Targeted biohybrid nanoplatform for spinal cord injury treatment: Restoring microglial mitophagy and alleviating oxidative stress

Abstract

Spinal cord injury (SCI) remains refractory because oxidative stress, impaired mitochondrial quality control, and maladaptive microglial activation form an interdependent cascade that sustains secondary degeneration. Here, we develop a modular biomimetic nanozyme platform (PM-Lipo@PB NPs) by integrating distinct functional components into a unified system. Platelet-derived membrane vesicles (PMVs) and M2 macrophage–derived extracellular vesicles (M2EVs) are fused to form a hybrid membrane (PM-Lipo) that mediates lesion-associated accumulation and provides immunoregulatory cues, while a Prussian blue (PB) nanozyme shell is grown in situ to confer catalytic redox activity.

This design enables concurrent vascular injury–associated enrichment, interaction with activated microglia, and localized ROS scavenging within the lesion microenvironment. PM-Lipo@PB NPs accumulate at the injury site and reduce oxidative burden, thereby contributing to the recovery of mitochondrial function and cellular homeostasis. These effects are accompanied by modulation of PI3K/AKT/mTOR signaling and restoration of mitophagic flux, together with a shift of microglia toward a reparative phenotype.

In cellular and murine SCI models, the platform exhibits lesion-preferential accumulation, effective ROS clearance, reduced inflammatory activation, and improved tissue preservation and motor function. Taken together, this study establishes a modular strategy that couples membrane-derived biological functionality with nanozyme-based redox regulation, providing a basis for microenvironment-directed therapy in SCI and related neuroinflammatory conditions.

Graphical abstract

Key Highlights

• •Biohybrid membrane engineering: platelet–M2EV membrane enables vascular targeting and inflammatory homing.
• •Catalytic mitochondrial rescue: PB nanozyme scavenges ROS and inhibits PI3K/AKT/mTOR to restore mitophagy.
• •Microglial reprogramming: Promotes anti-inflammatory transition and resolves chronic neuroinflammation.
• •Microenvironment reconstruction: Enhances angiogenesis and preserves neuronal integrity after SCI.
• •In vivo efficacy: Achieves significant neuroprotection and functional recovery.

1. Introduction

Spinal cord injury (SCI) triggers a rapidly evolving and self-perpetuating secondary cascade that extends far beyond the initial mechanical insult [[1], [2], [3]]. Excessive reactive oxygen species (ROS) generation, mitochondrial collapse, compromised mitophagy, and maladaptive activation of microglia together form a tightly coupled pathological network that drives progressive neuronal loss and functional decline. Because these processes reinforce one another—ROS impairs autophagic flux, defective mitochondria amplify oxidative stress, and inflammatory microglia sustain both—a single-target intervention is rarely sufficient to halt or reverse the chronic degenerative trajectory of SCI. Despite the urgent therapeutic need, current interventions remain limited in their efficacy. Methylprednisolone, the only widely used pharmacological treatment, is constrained by a narrow therapeutic window (≤8 h) and an increased risk of infection [4,5]. Regenerative strategies, such as stem cell transplantation, face challenges with poor cell survival post-engraftment, significantly hindering their clinical translation [6,7]. The inadequacy of existing treatments highlights the critical need for novel therapeutic approaches that address the core pathological mechanisms of SCI. Achieving meaningful repair requires therapeutic systems capable of simultaneously addressing metabolic, inflammatory, and organelle-level dysfunction (see Scheme 1).

Scheme 1.

PM-Lipo@PB NPs restore mitochondrial homeostasis via targeted ROS scavenging and mitophagy reactivation in SCI.

A central player in these mechanisms is microglia, the resident immune cells of the central nervous system [8]. After SCI, microglial number and function undergo significant alterations [9,10]. In the acute phase (<24 h), microglial proliferation occurs rapidly and remains elevated [11]. Concurrently, mitochondrial dysfunction within these cells drives excessive ROS production, which activates the NLRP3 inflammasome and triggers the release of pro-inflammatory cytokines (e.g., IL-1β, TNF-α) [12,13]. This results in a self-perpetuating cycle of inflammation and oxidative stress, further exacerbating mitochondrial damage via a positive feedback loop [14,15]. Reduced energy supply and sustained inflammatory signals predominantly drive the pro-inflammatory M1 phenotype, which exacerbates chronic neurodegeneration [[16], [17], [18]]. In contrast, the M2 phenotype, which is associated with anti-inflammatory responses, promotes essential repair mechanisms [19]. Notably, M2 macrophages and their secreted extracellular vesicles (M2EVs) are known for their anti-inflammatory properties, making them promising components for therapeutic intervention [20,21]. Single-cell RNA sequencing studies reveal that SCI induces lasting transcriptional reprogramming in microglia [22]. Although microglia are often simplified into M1 and M2 categories, their polarization is a dynamic process, representing a spectrum of functional states [23]. Additionally, microglia are involved in crucial post-SCI processes such as neuroprotection, angiogenesis, and axonal regeneration [[24], [25], [26]]. Thus, reducing ROS production and maintaining a healthy population of mitochondria within microglia are critical for effective SCI treatment.

Mitophagy, the selective autophagic degradation of damaged mitochondria, plays a fundamental role in maintaining microglial mitochondrial homeostasis and functional plasticity [27,28]. Efficient mitophagy prevents mitochondrial ROS (mtROS) leakage, reduces inflammation, favors an M2-polarized state in microglia, inhibits neuronal ferroptosis, and mitigates astrocyte-mediated cytotoxicity, thereby improving secondary SCI pathology [29,30]. Correcting mitochondrial quality control (MQC) dysfunction, particularly by alleviating suppressed mitophagy in microglia, represents a promising therapeutic strategy for SCI.

ROS serves as a key signal for initiating mitophagy [31,32]. By scavenging pathologically elevated ROS levels in the post-SCI microenvironment, we can directly counteract ROS-mediated suppression of autophagy machinery (e.g., ATG proteins) and promote autophagosome formation [33,34]. In parallel, ROS clearance inhibits the PI3K/AKT/mTOR signaling pathway, a major negative regulator of autophagy, further promoting mitophagy [35,36]. This dual-action strategy not only restores MQC but also increases the population of microglia with functional mitochondria.

Nanomedicine has offered promising strategies for SCI by enabling targeted delivery and microenvironment modulation [37,38]. Among these, Prussian blue (PB) nanozymes have gained attention for their potent catalytic ROS-scavenging activity and intrinsic biocompatibility [[39], [40], [41]]. However, most PB-based approaches rely on passive accumulation or simple surface functionalization, limiting their ability to reach and regulate the intracellular organelles—particularly mitochondria—where critical elements of SCI pathology originate. Likewise, existing biomimetic nanocarriers often employ single-component membrane coatings that enhance immune evasion or targeting but seldom combine biological signaling with catalytic function in a coordinated manner. Thus, a major challenge remains: how to design a therapeutic platform that integrates precise targeting, microglial immunomodulation, and restoration of mitochondrial quality control.

To address this, we developed a modular biohybrid nanoplatform (PM-Lipo@PB NPs) that combines biologically encoded membrane targeting with catalytic nanozyme activity and downstream engagement of mitochondrial repair pathways. The central materials innovation lies in a dual-component fused membrane, formed by directly integrating platelet-derived membrane vesicles (PMVs) with M2EVs, which is subsequently coated with an in situ–assembled Prussian blue shell to create a three-part biohybrid nanoarchitecture. Unlike traditional designs where PB forms the particle core, here PB is assembled as an outer shell coating the fused natural membrane. This configuration ensures that catalytic activity interfaces directly with extracellular ROS while remaining accessible during membrane-mediated uptake and intracellular trafficking—an essential prerequisite for alleviating mitochondrial dysfunction.

The hybrid membrane provides several synergistic advantages. Platelet-derived components supply vascular-injury–directed adhesion via surface molecules such as P-selectin, enabling early and lesion-selective accumulation [42]. M2EVs contribute endogenous pro-regenerative cargo—including regulatory proteins and miRNAs—and retain inflammatory tropism that favors homing to activated immune cells [43]. When fused into a single continuous cloak, these two natural membranes yield emergent properties unobtainable by single-membrane or synthetic coatings alone: multivalent targeting across vascular and immune compartments, improved immune evasion and circulation persistence, and localized presentation of biological signals that complement PB's catalytic ROS neutralization. Within this coordinated framework, the nanoplatform acts as a “nanoscale fire brigade”: once concentrated at the inflammatory epicenter of SCI, it releases its Prussian blue payload to extinguish the local “oxidative fire.”

Mechanistically, this integrated architecture is designed to intervene at multiple interdependent nodes of the SCI pathology network. PB nanozyme activity attenuates oxidative stress, alleviating a major barrier to mitophagic and autophagic function. Concurrently, M2EV-derived signals promote microglial polarization toward an anti-inflammatory phenotype, creating a microenvironment more permissive to mitochondrial recovery [44]. Targeted delivery ensures that these functions converge specifically within the lesion region, concentrating therapeutic activity where mitochondrial injury, ROS burden, and inflammatory activation are most severe. Through this coordinated multi-level regulation, PM-Lipo@PB NPs aim not only to detoxify ROS but also to restore mitochondrial quality control and disrupt the self-sustaining “oxidative stress–inflammation” loop that drives chronic SCI deterioration.

Here, we report the design, characterization, and therapeutic evaluation of this hybrid nanozyme platform in vitro and in vivo. We demonstrate that PM-Lipo@PB NPs exhibit enhanced microglial targeting, effective rescue of mitochondrial function, restoration of mitophagic flux, attenuation of inflammatory responses, and promotion of angiogenesis—collectively reconstructing a cellular and vascular environment conducive to neural repair. Beyond presenting a therapeutic candidate for SCI, this work introduces a conceptual framework for combining natural membrane fusion with catalytic nanomedicine to achieve organelle-level regeneration in complex CNS injuries.

2. Results

2.1. Post-SCI microglial expansion mediates mitophagy suppression via PI3K/AKT/mTOR hyperactivation

To establish the baseline cellular architecture of the uninjured spinal cord, we performed t-distributed stochastic neighbor embedding (t-SNE) analysis to map the major resident cell populations (Fig. 1a). Astrocytes (green), the predominant glial cell type, displayed a widespread distribution across the cord. Oligodendrocytes (red) were primarily localized within the white matter tracts. Microglia (blue) exhibited diffuse distribution, with a notable predominance in the gray matter. Neurons (yellow), although sparse, were concentrated primarily in the gray matter. This spatial distribution aligns with the known cytoarchitectural organization of the spinal cord.

Fig. 1.

Microglial Dynamics and Gene Expression Alterations Following SCI.

(a) t-SNE clustering reveals the major cellular populations and their spatial organization in the intact spinal cord.

(b) Comparative microglial quantification and distribution pre- and post-SCI.

(c) Differential expression of principal cellular markers between intact controls and 3-day post-SCI specimens.

(d) Quantitative assessment of cellular populations at 3 days post-SCI.

(e) KEGG pathway enrichment analysis of transcriptional changes post-SCI.

(f) Heatmap visualization of core mitophagy-related gene expression dynamics.

(g) Temporal expression profiles of autophagy-associated genes at 3, 7, and 14 days post-SCI.

(h) Boxplot quantification of autophagy-related gene expression across post-injury time points.

Following SCI, we quantified the dynamics of microglial activation using established markers, comparing the results to sham controls. Post-SCI t-SNE analysis revealed a dramatic expansion of microglia, indicating a significant increase in their numbers, along with marked centripetal migration toward the lesion site, where they formed a dense perilesional accumulation (Fig. 1b). This activated microglial phenotype was notably persistent at 3 days post-injury (dpi) (Fig. 1c and d), in agreement with canonical microglial activation timelines observed in SCI.

It should be noted that the t-SNE analysis employed in this study is based on transcriptional features to resolve cellular heterogeneity and cluster cell populations, rather than to infer their spatial localization within the spinal cord tissue. To further relate these transcriptomic findings to histopathological changes, we performed additional immunofluorescence staining in sham and SCI groups. Co-staining of iNOS and Arg-1 with DAPI revealed that, under physiological conditions, microglia exhibited a relatively dispersed distribution, whereas after SCI they became markedly activated and accumulated at the lesion site, accompanied by morphological changes such as enlarged soma and retracted processes (Fig. S15a). In parallel, immunofluorescence staining of NeuN and GFAP demonstrated that astrocytes became densely activated and overlapped within the lesion region following injury, while neurons showed evident loss and atrophic features, including reduced soma size and simplified morphology (Fig. S15b). These observations provide histological context supporting the transcriptional alterations identified by single-cell analysis.

Transcriptomic analysis of SCI-related gene expression from public databases revealed significant enrichment of the PI3K/AKT/mTOR pathway through KEGG pathway analysis (Fig. 1e), suggesting its involvement in the suppression of mitophagy. Temporal profiling of mitophagy-related genes at 3, 7, and 14 dpi, compared to sham controls, demonstrated dysregulated expression patterns indicative of MQC failure, with specific inhibition of mitophagy within the injured microenvironment (Fig. 1f, g, and 1h).

Together, these findings demonstrate a robust and sustained expansion and activation of microglia following SCI, coupled with a significant suppression of mitophagy. This suppression is mechanistically driven by hyperactivation of the PI3K/AKT/mTOR signaling pathway, highlighting the therapeutic potential of restoring mitophagic flux to alleviate secondary SCI pathology.

2.2. Fabrication, characterization, and antioxidant capacity of PM-Lipo@PB NPs

M0 macrophages were first induced to differentiate into the M2 phenotype by culturing RAW 264.7 cells with interleukin-4 (IL-4) for 24 h. The M2 macrophages were then cultured in exosome-depleted serum for an additional 24 h. The resulting M2-derived extracellular vesicles (M2EV) were collected by ultracentrifugation. These M2EV inherit the surface proteomic and lipidomic characteristics of their parent M2 macrophages, which are expected to enhance their targeting to inflammatory sites and modulate immune cell responses. The synthetic PMVs were obtained via a repeated freeze-thaw method, as described previously.

To fabricate the PM-Lipo platform, M2EV and PMVs were fused through serial extrusion, creating a hybrid membrane structure. This approach integrates the vascular targeting capabilities of platelet membranes, the innate inflammatory homing ability of M2EV, resulting in a robust and versatile delivery platform. The PB NPs were then incorporated onto the surface of the hybrid nanovesicles via an in situ biomineralization strategy (Fig. 2a).

Fig. 2.

Synthesis and Characterization of PM-Lipo@PB NPs and Evaluation of Their Antioxidant Properties

(a) Schematic illustration of the fabrication process for PM-Lipo@PB NPs.

(b) Transmission electron microscopy (TEM) image showing the core-shell structure of PM-Lipo@PB NPs. Scale bars: 50 nm

(c) Elemental mapping confirming the uniform distribution of Fe, O, N, and P in PM-Lipo@PB NPs.

(d, e) Dynamic light scattering (DLS) analysis showing changes in hydrodynamic diameter (d) and zeta potential (e) during synthesis.

(f) SDS-PAGE protein profile demonstrating retention of EV-derived proteins in PM-Lipo@PB NPs.

(g) Concentration-dependent scavenging efficiency of superoxide anion (O₂•^-).

(h) H₂O₂ elimination capacity compared to untreated control.

(i) Radical scavenging activities against ABTS radicals, with ascorbic acid (AA) as the positive control.

Data represent mean ± SD (n = 3). *p<0.05, **p<0.01 vs. control group.

Transmission electron microscopy (TEM) images revealed that the PM-Lipo@PB NPs exhibited a core-shell structure, with a shell thickness of approximately 12 nm (Fig. 2b). Elemental mapping and analysis confirmed the presence of iron (Fe), oxygen (O), nitrogen (N), and phosphorus (P) in the PM-Lipo@PB NPs, further validating their successful synthesis (Fig. 2c). The high-angle annular dark-field scanning transmission electron microscopy HAADF-STEM image provided complementary structural evidence, clearly delineating the core–shell architecture and confirming the uniform deposition of the electron-dense PB layer on the vesicular framework (Fig. S18). The structural integrity of the PM-Lipo@PB NPs was also corroborated by characteristic peaks of PB NPs in spectroscopic analyses (Fig. S1 and S2). The changes in particle size and surface charge during the synthesis process were monitored using dynamic light scattering (DLS), showing significant increases in both diameter and potential (Fig. 2d and e).

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis demonstrated that the majority of proteins derived from the M2EVs were retained in the final PM-Lipo@PB NPs (Fig. 2f). This result suggests that the PB mineralization process did not adversely affect the protein components of the M2EVs, allowing the preservation of key functional membrane proteins alongside the synthetic PMVs framework.

To evaluate the antioxidant capacity of PM-Lipo@PB NPs, several ROS and reactive nitrogen species (RNS) scavenging assays were conducted. These included H₂O₂, superoxide anion (O₂•^-), and the classical DPPH and ABTS radicals. The scavenging efficiency of O₂•^- increased with the concentration of PM-Lipo@PB NPs, reaching up to 80% at a concentration as low as 30 μg/mL (Fig. 2g). PM-Lipo@PB NPs also showed significant scavenging activity against H₂O₂, with higher concentrations leading to increased scavenging (Fig. 2h). Ascorbic acid (AA) was used as a positive control, and the results indicated that both PM-Lipo@PB NPs and AA were effective in scavenging DPPH and ABTS radicals at low concentrations, with over 80% of the radicals eliminated (Fig. 2i, Fig. S3). These findings demonstrate that PM-Lipo@PB NPs possess superior ROS scavenging capabilities compared to AA, with the coordination of Fe(III) ions not affecting their antioxidant properties.

2.3. PM-Lipo@PB NPs exhibit excellent biocompatibility

Following the successful synthesis and structural characterization of PM-Lipo@PB NPs, we performed comprehensive in vitro and in vivo assessments to evaluate their biocompatibility. Given the intended intravenous administration, we first assessed the compatibility of PM-Lipo@PB NPs with red blood cells, which is critical for intravenous applications. We performed hemolysis assays at various nanoparticle concentrations (1, 5, 10, 20, 50, 75, 100, and 200 μg/mL). Even at higher concentrations (up to 75 μg/mL), PM-Lipo@PB NPs exhibited minimal hemolytic activity, indicating a low risk of hemolysis (Fig. 3a).

Fig. 3.

Biocompatibility Profile of PM-Lipo@PB NPs

(a) Hemolytic response assessment at various nanoparticle concentrations (1–200 μg/mL; n = 6).

(b) BV2 cell viability following exposure to graded concentrations of PM-Lipo@PB NPs (n = 6).

(c-h) Serum biochemical parameters following 28-day administration:

(c) Alanine aminotransferase (ALT, U/L; IFCC method; n = 3)

(d) Aspartate aminotransferase (AST, U/L; aspartate substrate method; n = 3)

(e) Blood urea nitrogen (BUN, mmol/L; urease-GLDH method; n = 3)

(f) Total bilirubin (TBIL, μmol/L; diazo-2,4-dichloroaniline colorimetry; n = 3)

(g) Direct bilirubin (D-Bil, μmol/L; vanadate oxidation; n = 3)

(h) Creatinine (μmol/L; enzymatic colorimetry; n = 3)

(i) Representative bright-field micrographs of H&E-stained tissue sections with corresponding enhanced-color photomicrographs. Scale bars: 2000 μm (overview), 200 μm (detail). Data represent mean ± SD; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 (Tukey's test for parametric data, Dunn's test for nonparametric data following Kruskal-Wallis test). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

In addition to red blood cell compatibility, we evaluated the biocompatibility of PM-Lipo@PB NPs with microglial cells using the BV2 cell line. The results from a CCK-8 assay showed that PM-Lipo@PB NPs did not significantly affect the viability of BV2 cells across the tested concentrations (50 μg/mL), supporting the hemolysis assay findings (Fig. 3b).

To further assess the potential toxicity of PM-Lipo@PB NPs, we measured serum biochemical parameters in mice following a 28-day administration period. These included markers of liver and kidney function, which are commonly used to assess metabolic organ health. No significant differences in alanine aminotransferase (ALT), aspartate aminotransferase (AST), blood urea nitrogen (BUN), total bilirubin (TBIL), direct bilirubin (D-Bil), or creatinine levels were observed between the treated and sham groups, with all values remaining within normal physiological ranges (Fig. 3c–h).

Histopathological examinations of major metabolic organs, including the heart, spleen, liver, lungs, and kidneys, revealed no observable differences between the treated and sham groups (Fig. 3i). Representative hematoxylin and eosin (H&E)-stained tissue sections further confirmed the absence of toxicological changes in any of the examined organs.

Collectively, the in vitro and in vivo findings demonstrate that PM-Lipo@PB NPs exhibit excellent biocompatibility, with no evidence of hemolysis, cellular toxicity, or organ damage, supporting their safe use for therapeutic applications.

2.4. Precise targeting and superior cellular penetration of PM-Lipo@PB NPs

The success of the “firefighting crew” (PB NPs) is heavily dependent on the performance of the “firefighting vehicle” (PM-Lipo), which must ensure precise targeting and efficient delivery. To assess the cellular penetration of PM-Lipo@PB NPs, we conducted in vitro experiments using FITC-labeled PB NPs and PM-Lipo@PB NPs, followed by confocal microscopy. The results revealed significantly enhanced cellular penetration by PM-Lipo@PB NPs compared to PB NPs alone within the same timeframe (Fig. 4a and b), highlighting the improved maneuverability and internalization capability of the hybrid nanoplatform.

Fig. 4.

Cellular Internalization and Biodistribution of FITC-Conjugated Nanoparticles

(a-b) Confocal microscopy visualization and quantification of FITC-PB NPs vs. FITC-PM-Lipo@PB NPs cellular uptake (Scale bar: 20 μm).

(c) Temporal biodistribution (0–24 h post-IV administration) and lesion accumulation.

(d) In vivo fluorescence quantification at the spinal cord lesion site.

(e) Ex vivo fluorescence quantification in spinal tissue.

(f-g) Fluorescence intensity in metabolic organs for (f) FITC-PB NPs and (g) FITC-PM-Lipo@PB NPs.

Data represent mean ± SD, n = 3; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 (Tukey's/Dunn's tests for parametric and nonparametric data, respectively).

Beyond cellular penetration, precise navigation and blood-spinal cord barrier (BSCB) permeability are critical factors for successful SCI therapeutics [45]. To evaluate these parameters, we tracked the biodistribution of FITC-conjugated nanoparticles in mice using live imaging. The results showed that PM-Lipo@PB NPs, compared to PB NPs alone, achieved superior accumulation at the lesion site, demonstrated a longer blood circulation half-life, and exhibited enhanced targeting specificity. This targeted accumulation at the SCI site remained evident even after considering uptake by major metabolic organs, such as the liver and kidneys (Fig. 4c, d, e).

To further assess the in vivo targeting specificity toward IBA-1⁺ microglia, fluorescence microscopy was performed on spinal cord tissue sections. The images showed robust internalization of FITC-labeled nanoparticles by IBA-1⁺ cells within the lesion area. Quantitative co-localization analysis confirmed that FITC-PM-Lipo@PB NPs displayed markedly higher intracellular fluorescence than plain FITC-PB NPs, indicating superior microglial uptake. This enhanced targeting efficiency is attributable to the synergistic functionalization with platelet membranes and M2 macrophage–derived extracellular vesicles, which together promote immune evasion and microglia-directed homing (Figs. S4 and S5).

Furthermore, ex vivo fluorescence quantification confirmed that the majority of the NPs were metabolized through the liver and kidneys, with minimal accumulation in non-targeted organs, reaffirming their biosafety and biocompatibility (Fig. 4f and g). These findings collectively demonstrate the excellent targeting precision, prolonged circulation, and high specificity of PM-Lipo@PB NPs for SCI treatment.

2.5. PM-Lipo@PB NPs attenuate microglial mitochondrial dysfunction and apoptosis under oxidative stress

To assess the therapeutic potential of PM-Lipo@PB NPs against SCI-related oxidative injury, we first investigated their effects on H₂O₂-induced oxidative stress in BV2 microglial cells, which serve as a widely used model for studying post-SCI oxidative damage. We measured mitochondrial membrane potential (ΔΨm) using JC-1 staining, revealing that H₂O₂ exposure caused a significant collapse of ΔΨm, which was notably restored following treatment with PM-Lipo@PB NPs (Fig. 5a–f).

Fig. 5.

PM-Lipo@PB NPs Ameliorate H₂O₂-Induced Mitochondrial Impairment and Cytotoxicity in BV2 Microglia

(a, f) JC-1 staining demonstrating loss of ΔΨm after H₂O₂ (60 μM) exposure and its restoration upon PM-Lipo@PB NPs treatment (scale bar: 50 μm) 

(b, g) Apoptotic progression visualized by PI/Calcein-AM co-staining (scale bar: 50 μm).

(c) Quantitative apoptosis analysis by flow cytometry.

(d–e) Mitochondrial morphometric analysis: (e) branch length and (d) network integrity (scale bars: 5 μm [overview], 1 μm [detail]).

Data represent mean ± SD (n = 3); *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 (Tukey's or Dunn's tests based on distribution normality).

Morphological analysis of mitochondria further demonstrated that H₂O₂ treatment led to severe mitochondrial swelling, fragmentation, and disruption of cristae. In contrast, these pathological changes were substantially mitigated in the presence of PM-Lipo@PB NPs (Fig. 5d and e), indicating their protective effects on mitochondrial integrity.

Furthermore, consistent with the preservation of mitochondrial function, PM-Lipo@PB NPs effectively reduced H₂O₂-induced apoptosis, as shown by live-dead staining using PI/Calcein-AM (Fig. 5b–g) and confirmed by flow cytometric analysis (Fig. 5c). These results suggest that PM-Lipo@PB NPs provide robust protection to microglia under oxidative stress by preventing mitochondrial dysfunction and subsequent apoptosis.

Together, these preliminary findings underscore the potential of PM-Lipo@PB NPs to counteract a critical pathological event in secondary SCI, namely microglial mitochondrial dysfunction and apoptosis, thereby offering a promising therapeutic strategy for SCI recovery.

2.6. PM-Lipo@PB NPs effectively scavenge ROS generated in High-ROS environments

Having established that PM-Lipo@PB NPs protect against mitochondrial dysfunction and apoptosis, we next investigated whether these protective effects were due to their ability to directly neutralize elevated levels of ROS in high-ROS environments. Fluorescence microscopy provided direct visual evidence of ROS scavenging by both PB NPs and PM-Lipo@PB NPs, with the latter demonstrating significantly superior performance (Fig. 6a). This enhanced scavenging ability is likely attributed to the improved cellular penetration of PM-Lipo@PB NPs.

Fig. 6.

ROS Scavenging Efficacy and Cytoprotection in H₂O₂-Stressed BV2 Cells

(a, c) DCFH-DA fluorescence imaging showing ROS clearance (Scale bar: 50 μm) 

(b, d) Cleaved-caspase 3 immunofluorescence and corresponding quantification (Scale bar: 50 μm).

(e) Flow cytometric analysis of ROS levels.

(f, h) Antioxidant capacity: (f) GSH content (nmol/10⁴ cells) and (h) SOD activity.

(g) Measurement of lipid peroxidation (MDA levels).

(i) Cellular ATP content.

Data represent mean ± SD (n = 3); *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 (Tukey's/Dunn's tests based on distribution normality).

Quantitative analysis further supported the superior ROS-quenching capacity of PM-Lipo@PB NPs compared to PB NPs (Fig. 6c). Flow cytometric analysis provided additional confirmation of this enhanced ROS scavenging activity (Fig. 6e). Furthermore, the antioxidant capacity of the nanoparticles was assessed by measuring the activity of key antioxidant enzymes, including SOD and glutathione (GSH) both enzymatic (SOD activity) and non-enzymatic antioxidant indicators (GSH content). Both assays indicated that PM-Lipo@PB NPs significantly outperformed PB NPs in terms of antioxidant activity and redox-regulating capacity (Fig. 6f–h).

To further validate the protective effect of PM-Lipo@PB NPs, we measured malondialdehyde (MDA) levels, a marker of lipid peroxidation and membrane damage. PM-Lipo@PB NPs treatment resulted in a significant reduction in MDA levels, suggesting protection against oxidative damage (Fig. 6g). Additionally, we assessed the expression of cleaved caspase-3, a classical marker of apoptosis, and found that treatment with PM-Lipo@PB NPs significantly reduced its expression, with the most pronounced effect observed in the PM-Lipo@PB NPs group (Fig. 6b–d).

Finally, cellular ATP content was measured across all experimental groups, revealing a strong positive correlation between ROS scavenging activity and ATP levels (Fig. 6i), further emphasizing the critical role of ROS in mitochondrial dysfunction and cellular energy metabolism.

2.7. Hybrid membrane coating enhances the mitochondrial protective efficacy of PB NPs

To further validate the protective effects observed in our preliminary assessments and to directly evaluate the contribution of the hybrid membrane coating, we compared the mitochondrial recovery promoted by PM-Lipo@PB NPs with that of plain PB NPs under identical oxidative stress conditions. ΔΨm assays revealed that both PB NPs and PM-Lipo@PB NPs contributed to the restoration of ΔΨm; however, the effect was significantly more pronounced in the PM-Lipo@PB NPs group (Fig. 7a–d). These results were further corroborated by flow cytometry analysis (Fig. 7f), which confirmed that PM-Lipo@PB NPs provided superior mitochondrial protection.

Fig. 7.

Mitochondrial Preservation in H₂O₂-Challenged BV2 Cells

(a, d) JC-1 fluorescence imaging showing restoration of ΔΨm (Scale bar: 50 μm) 

(b, e) MitoNeoD quantification of intramitochondrial superoxide levels (Scale bars: 5 μm [overview], 1 μm [detail]).

(c) Transmission electron microscopy (TEM) visualization of mitochondrial ultrastructure.

(f) Flow cytometric analysis of ΔΨm.

Data represent mean ± SD (n = 3); *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 (Tukey's/Dunn's tests).

Transmission electron microscopy (TEM) imaging provided ultrastructural evidence of the mitochondrial preservation achieved by PM-Lipo@PB NPs treatment. Specifically, mitochondria from PM-Lipo@PB NPs-treated cells exhibited well-preserved cristae and overall morphology compared to the PB NPs group, which showed more severe mitochondrial damage (Fig. 7c).

To directly link the enhanced mitochondrial protection to the improved ROS scavenging capability conferred by the hybrid membrane coating, we measured intramitochondrial superoxide levels using MitoNeoD staining. The results indicated that superoxide levels were most effectively reduced in the PM-Lipo@PB NPs-treated cells (Fig. 7b–e), further supporting the notion that the hybrid membrane enhances the ROS-scavenging capacity of PB NPs.

Collectively, these findings demonstrate that the hybrid membrane coating not only retains the intrinsic mitochondrial protective activity of PB NPs but also substantially enhances their functional efficacy in restoring mitochondrial health under oxidative stress conditions.

2.8. PM-Lipo@PB NPs alleviate mitophagy suppression in BV2 cells and the underlying mechanism

Building on previous experiments that demonstrated the link between ROS, BV2 cells, and mitochondrial integrity, we next investigated whether PM-Lipo@PB NPs could alleviate mitophagy suppression and elucidate the relationship between mitophagy, ROS, and mitochondrial function.

Mitophagy, a selective autophagic degradation of damaged mitochondria, is often regulated by the co-localization of mitochondria and lysosomes [46]. Following treatment with PM-Lipo@PB NPs, we observed a significant increase in both lysosomal activation and mitochondria-lysosome co-localization, with the most substantial changes occurring in the PM-Lipo@PB NPs-treated cells (Fig. 8a–c). Additionally, we analyzed the co-localization of the mitochondrial outer membrane protein TOMM20 and LC3, a marker of autophagosomes. The results aligned with the lysosomal co-localization data, further confirming that PM-Lipo@PB NPs effectively promote mitophagy in BV2 cells (Fig. 8b–d).

Fig. 8.

Mitophagic Flux Modulation and Mechanistic Insights

(a, c) Live-cell imaging showing co-localization of mitochondria (green) and lysosomes (red) with Hoechst nuclear counterstaining (blue) (Scale bars: 5 μm [overview], 1 μm [detail]).

(b, d) Immunofluorescence showing co-localization of TOMM20 (red) and LC3 (green) with DAPI nuclear stain (blue) (Scale bars: 5 μm [overview], 1 μm [detail]). (e-h) Representative Western blots and quantification of PI3K/AKT/mTOR pathway components in spinal tissue 3 days post-treatment.

Data: mean ± SD (n = 3); *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 (Tukey's/Dunn's tests). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

These findings provided strong evidence that PM-Lipo@PB NPs alleviate mitophagy suppression in BV2 cells, suggesting that their ROS-scavenging properties help restore the autophagic process. To explore the underlying mechanism, we examined whether ROS scavenging-mediated suppression of the PI3K/AKT/mTOR pathway played a role. Western blot analysis revealed that treatment with PM-Lipo@PB NPs reduced ROS levels, which in turn led to decreased phosphorylation of PI3K, AKT, and mTOR—key regulators of autophagy.

To further determine whether modulation of the PI3K/mTOR axis is causally involved in the observed restoration of mitophagy, additional pharmacological intervention experiments were conducted. As shown in Supplementary Fig. S16, PM-Lipo@PB NPs markedly reduced phosphorylation levels of PI3K and mTOR while restoring autophagic activity, as evidenced by increased LC3B level and decreased p62 accumulation. Importantly, these effects were partially reversed upon treatment with the mTOR activator MHY1485, whereas co-administration with the PI3K inhibitor LY294002 further enhanced the pro-autophagic effect (Fig. S16). Consistently, immunofluorescence analysis revealed increased co-localization of LC3 with TOMM20 following nanoparticle treatment, which was attenuated by mTOR activation (Fig. S17). These findings collectively support a functional link between PI3K/mTOR inhibition and mitophagy restoration.

These results indicate that inhibition of the PI3K/AKT/mTOR pathway, in addition to directly counteracting ROS-induced inhibition of autophagy proteins, contributes to the alleviation of mitophagy suppression.

2.9. PM-Lipo@PB NPs promote M2 phenotype polarization in BV2 cells via ROS reduction

Accumulating evidence indicates that eliminating excessive ROS can interrupt the self-reinforcing “inflammation–oxidative stress” cycle, thereby reducing pro-inflammatory cytokine release and preventing further ROS accumulation. In addition, maintaining mitochondrial integrity—ensuring sufficient bioenergetic support—is essential for driving microglia toward an M2 anti-inflammatory phenotype. Given the potent ROS-scavenging and mitochondrial protective properties of PM-Lipo@PB NPs, we evaluated microglial polarization in BV2 cells following treatment with PB NPs or PM-Lipo@PB NPs.

Our results revealed a notable shift towards the M2 phenotype in both the PB NPs and PM-Lipo@PB NPs-treated groups. However, this polarization effect was significantly more pronounced in the PM-Lipo@PB NPs group (Fig. 9a–c, d). This shift was further supported by the expression analysis of inflammatory cytokines. Treatment with PM-Lipo@PB NPs resulted in a marked reduction in the levels of pro-inflammatory cytokines such as IL-1β and TNF-α compared to the other groups, indicating that the enhanced ROS scavenging and mitochondrial protection promoted by PM-Lipo@PB NPs effectively reduced the inflammatory response (Fig. 9b–e, f). To further evaluate treatment-induced changes in microglial activation in vivo, we conducted morphological analyses based on IBA-1 immunofluorescence staining of spinal cord sections. Quantitative assessment of process length and branching complexity demonstrated that both nanoparticle treatments shifted microglia toward a more ramified and less reactive morphology relative to the SCI group (Figs. S6 and S7).

Fig. 9.

PM-Lipo@PB NPs Promote M2 Polarization in BV2 Cells

(a, c, d) Immunofluorescence quantification of M1 (iNOS) and M2 (Arg-1) markers with DAPI counterstaining (Scale bar: 50 μm) 

(b, e, f) Inflammatory cytokine (IL-1β, TNF-α) expression analysis (Scale bar: 50 μm).

Data: mean ± SD (n = 3); *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 (Tukey's/Dunn's tests).

2.10. PM-Lipo@PB NPs promote spinal cord repair and preserve motor function in SCI mice

Building on the promising in vitro results showing the potent ROS scavenging, mitochondrial protection, and M2 polarization-promoting capabilities of PM-Lipo@PB NPs, we next investigated their therapeutic efficacy in vivo using a mouse model of secondary SCI. The in vivo studies were designed to assess both histopathological and functional outcomes.

Longitudinal spinal cord sections stained with H&E revealed significantly smaller lesion cavities and scar areas in the PM-Lipo@PB NPs treated group compared to the controls, indicating enhanced tissue preservation and reduced injury severity. Nissl staining further demonstrated a higher degree of preservation and better organization of Nissl bodies, which are indicative of viable neurons. The PM-Lipo@PB NPs group showed the most substantial improvement in neuronal integrity (Fig. 10a). Luxol Fast Blue (LFB) staining was used to assess neuroaxonal protection, and it showed reduced areas of demyelination in the PM-Lipo@PB NPs group, suggesting superior preservation of myelin sheaths (Fig. 10a).

Fig. 10.

Therapeutic Efficacy in SCI Murine Mode (a) Representative H&E, Nissl, and LFB-stained spinal sections (Scale bars: 1000 μm [overview], 200 μm [detail]).

(b, f, g) Immunofluorescence quantification of neuronal (NeuN) and axonal (NEFH) markers.

(c) Representative footprint patterns.

(d, e) Motor function assessment via Basso Mouse Scale (BMS) and inclined plane testing (n = 6/group).

Data: mean ± SD; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 (Tukey's/Dunn's tests).

Immunofluorescence staining for the neuronal marker NeuN and the neurofilament marker NEFH corroborated these findings, showing greater preservation of neurons and axons in the PM-Lipo@PB NPs group compared to other treatment groups (Fig. 10b). Quantitative analysis of these markers further supported the observation of more extensive neuronal and axonal preservation in the PM-Lipo@PB NPs-treated animals (Fig. 10f and g).

In terms of functional recovery, we assessed motor function using multiple tests. Representative footprint analyses showed significantly better locomotor recovery in the PM-Lipo@PB NPs group (Fig. 10c). This was further confirmed by comprehensive motor function assessments using the Basso Mouse Scale (BMS) and inclined plane test, both of which revealed that the PM-Lipo@PB NPs group had the highest level of preserved motor function compared to the control groups.

2.11. p.m.-Lipo@PB NPs promote angiogenesis to reconstruct a regenerative microenvironment

Successful neural repair relies on the establishment of a supportive microenvironment, in which neovascularization plays a central role by supplying nutrients and oxygen while facilitating metabolic waste removal. Both M2 macrophages and their extracellular vesicles (M2EVs) are recognized as critical regulators of angiogenesis, as they carry pro-angiogenic miRNAs (e.g., miR-210, miR-126) and proteins (e.g., VEGF, TGF-β) that stimulate endothelial proliferation, migration, and tube formation. In parallel, platelet membranes contain growth factors and angiogenesis-associated proteins that further support vascular regeneration. At sites of spinal cord injury, however, excessive oxidative stress and inflammation impair endothelial stability and compromise nascent vessel formation. The PB nanozyme component of PM-Lipo@PB NPs counteracts this effect by efficiently scavenging ROS, thereby protecting endothelial cells from oxidative injury and preserving their function. Additionally, the anti-inflammatory activity of the hybrid system reduces vascular inhibitory factors such as TNF-α, creating a more permissive biochemical environment for angiogenesis. Collectively, these features suggest that PM-Lipo@PB NPs act as a synergistic reservoir and targeted delivery platform for pro-angiogenic signals. To provide a more integrated view of the therapeutic cascade, a schematic illustration summarizing the proposed mechanism has been included (Fig. S14), depicting the sequential process from lesion-associated accumulation and ROS scavenging to PI3K/mTOR inhibition, mitophagy restoration, microglial polarization, and subsequent neurovascular repair.

To validate its angiogenic potential, we performed a series of in vitro assays using human umbilical vein endothelial cells (HUVECs) under oxidative stress. First, a tube formation assay was conducted. HUVECs were stained with Calcein AM, and their tubular networks were visualized via fluorescence microscopy after 6 h and 12 h of culture. Quantitative analysis of the number of branch points and total tube length revealed that both PB NPs and PM-Lipo@PB NPs treatment groups exhibited superior tube-forming capacity compared to the injury-only control, with the PM-Lipo@PB NPs group showing the most pronounced enhancement (Fig. S8 and S9). Consistent with this finding, confocal microscopy analysis of CD31 (an endothelial cell marker) expression in HUVECs demonstrated a significant increase in fluorescence intensity in the treatment groups, particularly in the PM-Lipo@PB NPs group, corroborating the pro-angiogenic effect (Fig. S10 and S11). Additionally, a scratch wound healing assay was employed to evaluate HUVEC migration. Quantification of the wound closure area at 12 h and 24 h post-scratch confirmed that the PM-Lipo@PB NPs treatment most effectively promoted endothelial cell migration under oxidative stress conditions (Fig. S12 and S13).

3. Discussion

Spinal cord injury (SCI) remains challenging to treat because secondary damage is driven by a tightly coupled network of pathological processes—including excessive oxidative stress, mitochondrial collapse, impaired mitophagy, and maladaptive microglial activation—that reinforce one another to sustain a degenerative microenvironment [47,48]. Current nanotherapeutic strategies typically target isolated pathological events (e.g., ROS scavenging or inflammation suppression), which limits their ability to restore integrated cellular homeostasis [49]. In this study, we designed PM-Lipo@PB NPs, a biohybrid platform that integrates biologically encoded targeting, catalytic ROS neutralization, and modulation of mitochondrial quality control within microglia.

In SCI, multiple reactive oxygen species—including superoxide (O₂•^-), hydrogen peroxide (H₂O₂), and hydroxyl radicals (•OH)—contribute to secondary injury. Among these, H₂O₂ is relatively stable and diffusible, often accumulating at micromolar levels in the lesion microenvironment, thereby sustaining oxidative damage and disrupting mitochondrial signaling. The broad-spectrum ROS-scavenging activity of Prussian blue nanozymes is therefore particularly advantageous for restoring redox balance at a systems level.

3.1. Innovation in the three-component hybrid membrane

At the core of this strategy is a fused hybrid membrane created by the direct integration of PMVs and M2EVs, followed by in situ assembly of a Prussian blue (PB) nanozyme shell on the membrane surface. This architecture enables PB to function as an exposed catalytic interface, maintaining ROS-scavenging activity during circulation, cellular uptake, and intracellular trafficking. Unlike synthetic ligand modification, the hybrid membrane conveys biologically encoded recognition and signaling. PM components guide lesion-anchored accumulation through vascular adhesion molecules such as P-selectin, while M2EVs offer endogenous pro-repair molecules and an intrinsic tropism toward inflamed immune cells [50,51]. The integration of these components enhances targeting efficiency, immune compatibility, and local immunomodulation in a coordinated manner.

3.2. Advancing beyond traditional PB nanozyme therapies

While PB nanozymes are widely appreciated for their robust catalytic detoxification of ROS, their clinical impact has been constrained by limited targeting and insufficient intracellular engagement [52]. PM-Lipo@PB NPs overcome these limitations by enhancing microglial uptake and enabling PB activity in proximity to mitochondria. Importantly, ROS clearance is coupled with regulation of mitophagy-related signaling pathways, including normalization of PI3K/AKT/mTOR activity, thereby extending the functional scope of PB nanozymes beyond extracellular antioxidation.

3.3. Mechanistic significance

The therapeutic benefits arise not from additive functions but from coordinated synergy. The hybrid membrane enriches nanozyme accumulation where ROS burden and mitophagy blockade are maximal. M2EV-derived biomolecules reinforce microglial transition toward reparative phenotypes. PB catalytically removes ROS, lifting a central barrier that impairs autophagic and mitophagic machinery. Together, these actions restore mitochondrial homeostasis, attenuate the self-perpetuating inflammatory cycle, and create a microenvironment permissive for angiogenesis and neuroprotection. Collectively, these findings support a coordinated therapeutic paradigm in which redox modulation, mitochondrial quality control, and immune phenotypic regulation are mechanistically interconnected rather than independently regulated processes. In this context, suppression of excessive ROS not only alleviates oxidative damage but also functions upstream to relieve aberrant activation of the PI3K/AKT/mTOR axis, thereby restoring mitophagic flux. The reestablishment of mitochondrial homeostasis subsequently provides a metabolic foundation for microglial phenotypic transition toward a reparative state. This cascade highlights a hierarchical regulatory framework in which metabolic and redox cues jointly orchestrate immune responses and tissue repair following spinal cord injury.

From a conceptual perspective, this study extends beyond conventional antioxidant strategies by demonstrating that nanozyme-mediated redox regulation can serve as an upstream regulatory node that coordinates multiple downstream repair processes. Unlike traditional approaches that target single pathological factors, the present platform integrates lesion-associated accumulation, enzymatic ROS scavenging, and immunomodulatory membrane functionality into a unified system. This modular design enables simultaneous intervention across oxidative stress, mitochondrial dysfunction, and immune imbalance, thereby offering a more holistic strategy for reshaping the post-injury microenvironment.

3.4. Translational perspective

From a translational standpoint, the biomimetic and modular nature of PM-Lipo@PB NPs provides several advantages. The use of biologically derived membrane components facilitates compatibility with the host environment, while the nanozyme core offers stable and sustained catalytic activity without the limitations associated with natural enzymes. Moreover, the systemic administration route and observed lesion-associated accumulation suggest potential feasibility for minimally invasive therapeutic strategies. These features collectively position this platform as a promising candidate for further preclinical development in neurotrauma and related disorders.

3.5. Limitations and future directions

Despite these promising findings, several limitations should be acknowledged. First, the scalability of hybrid membrane-based nanoplatforms remains a technical challenge, particularly with respect to maintaining vesicle integrity and compositional consistency during large-scale production. Second, the use of extracellular vesicle-derived components introduces potential variability due to differences in source cells and culture conditions, which may affect reproducibility. Third, although biosafety was systematically evaluated, the long-term biodistribution, degradation pathways, and potential immunogenicity of such biomimetic systems require further investigation. In addition, while BV2 cells provide a convenient and widely accepted in vitro model, they may not fully capture the complexity of primary microglia in vivo. Future studies incorporating primary cells and large-animal models will be essential to further validate the therapeutic potential and clinical applicability of this strategy.

3.6. Overall significance

In summary, PM-Lipo@PB NPs provide a multifunctional platform that integrates redox regulation, mitophagy restoration, and immune modulation to reshape the SCI microenvironment. This study establishes a mechanistic link between oxidative stress, mitochondrial quality control, and microglial polarization, offering a framework for the development of next-generation nanotherapeutics targeting complex CNS injuries.

4. Conclusion

In this work, we establish that simultaneous lesion targeting, catalytic ROS neutralization, and restoration of microglial mitophagic flux represents a compelling and mechanistically grounded therapeutic paradigm for spinal cord injury. The PM-Lipo@PB NPs system embodies this concept through a deliberately engineered, tri-component hybrid membrane fused from platelet-derived vesicles and M2 macrophage–derived EVs, overlaid with an in situ–assembled Prussian blue shell. This biohybrid configuration integrates vascular-injury tropism, inflammatory homing, endogenous immunoregulatory cues, and catalytic nanozyme activity into a single, modular platform.

The therapeutic efficacy of PM-Lipo@PB NPs arises from a coordinated, multi-level mechanism: membrane-guided delivery concentrates the nanozyme within the lesion niche; M2EV-derived signals contribute a permissive immunoregulatory environment; PB-mediated catalytic ROS scavenging alleviates the oxidative blockade that impairs mitochondrial quality control; and suppression of PI3K/AKT/mTOR hyperactivation restores regulated mitophagy. Collectively, these actions disrupt the self-amplifying “inflammation–oxidative stress” feedback loop that drives secondary degeneration in SCI.

By reestablishing organelle-level homeostasis and promoting a microenvironment favorable for repair, PM-Lipo@PB NPs yield meaningful neuroprotection and functional preservation in vivo. Beyond introducing a potent therapeutic candidate, this study highlights a broadly applicable design principle: biologically fused natural membranes can be integrated with catalytic nanomaterials to achieve targeted, multi-modal intervention in complex CNS injuries. This work therefore lays a foundation for future translational development and provides a conceptual framework for next-generation nanomedicines capable of orchestrating regenerative processes at molecular, cellular, and tissue scales.

Funding sources

This work was supported by the Liaoning Provincial Science and Technology Plan Joint Program (Natural Science Foundation-General Program) (2025-MSLH-219) and the Jinzhou City Guiding Science and Technology Plan Project (JZ2025B015).

CRediT authorship contribution statement

Yu Deng: Conceptualization, Formal analysis, Methodology, Software, Visualization, Writing – original draft. Haosen Zhao: Conceptualization, Formal analysis, Methodology, Software, Visualization. Mina Huang: Conceptualization, Methodology, Software, Visualization. Yue Guo: Conceptualization, Formal analysis, Validation. Feng Jin: Methodology, Software, Visualization. Baofeng Zhao: Formal analysis, Visualization. Hongkai Yang: Methodology, Software. Dongyang Li: Software. Xifan Mei: Conceptualization, Data curation, Funding acquisition, Resources, Supervision, Validation, Writing – review & editing. Sen Lin: Conceptualization, Data curation, Funding acquisition, Resources, Supervision, Validation, Writing – review & editing. Zhanpeng Guo: Conceptualization, Data curation, Funding acquisition, Resources, Supervision, Validation, Writing – review & editing.

Declaration of competing interest

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

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Data availability

Data will be made available on request.