated and 349 down-regulated metabolites (Supplementary Fig. 13B). A circular heatmap confirmed tight clustering of QC samples and distinct metabolite distribution alterations, indicating electrical stimulation's pronounced regulatory effect on the cellular metabolic landscape (Supplementary Fig. 13C). KEGG analysis revealed DEMs were primarily involved in lipid metabolism and amino acid metabolism (Fig. 5J), which are critical for regulating macrophage activation and function [62]. Metabolomic profiling showed marked alterations in lipid metabolism-related metabolites post-stimulation. Upregulation of choline and glycohyodeoxycholic acid suggests enhanced phospholipid synthesis and bile acid recycling; increase in the unsaturated fatty acid intermediate 5-dihydroxyhept-6-enoic acid indicates elevated fatty acid oxidation; downregulation of pro-inflammatory lysophosphatidylcholines (LPCs) reflects improved membrane lipid homeostasis and attenuated lipid-mediated inflammatory signaling. Collectively, these changes support a metabolic shift toward restored lipid balance (Supplementary Fig. 13B). Subsequently, we examined transcriptomic profiles pre- and post-electrical stimulation to examine metabolite alteration effects on gene expression. PCA demonstrated distinct global gene expression patterns between groups (Supplementary Fig. 13D). Compared with controls, stimulated macrophages exhibited 69 upregulated and 151 downregulated differentially expressed genes (DEGs) (Fig. 5K). Upregulated anti-inflammatory (Rgs16, Smad7), mitochondrial energy metabolism (OXA1L, ESRRA, Rab11fip5, CPT1A) and pro-regenerative (Klf11, Col15a1) genes, and downregulated inflammatory mediators (Jun, Tnfsf18, TNFAIP3, BCL6) (Fig. 5L, Supplementary Fig. 13E–F). These indicate 100 mV/mm for 30 min stimulation promotes mitochondrial energy restoration and suppresses inflammation. KEGG enrichment demonstrated electrical stimulation modulates mitochondrial function by lipid metabolism and inflammatory signaling (Fig. 5M). Furthermore, transcriptomic sequencing showed significant suppression of the cGAS-STING signaling pathway (Supplementary Fig. 14). Collectively, the data demonstrate that electrical stimulation induces metabolic reprogramming in macrophages, significantly enhancing mitochondrial oxidative capacity. Overall, multi-omics integration revealed a reduction in Arg1 (maintains cristae integrity, regulates mitochondrial/vascular function, restricts mtDNA release [63,60]) (Fig. 5O). Previous studies showed electrical stimulation improves disuse muscle atrophy [32] and cognitive function [64] in vascular dementia (VD) rats by activating PERK signaling and Parkin/PINK‒mediated mitophagy. Autophagy is a key for cellular homeostasis: reduced mitochondrial membrane potential leads to PINK1 accumulation on the outer mitochondrial membrane, activating E3 ubiquitin ligase Parkin and initiating mitophagy. Subsequently, autophagy receptor p62 binds ubiquitinated substrates and interacts with LC3‒II to deliver damaged mitochondria to autophagosomes for degradation (characterized by increased LC3‒II and reduced p62) [26]. Similarly, recent work reported piezoelectric nanosheets generate electrons under ultrasound stimulation, consuming outer mito chondrial membrane H+, disrupting matrix H+ supply, inducing mitochondrial membrane potential depolarization, thereby triggering mitophagy within inflamed regions to eliminate sources of ROS regeneration [59]. To investigate whether electrical stimulation modulates macrophage autophagy, we assessed autophagy-related protein levels via Western blot. Compared with the control, Miro1 expression showed no appreciable difference, indicating minimal impact of electrical stimulation on mitochondrial transfer. In contrast, Pink1, Parkin, and LC3-II levels were markedly increased, while LC3-I and p62 were significantly reduced in the stimulated group. These results suggest that electrical stimulation enhances mitophagy through the PINK1/Parkin pathway, thereby improving mitochondrial function and reducing damaged mitochondrial accumulation (Fig. 5N). Consistently, immunofluorescence analysis of PINK1 and Parkin before and after ES further supported this conclusion (Supplementary Fig. 15). Additionally, electrical stimulation upregulated CD31 and TOM20 expression in ECs, which may contribute to angiogenesis within diabetic wounds (Supplementary Fig. 16A–C). Finally, in vivo therapeutic evaluation showed electrical stimulation significantly accelerated wound closure in diabetic mice vs. control (Supplementary Fig. 17A–C). H&E and Masson staining further revealed a marked reduction of inflammatory cells in the stimulated group, while collagen deposition remained limited, which may limit overall tissue regeneration quality. Therefore, while electrical stimulation facilitates wound closure, additional synergistic strategies are required to enhance tissue regeneration and repair quality (Supplementary Fig. 17D–E). Fabrication and characterization of the MtNR MtNR was constructed by conjugating mABs to 3D piezoelectric short-fiber scaffolds (PSFS) via SPAAC click chemistry, ensuring efficient and bio-orthogonal attachment of mABs. PSFS was prepared as previously described [65,66]. A stable emulsion was obtained by high-speed stirring a 10% PLLA/collagen solution (4:6, v/v) at room temperature [67,68]. Aligned electrospun (ES) nanofiber membranes were prepared by electrospinning (16 kV, 50 μL/min feed rate, 1500 rpm collector rotation) (Fig. 6A and Supplementary Fig. 18A). Membranes were then mechanically homogenized at 12,000 rpm to obtain micron-scale short fibers, dispersed in tert-butanol, lyophilized for 24 h, and thermally crosslinked to form a 3D nanofiber network (Fig. 6B). showed scaffold shrinkage/stiffening when >30 min (likely due to collagen melting and structural collapse), so 30 min was selected. SEM showed aligned, uniform, smooth ES fibers (diameter – μm, length 10–110 μm, angle 10–60°) and loose, porous, randomly oriented short-fiber scaffolds (diameter ∼2–8 μm, likely due to freeze-drying swelling; length 6–36 μm, angle −100 to 50°) (Fig. 6C, Supplementary Fig. 18B and C). Synthesis and characterization of MtNR. A) Schematic illustration of MtNR preparation. B) Images of PSFS at different crosslinking times. C) SEM images of ES and PSFS. D) FTIR spectra of collagen, PLLA, ES, and PSFS. E) Piezoelectric curves of PSFS under different ultrasound intensities measured by an oscilloscope. Piezoelectric response curves of F) ES and G) PSFS under applied mechanical force measured by an electrometer. H) XRD and I) FTIR spectra of DBCO-PSFS, ABs@DBCO-PSFS, and N3-ABs@DBCO-PSFS. J) Compressive cycling curves of PSFS under different cycle numbers. K) Mapping images of PSFS. Piezoresponse force microscopy (PFM) analysis: L) amplitude image, M) phase images, N) D33 curves, O) phase curves, P) hysteresis loops, and Q) butterfly loops. R) SEM images and S) quantitative analysis of the remaining payload on ABs@PSFS and N3-ABs@DBCO-PSFS at different time points. T) Release profiles of conventionally loaded ABs and click-chemistry-loaded ABs under simulated physiological conditions (37 °C, SBF). The significance between the two groups was calculated using the two-tailed Student's t-test. Data are presented as mean ± SD (n = 5 independent samples for S, T). Fourier transform infrared spectroscopy (FTIR) revealed characteristic absorption peaks of collagen, PLLA, ES, and PSFS: 1092 cm−1 (PLLA, ES, PSFS) for PLLA –COC– vibration; 1545 cm−1 (collagen, ES, PSFS) for collagen C–N and N–H vibrations; 1759 cm−1 (PLLA, ES, PSFS) for PLLA C=O vibration (Fig. 6D). As an implantable mechanoelectrical conversion device for restoring mitochondrial mitophagy during wound healing, PSFS must have favorable piezoelectric properties. Previous studies indicate that the electric field strength in normal wounds ranges from approximately 100 to 200 mV/mm [69,70]. Electromechanical coupling efficiency was evaluated by open-circuit voltage under ultrasound activation: 59 mV at 0.5 W/cm2, 93 mV at 1 W/cm2, and 123 mV at 1.5 W/cm2 (Fig. 6E). The voltage and current responses of PSFS positively correlated with ultrasound intensity, indicating sensitivity to ultrasonic stimulation and potential for restoring electric field strength in diabetic wounds. Wounds are prone to external mechanical forces. PSFS's force-to-electricity conversion capability can transform such harmful loads into electrical stimulation to promote wound healing. PSFS piezoelectric response under mechanical pressure was evaluated: 0.5 N at 100 Hz generated 90.5 mV for ES vs. 130.9 mV for PSFS. The output waveform showed strong periodicity and excellent repeatability without obvious signal attenuation over multiple cycles, indicating stable piezoelectric behavior and good mechanical durability, suitable for wearable/biomechanical applications (Fig. 6F and G). X-ray diffraction (XRD) and FTIR analyses of SFS, ABs@DBCO-PSFS, and N3-ABs@DBCO-PSFS characterized crystalline structures and absorption peaks, confirming piezoelectric composite scaffold fabrication did not alter original structural properties (Fig. 6H). FTIR spectra showed that the characteristic azide (–N3) stretching peak at ∼2100 cm−1 was markedly weakened after the reaction, accompanied by a weakening of the alkyne stretching vibration peak, indicating the successful strain-promoted azide–alkyne cycloaddition (SPAAC) between Ac4ManNAz and DBCO. While absorption bands at 1456–1647 cm−1 were enhanced, corresponding to triazole-related C=N/C–N vibrations, demonstrating efficient N3-mABs grafting (Fig. 6I). Piezoelectric scaffolds convert mechanical force into electrical stimulation and can “remember” and quickly restore their original shape after repeated compression, likely due to their interconnected porous structure and high porosity. To verify this, we evaluated the scaffold mechanical properties: PSFS showed excellent compressive resilience even after 60 compression cycles, with maximum stress slightly decreasing from 15.06 kPa to 14.2 kPa (Fig. 6J). Thus, their outstanding mechanical stability ensures reversible shape memory and efficient mechanoelectrical conversion under repeated external compression. To further investigate their piezoelectric properties, piezoresponse force microscopy (PFM) amplitude and phase images showed good correspondence between bright and dark regions of PSFS under an alternating electric field (Fig. 6K–M). D33 curve analysis indicated a piezoelectric constant of ∼13.4 p.m. V−1. Phase loops, butterfly loops, and hysteresis loops further demonstrated PSFS polarization switching under applied electric fields, confirming the piezoelectric effect (Fig. 6N–Q and Supplementary Figs. 18D and 19A). Appropriate biodegradability is crucial for the wound healing process: overly rapid degradation impairs function, while excessively slow degradation hinders new tissue ingrowth. To evaluate this, PSFS (PLLA and collagen, 4:6) was thermally crosslinked for 30 min, and mABs were loaded via SPAAC click chemistry to form PSFS loaded with mABs (MtNR). The digital photos show that PSFS gradually deteriorates over time. By the 9th day, it had degraded by approximately 37.6%. By the 18th day, approximately 67.3% of it had degraded (Supplementary Fig. 19B). For in vitro mABs release assessment, MtNR was soaked in SBF, 37 °C. SEM and UV–vis results showed click-chemistry-modified ABs were gradually released (Fig. 6R and S). UV–vis measurements showed MtNR cumulative release reached 62.7% at 72 h, significantly lower than the 80.2% release from conventionally loaded ABs (Fig. 6T). MtNR exhibits excellent biocompatibility First, MtNR biocompatibility was evaluated. Cultured macrophages and endothelial cells separately with material from the five groups (control, ES, PSFS, mABs, MtNR), and cytotoxicity was assessed using live/dead cell staining. Inverted fluorescence microscopy showed no obvious cell growth abnormalities (e.g., uneven distribution, morphological deformation) in all groups. specifically, mABs- or MtNR-co-cultured groups contained predominantly live cells (green) and very few dead cells (red) (Supplementary Fig. 20A–B). Live/dead cell quantification revealed no significant differences in viable cell proportions between groups (Supplementary Fig. 20C–D). CCK-8 assay for macrophage proliferation showed none of the material groups significantly inhibited proliferation or induced cytotoxicity. Only a slight non-significant proliferation-promoting effect was observed. These results demonstrate that MtNR has excellent biocompatibility and great potential for medical applications (Supplementary Fig. 20E). M2 macrophage–to–fibroblast mitochondrial transfer occurs via TNTs Macrophages transfer healthy mitochondria to injured cells via TNTs or vesicle-mediated pathways. To investigate mitochondrial transfer between macrophages and fibroblasts, H2O2-pretreated (100 μM) DsRed(+) mouse embryonic fibroblasts (NIH3T3) were co-cultured with CFDA-SE(+) M2 macrophages (Fig. 7A and B) and observed using laser confocal microscopy (LSCM). Punctate mitochondrial fluorescence signals were detected within cytoskeleton-composed heterogeneous intercellular channels, and the colocalization of mitochondria with the nanotubes further verified TNT-mediated mitochondrial transfer from macrophages to fibroblasts (Fig. 7C). Mitochondrial transfer effect of MtNR. A) Diagram illustrating the direct co-culture system used to evaluate mitochondrial transfer between DsRed-labeled NIH3T3 fibroblasts and CFDA-SE-labeled M2 macrophages for 16 h. B) Mitochondria of macrophages were labeled with DsRed, and the cytoskeleton of fibroblasts was localized with CFDA-SE to track mitochondrial transfer between macrophages and fibroblasts. C) Images showing mitochondrial transfer between macrophages and fibroblasts. D) Representative fluorescence images and E) statistical analysis (n = 3) of mitochondrial transfer efficiency among different treatment groups. F) Flow cytometry images depicting mitochondrial transfer, and G) statistical analysis (n = 3) of mitochondrial transfer efficiency among different treatment groups. H) Representative Western blot results and I) statistical analysis (n = 3) of mitochondrial Parkin, TFAM, and TOM20 among different treatment groups. J) Schematic diagram illustrating the mechanism by which MtNR-induced mitochondrial transfer and autophagy. The significance between groups was calculated using the One-way ANOVA followed by Tukey's test. Data are presented as mean ± SD (n = 5 independent samples for E; n = 3 independent samples for G, I). Subsequently, mitochondrial transfer efficiency under different treatments was observed via LSCM. The MtNR group exhibited the highest levels of mitochondrial transfer to fibroblasts. However, pre-damaging mitochondria of macrophage-derived mABs with ethidium bromide (EtBr) significantly inhibited transfer ability, confirming that fibroblasts acquired mitochondria from macrophages (Fig. 7D and E). Furthermore, using DsRed to label macrophage mitochondria and CFDA-SE for fibroblast cytoskeletal proteins, mitochondrial transfer efficiency under different co-culture conditions was quantitatively analyzed by flow cytometry (Fig. 7F). Results indicated that macrophages treated with PSFS, mABs, or MtNR all had enhanced mitochondrial transfer to fibroblasts, with the MtNR group exhibiting the highest efficiency (Fig. 7G). Given that MtNR-group macrophages showed more efficient mitochondrial transfer to fibroblasts under oxidative stress, we hypothesized this transfer might alleviate mitochondrial damage in NIH3T3 cells. Western blot was used to assess Parkin, TFAM, and TOM20 expression (markers of mitophagy and mitochondrial content) (Fig. 7H). Results showed MtNR significantly upregulated Parkin, TFAM, and TOM20 expression (Fig. 7I), indicating it promotes mitophagy, enhances mitochondrial biogenesis, and increases mitochondrial content. Collectively, these findings demonstrate that MtNR not only facilitates macrophage-to-fibroblast mitochondrial transfer via enhanced mitochondrial- and Miro1-mediated mechanisms but also supports mitophagy and mitochondrial biogenesis (Fig. 7J). MtNRs facilitate M2 macrophage-mediated damaged cell repair via efficient mitochondrial transfer To assess viability of fibroblasts co-cultured with the engineered strategy, fluorescence-activated cell sorting (FACS) separated fibroblasts after 16 h co-culture with macrophages to determine if MtNR-treated fibroblasts have enhanced mitochondrial recovery. Raw 264.7 and NIH3T3 cells were exposed to 100 μM H2O2 for 6 h to simulate diabetic wound oxidative stress (Fig. 8A). ΔΨm changes (indicator: red/green fluorescence ratio; higher = preserved ΔΨm) showed H2O2 reduced the ratio to ∼23.5%, while PSFS, mABs and MtNR partially restored it; The impact of MtNR is the most significant (66.8%) (Fig. 8B and C). Compromising macrophage mitochondria with 0.5 μM ethidium bromide blocked recipient cell mitochondrial function restoration and increased apoptosis. These results indicate enhanced direct contact therapeutic effect relies on functional mitochondrial transfer rather than other cell-cell contact mechanisms; improving donor cell mitochondrial performance may promote recipient cell function restoration. Metabolic and mitochondrial protective effects of MtNRs. A) Diagram illustrating the in vitro experimental setup in which macrophages were labeled with F4/80, followed by FACS. B) Flow cytometry analysis and C) quantitative analysis (n = 3) of JC-1 signals in NIH3T3 cells across different treatment conditions. D) Analysis of oxidative phosphorylation levels among different groups via OCR experiment OM, oligomycin; FCCP, carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone; R&A, rotenone and antimycin A. E) Quantitative analysis of key mitochondrial respiration parameters, including basal respiration, maximal respiration, ATP production, and spare respiratory capacity. F) Representative fluorescence images and G) quantitative analysis (n = 5) of mtROS levels via MitoSOX staining in macrophages under various experimental conditions. H) Visualization of mitochondrial membrane potential in fibroblasts using TMRM and I) statistical quantification of fluorescence intensity (n = 5) under different treatment conditions. The significance between groups was calculated using the One-way ANOVA followed by Tukey's test. Data are presented as mean ± SD (n = 3 independent samples for C, E; n = 3 independent samples for G, I). Given the close association between mitochondrial respiratory function and overall mitochondrial health, cellular oxygen consumption rate (OCR) was evaluated via Seahorse analysis post H2O2 stimulation and MtNR treatment (Fig. 8D). H2O2 significantly inhibited mitochondrial respiration, while MtNR alleviated this inhibition and notably enhanced basal respiration, ATP production, maximal respiration, spare respiratory capacity, proton leak, and non-mitochondrial respiration. Compared with the control group, the reduction in respiratory reserve capacity indicates that the body is transitioning towards a state of heightened metabolic activity, during which the basal respiratory rate increases to meet the heightened energy demands (Fig. 8E and Supplementary Fig. 21). Subsequently, MtNR's capacity to scavenge mitochondrial reactive oxygen species (mtROS) for regulating mitochondrial homeostasis was assessed. MitoSox detection showed mtROS in the MtNR group decreased to 46.98% of the control, indicating significant reduction of mitochondrial oxidative stress (Fig. 8F and G). To further visualize ΔΨm changes, mitochondrial TMRM staining fluorescence in fibroblasts was observed. ΔΨm analysis revealed the greatest depolarization in the control macrophage group, caused by inflammatory factors and ROS secreted by fibroblasts under oxidative stress. Notably, MtNR partially counteracted H2O2-induced ΔΨm loss (Fig. 8H). Furthermore, the MtNR group outperformed the mABs group, with fibroblast ΔΨm restored to 3.813 times that of the control (Fig. 8I). In vivo evaluation of MtNRs in diabetic wound repair To evaluate MtNR's in vivo biological functions, a full-thickness skin defect model was generated in diabetic mice, with 10 mm-diameter and ∼2 mm-deep circular wounds. Biomaterial immunogenicity is mainly determined by chemical composition. PSFS comprises poly-L-lactic acid (PLLA) and collagen. PLLA is FDA-approved for clinical applications (e.g., sutures, skin fillers) [67,68]. Collagen, a natural protein with low immunogenicity, good biocompatibility and degradability, is widely used in medicine as a hemostatic agent, tissue engineering scaffold and orthopedic implant. Considering these properties and complying with animal welfare guidelines while ensuring experimental reliability and statistical validity, five experimental groups were established: Control, ES, PSFS, mABs, and MtNR. For ES, PSFS and MtNR groups, corresponding materials were implanted to close skin defects; the mABs group received subcutaneous injection. The in vivo experimental flowchart is shown in Fig. 9A. Seven days after surgery, major organs were harvested for H&E staining. The results showed no obvious organ rejection induced by materials in any group (Supplementary Fig. 22). MtNRs promote diabetic wound healing in vivo. A) Schematic diagram of the establishment of a full-thickness skin defect model in diabetic C57 mice and subsequent in vivo experimental procedures. B) Digital photographs of wound healing in mice from the Control, ES, PSFS, mABs, and MtNR groups. C) Comparison of healing conditions of the groups at different time points. D) Statistical analysis of the wound area of each group at different time points. E)In vivo small animal living imaging photos of mABs, ABs@PSFS, and MtNRs. F) Retention curves of mABs, ABs@PSFS, and MtNRs. G) Representative detection images, and H-I) 3D plots comparing wound potential in the Control (left) and MtNR group (right). On day 7, representative images of J) H&E staining and K) quantification of inflammatory cells in each experimental group. At day 7, L) representative Masson staining images and M) collagen content quantification in each experimental group. The significance between groups was calculated using the One-way ANOVA followed by Tukey's test. Data are presented as mean ± SD (n = 3 independent animal samples for D, F, n = 5 independent animal samples for K, M). Surgical photographs and wound area measurements showed that the PSFS, mABs, and MtNR groups significantly promoted wound healing (Fig. 9B). At the early proliferation stage (Day 7), the wound healing area ratio in the MtNR group was 48.07%, which was 1.12-fold, 1.83-fold, and 2.61-fold higher than that in the mABs monotherapy group (43.02%), PSFS monotherapy group (26.24%), and Control group (18.40%), respectively. Extending treatment to 21 days, most wounds in the MtNR group were successfully repaired (76.83%), which was markedly higher than in the PSFS group (62.49%), mABs group (59.19%), and Control group (35.17%) (Fig. 9C and D). Animals were treated with DiD-labeled naked mABs, conventionally loaded mABs@PSFS, and click chemistry-conjugated MtNRs, respectively. Dynamic tracking and systematic time-course analysis of the in vivo retention of various ABs were performed using in vivo small animal imaging technology at multiple time points (Fig. 9E). Results showed naked mABs had a short lifespan and were mostly cleared within 3 days of local injection. mABs simply adsorbed onto short-fiber scaffolds maintained retention for ∼7 days, while click chemistry-conjugated MtNRs sustained biological signals for ∼14 days (Fig. 9F).​ Compared with previous studies where MSCs were cleared within approximately 1–2 days after in vivo injection [39,43], our mABs maintained retention for ∼7 days. This may be attributed to local injection (vs. intravenous injection), reduces rapid phagocytic clearance by lung entrapment, hepatic Kupffer cells, and the splenic filtration. In addition, as apoptotic products of M2-macrophages, ABs exhibit low immunogenicity and do not immediately trigger intense inflammation or immune attack after local administration, ensuring relative stability. To confirm MtNR could deliver a stable, effective electric field to the wound, wound potential was continuously monitored pre- and post-MtNR treatment (Fig. 9G). In the diabetic rat wound model, normal wounds showed a “volcano-like” electric field distribution with a 50.16 mV potential difference. Electro-generating dressing therapy converts mechanical energy from periodic dressing deformation into electrical energy; the electric potential generated by built-in MtNRs was rectified and applied to the wound via electrodes, increasing the center-to-edge potential gradient and remodeling the wound's endogenous electric field. The average endogenous electric field intensity in the MtNR group was 156.18 mV, 3.11 times that of the control group (Fig. 9H and I). Wound healing is a complex process involving successive, overlapping inflammation, proliferation, and remodeling phases. In vivo tracking demonstrated that mito-GFP–labeled mitochondria delivered via mABs were primarily localized in CD206+ macrophages at days 1 and 3 post-injection, whereas by day 7 they were predominantly detected in FAP+ fibroblasts, indicating dynamic mitochondrial transfer from macrophages to fibroblasts during wound healing (Supplementary Fig. 23). To assess wound healing, regenerated tissues from five mouse groups were collected on postoperative day 7 for hematoxylin-eosin (H&E) and Masson staining. Compared with the control, ES, and PSFS groups, the mABs and MtNR groups showed reduced inflammatory cell infiltration, highlighting MtNR's anti-inflammatory effect (Fig. 9J and K). Further Masson staining analysis revealed that the MtNR group had increased collagen fiber deposition with denser, more regularly aligned bundles. Compared with other treatment groups, this group exhibited superior collagen remodeling. This indicates that MtNR not only alleviates the inflammatory response but also promotes collagen matrix restoration and reconstruction in diabetic wounds (Fig. 9L and M). Evaluation of mitochondrial function training by MtNRs in diabetic wounds Adequate energy metabolism is crucial for wound healing. However, in diabetic wounds, the interplay between insufficient energy metabolism and excessive inflammation further impairs necrotic tissue and pathogen clearance. This leads to inadequate synthesis and disorganized arrangement of newly formed collagen, which significantly compromises the mechanical and biological quality of regenerated tissue (Fig. 10A). TEM results showed that both PSFS and mABs groups improved mitochondrial morphological abnormalities in the diabetic rat full-thickness skin defect model. These abnormalities included swelling, vacuolization, cristae reduction, and length shortening. However, compared with PSFS and mABs groups, mitochondrial morphology in tissue cells of the MtNR group was more intact and healthier. The healthy-to-damaged mitochondria ratio was significantly increased, indicating that MtNRs contribute to mitochondrial function restoration (Fig. 10B–G). As previously mentioned, Miro1-mediated mitochondrial transfer is a key factor limiting M2 macrophage therapeutic efficacy. Immunofluorescence staining was performed to assess Miro1 expression during diabetic wound repair. The MtNR group exhibited the highest Miro1 levels, suggesting enhanced mitochondrial transfer capacity (Fig. 10C–H). Evaluation of mitochondrial transfer and autophagy in diabetic wounds. A) Diagram illustrating the role of MtNR in promoting diabetic wound healing. B) Representative TEM images of mitochondria in wounds from each group on day 7. C) Representative immunofluorescence staining images of Miro1 expression in wounds from each group on day 7.​ D-E) Representative immunohistochemical images of IL-1β and TNF-α on day 7. F) Representative immunofluorescence staining of CD31 and COL3 on day 14. G) Statistical analysis of the proportion of swollen/healthy mitochondria in the TEM images. H) Statistical analysis of Miro1 expression at the wound site on day 7. I) Statistical analysis of the expression of IL-1β and TNF-α on day 7. J) Western blot analysis and K) statistical analysis.​ L) Representative double immunofluorescence staining of P62 and Parkin on day 7. M) Quantitative analysis of the expression of CD31 and COL3 on day 14. N) Statistical analysis of the expression of P62 and Parkin on the 7th day. O) Schematic diagram of mitochondrial autophagy reflecting local wound repair following MtNR application. The significance between groups was calculated using the One-way ANOVA followed by Tukey's test. Data are presented as mean ± SD (n = 5 independent animal samples for G, H, I, M; n = 3 independent animal samples for K, N). Chronic diabetic foot ulcers (DFUs) exhibit insufficient local energy metabolism. This leads to normal cell dysfunction or necrosis, impaired metabolic waste clearance, prolonged inflammation, and sustained local accumulation of cytokines (e.g., IL-1β, TNF-α) that hinder tissue repair. Based on MtNR's in vitro advantages (clearing excessive mitochondrial ROS, supplying energy substrates), we further evaluated their multifaceted in vivo wound repair effects. Immunohistochemical analysis of IL-1β and TNF-α in wound tissues on postoperative day 7 showed MtNR effectively reduced their expression, indicating facilitation of chronic wound local inflammation resolution (Fig. 10D–E, I). On day 14, CD31 and COL3 staining of regenerated tissues to assess angiogenesis and collagen deposition showed marked increases in CD31 (neovascular marker) and COL3 (collagen marker) fluorescence signals, indicating improved angiogenesis and collagen deposition (Fig. 10F–M). To further validate the effect of MtNR on tissue energy metabolism in vivo, wound tissues were harvested on day 5 and homogenized for ATP quantification using the CellTiter-Glo Luminescent Cell Viability Assay. The results showed that the MtNR group exhibited significantly higher ATP levels than the control and other treatment groups, indicating enhanced local energy metabolism. Furthermore, immunofluorescence staining for Ki67 and α-smooth muscle actin (α-SMA) was performed on day-7 wound sections to assess cell proliferation and myofibroblast activation during wound repair. Compared with other groups, the MtNR group showed more Ki67-positive cells and stronger α-SMA expression, suggesting enhanced cellular proliferation and fibroblast activation during tissue regeneration (Supplementary Fig. 24A–E). Western blot analysis on Day 14 revealed that Miro1 expression levels and mitophagy activity in the ES and PSFS groups were comparable to those in the control group, but significantly elevated in the mABs and MtNR groups (Fig. 10J). Elevated Miro1 indicates enhanced intercellular mitochondrial transport. Additionally, in the MtNR group, P62 expression decreased significantly, Parkin increased, and LC3-II/LC3-I ratio elevated, indicating significant mitophagy activation. Furthermore, increased Tom20 protein (Fig. 10K) suggested MtNRs significantly raised local wound mitochondrial levels, possibly associated with enhanced mitochondrial biogenesis. Mitophagy was assessed via double-labeled immunofluorescence (Parkin: green; P62: red). Control group showed decreased Parkin with P62 accumulation (Fig. 10L), indicating impaired mitophagy. In contrast, PSFS/mABs groups had upregulated Parkin and downregulated P62, suggesting restored active mitophagy. Notably, the MtNR group exhibited the strongest Parkin expression and lowest P62 content, indicating the most active mitophagy at the wound site (Fig. 10N). Overall, MtNR not only effectively suppresses excessive inflammation in diabetic wounds but also promotes angiogenesis and collagen matrix remodeling, thereby accelerating DFU progression from inflammatory to proliferative phase. This reparative effect is likely closely associated with local energy support provision and tissue regenerative microenvironment improvement. Parkin provides a “targeting signal” by adding ubiquitin tags to outer mitochondrial membrane proteins. Subsequently, the autophagic receptor P62 (SQSTM1) specifically recognizes and binds to these ubiquitinated damaged mitochondria, and simultaneously interacts with LC3-II-labeled autophagosomes, targeting and transporting damaged mitochondria to lysosomes for degradation [26]. Additionally, M2 macrophages transfer healthy mitochondria to damaged fibroblasts via TNTs, with the mitochondrial surface-localized Miro1 protein mediating this transcellular mitochondrial transport process (Fig. 10O). Conclusion In summary, we developed a mitochondrial micro-nano reactor (MtNR) that promotes mitochondrial transfer and mitophagy to restore cellular energy metabolism. MtNR induces M2 macrophages to overexpress Miro1 and promotes mitochondrial enrichment via hypoxic training and lentiviral transduction, then generates mitochondria-rich apoptotic bodies through STS-induced apoptosis. These bodies enhance intercellular mitochondrial transfer via the Miro1-mitochondria-transporter-microtubule complex. Concurrently, piezoelectric electrospun fibers stimulate mitophagy through the Pink1-Parkin-P62/SQSTM1-LC3 pathway. SPAAC copper-free click chemistry enables sustained energy modulation and long-term mitochondrial function. By targeting M2 macrophages, MtNR delivers energy to surrounding fibroblasts, restoring their metabolic activity and accelerating tissue repair in chronic diabetic wounds. This work establishes a modular micro-nano platform that integrates mitochondrial bioenergetics, nanomedicine, and regenerative therapy, providing a novel strategy for chronic wound repair. Discussion This study demonstrates the therapeutic potential of the MtNR platform for metabolic reprogramming and tissue repair in diabetic wounds. By integrating mitochondrial transfer and mitophagy regulation, MtNR provides a synergistic and efficient strategy to restore cellular energy homeostasis. Notably, its modular design enables the coordinated delivery of bioactive mitochondrial components and precise regulation of intracellular quality control pathways, thereby overcoming key limitations of conventional therapies. Through this multi-functional and biomimetic approach, MtNR effectively enhances cellular metabolic activity and promotes tissue regeneration, highlighting its significant promise in the field of regenerative medicine. However, the clinical translation of MtNR still faces several challenges that warrant further investigation. First, the sample size was relatively small, and part of the analysis relied on public datasets, highlighting the need for validation in larger, multicenter cohorts. Second, technical constraints limited comprehensive in vivo evaluation, including mitochondrial tracing using Cre–LoxP reporter systems, systematic validation of molecular mechanisms with knockout models, and single-cell transcriptomic assessment of cellular heterogeneity and intercellular communication. Lastly, although MtNR shows advantages over conventional delivery systems, further studies are required to perform systematic comparisons with existing carriers, validate efficacy and safety in large-animal models, and assess long-term effects and challenges related to large-scale production and clinical translation. Taken together, MtNR represents a versatile and promising platform for chronic wound repair by integrating mitochondrial bioenergetics with advanced biomaterials design. With further optimization and rigorous validation, this strategy is expected to advance the development of next-generation mitochondrial therapies and facilitate its translation toward clinical applications. Experimental section Integrated analysis of single-cell sequencing data Sample information We obtained single-cell RNA-seq data of diabetic wounds from the GEO database (NCBI), dataset GSE165816. This dataset was generated using the Illumina NovaSeq 6000 platform (Homo sapiens). Cell filtering, clustering, and annotation were performed with Seurat (v5.0.1) in R. To enable comparison, three pairs of healing and non-healing diabetic wounds were selected. Detailed information of the three pairs of diabetic wound patients is provided in Supplementary Tables 1 and 2. Sample quality control was first performed, and the ages of the two groups were 50.33 ± 14.19 and 49.33 ± 13.61 years, respectively. Single-cell transcriptome analysis using Seurat Seurat (v4.0.4) and R (v4.1.0) were used for data quality control, log-normalization, and clustering. Abnormal cells were filtered out based on the number of genes expressed per cell (nGene), UMI counts (nUMI), mitochondrial and ribosomal gene expression, and doublet detection (using DoubletFinder), generally applying the two-standard-deviation rule by default. This ensures the reliability and accuracy of downstream analyses. After dimensionality reduction, batch effects were further corrected before performing subsequent clustering analyses. For visualization of clustering results, nonlinear dimensionality reduction methods were employed. In this study, two approaches were used: t-SNE and UMAP, and the relationships between clusters were also assessed. Following clustering, cells were classified into distinct clusters. Seurat can then perform comparative analysis between each cluster and the remaining clusters to identify differentially expressed genes (DEGs) among cell subpopulations. DEG was calculated using the FindAllMarkers function in Seurat, typically with default parameters, with adjustments made as needed based on sequencing depth or genomic features. DEGs identified in each cluster were used to further screen for marker genes and were visualized in various forms. Cell type annotation was performed using the SingleR package (Single-cell Recognition of cell types). Cell-cell communication and functional enrichment analysis Cell-cell communication among cell populations was assessed using CellChat (R-based), which evaluates intercellular regulatory or signaling interactions primarily mediated through ligand-receptor binding. In addition, the ligand-receptor database of CellPhoneDB was used to validate and obtain more accurate ligand-receptor annotations. Differentially expressed genes/transcripts were analyzed for functional enrichment using GO, KEGG and GSEA pathways. The hypergeometric distribution algorithm was applied to identify significantly enriched functions within the gene sets. P-values were corrected with the Benjamini-Hochberg (BH) method, and a P–adjust <0.05 indicated significant enrichment. Methods for establishing diabetic rat/mouse wound models The animal study was approved by the Ethics Committee of Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine (Approval No. SH9H-2022-A96-1) and conducted following U.S. NIH guidelines. Diabetic rat or mouse wound models are typically established by chemical induction, commonly using streptozotocin (STZ), with doses of 50–60 mg/kg for rats and 100 mg/kg for mice. Blood glucose is measured 3–7 days post-injection, and persistent glucose ≥16.7 mmol/L indicates successful diabetes induction. Once the diabetic model is established, animals are anesthetized, shaved, disinfected, and full-thickness skin defects (6–8 mm for mice, 10 mm for rats) are created on the back. Wounds can be covered with a transparent dressing to prevent infection and dehydration. Wound tissues can be harvested at various time points for histological or molecular analyses. Hematoxylin and eosin (H&E) and Masson's trichrome staining Samples were fixed in 4% paraformaldehyde for 24 h, paraffin-embedded, and cut into 5 μm sections. Sections were baked at 60 °C for 1 h, deparaffinized in xylene, and rehydrated in graded ethanol. Hematoxylin staining was applied for 5–10 min, followed by tap water rinsing to achieve blue nuclear staining, a short differentiation in 1% acid alcohol, and bluing under running water. Eosin staining was performed for 1–3 min, followed by dehydration, clearing, and mounting with neutral resin. Histological morphology was observed and recorded under a light microscope. Deparaffinized 5-μm sections were rehydrated and stained using a commercial Masson's trichrome kit. Briefly, sections were treated with Bouin's solution (60 °C, 60 min) and washed well. Weigert's iron hematoxylin-stained nuclei (5–10 min), acid fuchsin stained cytoplasm, and collagen was labeled with phosphomolybdic acid and aniline blue. After dehydration and clearing, the sections were resin mounted. Collagen deposition as well as tissue morphology were examined with a light microscope, and images were captured with an automated imaging system. Assessment of Miro1, LC3P62, COL1, FAP, and CD68 expression in diabetic skin tissue sections Skin sections from normal and diabetic samples were used to evaluate the expression of Miro1, LC3, and P62. Immunofluorescence staining was conducted following established protocols. Briefly, paraffin sections were deparaffinized, rehydrated, and subjected to heat-induced antigen retrieval. After blocking with 3% BSA for 30 min at room temperature, the sections were incubated with primary antibodies against Miro1 (84055-5-RR, 1:100, Proteintech, China), LC3 (ab192890, 1:100, Abcam, UK), P62 (ab109012, 1:100, Abcam, UK), COL1 (83752-5-RR, 1:100, Proteintech, China), FAP (ab314456, 1:100, Abcam, UK), CD68 (ab955, 1:100, Abcam, UK). Sections were then treated with fluorescence-labeled secondary antibodies (1:500, ABclonal, China) for 1 h at room temperature, followed by DAPI counterstaining for 5 min. Images were captured using an Olympus BX15 microscope (Tokyo, Japan). Colocalization characteristics of the immunofluorescence images were analyzed using a colocalization finder tool. Cell culture Raw 264.7 macrophages (AW-CM0088, ANWEI Biology, Shanghai, China) and HUVECs (AW-CH0165, ANWEI Biology, Shanghai, China) were cultured in DMEM containing 10% heat-inactivated FBS and 1% penicillin–streptomycin–glutamine (PSG, Thermo Fisher Scientific). NIH-3T3 cells (AW-CM0035, ANWEI Biology, Shanghai, China) were maintained in DMEM containing 10% (v/v) heat-inactivated calf serum (CS, Thermo Fisher Scientific) and 1% (v/v) PSG. All cell lines tested negative for mycoplasma. Fluorescence labeling was achieved either by DsRed transfection or CFDA-SE staining. Briefly, macrophages were transfected with pCMV-C-DsRed (D2624, Beyotime, Shanghai, China) using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. After 6 h of incubation, the medium was replaced with fresh complete medium, and cells were further cultured for 24–48 h to allow stable DsRed expression. Alternatively, fibroblasts or HUVECs were incubated with 5.0 μM CFDA-SE (C1031, Beyotime, Shanghai, China) at 37 °C in a 5% CO₂ incubator. Following staining, cells were washed twice and allowed to recover in complete medium for 1 h to remove residual dye. For co-culture experiments, macrophages were mixed with ECs or fibroblasts at a 1:1 ratio in a blend of their complete media and maintained for the indicated time periods. Subsequent analyses were performed by fluorescence microscopy or flow cytometry. Lentiviral transduction Macrophages were plated at 1 × 10^5 cells/mL in 96-well plates and infected once they reached 30–50% confluence. According to the experimental design, lentiviral vectors carrying the Miro1 gene (CV186, GeneChem, Shanghai, China) were used. Following the manufacturer's guidelines, viral particles were added at a MOI of 100 in culture medium supplemented with 8 μg/mL HiTransG P. After 12–16 h at 37 °C and 5% CO2, cells were refreshed with complete medium. Fluorescent reporter expression (GFP/RFP) was examined 48–72 h post-infection, and positive cells were selected using puromycin (2 μg/mL) over 3–5 days. Western blotting and qPCR were used to confirm transduction efficiency and gene overexpression. CoCl2-induced hypoxia model A hypoxic microenvironment was established in vitro using cobalt chloride (CoCl2), as described previously [71]. Miro1 OE@M2 macrophages were cultured at 37 °C in 5% CO2. CoCl2·6H2O (100 μM; Thermo Fisher Scientific, St. Louis, MO, USA) was freshly diluted in DMEM immediately before use and added to the cells to induce chemical hypoxia. Following 24 h of incubation, the culture medium was refreshed for downstream assays. Immunofluorescence staining of cells Cells were cultured on coverslips or culture dishes until reaching an appropriate density, then fixed in 4% PFA for 15 min at room temperature and washed three times with PBS. After permeabilization with Triton X-100, cells were washed and blocked with 5% BSA for 1 h at room temperature. Primary antibodies were applied overnight at 4 °C, followed by three PBS washes and 1 h incubation with fluorescent secondary antibodies in the dark. Following nuclear staining with DAPI (1 μg/mL) for 5 min, coverslips were washed, mounted, and visualized on a laser scanning confocal microscope (Nikon A1 or Zeiss LSM880). Collection of extracellular mitochondria RAW264.7 macrophages were cultured until reaching >95% confluency. Cells were washed with PBS and incubated in serum-free medium for 24 h. The collected medium was sequentially centrifuged at 500 g and 2000 g to remove cells and debris, producing RAW264.7-CM. The supernatants from twenty T75 flasks were combined and centrifuged at 15,000 g for 30 min at 4 °C. The resulting pellet, containing highly concentrated extracellular mitochondria, was resuspended in pre-chilled 2.5% glutaraldehyde fixative and stored at 4 °C until TEM processing. Transmission electron microscopy (TEM) observation of vesicle morphology Purified apoptotic bodies (10 μL) were applied to Formvar/carbon-coated grids and adsorbed for 10 min at room temperature. The grids were rinsed with distilled water and negatively stained with 2% phosphotungstic acid (pH 7.0) for 1 min. Excess stain was carefully blotted away, and the grids were allowed to air-dry at room temperature. The samples were imaged with a TEM at 80–120 kV accelerating voltage. Images were acquired using a digital imaging system, and vesicle morphology and size distribution were analyzed with ImageJ software. Molecular dynamics analysis simulates the click chemical formation process of SPAAC To simulate the potential, click-binding interface between the ligand and target protein, molecular dynamics analysis was performed to predict the specific reaction sites. Protein Data Bank (PDB) derived 3D structure of the target protein was processed in AutoDock Tools 1.5.6 by removing water molecules, adding polar hydrogens, and assigning Gasteiger charges. Ligand structures containing clickable functional groups (alkyne or azide) were drawn in Chem3D, energy-minimized using the MM2 force field, and exported as PDB format. Docking simulations were conducted using AutoDock Vina or Schrödinger Glide. The grid box was set to cover potential nucleophilic/electrophilic reactive residues (e.g., pockets adjacent to Cys, Lys, or His). Ligands were aligned such that the clickable group was oriented toward possible reaction sites on the protein surface. After docking, binding affinity and structural conformations were analyzed. Complexes with the lowest predicted binding energies and reasonable steric configurations were retained. Hydrogen bonding, hydrophobic interactions, and the distance between clickable groups and reactive residues (typically <6 Å) were visualized using PyMOL or Chimera. Metabolic glycoengineering and detection in macrophages Macrophage surface sialoglycoproteins were metabolically engineered by introducing azide groups as previously described [57]. Briefly, a 100 mM stock solution of Ac4ManNAz was prepared in DMSO, sterilized using a 0.2-μm filter, and kept at −20 °C until required. Macrophages were incubated in standard culture medium containing varying concentrations of Ac4ManNAz (10, 25, 50, or 100 μM) for 24 h, while untreated cells serving as the negative control. After incubation, cells were rinsed twice with pre-warmed Dulbecco's PBS (dPBS, pH 7.4) and treated with the azide-reactive fluorescent probe DBCO-PEG4-rhodamine110 for 1 h. Following two washes with dPBS, cells were stained with DAPI for 5 min to label nuclei. Following three additional washes, fluorescence images were captured using an inverted microscope (Carl Zeiss Microscopy), and subsequently analyzed using ImageJ software. Approximately 2 × 104 DBCO-PEG4-rhodamine110–positive cells were sorted on a BD Accuri C6 Plus and analyzed in FlowJo. Dynamic light scattering (DLS) analysis of vesicle size distribution To determine vesicle size and distribution, engineered apoptotic bodies were analyzed using dynamic light scattering (DLS). Samples were diluted to 500 μg/mL in PBS, loaded into low-particle cuvettes, and analyzed on a Malvern Zetasizer Nano ZS at 25 °C. Samples were run in triplicate with scattering recorded at 173°. Zetasizer software was used to obtain the Z-average size and PDI, and data were exported for downstream analysis. Zeta potential measurement by Zetasizer Nano ZS analyzer Vesicle samples were resuspended in PBS (or 1 mM KCl buffer) and diluted to an appropriate concentration. Zeta potential was analyzed with a Zetasizer Nano ZS and DTS1070 capillary cells (Malvern Instruments, UK). Samples were analyzed at 25 °C, and particle electrophoretic mobility was determined by electrophoretic light scattering, followed by automatic conversion to Zeta potential values (mV). All measurements were performed in triplicate, and data are presented as mean ± standard deviation. Measurement of intracellular ATP levels Cells were washed, lysed with 200 μL buffer, spun at 12,000×g for 5 min at 4 °C, and the supernatants were collected. ATP levels were measured using an Enhanced ATP Assay Kit (S0027, Beyotime, Shanghai, China) according to the manufacturer’s instructions. Briefly, 100 μL of ATP working solution was added per well, with blank wells containing only the working solution. Following a 3–5 min incubation at room temperature (25 °C), 20 μL of each sample was added and gently mixed. Relative luminescence units (RLU) were recorded using a multimode microplate reader (SpectraMax M2e, Molecular Devices) at an interval of at least 2 s. ATP concentration was calculated from a standard curve. Piezoelectric performance test The piezoelectric response of the samples was evaluated by measuring the output voltage under external mechanical stimulation. Each sample was sandwiched between two gold electrodes, one at the bottom and one at the top. The electrodes were connected to a digital storage oscilloscope (UPO2202, UNI-T, Hong Kong, China) to construct the piezoelectric measurement circuit. At room temperature, the samples were subjected to ultrasonic mechanical excitation and periodic compressive loading, and the oscilloscope recorded the real-time output voltage signals. The sampling frequency was set to 1–10 kHz, with a time window covering the entire deformation cycle. The output voltage amplitude, waveform characteristics, and signal reproducibility were used to assess piezoelectric response performance. All measurements were conducted in an electromagnetically shielded environment to minimize background noise, and each test was repeated at least three times for reliability. The piezoelectric properties of the PSFS were further characterized using an MFP-3D atomic force microscope (Asylum Research) with a conductive probe in piezoelectric force microscopy (PFM) mode to obtain surface morphology and butterfly amplitude loops. Polarization–electric field hysteresis loops were recorded using an RTI-MultiFerroic ferroelectric tester (Radiant Technologies, USA). Mechanical performance test of the PSFS Scaffold mechanical properties were measured under cyclic compression with a universal testing machine (HY-940FS, China). Each scaffold was placed between two compression platens, and the initial height was set before loading. The samples were compressed to 50% strain at a loading rate of 1 mm/min, followed by unloading to their original position. This loading–unloading process was repeated for 60 consecutive cycles. Load–displacement curves were recorded throughout the test, from which stress–strain profiles were obtained. The elastic recovery and compressive resistance of the scaffolds were analyzed based on cyclic mechanical behavior. Scanning electron microscopy (SEM) is used to observe electrospun fiber membranes, short fiber scaffolds, AB-loaded short fiber scaffolds, and release behavior To visualize the surface morphology of electrospun fiber membranes, short fiber scaffolds, short fiber scaffolds loaded with apoptotic bodies (ABs), scanning electron microscopy (SEM) was performed. After fixation in 4% paraformaldehyde for 30 min, scaffolds containing native or click-modified ABs were dehydrated in graded ethanol (30%–100%, 10 min each). After dehydration, the samples were dried using critical point drying or hexamethyldisilazane (HMDS) treatment. Samples were gold- or platinum-coated (∼10 nm) and observed by SEM. The imaging analysis focused on: (i) the presence of ABs (500–2000 nm) attached to the scaffold surface or within the pores; (ii) the distribution and interaction between ABs and fiber surfaces; and (iii) structural changes of the scaffold at different release time points, potentially associated with cellular interactions or material degradation. Flow cytometry After treatment, cells were washed twice and resuspended in FACS buffer at 1–5 × 10^6 cells/mL. According to the experimental design, the appropriate live-cell fluorescent dyes were added for staining. Cells were washed, resuspended in 2% FBS–PBS, and stored on ice in the dark for analysis. Samples were analyzed within 2 h using a Beckman CytoFLEX flow cytometer and subsequently analyzed with FlowJo software (v10.7.1). Immunofluorescence staining of cells and tissue sections Paraffin-embedded tissue sections were baked at 60 °C for 1 h, then underwent standard deparaffinization and rehydration. S Antigen retrieval was performed in citrate buffer (pH 6.0) by microwave boiling for 10–15 min, followed by cooling and PBS washes. Permeabilization was carried out using 0.3% Triton X-100 for 10 min (optional for frozen sections). Sections were then blocked with 5% BSA at room temperature for 1 h, followed by overnight incubation at 4 °C with primary antibodies, including anti-Miro1 (ab188029, 1:100, Abcam, UK), anti-Parkin (ab77924, 1:100, Abcam, UK), anti-P62 (ab109012, 1:100, Abcam, UK), anti-CD31 (ab28364, 1:100, Abcam, UK) and anti-Collagen III (ab23445, 1:100, Abcam, UK). After washing, sections were incubated with Alexa Fluor 594 secondary antibodies (1:500) for 1 h at room temperature in the dark. Nuclei were stained with DAPI, and slides were mounted for confocal imaging. Immunohistochemistry (IHC) Paraffin sections were deparaffinized in xylene, rehydrated through graded ethanol, and subjected to antigen retrieval in sodium citrate buffer using a pressure cooker. Sections were treated with 3% H2O2 to block endogenous peroxidase, then with 5% normal serum, and incubated overnight at 4 °C with primary antibodies against IL-1β (ab178846, 1:200, Abcam, UK) and TNF-α (ab220210, 1:200, Abcam, UK). HRP-labeled secondary antibody polymer was applied after PBS washing and incubated at room temperature. Visualization was achieved using DAB chromogenic reagent, and counterstained with hematoxylin. After dehydration and clearing, coverslips were applied, and the slides were photographed. Quantification was carried out using ImageJ. In vivo toxicity evaluation Male C57BL/6 mice (6–8 weeks) were randomly divided into five groups: (i) Control, (ii) ES, (iii) PSFS, (iv) mABs, and (v) MtNR. Mice in the ABs group received treatment every three days for 21 days. Mice were sacrificed after treatment, and major organs were harvested for histology. H&E staining was performed on tissue sections to assess any systemic toxic effects. Statistics Data analysis was performed using GraphPad Prism 8.0 (San Diego, CA, USA) and R (4.2.2). Results are shown as mean ± standard deviation (SD), with one- or two-way ANOVA and Tukey's test used for multiple group comparisons. Two-tailed Student's t-test was used to compare differences between two groups. Statistical analyses were performed with GraphPad Prism 8.0. Significance levels were set as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. CRediT authorship contribution statement Jiayi Mao: Writing – original draft, Methodology, Investigation, Data curation. Wenzheng Xia: Writing – original draft, Methodology, Investigation, Data curation. Minxiong Li: Writing – original draft, Methodology, Investigation, Data curation. Xin Huang: Writing – original draft, Data curation. Yun Zhao: Methodology, Conceptualization. Zheyuan Hu: Writing – original draft, Data curation. Yinghong Su: Writing – original draft, Data curation. Juan Wang: Writing – review & editing, Methodology, Investigation. Wenguo Cui: Writing – review & editing, Supervision, Resources, Methodology, Investigation. Tao Zan: Writing – review & editing, Supervision, Resources, Methodology, Investigation. Ethics approval and consent to participate All animal procedures were approved by the Ethics Committee of Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine (Approval No. SH9H-2022-A96-1) and conducted following U.S. NIH guidelines. Declarations of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements J.M., W.X. and M.L. contributed equally to this work. This research was supported by grants from the National Natural Science Foundation of China (82471589, 82572447, 82072177, 82272264, and 82503011); Shanghai Oriental Talents Program Youth Project (QNWS2024075), Seed Program for Medical New Technology Research and Transformation of Shanghai Municipal Health Commission (2025ZZ2042); Shanghai Natural Science Foundation (25ZR1402458); the "Two Hundred Talent" Program and the first Continuous Outstanding Special Technology Supporting Project and Shanghai Municipal Key Clinical Specialty (shslczdzk00901); China Postdoctoral Science Foundation (2025M772089) and the Science and Technology Commission of Shanghai Municipality (23015820800). Footnotes Peer review under the responsibility of editorial board of Bioactive Materials. Supplementary data to this article can be found online at https://doi.org/10.1016/j.bioactmat.2026.03.048. Contributor Information Juan Wang, Email: juanwang1006@126.com. Wenguo Cui, Email: wgcui@sjtu.edu.cn. Tao Zan, Email: zantao@sjtu.edu.cn. Supplementary data The following are the Supplementary data to this article: Multimedia component 1 Multimedia component 2 Multimedia component 3 Multimedia component 4 Data availability The authors declare that all data supporting the results of this study are available within the paper and its Supplementary Information, or from the corresponding authors upon request. Source data are provided with this paper. References Associated Data Supplementary Materials Multimedia component 1 Multimedia component 2 Multimedia component 3 Multimedia component 4 Data Availability Statement The authors declare that all data supporting the results of this study are available within the paper and its Supplementary Information, or from the corresponding authors upon request. Source data are provided with this paper.