A nanozyme-augmented peripheral nerve-targeting liposome alleviates local immune inflammation and ferroptosis for enhanced facial nerve repair

Abstract

Facial nerve injury (FNI) is a common form of peripheral nerve damage that often results in incomplete functional recovery due to a hostile microenvironment characterized by excessive oxidative stress and inflammatory activation. These pathological conditions disrupt Schwann cell homeostasis, impair myelin maintenance, and hinder nerve regeneration. Although ferroptosis, an iron-dependent lipid peroxidation-driven form of regulated cell death, has been implicated in peripheral nerve disorders, its involvement in acute FNI remains not fully defined.

In this study, transcriptomic analysis of injured facial nerve tissue suggested ferroptosis-related transcriptional alterations and lipid peroxidation-associated cellular injury during the acute phase of FNI, including changes in ferroptosis-related gene expression, decreased GPX4 levels, and increased 4-HNE accumulation. To modulate this complex microenvironment, we developed a peripheral nerve-targeted, ROS-responsive liposomal nanoplatform (F-MHC@PNRLs) co-delivering Ferrostatin-1 (Fer-1) and Mn-doped CeO₂ nanozymes (MHC). The designed system integrates complementary mechanisms, where MHC provides catalytic ROS-scavenging activity to regulate upstream oxidative stress, while Fer-1 acts downstream to suppress lipid peroxidation propagation. In addition, NP41 peptide modification enhances peripheral nerve-associated retention and lesion-site accumulation following local administration. In a mouse FNI model, F-MHC@PNRLs effectively reduced oxidative stress, modulated macrophage polarization toward a reparative phenotype, and was associated with activation of NRF2/HO-1 antioxidant signaling while inhibiting NF-κB/MAPK-mediated inflammatory responses. Furthermore, F-MHC@PNRLs alleviated Schwann cell injury, preserved mitochondrial homeostasis, and promoted remyelination, leading to improved functional and electrophysiological recovery. Importantly, the formulation exhibited favorable preliminary biosafety in vivo.

Collectively, these findings suggest that ferroptosis-related lipid peroxidation and oxidative inflammatory imbalance are involved in acute FNI pathology, and demonstrate that NP41-modified ROS-responsive liposomes co-delivering Fer-1 and Mn-doped CeO₂ nanozymes represent a promising strategy for enhancing facial nerve repair.

Graphical abstract

1. Introduction

Facial nerve injury (FNI) is a common type of peripheral nerve damage caused by trauma, tumor compression, viral infection, or iatrogenic injury [1]. Clinically, FNI is mainly manifested as unilateral facial muscle dysfunction, including facial paralysis, incomplete eyelid closure, impaired blink reflex, and abnormal facial expressions [2]. These functional deficits not only compromise facial symmetry and quality of life, but also impose long-term psychological and social burdens on patients, and severe cases may progress to irreversible functional impairment [1]. After FNI, the distal nerve segment rapidly undergoes Wallerian degeneration (WD), characterized by axonal disintegration, myelin sheath breakdown, and inflammatory cell recruitment [3]. During the subsequent repair process, Schwann cells are indispensable for nerve regeneration, as they dedifferentiate, proliferate, clear myelin debris, secrete neurotrophic factors, and eventually support axonal regrowth and remyelination [4,5]. However, the regenerative capacity of injured facial nerves is often compromised by an adverse local microenvironment dominated by excessive oxidative stress, persistent inflammatory activation, and regulated cell death. Emerging evidence has identified ferroptosis, a form of iron-dependent regulated cell death driven by lipid peroxidation, as an important pathological mechanism in peripheral nerve injury [6,7]. Excessive ferroptosis-related lipid peroxidation may impair Schwann cell survival, aggravate neuronal damage, disrupt myelin reconstruction, and ultimately delay axonal regeneration and functional recovery [6]. Oxidative stress represents a central upstream driver of ferroptosis. Following nerve injury, excessive accumulation of reactive oxygen species (ROS) in the local microenvironment promotes lipid peroxidation and triggers ferroptotic cascades [8]. Meanwhile, ferroptosis can further amplify ROS generation and mitochondrial dysfunction, forming a self-reinforcing pathological loop between oxidative stress and ferroptotic damage [8,9]. This vicious cycle aggravates intracellular oxidative injury, impairs mitochondrial homeostasis, and contributes to irreversible neuronal and glial cell damage [10]. Therefore, the tightly coupled oxidative stress–ferroptosis axis, together with inflammation-driven microenvironmental deterioration, constitutes a major obstacle to effective facial nerve regeneration.

Based on these pathological mechanisms, effective therapeutic strategies for FNI should aim to interrupt the “oxidative stress–inflammation–ferroptosis” axis at an early stage and reconstruct a regenerative microenvironment favorable for neural repair. This requires therapeutic systems capable of simultaneously regulating ROS overload, inflammatory activation, and ferroptotic injury, rather than targeting a single pathological event. Ferrostatin-1 (Fer-1), a classical ferroptosis inhibitor, functions as a radical-trapping antioxidant that interrupts lipid peroxidation propagation [11]. By suppressing phospholipid hydroperoxide accumulation and lipid-derived radical chain reactions, Fer-1 directly targets one of the central execution steps of ferroptotic cell death. In peripheral nerve injury, where Schwann cells are particularly vulnerable to oxidative stress and membrane lipid peroxidation, Fer-1-mediated ferroptosis inhibition may help preserve Schwann cell viability, maintain myelin-supporting functions, and provide a cellular basis for axonal regeneration and remyelination [12]. Nevertheless, Fer-1 mainly acts as a radical-trapping antioxidant to interrupt lipid peroxidation propagation; however, it does not broadly eliminate upstream ROS overproduction, suppress Fe²⁺-driven Fenton chemistry, or restore global redox homeostasis. These unresolved oxidative processes may continuously generate highly reactive radicals and sustain oxidative damage, thereby limiting the durability of Fer-1-mediated ferroptosis inhibition in complex injury microenvironments [13,14]. In injured peripheral nerves, excessive ROS accumulation can further activate pro-inflammatory signaling pathways, promote macrophage inflammatory polarization, impair Schwann cell function, and aggravate demyelination [6,15]. Thus, even when lipid peroxidation is partially interrupted by Fer-1, persistent ROS production and inflammatory activation may continue to drive tissue damage and weaken durable ferroptosis inhibition. These limitations suggest that ferroptosis blockade alone is unlikely to provide sufficient and long-lasting protection in pathological niches characterized by simultaneous oxidative stress, iron dysregulation, and sustained inflammation.

Nanozymes, a class of nanomaterials with enzyme-like catalytic activities, have recently emerged as powerful tools for oxidative microenvironment regulation [16]. Compared with conventional small-molecule antioxidants, nanozymes can catalytically and repeatedly eliminate ROS, thereby providing broader and more durable redox modulation [17,18]. Among them, CeO₂-based nanozymes have attracted considerable attention owing to their reversible Ce³⁺/Ce⁴⁺ redox cycling, which endows them with superoxide dismutase- and catalase-mimicking activities. Through these enzyme-like catalytic functions, CeO₂ nanozymes can convert superoxide anions and hydrogen peroxide into less harmful products, reduce oxidative stress burden, and restore redox homeostasis [19]. More importantly, by reducing upstream ROS accumulation, CeO₂-based nanozymes may further suppress ROS-amplified inflammatory signaling and inhibit pro-inflammatory macrophage activation [20]. Mn doping provides an additional opportunity to enhance the catalytic performance and biological functionality of CeO₂ nanozymes. The incorporation of Mn can modulate the oxygen vacancy structure and redox cycling capacity of CeO₂, thereby improving ROS-scavenging efficiency and broadening antioxidative activity [18]. Therefore, Mn-doped CeO₂ nanozymes may serve as upstream microenvironmental regulators in FNI by continuously scavenging excessive ROS, attenuating oxidative stress-driven inflammation, and protecting repair-associated cells from redox imbalance. In addition to regulating oxidative stress and inflammation, efficient lesion-site retention and peripheral nerve-associated delivery are also critical for improving therapeutic efficacy after FNI. Local administration can increase drug exposure around the injured nerve, but conventional nanocarriers may still diffuse into surrounding tissues, resulting in insufficient retention at the nerve lesion site. NP41 is a peripheral nerve-affinitive peptide that has been used to enhance the association of delivery systems with peripheral nerve tissues. Surface modification with NP41 may improve the local retention of liposomal formulations around injured peripheral nerves and increase their interaction with nerve-associated structures after local administration. Therefore, NP41-mediated modification provides a rational strategy to improve the spatial delivery efficiency of ROS-responsive liposomal systems for peripheral nerve repair.

Based on these considerations, Fer-1 and MHC were integrated to achieve complementary regulation of ferroptosis-related lipid peroxidation and oxidative inflammation. Fer-1 blocks the downstream execution phase of ferroptosis-related injury by suppressing lipid peroxidation, whereas Mn-doped hollow mesoporous ceria nanozymes (MHC) regulate the upstream oxidative-inflammatory microenvironment through catalytic ROS scavenging and redox homeostasis restoration. To further improve lesion-site retention, ROS-responsive release, and peripheral nerve-associated delivery, we constructed a NP41-modified ROS-responsive liposomal nanoplatform, termed F-MHC@PNRLs. As shown in Scheme 1, Fer-1 was loaded into MHC to form F-MHC, which was subsequently encapsulated into NP41-modified ROS-responsive liposomes (PNRLs). After local administration in the FNI model, F-MHC@PNRLs are designed to enhance peripheral nerve-associated accumulation and local retention around the injured nerve region while enabling ROS-responsive F-MHC release. The released F-MHC is expected to scavenge ROS, be associated with activation of NRF2/HO-1 antioxidant signaling, suppress NF-κB/MAPK-mediated inflammatory responses, promote reparative macrophage polarization, and alleviate ferroptosis-related Schwann cell injury, thereby improving mitochondrial homeostasis, Schwann cell repair responses, remyelination, and early functional facial nerve recovery.

Scheme 1.

Construction of peripheral nerve-targeted ROS-responsive F-MHC@PNRLs and their therapeutic mechanism for FNI. F-MHC@PNRLs were fabricated by encapsulating Fer-1-loaded Mn-doped CeO₂ nanoparticles into NP41-modified ROS-responsive liposomes. After local administration in the facial nerve injury model, F-MHC@PNRLs preferentially target injured peripheral nerves and release therapeutic nanoparticles in response to excessive ROS. The released F-MHC scavenges ROS, activates the NRF2/HO-1 pathway, suppresses NF-κB/MAPK-mediated inflammation, promotes M2 macrophage polarization, and inhibits ferroptosis in Schwann cells, thereby improving mitochondrial homeostasis, Schwann cell proliferation and migration, and ultimately facilitating facial nerve repair.

2. Experimental section

2.1. Materials and reagents

DSPE-TK-PEG₂₀₀₀-COOH (abbreviated as DSPE-TK-PEG-COOH) was purchased from Xi'an Ruixi Biological Technology Co., Ltd. (Xi'an, China), and the NP41 peptide (sequence: NH2-NTQTLAKAPEHT-OH) was purchased from Hefei Novabio-Chem Biotechnology Co., Ltd. (Hefei, China). Cerium nitrate hexahydrate, polyvinylpyrrolidone K30 (PVP K30), glacial acetic acid, and ethylene glycol were purchased from Macklin Biochemical Co., Ltd. (Shanghai, China). Manganese chloride was purchased from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Ferrostatin-1 (Fer-1; Cat# A4371) and RSL3 (Cat# B6095) were purchased from APExBIO (Houston, TX, USA). ELISA kits for mouse IL-1β (Cat# ZC-37974) and TNF-α (Cat# ZC-39024) detection were obtained from Shanghai Zhuocai Biotechnology Co., Ltd. (Shanghai, China). Cell Counting Kit-8 (CCK-8; Cat# YZ-CK04), Calcein/PI Cell Viability/Cytotoxicity Assay Kit (Cat# CA5640) were purchased from Beyotime (Shanghai, China). Reactive Oxygen Species Assay Kit (Cat# S0033S), Hoechst 33342 (Cat# C1022), DAPI (Cat# C1002), Malachite Green Phosphate Detection Kit (Cat# S0196), Cellular Ferrous Iron Red Fluorescence Detection Kit (RhoNox-6; Cat# S1070), Lipid Peroxidation Assay Kit (Cat# S0043S), and qPCR primer pairs (Table 1) were purchased from Beyotime (Shanghai, China). Assay kits for superoxide dismutase (SOD), malondialdehyde (MDA), and catalase (CAT) were purchased from Solarbio (Beijing, China). Phosphate-buffered saline (PBS, pH 7.4) and cell culture-related media were obtained from Gibco (California, USA) and Servicebio (Wuhan, China). Nuclear protein and plasma protein extraction kit (Cat# PH0324) was purchased from Phygene (Fujian, China). Primary antibodies against NF-κB p65 (1:1500; Cat# 8242), phospho-NF-κB p65 (1:1500; Cat# 3033), IκBα (1:1000; Cat# 4814), phospho-IκBα (1:1000; Cat# 2859), ERK1/2 (1:1500; Cat# 4695), phospho-ERK1/2 (1:1500; Cat# 4370), JNK (1:1500; Cat# 9252), phospho-JNK (1:1500; Cat# 4668), p38 MAPK (1:1500; Cat# 8690), and phospho-p38 MAPK (1:1500; Cat# 9211) were purchased from Cell Signaling Technology (Danvers, MA, USA). Primary antibodies against NRF2 (1:1500; Cat# HA723302), HO-1 (1:1500; Cat# HA721854), GPX4 (1:1500; Cat# ET1706-45), β-actin (1:15000; Cat# HA722023), and Lamin B1 (1:10000; Cat# R1508-1) were obtained from HUABIO (Hangzhou, China). Primary antibodies against iNOS (1:1000; Cat# AF0199) and CD206 (1:1000; Cat# DF4149) were purchased from Affinity Biosciences (Jiangsu, China). MitoSOX Red Mitochondrial Superoxide Indicator (Cat# 40778ES50) and JC-1 Mitochondrial Membrane Potential Assay Kit (Cat# 40706ES60) were purchased from Yeasen Biotechnology Co., Ltd. (Shanghai, China).

Table 1.

Primer pairs used for quantitative real-time PCR analysis.

Primers	Catalog number	Distributor
Mouse Tnf qPCR Primer Pair	QM05522S	Beyotime
Mouse Il6 qPCR Primer Pair	QM03482S	Beyotime
Mouse Nos2 qPCR Primer Pair	QM04186S	Beyotime
Mouse Il1b qPCR Primer Pair	QM03422S	Beyotime
Mouse Arg1 qPCR Primer Pair	QM09790S	Beyotime
Mouse Actb qPCR Primer Pair	QM00002S	Beyotime

2.2. Facial nerve crush injury model

All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee of Tongji University (Approval No.: SHDSYY-2026-P1817-1) and were reported in accordance with the ARRIVE guidelines [21]. Before the formal experiment, the sample size was estimated using the degree of freedom (E-value) method [22]. Eight-week-old female C57BL/6 mice were obtained from the Animal Center of Tenth People's Hospital Affiliated to Tongji University (Shanghai, China). All animals were housed under specific pathogen-free (SPF) conditions at 20–25 °C and 40–70% humidity with a 12 h light/dark cycle and were given free access to sterile food and purified water. Mice were acclimatized for 7 days before surgery.

All surgical procedures were performed under anesthesia with 1.25% tribromoethanol at a dose of 200 μL/10 g body weight. Under sterile conditions, the right facial nerve was exposed at the mastoid foramen and crushed twice at different angles using microforceps for 30 s each time to establish the facial nerve crush injury model. After surgery, the incision was sutured layer by layer, and the mice were placed on a constant-temperature heating pad until recovery from anesthesia. In the sham-operated group, the right facial nerve was exposed without crush injury. All procedures were performed by the same investigator to minimize inter-individual and technical variability. Postoperative animals were monitored daily for general condition, wound healing, and signs of distress.

2.3. Assessment of ferroptosis-related molecular changes after FNI by RNA-seq and validation

To investigate ferroptosis-related molecular changes after facial nerve injury (FNI), a total of 12 mice were used, including 6 sham-operated mice and 6 FNI mice. For RNA-seq analysis, facial nerve tissues were collected from sham-operated mice and FNI mice at day 1 after injury, with three biological replicates per group. For immunofluorescence validation, facial nerve tissues were collected from FNI mice at day 1 after injury, with three mice per time point. Additional sham-operated facial nerve tissues were collected as baseline controls for immunofluorescence staining.

RNA-seq was commissioned to Shanghai Personal Biotechnology Co., Ltd. (Shanghai, China), including mRNA enrichment, library construction, and high-throughput sequencing. Subsequent bioinformatics analysis was erformed using R software (version 4.3.3), and the analysis workflow and visualization were conducted according to previously reported methods with minor modifications [23,24]. To quantitatively evaluate the ferroptosis tendency of each sample, a ferroptosis potential index (FPI) was calculated based on single-sample gene set enrichment analysis (ssGSEA) [25]. Briefly, ferroptosis-related genes were divided into a ferroptosis-positive gene set and a ferroptosis-negative gene set according to their reported regulatory roles in ferroptosis. The ferroptosis-positive gene set included Acsl4, Ncoa4, Sat1, Tfrc, Alox15, Lpcat3, Nox1, Steap3, and Slc11a2, whereas the ferroptosis-negative gene set included Gpx4, Slc7a11, Fth1, Ftl, Gclm, Gss, Slc3a2, Prnp, and Pcbp1. These two gene sets were constructed using the GeneSet function and integrated into a GeneSetCollection object. ssGSEA was then performed on the normalized gene expression matrix using the GSVA package in R. The enrichment scores of the ferroptosis-positive and ferroptosis-negative gene sets were calculated for each sample. Finally, the FPI was defined as the difference between the ssGSEA score of the ferroptosis-positive gene set and that of the ferroptosis-negative gene set, as follows: FPI = ssGSEA score_positive − ssGSEA score_negative. A higher FPI indicates a stronger ferroptosis tendency, whereas a lower FPI suggests reduced ferroptosis potential. Finally, the harvested tissues were fixed and subjected to immunofluorescence staining for GPX4, 4-HNE, S100β, and DAPI, thereby enabling assessment of temporal alterations in ferroptosis-related antioxidant defense, lipid peroxidation, and Schwann cell-associated responses after FNI.

2.4. Preparation and characterization of DSPE-TK-PEG-NP41

200 mg of DSPE-TK-PEG-COOH was accurately weighed and dissolved in 20 mL of dimethyl sulfoxide (DMSO). Then, 100 mg of EDC was added to the solution, and the mixture was incubated in a shaker at room temperature for 30 min in the dark to activate the carboxyl groups. Subsequently, 50 mg of NP41 peptide was added into the activated solution, followed by sonication to achieve complete dissolution. The reaction mixture was stirred continuously at 4 °C overnight to facilitate the coupling reaction. After the reaction, the crude product was transferred into a 1000 kDa dialysis bag and dialyzed against ultrapure water for 3 days at 4 °C to remove residual DMSO, EDC, and unreacted small molecules. The dialysate was refreshed every 4 h during the dialysis process. Finally, the purified solution was freeze-dried to obtain the final DSPE-TK-PEG-NP41 conjugate as a white fluffy powder. The successful synthesis of DSPE-TK-PEG-NP41 was further verified by Fourier-transform infrared spectroscopy (FTIR), with particular attention to the characteristic absorption peak around 1100 cm⁻¹, corresponding to the C–O–C stretching vibration of PEG, and the absorption band around 1650 cm⁻¹, corresponding to the amide I band, indicating successful coupling between DSPE-TK-PEG-COOH and NP41.

2.5. Preparation and characterization of peripheral nerve-targeted ROS-responsive liposomes (PNRLs)

Egg yolk lecithin (12 mg), cholesterol (4 mg), and DSPE-TK-PEG-NP41 (4 mg) were dissolved in 10 mL of chloroform. The solvent was removed using a rotary evaporator at 40 °C with a condenser temperature of −10 °C and a rotation speed of 70 rpm for 10 min to form a thin lipid film. The film was then hydrated with 10 mL of PBS under high-speed stirring overnight. The resulting liposomal suspension was sonicated at 220 W with an alternating cycle of 5 s on and 5 s off for a total of 5 min, followed by sequential extrusion through 0.45 μm and 0.22 μm membranes to obtain peripheral nerve-targeted ROS-responsive liposomes (PNRLs). In parallel, non-targeted ROS-responsive liposomes (RLs) were prepared using the same procedure, except that DSPE-TK-PEG-NP41 was replaced with DSPE-TK-PEG-COOH (4 mg). The particle size, polydispersity index (PDI), and zeta potential of the liposomes were determined by dynamic light scattering. The morphology of the liposomes was observed by TEM.

2.6. Synthesis of Mn-doped CeO₂ nanozymes and preparation of Fer-1-loaded MHC (F-MHC)

Briefly, 1 g of cerium nitrate hexahydrate was dissolved in a mixed solvent containing 10 mL of deionized water and 20 mL of ethylene glycol in a glass beaker under magnetic stirring at room temperature. Subsequently, 1 mL of glacial acetic acid was added dropwise, followed by the slow addition of 0.8 g of polyvinylpyrrolidone (PVP). The mixture was continuously stirred until a homogeneous solution was obtained. The resulting solution was then transferred into a 50 mL Teflon-lined stainless-steel autoclave and heated at 180 °C for 22 h. After naturally cooling to room temperature, anhydrous manganese chloride (29.0 mg) was added, and the mixture was subjected to a further hydrothermal reaction for 2 h. After cooling to room temperature, the obtained precipitate was collected by centrifugation and washed twice with ethanol, methanol, and deionized water. The washed product was vacuum-dried at 50 °C to obtain a light gray precursor powder. Finally, the precursor was calcined in a muffle furnace at 500 °C for 2 h to yield light yellow manganese-doped cerium oxide powder.

Subsequently, 10 mg of as-prepared MHC was weighed and dispersed in 10 mL of dimethyl sulfoxide (DMSO) under sonication to form a homogeneous suspension. Subsequently, Ferrostatin-1 (Fer-1) solution was added into the above suspension at a mass ratio of Fer-1 to MHC of 1:10 (w/w), corresponding to 1 mg of Fer-1. The mixture was magnetically stirred at room temperature in the dark for 12–24 h to facilitate sufficient drug loading. After incubation, the resulting suspension was centrifuged at 12, 000 rpm for 10 min, and the precipitate was collected and washed repeatedly with deionized water to remove unloaded free Fer-1. Finally, the purified product was freeze-dried to obtain a solid powder of Fer-1-loaded MHC, hereafter denoted as F-MHC. The morphology, microstructure, and elemental composition of F-MHC were characterized by TEM, scanning electron microscopy (SEM), and X-ray energy dispersive spectroscopy (EDS). UV–vis absorption spectroscopy was applied to quantify the drug loading content and encapsulation efficiency. Zeta potential and dynamic light scattering (DLS) were performed to evaluate the surface charge and hydrodynamic size distribution of the nanocomposites.

2.7. Preparation and Characterization of F-MHC-Loaded Peripheral Nerve-Targeted ROS-responsive liposomes (F-MHC@PNRLs)

F-MHC@PNRLs were prepared by mixing the pre-synthesized peripheral nerve-targeted ROS-responsive liposomes (PNRLs) with F-MHC at a 1:1 mass ratio, followed by sonication until a homogeneous mixture was obtained. The resulting mixture was then sequentially extruded through 800-mesh, 400-mesh, and 200-mesh filters to obtain the final F-MHC@PNRLs product. In parallel, non-targeted F-MHC-loaded ROS-responsive liposomes (F-MHC@RLs) were prepared using the same procedure, except that non-targeted ROS-responsive liposomes were used instead of PNRLs. The resulting liposomal formulations were characterized for their particle size, morphology, drug loading efficiency, and ROS-responsive release behavior using dynamic light scattering, transmission electron microscopy, and in vitro release studies.

2.8. Cell culture

RAW264.7 and RSC96 cells were purchased from Shanghai Chenying Biotechnology Co., Ltd. (Shanghai, China). The cells were cultured in high-glucose Dulbecco's Modified Eagle Medium (DMEM; Gibco, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 1% penicillin/streptomycin (P/S; Gibco, USA) at 37 °C in a humidified incubator with 5% CO₂. The medium was refreshed every 1–2 days, and cells were passaged at 70–80% confluence.

2.9. Evaluation of cytotoxicity assays in vitro

Cell viability was evaluated using a CCK-8 assay and Calcein-AM/PI live/dead staining. Briefly, RAW264.7 and RSC96 cells were seeded into 96-well plates at a density of 5 × 10³ cells/well and allowed to adhere overnight before subsequent treatments. To establish an RSL3-induced ferroptosis model in RSC96 cells, the cells were treated with different concentrations of RSL3 (1, 5, 10, 25, and 50 μM) for 24 h, and cell viability was determined by CCK-8 assay. To evaluate the cytotoxicity of MHC, RAW264.7 and RSC96 cells were incubated with different concentrations of MHC (10, 25, 50, 100, and 200 μg/mL) for 24 h. To assess the cytotoxicity of Fer-1, RAW264.7 and RSC96 cells were exposed to different concentrations of Fer-1 (1, 2.5, 5, 10, and 25 μg) for 24 h. To determine the rescue effect of Fer-1 on RSL3-induced ferroptosis, RSC96 cells were co-treated with RSL3 (10 μM) and different concentrations of Fer-1 (1, 2, 5, and 10 μg) for 24 h. After treatment, the medium was replaced with fresh medium containing 10% CCK-8 reagent and incubated at 37 °C for 1–2 h. The absorbance at 450 nm was measured using a microplate reader. In addition, cell viability and cytotoxicity in different treatment groups were further assessed using a Calcein-AM/PI Cell Viability/Cytotoxicity Assay Kit according to the manufacturer's instructions. Fluorescence images were captured under a fluorescence microscope.

2.10. In vitro evaluation of antioxidant and anti-inflammatory effects

Murine macrophage-like RAW 264.7 cells were used as an in vitro model to investigate LPS-induced inflammatory responses associated with FNI. RAW 264.7 cells were seeded in 6-well plates at a density of 5 × 10⁵ cells per well and incubated overnight at 37 °C in a humidified atmosphere containing 5% CO₂ to allow cell attachment. According to the experimental grouping described above, cells were divided into the following groups: G1, blank control; G2, LPS-only treatment; and G3–G6, LPS combined with different treatments. Except for the blank control group, cells were stimulated with LPS at a final concentration of 1 μg/mL and co-cultured with the corresponding formulations for 24 h. Total RNA was then extracted from the collected cells using TRIzol reagent. After RNA purity and concentration were determined using a nucleic acid/protein analyzer, cDNA was synthesized using a reverse transcription kit. RT-qPCR amplification was performed with specific primers targeting pro-inflammatory genes (Tnf, Il1b, Il6, Nos2) and anti-inflammatory genes (Arg1, Il10), with Actb as the internal reference gene. The relative expression levels of each gene were calculated using the 2^−ΔΔCt method. The treated macrophages were harvested, washed twice with PBS, fixed with 4% paraformaldehyde for 30 min, and permeabilized with 0.1% Triton X-100 for 15 min. FITC-conjugated anti-CD86 antibody (M1 macrophage marker) and APC-conjugated anti-CD206 antibody (M2 macrophage marker) were added separately, followed by incubation at 4 °C in the dark for 30 min. After washing, the cells were resuspended in PBS, and the proportions of CD86-positive and CD206-positive cells were detected using a BD LSRFortessa flow cytometer (BD Biosciences, San Jose, CA, USA). Data were analyzed with FlowJo software. The cell culture supernatants of each group were collected, and the protein concentrations of TNF-α and IL-1β in the supernatants were measured using ELISA kits following the manufacturer's instructions. The absorbance at the corresponding wavelength was measured with a microplate reader, and the cytokine content was calculated based on the standard curve. The protein expression levels of iNOS and CD206 were detected by western blot. Briefly, treated RAW264.7 cells were lysed with RIPA buffer, and total protein was quantified using a BCA assay. Equal amounts of protein were separated by SDS-PAGE, transferred to PVDF membranes, blocked with 5% non-fat milk, and incubated with primary antibodies against iNOS, CD206, and β-actin overnight at 4 °C. After incubation with HRP-conjugated secondary antibodies, protein bands were detected using ECL reagent on Odyssey imaging system (Tanon, Shanghai, China) and quantified with ImageJ. β-actin was used as the internal control.

2.11. In vitro evaluation of antioxidant and anti-inflammatory mechanisms

Different RAW 264.7 cell treatment groups, including G1, G2 and G6 (5 × 10⁶ cells per tube), were collected, and RNA-seq was commissioned to Shanghai Paisenuo Biotechnology Co., Ltd., including mRNA enrichment, library construction, and high-throughput sequencing. Raw sequencing data were filtered by removing adapters and low-quality sequences to obtain high-quality clean data. Data are presented in Supplementary file 1. Subsequent bioinformatics analysis was performed using R software (version 4.3.3). Comparative transcriptomic analysis between G2 and G1 was conducted to characterize transcriptomic alterations under LPS-induced inflammatory conditions, while comparisons between G6 and G2 were performed to reveal transcriptomic changes modulated by F-MHC@PNRLs intervention. The analytical pipeline included identification of differentially expressed genes (DEGs) with thresholds of FDR <0.05 and |log₂FC| > 0.5, Gene Ontology (GO) functional enrichment analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) signaling pathway enrichment analysis, and gene set enrichment analysis (GSEA) to identify significantly enriched biological processes and pathways at the gene-set level. Furthermore, multiple visualization strategies were applied, including Venn diagrams of DEGs, GO/KEGG enrichment bubble plots, GSEA enrichment plots, and gene expression heatmaps to illustrate transcriptomic differences among groups.

Based on the key pathways (NF-κB, MAPK, etc.) identified by RNA-seq, core proteins and their phosphorylated forms were selected for Western blot validation. The treated cells were lysed with RIPA lysis buffer containing protease and phosphatase inhibitors on ice for 30 min, followed by centrifugation at 12,000 rpm for 15 min at 4 °C to obtain the total protein supernatant. Protein concentration was determined by the BCA method, and equal amounts of protein (30 μg per well) were loaded onto a 10% SDS-PAGE gel. After electrophoresis at 80 V for 30 min and 120 V for 90 min, the proteins were wet-transferred to a PVDF membrane (activated with methanol). The membrane was blocked with 4% skim milk at room temperature for 1.5 h, then incubated with primary antibodies (IκBα, p-IκBα, P65, p-P65, P38, p-P38, ERK, p-ERK, JNK, p-JNK, and β-actin, all diluted according to the instructions) overnight at 4 °C. After washing three times with TBST, the membrane was incubated with the corresponding secondary antibody at room temperature for 1 h, washed again, and the fluorescent signals were detected using an Odyssey imaging system (Tanon, Shanghai, China). The gray values of the bands were analyzed with ImageJ software, and the relative protein expression levels were calculated with β-actin as the internal reference.

2.12. Evaluation of anti-ferroptotic capacity in vitro

To balance model robustness and mechanistic interpretability, different RSL3 concentrations were used for different experimental purposes. For cytotoxicity-related assays, 25 μM RSL3 was selected because it induced obvious RSC96 cell injury, thereby providing a suitable window to evaluate the cytoprotective effects of F-MHC@PNRLs. For mechanistic assays, 10 μM RSL3 was selected to avoid excessive cell death and secondary nonspecific damage, which may obscure the detection of ferroptosis-related molecular events. This lower concentration allowed more reliable assessment of GPX4 expression, lipid peroxidation, mitochondrial membrane potential, and related signaling changes.

To evaluate the anti-ferroptotic effects of different formulations, RSC96 cells were divided into six groups: control group (G1), RSL3 group (G2), RSL3 + MHC group (G3), RSL3 + Fer-1 group (G4), RSL3 + F-MHC group (G5), and RSL3 + F-MHC@PNRLs group (G6). Except for the control group, cells were stimulated with RSL3 to establish an in vitro ferroptosis-related injury model. Specifically, 25 μM RSL3 was used for cytotoxicity-related assays, including cell viability and live/dead staining, whereas 10 μM RSL3 was used for mechanistic assays, including ROS detection, ferrous iron measurement, lipid peroxidation analysis, mitochondrial membrane potential assessment, antioxidant capacity evaluation, and western blot analysis. Cells were treated with the corresponding formulations according to the experimental design. Intracellular reactive oxygen species (ROS) levels were detected using a DCFH-DA fluorescent probe according to the manufacturer's instructions. Fluorescence images were captured under a fluorescence microscope, and relative fluorescence intensity was quantified using ImageJ software. Intracellular ferrous iron (Fe²⁺) accumulation was detected using the Cellular Ferrous Iron Red Fluorescence Detection Kit based on RhoNox-6. After staining, fluorescence images were obtained under a fluorescence microscope, and the mean fluorescence intensity was quantified. Lipid peroxidation was evaluated using a lipid peroxidation fluorescent probe. If C11-BODIPY 581/591 was used, the reduced and oxidized fluorescence signals were observed under a fluorescence microscope, and relative fluorescence intensities were analyzed to assess lipid ROS accumulation. The intracellular antioxidant capacity and lipid peroxidation status were further assessed by measuring superoxide dismutase (SOD), catalase (CAT), malondialdehyde (MDA), glutathione (GSH), and oxidized glutathione (GSSG) levels using corresponding commercial assay kits according to the manufacturers' protocols. To investigate ferroptosis-related antioxidant and anti-lipid peroxidation markers, the protein levels of NRF2, HO-1, nuclear NRF2 (n-NRF2), GPX4, and SLC7A11 were determined by western blot analysis. β-actin was used as the internal control for total protein, while Lamin B1 was used as the nuclear protein loading control. Protein band intensities were quantified by densitometric analysis.

2.13. Evaluation of mitochondrial protective effects in vitro

To evaluate the mitochondrial protective effects of different formulations, RSC96 cells were divided into six groups: control group (G1), RSL3 group (G2), RSL3 + MHC group (G3), RSL3 + Fer-1 group (G4), RSL3 + F-MHC group (G5), and RSL3 + F-MHC@PNRLs group (G6). Except for the CCK-8 and live/dead staining assays, in which 25 μM RSL3 was used to evaluate cytotoxic injury, all other mechanistic experiments were performed using 10 μM RSL3 to establish an in vitro ferroptosis model. After the indicated treatments, mitochondrial function and ultrastructural integrity were assessed by MitoSOX staining, JC-1 staining, and transmission electron microscopy (TEM). Mitochondrial reactive oxygen species (mtROS) levels were detected using MitoSOX Red Mitochondrial Superoxide Indicator according to the manufacturer's instructions. After staining, fluorescence images were captured under a fluorescence microscope, and the mean fluorescence intensity was quantified to evaluate mitochondrial oxidative stress. Mitochondrial membrane potential (MMP) was assessed using a JC-1 Mitochondrial Membrane Potential Assay Kit. After treatment, cells were incubated with JC-1 working solution according to the manufacturer's protocol. Fluorescence images of JC-1 monomers (green) and JC-1 aggregates (red/orange) were obtained under a fluorescence microscope, and the ratio of aggregate-to-monomer fluorescence intensity was calculated to evaluate changes in mitochondrial membrane potential. For ultrastructural observation, treated cells were collected and fixed with 2.5% glutaraldehyde at 4 °C overnight, followed by postfixation with 1% osmium tetroxide. The samples were then dehydrated through a graded ethanol series, embedded in resin, sectioned into ultrathin slices, stained with uranyl acetate and lead citrate, and observed under a transmission electron microscope. Mitochondrial morphology and ultrastructural damage were evaluated among different groups.

2.14. In vivo fluorescence imaging of NP41-mediated peripheral nerve-associated retention

To directly evaluate the effect of NP41 modification on the local retention and peripheral nerve-associated accumulation of the liposomal formulation after FNI, near-infrared fluorescence imaging was performed using Cy5.5-labeled RLs and PNRLs. Briefly, Cy5.5 was incorporated into non-targeted ROS-responsive liposomes (RLs) and NP41-modified peripheral nerve-targeted ROS-responsive liposomes (PNRLs) during liposome preparation to obtain Cy5.5@RLs and Cy5.5@PNRLs, respectively. Free Cy5.5 was removed by repeated washing and ultrafiltration/centrifugation until no obvious free fluorescence signal was detected in the filtrate. The obtained fluorescently labeled liposomes were resuspended in sterile PBS for subsequent in vivo imaging.

After establishment of the facial nerve crush injury model, mice were randomly divided into two groups and locally administered 20 μL of Cy5.5@RLs or Cy5.5@PNRLs around the injured facial nerve, with three mice in each group. The two formulations were administered at the same liposome-equivalent dose to ensure comparability between the non-targeted and NP41-modified groups.In vivo fluorescence imaging was performed at 3, 6, 12, 24, and 48 h after local administration using an IVIS imaging system. Briefly, mice were anesthetized and placed in the imaging chamber, and fluorescence signals at the injured facial nerve region were acquired under identical exposure settings for all groups. Quantitative analysis of fluorescence intensity was performed using Living Image software.

2.15. Evaluation of therapeutic efficacy in the FNI model

For preliminary dose exploration, mice with FNI were randomly divided into five groups and locally administered F-MHC@PNRLs at F-MHC-equivalent doses of 0, 0.5, 1.0, 2.5, and 5.0 mg/kg in a fixed volume of 20 μL per mouse. Immediately after facial nerve crush injury, local perineural administration was performed under a surgical microscope. Briefly, after the injured facial nerve segment was exposed, 20 μL of the corresponding formulation was slowly applied around the crush site using a calibrated microsyringe. Care was taken to cover the injured nerve segment and surrounding perineural tissue evenly, while avoiding direct intraneural injection or mechanical compression of the nerve. After administration, the formulation was allowed to remain in contact with the injured facial nerve region for approximately 2–3 min to reduce fluid leakage and promote local retention. The surgical field was then gently checked to ensure that no obvious leakage occurred. The muscle and skin layers were closed separately using absorbable sutures or wound clips. Mice were placed on a heating pad until recovery from anesthesia and then returned to their cages. All local administrations were performed by the same operator under sterile conditions to minimize procedural variability.

For mice receiving F-MHC@RLs or F-MHC@PNRLs, the formulations were prepared in sterile PBS at a concentration corresponding to an F-MHC-equivalent dose of 1 mg/kg in a final volume of 20 μL per mouse. For a mouse weighing approximately 20 g, this dose corresponded to approximately 20 μg F-MHC, including 18.44 μg MHC and 1.56 μg Fer-1. The FNI + PBS group received 20 μL of sterile PBS using the same procedure.

To evaluate the therapeutic efficacy of F-MHC@PNRLs and the contribution of NP41-mediated peripheral nerve-targeting modification, mice were randomly assigned to four groups: Sham group (G1), FNI + PBS group (G2), FNI + F-MHC@RLs group (G3), and FNI + F-MHC@PNRLs group (G4). A total of 18 mice were included in each group. Among them, six mice per group were used for continuous facial nerve functional assessment throughout the 14-day observation period, and tissues from these mice were harvested on day 14 for LFB and H&E staining. Another six mice per group were sacrificed on day 3 for ROS detection and immunofluorescence staining, while the remaining six mice per group were sacrificed on day 7 for TEM observation and immunofluorescence staining.

The facial nerve crush injury model was established as described above. Immediately after modeling, 20 μL of the corresponding formulation was locally administered around the injured facial nerve. The FNI + PBS group received the same volume of sterile PBS, and the sham-operated group underwent facial nerve exposure without crush injury. F-MHC@RLs were used as the non-targeted ROS-responsive liposomal control, whereas F-MHC@PNRLs were modified with NP41 to determine whether peripheral nerve-targeting modification could enhance local therapeutic efficacy after FNI. F-MHC@RLs and F-MHC@PNRLs were administered at the same F-MHC-equivalent dose of 1 mg/kg. For a mouse weighing approximately 20 g, this dose corresponded to approximately 20 μg F-MHC per mouse, including 18.44 μg MHC and 1.56 μg Fer-1. Therefore, the two formulations were administered at equivalent F-MHC-, MHC-, and Fer-1-equivalent doses, ensuring that the comparison between F-MHC@RLs and F-MHC@PNRLs reflected the contribution of NP41-mediated peripheral nerve-targeting modification rather than differences in nanozyme or Fer-1 dosage.

Facial nerve functional recovery was evaluated every 2 days after surgery until day 14 by investigators blinded to the experimental groups. Functional assessment included whisker movement, vibrissae position, eyelid closure, and nasal deviation, with higher scores indicating better recovery of facial nerve function. Electrophysiological assessment was performed on day 14 after FNI to objectively evaluate facial nerve conduction recovery. Briefly, mice were anesthetized, and the facial nerve was electrically stimulated at the proximal side of the injury site. Compound muscle action potentials were recorded from the ipsilateral orbicularis oculi muscle. The amplitude and latency of compound muscle action potentials were analyzed to assess facial nerve conduction recovery. Higher amplitudes and shorter latencies were considered to indicate better electrophysiological recovery. For mice scheduled for tissue collection, electrophysiological recording was performed before euthanasia and sample harvesting.

Facial nerve tissues were harvested at predefined time points for histological, ultrastructural, and immunofluorescence analyses. On day 3, ROS accumulation was evaluated by DCF staining, and inflammatory polarization was assessed by immunofluorescence staining of iNOS and CD206. Schwann cell-associated remyelination was examined by MBP and S100β staining. On day 7, TEM was performed to evaluate myelin ultrastructure and axonal recovery, together with corresponding immunofluorescence staining. On day 14, LFB was performed using tissues harvested from the functional assessment cohort to evaluate myelin integrity and general histopathological recovery. Quantitative analyses were performed using ImageJ software by investigators blinded to the treatment groups.

2.16. Assessment of functional recovery

Facial nerve crush injury can lead to impaired blink reflex, abnormal whisker movement, and nasal deviation. Starting from the first day after surgery, facial nerve functional recovery was evaluated three times per week by investigators blinded to the experimental groups. The blink reflex was assessed by gently blowing air toward the eyeball using a 5-mL syringe at a distance of approximately 3 cm. Referring to previously reported scoring methods [5], eyelid closure and whisker movement were each scored as follows: 0 = no movement; 1 = slight movement; 2 = obvious active movement; and 3 = symmetrical movement comparable to the healthy side. Nasal deviation was scored as follows: 0 = obvious nasal deviation toward the healthy side; and 1 = nasal tip located near the midline. The total facial nerve functional score was calculated as the sum of eyelid closure, whisker movement, and nasal deviation scores, with a maximum possible score of 7. Higher total scores indicated better recovery of facial nerve function.

2.17. In vivo toxic assessment

Fourteen days after administration, mice were euthanized, and major organs including heart, liver, spleen, lung and kidney were collected and stained with H&E to observe histomorphological changes and evaluate the systemic biological safety of different groups. Meanwhile, peripheral blood was collected for routine blood and biochemical analysis, including hemoglobin (HGB), red blood cells (RBC), platelets (PLT), white blood cells (WBC), monocytes (MON), neutrophils (NEUT), aspartate aminotransferase (AST), alanine aminotransferase (ALT), urea (UREA) and blood urea nitrogen (BUN).

2.18. Histopathological evaluation, immunofluorescence staining, and ultrastructural analysis

At the predefined time points, mice were euthanized by carbon dioxide asphyxiation, and the injured facial nerve segments were carefully dissected. For histological analysis, facial nerve tissues were fixed in 4% paraformaldehyde, dehydrated through a graded ethanol series, cleared in xylene, embedded in paraffin, and sectioned into 5-μm-thick slices. Hematoxylin and eosin (H&E) staining was performed to evaluate general histopathological changes in the injured facial nerve tissues. To assess ROS accumulation in vivo, freshly harvested facial nerve tissues were embedded in optimal cutting temperature compound and prepared as frozen sections. The sections were incubated with DCFH-DA according to the manufacturer's instructions to detect intracellular ROS levels. Fluorescence images were captured using a fluorescence microscope or laser confocal scanning microscope under identical imaging parameters.

For immunofluorescence staining, tissue sections were blocked with blocking buffer and incubated with primary antibodies against iNOS and CD206 to evaluate macrophage inflammatory polarization. iNOS was used as an M1-associated inflammatory marker, whereas CD206 was used as an M2-associated reparative marker. To evaluate Schwann cell-associated myelin repair, sections were incubated with primary antibodies against myelin basic protein (MBP, 1:500) and S100β (1:1000). MBP was used to assess myelin integrity, and S100β was used to identify Schwann cell-associated areas. After incubation with appropriate fluorescent secondary antibodies, nuclei were counterstained with DAPI. The sections were then observed and photographed using a laser confocal scanning microscope.

Luxol fast blue (LFB) staining was performed to further evaluate myelin integrity in facial nerve tissues. To observe the ultrastructure of myelinated nerve fibers, facial nerve samples were processed for transmission electron microscopy (TEM). Briefly, tissue samples were fixed, dehydrated, embedded, ultrathin-sectioned, and examined using a transmission electron microscope. Myelin sheath morphology and myelinated fiber ultrastructure were evaluated.

Quantitative analyses of fluorescence intensity, positive staining area, and abnormal myelinated fibers were performed using ImageJ software by investigators blinded to the experimental groups. Together, histological staining, immunofluorescence analysis, and TEM observation were used to evaluate the effects of different treatments on oxidative stress, inflammatory polarization, Schwann cell-associated myelin repair, and facial nerve tissue recovery.

2.19. Statistical analyses

Statistical analyses were performed using GraphPad Prism 9.0 and R software. Data are presented as mean ± standard deviation (SD). Comparisons between two groups were performed using Student's t-test. Comparisons among multiple groups were performed using one-way ANOVA followed by Tukey's post hoc test. For repeated behavioral measurements over time, two-way repeated-measures ANOVA followed by appropriate post hoc multiple-comparison tests was used. For transcriptome sequencing data, multiple hypothesis correction was performed using the Benjamini–Hochberg method. A value of p < 0.05 was considered statistically significant.

3. Results and discussion

3.1. FNI is associated with ferroptosis-related transcriptional changes and lipid perox

Ferroptosis has increasingly been recognized as an important form of regulated cell death involved in secondary damage after peripheral nerve injury, including FNI [6,12]. Following nerve injury, excessive reactive oxygen species (ROS) accumulation, inflammatory activation, iron metabolic disturbance, and lipid peroxidation may collectively impair Schwann cell function, aggravate myelin disruption, and hinder axonal regeneration [26]. Although ferroptosis has been implicated in several models of peripheral nerve injury, its involvement in FNI, particularly during the acute phase, remains insufficiently characterized. Therefore, to investigate whether ferroptosis-related molecular changes and lipid peroxidation-associated injury are involved shortly after FNI, facial nerve tissues were collected at day 1 after injury and subjected to RNA-seq analysis. The overall workflow for tissue collection, RNA extraction, sequencing, and bioinformatic analysis is shown in Fig. 1A. After normalization, the overall distribution of gene expression was comparable among samples, indicating acceptable data quality and inter-sample consistency (Fig. 1B). Principal component analysis (PCA) revealed a clear separation between the Normal and FNI groups, suggesting that FNI induced marked transcriptomic remodeling as early as day 1 after injury (Fig. 1C). Differential expression analysis further identified 179 upregulated and 159 downregulated genes in the FNI group compared with the Normal group, confirming substantial injury-induced transcriptional alterations during the acute phase of FNI (Fig. 1D).

Fig. 1.

Facial nerve injury (FNI) induces ferroptosis-related transcriptional remodeling and lipid peroxidation-associated Schwann cell damage. (A) Schematic illustration of the RNA-seq workflow using facial nerve tissues from sham and FNI mice at 1 day after injury (n = 3). (B) Boxplot showing comparable expression distribution among sequenced samples after normalization. (C) Principal component analysis (PCA) showing distinct transcriptional profiles between the two groups. (D) Volcano plot of differentially expressed genes (DEGs). Red indicates significantly upregulated genes, blue indicates significantly downregulated genes, and gray indicates genes without significant differential expression. (E) Gene Set Variation Analysis (GSVA) analysis showing a marked increase in the ferroptosis potential index (FPI) in FNI tissues. (F) Gene set enrichment analysis (GSEA) plot for the ferroptosis pathway. (G) Bubble heat map of ferroptosis-related genes. (H, I) Representative immunofluorescence staining of GPX4/S100β and 4-HNE/S100β in facial nerve sections from Sham mice and FNI mice at 1 day after injury. DAPI was used for nuclear counterstaining. Scale bar = 100 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

To further determine whether ferroptosis-related transcriptional programs were altered after FNI, Gene Set Variation Analysis (GSVA) was performed to calculate the ferroptosis potential index (FPI). A higher FPI indicates a stronger ferroptosis tendency, whereas a lower FPI suggests reduced ferroptosis potential. As shown in Fig. 1E, the FNI group exhibited a significantly higher FPI score than the Sham group, indicating enhanced ferroptosis-related transcriptional activity in injured facial nerve tissues. Consistently, Gene Set Enrichment Analysis (GSEA) showed a positive enrichment trend of the ferroptosis-related gene set in the FNI group (NES = 1.05, FDR = 0.24; Fig. 1F). These findings suggest that ferroptosis-related molecular changes may be involved in the early pathological response to facial nerve injury. Although the GSEA result did not reach statistical significance, the increased FPI score and positive enrichment trend collectively indicate that ferroptosis-associated genes are transcriptionally activated after FNI. This observation is biologically meaningful because Schwann cells are highly susceptible to oxidative stress and lipid peroxidation after peripheral nerve injury, and their survival and functional phenotype are essential for axonal guidance, remyelination, and neurological recovery. In line with our findings, It is reported that c-Jun overexpression inhibited erastin-induced ferroptosis in Schwann cells and promoted facial nerve functional recovery, highlighting the close association between ferroptosis suppression, Schwann cell protection, and facial nerve repair [6]. Therefore, the observed ferroptosis-related transcriptional changes and lipid peroxidation may be associated with Schwann cell dysfunction, impaired myelin repair, and delayed functional recovery after injury. These results provide a transcriptomic basis for targeting ferroptosis as a potential therapeutic strategy in facial nerve regeneration. The expression patterns of representative ferroptosis-related genes were further visualized using a bubble heatmap. Several genes involved in iron metabolism, glutathione metabolism, antioxidant defense, and lipid peroxidation regulation, including Ncoa4, Gpx4, Slc3a2, Fth1, Alox15, Gclm, Acsl4, Slc11a2, and Tfrc, exhibited differential expression patterns between Normal and FNI tissues (Fig. 1G). Among these genes, Gpx4 is of particular importance because it encodes glutathione peroxidase 4, a central anti-ferroptotic enzyme that converts phospholipid hydroperoxides into non-toxic lipid alcohols in a glutathione-dependent manner. By limiting the accumulation of lipid peroxides, GPX4 maintains membrane redox homeostasis and prevents ferroptotic cell death. Therefore, alterations in Gpx4 expression may directly reflect changes in the intrinsic anti-ferroptotic defense capacity of injured facial nerve tissues [27,28]. To validate the transcriptomic findings at the tissue level, immunofluorescence staining was performed to examine GPX4, 4-HNE, and S100β expression in facial nerve tissues after FNI. Compared with the Sham group, GPX4 fluorescence was markedly reduced in the FNI group, indicating that the antioxidant defense system against lipid peroxidation was weakened after injury (Fig. 1H). Meanwhile, S100β staining became disorganized and structurally disrupted in the injured nerve, suggesting Schwann cell disturbance and impaired nerve architecture. The merged images further showed that the reduction of GPX4 occurred within S100β-positive nerve regions, supporting a close association between ferroptosis-related antioxidant impairment and Schwann cell-associated pathological changes after FNI. In addition, 4-HNE staining was introduced to evaluate lipid peroxidation, a key biochemical feature of ferroptosis. As shown in Figs. 1I and 4-HNE fluorescence was weak in the Sham group but markedly increased after FNI. This increase was accompanied by disrupted S100β-positive structures, indicating that lipid peroxidation accumulated within injured facial nerve tissues and was spatially associated with Schwann cell-related damage. Since 4-HNE is a major reactive aldehyde generated during lipid peroxidation, its increased expression further supports the occurrence of ferroptosis-related oxidative lipid damage after FNI.

Taken together, these transcriptomic and histological findings suggest that FNI is accompanied by early ferroptosis-related molecular changes during the acute phase. The increased FPI score, positive enrichment trend of the ferroptosis pathway, altered expression of ferroptosis-associated genes, decreased GPX4 expression, and enhanced 4-HNE accumulation collectively indicate that impaired anti-ferroptotic defense and aggravated lipid peroxidation may contribute to Schwann cell dysfunction and early deterioration of the nerve microenvironment after FNI. These results provide a rationale for targeting ferroptosis-related lipid peroxidation as a potential therapeutic strategy for promoting facial nerve repair. Therefore, a therapeutic approach capable of simultaneously regulating oxidative stress, suppressing lipid peroxidation, and improving local delivery to injured peripheral nerves may be beneficial for facial nerve regeneration.

3.2. Preparation and characterization of F-MHC@PNRLs

To construct a multifunctional nanoplatform integrating ROS scavenging, ferroptosis inhibition, ROS-responsive release, and peripheral nerve targeting, F-MHC@PNRLs were prepared through a stepwise strategy. Based on the above findings that ferroptosis-related lipid peroxidation and impaired antioxidant defense were involved in FNI progression, we first designed an anti-ferroptotic nanozyme system by loading Fer-1 onto Mn-doped CeO₂ nanoparticles. As illustrated in Fig. 2A, CeO₂ nanoparticles were synthesized by a hydrothermal method using Ce(NO₃)₃, polyvinylpyrrolidone (PVP), and ethylene glycol as precursors. Subsequently, Mn doping was introduced by MnCl₂ treatment to obtain Mn-doped CeO₂ nanoparticles, defined as MHC. The introduction of Mn was expected to modulate the redox activity of ceria-based nanozymes and improve their catalytic ROS-scavenging capacity [29]. Fer-1, a classical ferroptosis inhibitor, was then loaded onto MHC to form F-MHC. To quantitatively determine Fer-1 loading and release, the UV–vis absorbance of Fer-1 was first characterized. Fer-1 showed a characteristic absorption peak at approximately 320 nm, and a standard calibration curve with good linearity was established at this wavelength (R² = 0.9991; Fig. S1). When 50 mg of MHC and 5 mg of Fer-1 were used for the loading process, the amount of Fer-1 successfully loaded onto MHC was determined to be 3.4 mg. Accordingly, the encapsulation efficiency and drug loading capacity of F-MHC were calculated to be 68.0% and 6.36%, respectively, indicating the efficient incorporation of Fer-1 into the Mn-doped CeO₂ nanozyme system. This design aimed to combine the catalytic antioxidant activity of Mn-doped CeO₂ nanozymes with the lipid peroxidation-blocking effect of Fer-1, thereby establishing a complementary anti-ferroptotic system. Dynamic light scattering showed that the average particle sizes of CeO₂, MHC, and F-MHC were 142.08 nm, 164.13 nm, and 170.72 nm, respectively, indicating a gradual increase in particle size after Mn doping and Fer-1 loading (Fig. 2B). Meanwhile, the zeta potential became more negative after modification, suggesting that Mn doping and Fer-1 loading altered the surface charge of the nanoparticles and may contribute to improved colloidal stability (Fig. 2C). TEM observation showed that F-MHC exhibited a nanoscale spherical morphology with a rough and porous structure, which may provide abundant surface area and internal space for Fer-1 loading (Fig. 2D). STEM-EDS elemental mapping further revealed the homogeneous distribution of characteristic elements within F-MHC, including Ce, Mn, O, C, and N-related signals, supporting the successful formation of Mn-doped CeO₂ nanoparticles and the incorporation of Fer-1-associated components (Fig. 2E). In addition, XPS analysis was performed to further verify the elemental composition and chemical states of F-MHC. The survey spectrum confirmed the presence of Ce, O, Mn, C, and N elements, while the high-resolution Mn 2p and Ce 3d spectra verified the coexistence of Mn and Ce species in F-MHC (Fig. S2). Notably, no obvious P2p signal was observed, consistent with the absence of phospholipid components before liposomal encapsulation. These results demonstrated that F-MHC was successfully prepared as a Fer-1-loaded Mn-doped CeO₂ nanozyme.

Fig. 2.

Preparation and Characterization of F-MHC-Loaded Peripheral Nerve-Targeted ROS-Responsive Liposomes (F-MHC@PNRLs). (A) Schematic illustration of the synthesis of F-MHC. CeO₂ nanoparticles were prepared by hydrothermal synthesis, followed by Mn doping to obtain Mn-doped CeO₂ nanozymes (MHC) and subsequent Ferrostatin-1 (Fer-1) loading to form F-MHC. (B) Hydrodynamic size distributions of CeO₂, MHC, and F-MHC measured by dynamic light scattering (DLS). (C) Zeta potentials of CeO₂, MHC, and F-MHC. (D) Representative transmission electron microscopy (TEM) image of F-MHC. (E) Scanning transmission electron microscopy (STEM) image and corresponding elemental mapping of F-MHC. (F) Schematic illustration of DSPE-TK-PEG-NP41 synthesis and the construction of F-MHC@PNRLs. DSPE-TK-PEG-COOH was conjugated with NP41 via EDC-mediated coupling, followed by liposome assembly with lecithin and cholesterol and subsequent F-MHC loading. (G) Fourier-transform infrared spectroscopy (FTIR) spectra of DSPE-TK-PEG-COOH, NP41, and DSPE-TK-PEG-NP41. (H, I) Representative TEM and scanning electron microscopy (SEM) images of F-MHC@PNRLs. (J) STEM image and corresponding elemental mapping of F-MHC@PNRLs. (K) XPS survey spectrum of F-MHC@PNRLs. (L–Q) High-resolution XPS spectra of C 1s (L), N 1s (M), O 1s (N), Mn 2p (O), Ce 3d (P), and P 2p (Q). (R) ROS-responsive Fer-1 release profiles from F-MHC@PNRLs under different H₂O₂ concentrations.

Next, to endow F-MHC with peripheral nerve-targeting and ROS-responsive delivery properties, F-MHC was further incorporated into NP41-modified ROS-responsive liposomes. As shown in Fig. 2F, DSPE-TK-PEG-COOH was first conjugated with the peripheral nerve-targeting peptide NP41 through EDC-mediated coupling to obtain DSPE-TK-PEG-NP41. The thioketal (TK) linker served as a ROS-cleavable unit, enabling the liposomal structure to respond to the oxidative microenvironment after nerve injury. FTIR analysis was performed to verify the successful synthesis of DSPE-TK-PEG-NP41. Compared with DSPE-TK-PEG-COOH and NP41, DSPE-TK-PEG-NP41 showed a characteristic absorption peak around 1100 cm⁻¹, corresponding to the C–O–C stretching vibration of PEG, and an absorption band around 1650 cm⁻¹, corresponding to the amide I band, indicating successful coupling between DSPE-TK-PEG-COOH and NP41 (Fig. 2G). Subsequently, DSPE-TK-PEG-NP41, phospholipid components, and cholesterol were assembled into peripheral nerve-targeted ROS-responsive liposomes (PNRLs) through hydration and ultrasonication. Before F-MHC loading, blank PNRLs were first characterized to verify the successful formation of the peripheral nerve-targeted ROS-responsive liposomal carrier. Dynamic light scattering showed that PNRLs exhibited an average hydrodynamic diameter of 193.5 nm with a low PDI of 0.158, indicating a relatively uniform size distribution (Fig. S3A). The zeta potential of PNRLs was −3.1 mV, suggesting a nearly neutral surface charge, which may be attributed to PEGylation and NP41 modification. TEM observation further confirmed that PNRLs displayed a spherical vesicle-like morphology with good dispersibility (Fig. S3B). These results demonstrated the successful construction of blank PNRLs, providing a stable liposomal carrier for subsequent F-MHC loading. F-MHC was then incorporated into PNRLs to obtain F-MHC@PNRLs (Fig. 2F). TEM images showed that F-MHC@PNRLs maintained a relatively uniform nanospherical morphology. Compared with free F-MHC, the liposomal formulation displayed an integrated nanoparticle structure, suggesting successful incorporation of F-MHC into the liposomal system (Fig. 2H). SEM observation further showed that F-MHC@PNRLs were well-dispersed spherical nanoparticles with relatively uniform morphology (Fig. 2I). STEM-EDS elemental mapping of F-MHC@PNRLs showed the co-localized distribution of elements derived from both the F-MHC core and the liposomal shell, including Ce, Mn, O, C, N, and P, further confirming the successful integration of F-MHC into PNRLs without obvious structural collapse or aggregation (Fig. 2J). DLS analysis further showed that F-MHC@PNRLs exhibited an average hydrodynamic diameter of 184.9 nm with a PDI of 0.213, indicating a relatively uniform nanoscale distribution (Fig. S4A). The zeta potential of F-MHC@PNRLs was −32.3 mV, which was markedly more negative than that of blank PNRLs and close to that of F-MHC (Fig. S4B). This shift may be attributed to the incorporation of negatively charged F-MHC and its possible association with the liposomal membrane during hydration-assisted loading. Together with TEM and elemental mapping results, these findings support the successful construction of F-MHC-loaded PNRLs. Furthermore, XRD analysis was performed to evaluate whether liposomal incorporation affected the crystalline structure of F-MHC. The characteristic diffraction peaks of F-MHC were retained in F-MHC@PNRLs, indicating that the Mn-doped CeO₂ crystalline structure was preserved after incorporation into PNRLs (Fig. S5). These results further suggested that the liposomal loading process did not cause obvious crystal-phase disruption of the F-MHC nanozyme. X-ray photoelectron spectroscopy (XPS) was then used to analyze the elemental composition and chemical states of F-MHC@PNRLs. The full survey spectrum showed characteristic signals of C, N, O, Mn, Ce, and P, which corresponded to the organic liposomal components, NP41 peptide, Mn-doped CeO₂ core, and phospholipid structure (Fig. 2K). High-resolution C 1s, N 1s, and O 1s spectra further confirmed the presence of organic functional groups, including C–C/C=C, C–N/C–O, amide-related nitrogen, and oxygen-containing bonds, supporting the successful introduction of DSPE-TK-PEG-NP41, lipid components, and Fer-1 into the nanoplatform (Fig. 2L–N). In addition, the Mn 2p and Ce 3d spectra verified the coexistence of Mn and Ce species within F-MHC@PNRLs, confirming the successful incorporation of Mn-doped CeO₂ nanozymes (Fig. 2O and P). The P 2p signal was consistent with the phospholipid component of the liposomal shell (Fig. 2Q). Together, these XPS results provided chemical evidence for the successful construction of F-MHC@PNRLs. The ROS-responsive release behavior of F-MHC@PNRLs was subsequently evaluated under different H₂O₂ concentrations.

Under physiological conditions without H₂O₂, Fer-1 release remained relatively slow, indicating favorable retention stability of the liposomal system. In contrast, Fer-1 release was markedly accelerated in the presence of H₂O₂, and a higher H₂O₂ concentration induced faster and more complete release. At 100 μM H₂O₂, the cumulative drug release reached approximately 80%–85%, whereas release under 10 μM H₂O₂ was moderate and release in the absence of H₂O₂ remained limited (Fig. 2R). This ROS-responsive release profile can be attributed to oxidative cleavage of TK bonds, which destabilizes the liposomal structure and promotes Fer-1 release. Since FNI is accompanied by excessive ROS accumulation in the injured nerve microenvironment, this ROS-triggered release behavior may allow F-MHC@PNRLs to preferentially release therapeutic components at the lesion site while reducing nonspecific leakage under normal conditions. Finally, to evaluate the colloidal stability of F-MHC@PNRLs, their hydrodynamic size and PDI were monitored over 7 days. As shown in Fig. S6, the particle size of F-MHC@PNRLs remained within a narrow range of approximately 165–180 nm, with no obvious increase during storage. Meanwhile, the PDI values remained below 0.30 throughout the observation period, indicating a relatively uniform size distribution. These results suggest that F-MHC@PNRLs possessed favorable short-term stability without apparent aggregation.

3.3. In vitro antioxidant and anti-inflammatory effects of F-MHC@PNRLs

Before evaluating the biological effects of F-MHC@PNRLs, the cytocompatibility of MHC and Fer-1 was first examined in RSC96 Schwann cells and RAW 264.7 macrophages. As shown in Fig. S7A and B, MHC showed good cytocompatibility in both cell types within the tested concentration range, with cell viability remaining close to or above 90% even at relatively high concentrations. Similarly, Fer-1 exhibited no obvious cytotoxicity toward RSC96 or RAW 264.7 cells within the tested concentration range (Fig. S7C and D). Next, different concentrations of RSL3 were used to establish an in vitro ferroptosis injury model in RSC96 cells. RSL3 reduced RSC96 cell viability in a concentration-dependent manner, with 10 μM causing moderate cell damage and 25 μM inducing more obvious cytotoxic injury (Fig. S7E). Therefore, 25 μM RSL3 was selected for cytotoxicity-related assays, such as CCK-8 and live/dead staining, whereas 10 μM RSL3 was used for mechanistic assays to avoid excessive cell death and better capture ferroptosis-related molecular changes. Under 25 μM RSL3 stimulation, MHC alone did not significantly rescue RSC96 cell viability within the tested concentration range (Fig. S7F), suggesting that ROS-scavenging nanozyme activity alone may be insufficient to reverse severe RSL3-induced ferroptotic injury. In contrast, Fer-1 significantly improved RSC96 cell viability under RSL3 stimulation, and the protective effect was already evident at 1 μg/mL, with no further marked increase at higher concentrations (Fig. S7G). Therefore, 1 μg/mL Fer-1 was selected as the effective therapeutic concentration for subsequent experiments. Accordingly, MHC (14.7 μg/mL), Fer-1 (1 μg/mL), F-MHC (15.7 μg/mL), and F-MHC@PNRLs (equivalent to 15.7 μg/mL of F-MHC) were diluted based on their Fer-1-equivalent concentration to ensure equivalent Fer-1 exposure among different treatment groups.

To evaluate the antioxidant and anti-inflammatory properties of F-MHC@PNRLs, an LPS-stimulated RAW264.7 macrophage model was established. Cells were divided into six groups: control (CON, G1), LPS (G2), LPS + MHC (G3), LPS + Fer-1 (G4), LPS + F-MHC (G5), and LPS + F-MHC@PNRLs (G6). MHC, Fer-1, F-MHC, and F-MHC@PNRLs were administered at corresponding concentrations to compare the therapeutic contribution of each component and the integrated nanoplatform (Fig. 3A). DCFH-DA staining showed that LPS stimulation markedly increased intracellular ROS levels, indicating successful establishment of an oxidative inflammatory microenvironment. MHC treatment significantly reduced ROS accumulation, consistent with the intrinsic nanozyme activity of ceria-based materials. Ceria-based nanozymes are known to exhibit reversible Ce³⁺/Ce⁴⁺ redox cycling, which confers superoxide dismutase- and catalase-like activities and enables continuous ROS scavenging. In the present MHC system, Mn doping was introduced to further optimize the redox microenvironment of CeO₂ by modulating Ce valence states, increasing oxygen-vacancy-related catalytic sites, and providing additional Mn-associated redox centers. This design was expected to improve the ROS-scavenging efficiency of the nanozyme and provide antioxidant support for ferroptosis inhibition [18]. This ROS-scavenging property has also been linked to anti-inflammatory effects, as excessive ROS can amplify NF-κB/MAPK signaling and promote cytokine production in activated macrophages. In contrast, Fer-1 alone only partially decreased ROS levels [30]. This result does not indicate that Fer-1 lacks bioactivity; rather, it is consistent with its pharmacological profile. Fer-1 is a classical ferroptosis inhibitor that primarily functions as a radical-trapping antioxidant to block lipid peroxidation chain reactions, especially in ferroptotic membranes, rather than acting as a broad-spectrum ROS-scavenging nanozyme [13,14]. Therefore, in an LPS-induced macrophage inflammation model dominated by intracellular ROS accumulation and inflammatory signaling activation, MHC exerted a stronger antioxidant effect than Fer-1 alone. Notably, both F-MHC and F-MHC@PNRLs markedly suppressed LPS-induced ROS generation (Fig. 3B and C).

Fig. 3.

F-MHC@PNRLs attenuate LPS-induced oxidative stress and inflammatory responses in macrophages in vitro. (A) Experimental groups: G1, control (CON); G2, LPS; G3, LPS + MHC; G4, LPS + Fer-1; G5, LPS + F-MHC; G6, LPS + F-MHC@PNRLs. (B) (B) Representative bright-field and fluorescence images of intracellular reactive oxygen species (ROS) detected by DCFH-DA staining. Scale bar = 200 μm. (C) Quantitative analysis of intracellular ROS fluorescence intensity normalized to G1. (D–H) Relative mRNA expression levels of inflammation- and polarization-related genes, including Tnf, Nos2, Il6, Arg1 and Mrc1, normalized to G1 as indicated. (I, J) Flow cytometry histograms showing the expression of CD86, an M1-associated marker, and CD206, an M2-associated marker, in each group. (K, L) Quantitative analysis of the relative mean fluorescence intensity (MFI) of CD86 and CD206, normalized to G1. (M, N) Enzyme-linked immunosorbent assay (ELISA) quantification of TNF-α and IL-1β secretion in the culture supernatant. (O) Western blot analysis of CD206 and iNOS protein expression in each group. (P, Q) Quantitative analysis of CD206 and iNOS protein levels, normalized to β-actin and presented relative to G1.

Given the close association between oxidative stress and macrophage inflammatory activation, we next investigated whether ROS suppression by MHC-containing formulations could translate into an anti-inflammatory and immunomodulatory effect. Since macrophage polarization is a key determinant of the inflammatory microenvironment after tissue injury, representative M1-associated markers and M2-associated markers were examined using qPCR, flow cytometry, ELISA, and western blotting. This multi-level validation strategy allowed us to evaluate not only the transcriptional changes of inflammatory genes but also macrophage phenotype switching and cytokine secretion, thereby providing a more comprehensive assessment of the immunoregulatory effects of different formulations. The qPCR results further supported the differential roles of MHC and Fer-1. LPS stimulation significantly upregulated the mRNA expression of pro-inflammatory genes, including Tnf, Nos2, and Il6. MHC markedly suppressed these inflammatory genes, whereas Fer-1 alone showed a relatively moderate inhibitory effect, particularly for Tnf and Nos2. This difference may be explained by the fact that MHC directly attenuates upstream ROS accumulation, thereby interrupting the oxidative stress–inflammation feedback loop, while Fer-1 is more specialized in suppressing lipid peroxidation and ferroptotic injury. Meanwhile, anti-inflammatory and M2-associated genes, including Arg1 and Il10, were increased after MHC, F-MHC, and F-MHC@PNRLs treatment, indicating that MHC-containing formulations could not only reduce inflammatory activation but also promote macrophage repolarization toward a reparative phenotype (Fig. 3D–H). Flow cytometry analysis confirmed this immunoregulatory trend. LPS increased the expression of CD86, an M1-associated marker, whereas MHC, F-MHC, and F-MHC@PNRLs reduced CD86 expression to varying degrees. Conversely, CD206, an M2-associated marker, was increased after MHC-containing treatments. Compared with Fer-1 alone, F-MHC and F-MHC@PNRLs showed stronger effects in reducing CD86 and enhancing CD206 expression, suggesting that the incorporation of Fer-1 into MHC-based nanostructures did not weaken the anti-inflammatory activity of MHC but instead supported a more favorable immunomodulatory profile (Fig. 3I–L). ELISA analysis showed a similar pattern: LPS markedly increased the secretion of TNF-α and IL-1β, whereas MHC, F-MHC, and F-MHC@PNRLs significantly reduced these pro-inflammatory cytokines. Fer-1 alone also decreased cytokine secretion to some extent, but its effect was less pronounced than that of MHC-containing formulations (Fig. 3M and N). Consistently, western blot analysis demonstrated that LPS increased iNOS expression, whereas F-MHC@PNRLs markedly suppressed iNOS and enhanced CD206 protein expression (Fig. 3O–Q). Taken together, these results indicate that F-MHC@PNRLs possess potent antioxidant and anti-inflammatory activities in vitro. Importantly, the comparative component analysis suggests that MHC mainly contributes to broad-spectrum ROS scavenging and inflammatory suppression.

3.4. F-MHC@PNRLs remodel LPS-induced inflammatory transcriptomic programs by suppressing NF-κB/MAPK signaling

To further elucidate the molecular mechanisms underlying the antioxidant and anti-inflammatory effects of F-MHC@PNRLs, RNA sequencing (RNAseq) was performed on RAW264.7 macrophages from the control group (G1), LPS group (G2), and LPS + F-MHC@PNRLs group (G6). The overall workflow of RNA extraction, sequencing, bioinformatic analysis, and validation is shown in Fig. 4A. Principal component analysis (PCA) demonstrated clear separation among G1, G2, and G6, indicating that LPS stimulation induced substantial transcriptomic remodeling, while F-MHC@PNRLs treatment reshaped the LPS-induced transcriptional profile (Fig. 4B). Differential expression analysis showed that LPS stimulation resulted in extensive changes in gene expression compared with the control group, with 1308 upregulated and 1650 downregulated genes identified in G2 versus G1 (Fig. 4C). In contrast, comparison between G6 and G2 revealed 1125 upregulated and 1277 downregulated genes after F-MHC@PNRLs treatment, suggesting that F-MHC@PNRLs markedly reversed the inflammatory transcriptional changes induced by LPS (Fig. 4D). Consistently, heatmap analysis showed that LPS increased the expression of inflammation-related genes, including Cd86, Il6, Il1b, Nos2, and Tnf, whereas F-MHC@PNRLs treatment reduced their expression and increased the expression of the M2-associated gene Arg1 (Fig. 4E). These results were consistent with the qPCR and protein-level findings described above, further confirming the anti-inflammatory and macrophage-repolarizing effects of F-MHC@PNRLs. GSEA-based ridge plot analysis was performed to further characterize the pathway-level transcriptional changes associated with LPS-induced macrophage activation and F-MHC@PNRLs-mediated immunomodulation. Compared with the blank control group, LPS stimulation induced a broad positive enrichment of inflammation- and oxidative stress-related pathways at the transcriptomic level, including reactive oxygen species, MAPK signaling, NF-κB signaling, JAK-STAT signaling, Toll-like receptor signaling, NOD-like receptor signaling, TNF signaling, IL-17 signaling, neutrophil extracellular trap formation, and cytokine–cytokine receptor interaction pathways (Fig. 4F). These results indicate that LPS challenge globally activates pro-inflammatory transcriptional programs and establishes a typical oxidative inflammatory phenotype in macrophages. In contrast, when F-MHC@PNRLs-treated cells were compared with LPS-stimulated cells, these inflammation-associated pathways exhibited an overall negative enrichment tendency, particularly NF-κB signaling, MAPK signaling, TNF signaling, cytokine–cytokine receptor interaction, Toll-like receptor signaling, and reactive oxygen species-related pathways (Fig. 4G). This opposite enrichment pattern suggests that F-MHC@PNRLs broadly suppress LPS-activated inflammatory and oxidative stress transcriptional programs rather than regulating only isolated inflammatory mediators. Further reversal analysis demonstrated that a subset of genes upregulated by LPS was downregulated after F-MHC@PNRLs treatment, while part of the genes suppressed by LPS was restored by F-MHC@PNRLs, indicating a bidirectional correction of LPS-induced transcriptional dysregulation (Fig. 4H). GO and KEGG enrichment analyses of the overlapping reversed genes further supported this interpretation. The enriched biological processes and pathways were mainly associated with immune response, cytokine activity, cytokine receptor binding, response to lipopolysaccharide, TNF signaling pathway, NF-κB signaling pathway, MAPK signaling pathway, and cytokine–cytokine receptor interaction (Fig. 4I). These findings suggest that F-MHC@PNRLs do not simply reduce a single inflammatory mediator, but broadly regulate the upstream inflammatory network activated by LPS.

Fig. 4.

F-MHC@PNRLs reverses LPS-induced inflammatory transcriptional remodeling and suppresses NF-κB/MAPK signaling. (A) Experimental workflow, including RNA extraction, sequencing, data analysis, and output. (B) PCA plot showing the transcriptional profiles of G1, G2, and G6 samples. (C, D) Volcano plot of differentially expressed genes (DEGs) in G2 vs G1 and G6 vs G2. Red, significantly upregulated genes; blue, significantly downregulated genes; gray, genes with no significant differential expression. (E) Heatmap showing expression profiles of key inflammation-related genes. (F, G) GSEA ridge plots showing enriched inflammation-related signaling pathways in G2 versus G1 and G6 versus G2. (H) Venn diagram showing the overlap among upregulated and downregulated DEGs from the G2 versus G1 and G6 versus G2 comparisons. (I) Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of overlapping DEGs. (J) GSEA plots of representative inflammatory pathways, including the NF-κB signaling pathway, MAPK signaling pathway, and TNF signaling pathway, in G2 versus G1 and G6 versus G2. (K) Western blot analysis of key proteins in the NF-κB signaling pathway, including IκBα, p-IκBα, p65, and p-p65. (L, M) Quantitative analysis of p-IκBα/IκBα and p-p65/p65, normalized to G1.(N) Western blot analysis of key proteins in the MAPK signaling pathway, including p38, p-p38, ERK, p-ERK, JNK, and p-JNK. (O–Q) Quantitative analysis of p-p38/p38, p-ERK/ERK, and p-JNK/JNK, normalized to G1. Data are shown as mean ± SD (n = 3). Statistical significance: ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 compared with G2 group. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Given that NF-κB and MAPK signaling pathways are central regulators of macrophage inflammatory activation, GSEA was further performed to examine these pathways in detail. In G2 versus G1, NF-κB, MAPK, and TNF signaling pathways were positively enriched, confirming that LPS robustly activated classical inflammatory signaling cascades. Conversely, in G6 versus G2, these pathways showed negative enrichment, indicating that F-MHC@PNRLs suppressed LPS-induced inflammatory pathway activation at the transcriptomic level (Fig. 4J). This result is mechanistically consistent with the antioxidant role of MHC. Excessive ROS is known to act as an upstream amplifier of inflammatory signaling, promoting IκBα phosphorylation and degradation, p65 activation, and MAPK phosphorylation. Therefore, the ability of F-MHC@PNRLs to scavenge ROS may contribute to the inhibition of NF-κB/MAPK-mediated inflammatory responses. To validate the transcriptomic results, western blot analysis was performed to examine key proteins in the NF-κB and MAPK pathways. LPS stimulation markedly increased the phosphorylation levels of IκBα and p65, indicating activation of the canonical NF-κB pathway. Treatment with MHC-containing formulations reduced p-IκBα/IκBα and p-p65/p65 ratios, with F-MHC and F-MHC@PNRLs showing stronger inhibitory effects than Fer-1 alone (Fig. 4K–M). This finding further supports the notion that MHC-mediated ROS scavenging plays a major role in blocking inflammatory signal transduction, whereas Fer-1 alone has relatively limited anti-inflammatory activity in this LPS-driven macrophage model. Similarly, LPS markedly increased the phosphorylation of p38, ERK, and JNK, confirming activation of the MAPK signaling pathway. F-MHC and F-MHC@PNRLs significantly reduced the phosphorylation levels of these MAPK proteins, especially p-p38/p38 and p-JNK/JNK, whereas Fer-1 alone showed a weaker or partial inhibitory effect (Fig. 4N–Q). These results indicate that F-MHC@PNRLs can effectively suppress MAPK pathway activation, thereby reducing downstream inflammatory gene expression and cytokine production.

3.5. Anti-ferroptosis activity of F-MHC@PNRLs on Schwann cells (SCs) in vitro

Schwann cell survival is essential for peripheral nerve regeneration because SCs provide trophic support, participate in myelin clearance and remyelination, and guide axonal regrowth after nerve injury [26]. Ferroptosis, a regulated cell death modality characterized by iron-dependent lipid peroxidation, has been increasingly implicated in peripheral nerve injury and Schwann cell dysfunction [6,27]. Therefore, after confirming the antioxidant and anti-inflammatory properties of F-MHC@PNRLs in macrophages, we further investigated whether this system could protect RSC96 Schwann cells against ferroptotic injury. RSL3, a classical GPX4 inhibitor, was used to establish an in vitro ferroptosis model. The experimental groups included control (G1), RSL3 (G2), RSL3 + MHC (G3), RSL3 + Fer-1 (G4), RSL3 + F-MHC (G5), and RSL3 + F-MHC@PNRLs (G6) (Fig. 5A). Based on the CCK-8 assay, RSL3 reduced RSC96 cell viability in a concentration-dependent manner. Therefore, 10 μM and 25 μM RSL3 were selected as representative concentrations to establish moderate and severe ferroptotic injury models, respectively, in subsequent experiments. Intracellular ROS levels were first evaluated using DCFH-DA staining. Compared with the control group, RSL3 (10 μM) stimulation markedly increased intracellular ROS accumulation after 24 h, indicating severe oxidative stress in Schwann cells. MHC treatment reduced ROS levels to a certain extent, consistent with its nanozyme-like ROS-scavenging activity. Fer-1 treatment showed a stronger inhibitory effect than MHC in this RSL3-induced ferroptosis model, which is consistent with the pharmacological feature of Fer-1 as a ferroptosis inhibitor that blocks lipid radical propagation. Notably, F-MHC and F-MHC@PNRLs further reduced ROS accumulation, with F-MHC@PNRLs showing the most evident suppression of ROS fluorescence (Fig. 5B and C). These results suggest that, unlike the LPS-induced macrophage inflammatory model where MHC dominated ROS scavenging, Fer-1 contributed more prominently under RSL3-induced ferroptotic stress, while the integrated F-MHC@PNRLs provided combined antioxidant and anti-ferroptotic protection. We next examined whether F-MHC@PNRLs regulated the NRF2/HO-1 antioxidant pathway. Western blot results showed that RSL3 (10 μM) increased total NRF2 expression and nuclear NRF2 to some extent, suggesting that Schwann cells may initiate a compensatory antioxidant response under ferroptotic stress. However, this response appeared insufficient to prevent oxidative damage. MHC alone did not markedly restore NRF2/HO-1 signaling under RSL3 stimulation, whereas Fer-1, F-MHC, and F-MHC@PNRLs significantly increased HO-1 and nuclear-NRF2 expression (Fig. 5D–G). This indicates that Fer-1-containing formulations can reinforce endogenous antioxidant defense during ferroptotic injury. Among these groups, F-MHC and F-MHC@PNRLs maintained strong activation of the NRF2/HO-1 axis, suggesting that the combination of Fer-1-mediated lipid peroxide inhibition and MHC-mediated redox regulation may cooperatively enhance cellular resistance to oxidative stress.

Fig. 5.

F-MHC@PNRLs protect RSC96 Schwann cells against RSL3-induced ferroptosis by suppressing oxidative stress, iron accumulation, and lipid peroxidatio. (A) Experimental groups: G1, control (CON); G2, RSL3; G3, RSL3 + MHC; G4, RSL3 + Fer-1; G5, RSL3 + F-MHC; G6, RSL3 + F-MHC@PNRLs. (B) Representative fluorescence images of intracellular ROS levels by DCFH-DA across all groups (C) Quantitative analysis of ROS fluorescence intensity, normalized to G1. (D) Western blot analysis of NRF2, HO-1, and nuclear NRF2 (n-NRF2) protein expression. β-actin and Lamin B1 were used as loading controls for total and nuclear proteins, respectively. (E–G) Quantitative analysis of NRF2, HO-1 and n-NRF2 protein levels, normalized to G1. (H–L) Quantitative analysis of antioxidant and oxidative stress-related indicators, including superoxide dismutase (SOD) activity, malondialdehyde (MDA) levels, glutathione (GSH) levels, oxidized glutathione (GSSG) levels, and catalase (CAT) activity.M) Representative fluorescence images of intracellular Fe²⁺ levels detected by RhoNox-6 staining. Hoechst 33342 was used for nuclear counterstaining. Scale bar = 500 μm. (N) Quantitative analysis of RhoNox-6 fluorescence intensity, normalized to G1. (O) Representative fluorescence images of lipid peroxidation detected by C11-BODIPY staining. Reduced and oxidized C11-BODIPY signals are shown in red and green, respectively. Scale bar = 200 μm. (P, Q) Quantitative analysis of reduced C11-BODIPY fluorescence intensity and oxidized C11-BODIPY fluorescence intensity, normalized to G1. (S, T) Quantitative analysis of GPX4 (S) and SLC7A11 (T) protein levels, normalized to β-actin and presented relative to G1. Data are shown as mean ± SD (n = 3). Statistical significance: ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 compared with G2 group. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Consistent with the above findings, RSL3 markedly disrupted the intracellular redox balance, as reflected by decreased superoxide dismutase (SOD), catalase (CAT), and glutathione (GSH) levels, together with increased malondialdehyde (MDA) and oxidized glutathione (GSSG) levels. These changes indicate impaired antioxidant capacity, enhanced lipid peroxidation, and glutathione redox imbalance in Schwann cells. MHC treatment partially restored antioxidant enzyme activity and reduced MDA accumulation, but its effect on GSH/GSSG homeostasis was relatively limited. In contrast, Fer-1 treatment more effectively reduced MDA and improved the glutathione redox state, supporting its specific role in suppressing lipid peroxidation. Importantly, F-MHC and F-MHC@PNRLs significantly restored SOD, CAT, and GSH levels while reducing MDA and GSSG levels, indicating a more comprehensive recovery of antioxidant homeostasis (Fig. 5H–L). These results further confirm that MHC and Fer-1 exert complementary effects: MHC mainly contributes to enzymatic ROS detoxification, whereas Fer-1 mainly limits lipid peroxidation-mediated ferroptotic damage. Iron accumulation is another key event in ferroptosis. RhoNox-6 staining showed that RSL3 markedly increased intracellular Fe²⁺ levels in RSC96 cells. Although MHC treatment reduced Fe²⁺ accumulation to some degree, Fer-1, F-MHC, and F-MHC@PNRLs displayed stronger inhibitory effects, with F-MHC@PNRLs showing the lowest RhoNox-6 fluorescence intensity among the treatment groups (Fig. 5M and N). These data suggest that F-MHC@PNRLs can effectively alleviate iron-associated oxidative stress in Schwann cells. This effect may be attributed to the combined reduction of ROS accumulation, lipid peroxide propagation, and ferroptotic amplification. Lipid peroxidation was further assessed using C11-BODIPY staining. In the control group, strong red fluorescence and weak green fluorescence were observed, indicating low lipid peroxidation. RSL3 treatment markedly decreased the reduced C11-BODIPY signal and increased the oxidized C11-BODIPY signal, confirming excessive lipid peroxide accumulation. MHC partially reduced lipid oxidation, whereas Fer-1 showed a more pronounced protective effect, consistent with its role as a lipid radical-trapping antioxidant. F-MHC and F-MHC@PNRLs strongly restored the reduced C11-BODIPY signal and suppressed the oxidized signal, indicating potent inhibition of lipid peroxidation (Fig. 5O–Q). These findings demonstrate that F-MHC@PNRLs effectively block the lipid peroxidation cascade, which is central to ferroptosis execution. Finally, the expression of key anti-ferroptotic proteins was evaluated. Western blot analysis showed that RSL3 markedly reduced GPX4 and SLC7A11 protein levels, indicating collapse of the SLC7A11/GSH/GPX4 antioxidant defense system. MHC alone showed limited restoration of GPX4 and SLC7A11 expression, whereas Fer-1 significantly increased their expression, likely by alleviating lipid peroxidation-associated cellular stress. F-MHC and F-MHC@PNRLs further restored GPX4 and SLC7A11 protein levels, suggesting that the integrated platform can preserve the core anti-ferroptotic defense pathway in Schwann cells (Fig. 5R–T). Since GPX4 directly detoxifies lipid hydroperoxides and SLC7A11 supports cystine uptake and GSH biosynthesis, the restoration of these proteins provides mechanistic evidence for the anti-ferroptotic activity of F-MHC@PNRLs.

3.6. F-MHC@PNRLs protect the mitochondrial homeostasis of SCs in vitro

Mitochondrial dysfunction is closely associated with ferroptotic injury and Schwann cell impairment [31]. Because Schwann cells (SCs) are essential for peripheral nerve repair by supporting axonal regeneration, remyelination, and trophic factor secretion, maintaining SC viability and mitochondrial homeostasis is critical for facial nerve recovery. Therefore, after confirming the anti-ferroptotic activity of F-MHC@PNRLs, we further evaluated whether this system could rescue RSL3-induced SC dysfunction. CCK-8 assays showed that RSL3 markedly decreased the viability of RSC96 Schwann cells at both 12 h and 24 h compared with the control group. MHC alone did not significantly restore cell viability, indicating that MHC-mediated ROS scavenging was insufficient to fully rescue RSL3-induced ferroptotic cytotoxicity. In contrast, Fer-1, F-MHC, and F-MHC@PNRLs significantly improved cell viability, with F-MHC@PNRLs showing a strong protective effect, especially at 24 h (Fig. 6A and B). These results suggest that Fer-1 plays a major role in counteracting RSL3-induced ferroptotic cell injury. Live/dead staining further confirmed this protective effect. RSL3 exposure led to an obvious increase in dead cells, whereas Fer-1-containing treatments markedly reduced cell death. Quantitative analysis showed that the cell death rate was significantly lower in the Fer-1, F-MHC, and F-MHC@PNRLs groups than in the RSL3 group (Fig. 6C and D). Notably, MHC alone showed limited protection, which is consistent with the CCK-8 results and further supports the idea that Fer-1 is the key anti-ferroptotic component in this model. Because Schwann cell migration is required for guiding axonal regrowth and organizing the regenerative microenvironment after peripheral nerve injury, we next assessed cell migration using wound-healing and Transwell assays. RSL3 substantially inhibited Schwann cell migration, as shown by delayed wound closure and reduced numbers of migrated cells. Fer-1 treatment partially restored migration capacity, whereas F-MHC and F-MHC@PNRLs further enhanced Schwann cell migration (Fig. 6E–H). These results indicate that ferroptotic stress impairs not only SC survival but also their regenerative behavior, and F-MHC@PNRLs can effectively preserve SC migration under ferroptotic conditions.

Fig. 6.

Effects of F-MHC@PNRLs on Schwann cell viability, migration, and mitochondrial homeostasis. (A, B) Quantification of RSC96 Schwann cell viability in different groups at 12 h and 24 h. Experimental groups: G1, control (CON); G2, RSL3; G3, RSL3 + MHC; G4, RSL3 + Fer-1; G5, RSL3 + F-MHC; G6, RSL3 + F-MHC@PNRLs. (C, D) Representative live/dead staining images and quantification of cell death rate at 24 h. Scale bar = 200 μm. (E, F) Representative images of wound-healing assays at 0 h and 24 h and quantification of migration rate. Scale bar = 500 μm. (G, H) Representative images of Transwell migration assays at 48 h and quantification of migrated cell number. Scale bar = 200 μm.(I, J) Representative MitoTracker and MitoSOX fluorescence staining images and quantification of relative MitoSOX fluorescence intensity. Scale bar = 200 μm. (K, L) Representative JC-1 staining images and quantification of the JC-1 aggregate/monomer ratio. Scale bar = 200 μm. (M) Representative transmission electron microscopy (TEM) images of mitochondrial ultrastructure in Schwann cells from different groups. Upper scale bar = 2 μm; lower scale bar = 0.5 μm.Data are shown as mean ± SD (n = 3). Statistical significance: ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 compared with G2 group.

Mitochondrial morphology and mitochondrial ROS accumulation were evaluated using MitoTracker and MitoSOX staining, respectively. RSL3 treatment caused a marked increase in mitochondrial ROS, suggesting severe mitochondrial oxidative damage. MHC alone slightly reduced MitoSOX fluorescence, whereas Fer-1, F-MHC, and F-MHC@PNRLs significantly suppressed mitochondrial ROS accumulation (Fig. 6I and J). This result is consistent with the role of Fer-1 in blocking lipid radical propagation and suggests that inhibition of ferroptotic lipid peroxidation also contributes to mitochondrial protection. JC-1 staining was used to assess mitochondrial membrane potential. In the control group, strong JC-1 aggregate fluorescence was observed, indicating intact mitochondrial membrane potential. RSL3 markedly decreased the JC-1 aggregate/monomer ratio, reflecting mitochondrial depolarization. MHC alone only partially restored mitochondrial membrane potential, whereas Fer-1, F-MHC, and F-MHC@PNRLs significantly increased the aggregate/monomer ratio (Fig. 6K and L). These findings demonstrate that F-MHC@PNRLs effectively preserve mitochondrial membrane potential under RSL3-induced ferroptotic stress. Finally, transmission electron microscopy (TEM) was performed to directly observe mitochondrial ultrastructure. Control cells displayed relatively intact mitochondria with recognizable cristae and preserved morphology. RSL3-treated cells exhibited typical mitochondrial damage, including swollen or structurally disrupted mitochondria, reduced cristae, and increased intracellular vacuolization. MHC treatment provided limited ultrastructural protection, whereas Fer-1-containing treatments markedly improved mitochondrial morphology. In particular, F-MHC@PNRLs-treated cells showed better-preserved mitochondrial structures and fewer obvious ultrastructural abnormalities (Fig. 6M).

3.7. Evaluation of therapeutic efficacy of F-MHC@PNRLs in vivo based on FNI model

Previous studies have demonstrated that NP41, a nerve-binding peptide with the sequence NTQTLAKAPEHT, can preferentially associate with peripheral nerves and has been widely used for peripheral nerve visualization and nerve-guided delivery [32]. Whitney et al. identified NP41 through phage display screening and showed that fluorescently labeled NP41 clearly delineated peripheral nerves in mice within 2 h after systemic administration, with high nerve-to-background contrast and no obvious behavioral toxicity [33]. Before conducting the formal in vivo therapeutic evaluation, a preliminary dose-screening experiment was performed to determine a suitable working dose of F-MHC@PNRLs for facial nerve repair. Facial nerve function was evaluated using the FNI scoring system at different time points after surgery. As shown in Fig. S8, all groups exhibited a sharp decrease in FNI scores on day 1 after injury, indicating successful establishment of the facial nerve crush injury model. In the untreated group, facial nerve function gradually recovered over time but remained incomplete by the late stage of observation. In contrast, treatment with F-MHC@PNRLs accelerated functional recovery in a dose-dependent manner. The 0.5 mg/kg group showed moderate improvement compared with the untreated group, whereas the 1.0, 2.5, and 5.0 mg/kg groups exhibited more pronounced recovery, with FNI scores approaching 6–7 points by the late observation stage. Notably, the recovery curves of the 2.5 and 5.0 mg/kg groups were relatively close, suggesting that the therapeutic effect tended to reach a plateau at higher concentrations. Considering both therapeutic efficacy and dose economy, 1.0 mg/kg was selected as the working dose for subsequent in vivo experiments. Based on the optimized dose and confirmed biosafety, we next evaluated the therapeutic efficacy of F-MHC@PNRLs in vivo using a mouse facial nerve injury model. The animals were divided into four groups: Sham (G1), FNI + PBS (G2), FNI + F-MHC@RLs (G3), and FNI + F-MHC@PNRLs (G4). F-MHC@RLs were used as the non-targeted ROS-responsive liposomal counterpart to clarify the contribution of NP41-mediated peripheral nerve-associated delivery. After FNI modeling on day 0, local treatments were administered around the injured facial nerve, and tissues were collected at different time points for ROS detection, immunofluorescence staining, functional scoring, electrophysiological recording, histological staining, and ultrastructural evaluation (Fig. 7A).

Fig. 7.

F-MHC@PNRLs promote early functional recovery, suppress oxidative inflammation, and enhance remyelination after facial nerve injury in vivo. (A) Schematic illustration of the in vivo experimental design. Mice were divided into four groups: G1, Sham; G2, FNI + PBS; G3, FNI + F-MHC@RLs; and G4, FNI + F-MHC@PNRLs. Facial nerve injury was established on day 0, followed by local treatment. Facial nerve tissues were collected at predefined time points for ROS detection, immunofluorescence staining, transmission electron microscopy (TEM), compound muscle action potential (CMAP) recording, Luxol fast blue (LFB) staining, and hematoxylin and eosin (H&E) staining. (B) Representative in vivo Cy5.5 fluorescence images showing the local retention of Cy5.5@RLs and Cy5.5@PNRLs around the injured facial nerve region at 3, 6, 12, 24, and 48 h after local administration. (C) Quantitative analysis of the fluorescence area in the Cy5.5@RLs and Cy5.5@PNRLs groups. (D) Facial nerve functional scores of mice in each group during the 14-day observation period. (E) Representative fluorescence images of ROS accumulation in facial nerve tissues detected by DCF staining. Nuclei were stained with DAPI. Scale bar = 200 μm. (F) Quantitative analysis of DCF fluorescence intensity, normalized to the Sham group. (G) Representative immunofluorescence images of iNOS and CD206 expression in facial nerve tissues. iNOS was used as an M1-associated inflammatory marker, and CD206 was used as an M2-associated reparative marker. Nuclei were stained with DAPI. Scale bar = 200 μm. (H, I) Quantitative analysis of iNOS-positive area (H) and CD206-positive area (I). (J) Representative immunofluorescence images of MBP and S100β expression in facial nerve tissues. MBP was used to evaluate myelin integrity, and S100β was used to identify Schwann cell-associated areas. Nuclei were stained with DAPI. Scale bar = 200 μm. (K, L) Quantitative analysis of MBP-positive area (K) and S100β-positive area (L). (M) Representative TEM images showing the ultrastructure of myelinated nerve fibers in each group. Scale bar = 1 μm. (N) Quantitative analysis of abnormal myelinated fibers based on TEM images. (O) Representative compound muscle action potential (CMAP) traces recorded from the ipsilateral orbicularis oculi muscle. (P) Quantitative analysis of CMAP amplitude in each group. (Q) Representative Luxol fast blue (LFB) staining images showing myelin morphology in facial nerve tissues. Scale bar = 20 μm. Data are presented as mean ± SD. ns, not significant; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 versus the FNI + PBS group; #P < 0.05, ##P < 0.01, and ####P < 0.0001 versus the FNI + F-MHC@RLs group. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

To directly verify whether NP41 modification enhanced local retention of liposomes around the injured facial nerve, in vivo Cy5.5 fluorescence imaging was performed using Cy5.5@RLs and Cy5.5@PNRLs. As shown in Fig. 7B, the fluorescence signal in the Cy5.5@RLs group gradually decreased over time, indicating progressive dispersion or clearance of non-targeted liposomes from the local injury region. In contrast, Cy5.5@PNRLs showed stronger and more sustained fluorescence retention around the injured facial nerve region from 3 to 48 h after local administration. Quantitative analysis further confirmed that the fluorescence area in the Cy5.5@PNRLs group remained significantly higher than that in the Cy5.5@RLs group at later time points, especially at 24 and 48 h (Fig. 7C). These results provide direct imaging evidence that NP41 modification enhances local retention and peripheral nerve-associated accumulation of the liposomal formulation after FNI. This enhanced retention may increase the effective exposure of F-MHC at the injured nerve site, thereby improving therapeutic efficiency while limiting nonspecific diffusion into surrounding tissues. Facial nerve function was then assessed by behavioral scoring during the 14-day observation period. The Sham group maintained normal facial nerve function, whereas the FNI + PBS group exhibited severe functional impairment after injury, followed by only partial spontaneous recovery. Compared with the FNI + PBS group, both F-MHC@RLs and F-MHC@PNRLs promoted functional recovery, as indicated by progressively increased FNI scores. Notably, the F-MHC@PNRLs group exhibited faster and more pronounced improvement than the non-targeted F-MHC@RLs group, particularly during the middle-to-late repair stage (Fig. 7D). This result suggests that NP41-mediated peripheral nerve-associated delivery can enhance the therapeutic efficacy of F-MHC in vivo. Since behavioral recovery after facial nerve injury depends on coordinated reduction of local tissue damage, preservation of Schwann cell-associated repair functions, remyelination, and restoration of neuromuscular signal transmission, the superior functional recovery observed in the F-MHC@PNRLs group indicates a broader reparative effect beyond simple symptom improvement. Because excessive ROS accumulation is a major contributor to secondary damage after peripheral nerve injury, DCF staining was performed to evaluate oxidative stress in injured facial nerve tissues. Compared with the Sham group, the FNI + PBS group showed markedly increased DCF fluorescence, indicating substantial ROS accumulation at the injury site. Treatment with F-MHC@RLs reduced ROS levels to some extent, whereas F-MHC@PNRLs further decreased DCF fluorescence intensity (Fig. 7E and F). These findings demonstrate that F-MHC@PNRLs effectively attenuate oxidative stress in vivo. Mechanistically, this effect may be attributed to the catalytic ROS-scavenging activity of MHC, which can reduce upstream oxidative burden through ceria-based redox cycling and Mn-enhanced nanozyme activity. Importantly, the stronger ROS suppression observed in the F-MHC@PNRLs group compared with the F-MHC@RLs group further supports the functional relevance of NP41-mediated local retention. Excessive ROS after nerve injury can amplify lipid peroxidation, activate inflammatory signaling, and impair mitochondrial homeostasis; therefore, reducing ROS overload is critical for interrupting the oxidative stress–inflammation–ferroptosis-related injury loop during early facial nerve repair [6,8,15].

The inflammatory microenvironment after FNI was further evaluated by immunofluorescence staining of iNOS and CD206. The FNI + PBS group exhibited a markedly increased iNOS-positive area, indicating enhanced M1-like inflammatory activation, whereas CD206 expression remained relatively low. F-MHC@RLs partially suppressed iNOS expression and increased CD206 expression. In comparison, F-MHC@PNRLs produced a stronger immunomodulatory effect, as reflected by a lower iNOS-positive area and a significantly higher CD206-positive area (Fig. 7G–I). These results indicate that F-MHC@PNRLs can suppress pro-inflammatory activation while promoting reparative macrophage polarization in injured facial nerve tissues. This effect is consistent with the in vitro findings showing that F-MHC@PNRLs inhibited LPS-induced inflammatory responses and shifted macrophages toward an M2-like phenotype. Macrophage phenotype switching plays an essential role in peripheral nerve repair because excessive M1-like activation sustains inflammatory injury, whereas M2-like macrophages contribute to debris clearance, resolution of inflammation, Schwann cell support, and tissue remodeling [34]. Thus, the ability of F-MHC@PNRLs to reshape the local inflammatory milieu may provide an important microenvironmental basis for subsequent remyelination and functional recovery.

To further assess Schwann cell-associated myelin repair, MBP and S100β immunofluorescence staining was performed. MBP was used to evaluate myelin integrity, whereas S100β was used to identify Schwann cell-associated areas. In the Sham group, MBP and S100β signals were abundant and well organized. After FNI, MBP-positive and S100β-positive areas were markedly reduced, indicating disruption of the Schwann cell–myelin unit. F-MHC@RLs partially restored MBP and S100β expression, whereas F-MHC@PNRLs resulted in a more pronounced recovery of both markers (Fig. 7J–L). These results suggest that F-MHC@PNRLs protect Schwann cell-associated structures and promote early remyelination after FNI. Schwann cells are indispensable for peripheral nerve repair because they support axonal regrowth, remyelination, and trophic factor secretion after injury [4,5]. However, Schwann cells are highly susceptible to oxidative stress, mitochondrial dysfunction, and lipid peroxidation-related damage. Therefore, the improved MBP and S100β signals in the F-MHC@PNRLs group may be associated with the combined effects of ROS scavenging, inflammatory regulation, and ferroptosis-related lipid peroxidation inhibition. Ultrastructural analysis by TEM further confirmed the protective effect of F-MHC@PNRLs on myelinated nerve fibers. The Sham group showed compact and regular myelinated fibers with preserved myelin sheath morphology. In contrast, the FNI + PBS group displayed obvious ultrastructural abnormalities, including disrupted myelin sheaths, irregular myelin morphology, and an increased proportion of abnormal myelinated fibers. F-MHC@RLs partially alleviated these pathological changes, whereas F-MHC@PNRLs further improved myelin ultrastructure and significantly reduced the percentage of abnormal myelinated fibers (Fig. 7M and N). These findings provide morphological evidence that F-MHC@PNRLs contribute to early structural preservation and remyelination after FNI. Since myelin integrity is essential for efficient nerve conduction, the improvement in myelin ultrastructure also provides a structural explanation for the subsequent electrophysiological recovery observed in the F-MHC@PNRLs group.

To provide an objective functional readout beyond behavioral scoring, electrophysiological assessment was performed by recording compound muscle action potentials (CMAPs) from the ipsilateral orbicularis oculi muscle. As shown in Fig. 7O, the Sham group exhibited a robust CMAP response, whereas FNI markedly reduced the CMAP amplitude, indicating impaired facial nerve conduction after injury. F-MHC@RLs partially restored CMAP amplitude, while F-MHC@PNRLs produced a more pronounced recovery compared with the non-targeted formulation (Fig. 7P). These electrophysiological results are consistent with the behavioral scoring, MBP/S100β staining, and TEM findings, collectively supporting that F-MHC@PNRLs improve early facial nerve functional recovery. Because CMAP amplitude reflects the ability of regenerated or preserved nerve fibers to conduct impulses and activate target muscles, the enhanced CMAP recovery in the F-MHC@PNRLs group further indicates that the improved local microenvironment and remyelination translated into functional nerve conduction restoration. Luxol fast blue staining was further performed to evaluate myelin morphology in facial nerve tissues. Dense and continuous LFB staining was observed in the Sham group, whereas the FNI + PBS group showed weakened and disorganized myelin staining, reflecting substantial demyelination. F-MHC@RLs improved LFB staining to some extent, while F-MHC@PNRLs more effectively restored myelin staining intensity and structural organization (Fig. 7Q). Together with the MBP immunofluorescence and TEM results, LFB staining further supports the conclusion that F-MHC@PNRLs promote early myelin preservation and remyelination after FNI.

The systemic biosafety of the different treatments was further evaluated by routine blood analysis, serum biochemical assays, and histological examination of major organs [35]. As shown in Fig. S9A and B, hematological parameters, including red blood cells, hemoglobin, platelets, white blood cells, neutrophils, and monocytes, showed no significant differences among groups. Similarly, liver function-related indicators, including ALT and AST, and renal function-related indicators, including urea and BUN, remained comparable among the Sham, FNI + PBS, FNI + F-MHC@RLs, and FNI + F-MHC@PNRLs groups (Fig. S9C and D). H&E staining of major organs, including the heart, liver, spleen, lung, and kidney, showed no obvious inflammatory infiltration, tissue necrosis, structural disruption, or pathological injury in any treatment group (Fig. S9E). These results indicate that local administration of F-MHC@PNRLs did not cause apparent systemic toxicity, supporting its favorable preliminary biosafety for subsequent therapeutic evaluation.

Collectively, these in vivo results demonstrate that NP41-modified F-MHC@PNRLs enhance local retention around the injured facial nerve, attenuate ROS accumulation, regulate macrophage inflammatory polarization, preserve the Schwann cell–myelin unit, improve myelin ultrastructure, and promote early functional and electrophysiological recovery after FNI. Compared with non-targeted F-MHC@RLs, F-MHC@PNRLs showed superior therapeutic efficacy under equivalent F-MHC-, MHC-, and Fer-1-equivalent doses, highlighting the importance of NP41-mediated peripheral nerve-associated delivery in enhancing the local therapeutic availability of the integrated anti-oxidative and anti-lipid peroxidation nanoplatform.

4. Conclusion

In this study, we first demonstrated that ferroptosis-related molecular events are activated during the acute phase of facial nerve injury. RNA-seq analysis revealed significant transcriptomic remodeling after FNI, accompanied by increased ferroptosis potential, altered expression of ferroptosis-associated genes, reduced GPX4 expression, and enhanced 4-HNE accumulation in injured facial nerve tissues. These findings indicate that impaired anti-ferroptotic defense and aggravated lipid peroxidation are closely associated with Schwann cell disturbance and early nerve microenvironment deterioration after FNI. Based on this pathological mechanism, we successfully constructed F-MHC@PNRLs, a peripheral nerve-targeted ROS-responsive liposomal nanoplatform integrating Mn-doped CeO₂ nanozymes and Ferrostatin-1. The system was designed to combine catalytic ROS scavenging, ferroptosis inhibition, ROS-triggered release, and NP41-mediated peripheral nerve targeting. Comprehensive physicochemical characterization confirmed the successful preparation of F-MHC@PNRLs, including stable nanoscale morphology, verified elemental composition, preserved nanozyme crystal structure, favorable colloidal stability, and ROS-responsive Fer-1 release. Functionally, F-MHC@PNRLs exerted strong antioxidant and anti-inflammatory effects in LPS-stimulated macrophages by suppressing ROS accumulation, reducing inflammatory cytokine expression, promoting reparative macrophage polarization, and inhibiting NF-κB/MAPK pathway activation. In RSL3-induced Schwann cell ferroptosis models, F-MHC@PNRLs effectively restored redox homeostasis, inhibited Fe²⁺ accumulation and lipid peroxidation, preserved the SLC7A11/GPX4 antioxidant axis, and protected mitochondrial function and ultrastructure. These in vitro results confirmed the complementary therapeutic roles of MHC-mediated ROS scavenging and Fer-1-mediated ferroptosis inhibition. In vivo, local administration of F-MHC@PNRLs significantly promoted facial nerve functional recovery after FNI. Compared with non-targeted F-MHC@RLs, F-MHC@PNRLs showed superior therapeutic efficacy in reducing local ROS accumulation, reshaping the inflammatory microenvironment, restoring MBP and S100β expression, improving LFB-positive myelin integrity, and preserving myelinated fiber ultrastructure. Moreover, hematological, biochemical, and major organ histological analyses indicated favorable preliminary biosafety. Taken together, this study identifies ferroptosis-related oxidative lipid damage as an important pathological mechanism in FNI and establishes F-MHC@PNRLs as a multifunctional therapeutic platform for facial nerve repair. By integrating antioxidative, anti-ferroptotic, immunomodulatory, ROS-responsive, and peripheral nerve-targeting properties, F-MHC@PNRLs provide a promising biomaterial-based strategy for improving peripheral nerve regeneration after injury.

Nevertheless, several limitations should be acknowledged. First, although the increased FPI score, positive ferroptosis enrichment trend, altered ferroptosis-related gene expression, reduced GPX4 expression, and enhanced 4-HNE accumulation collectively suggest the involvement of ferroptosis-related lipid peroxidation in early FNI, the current transcriptomic evidence should be interpreted cautiously because the GSEA result showed a positive trend rather than strong statistical significance. In addition, bulk RNA-seq analysis cannot precisely define the cellular origin of ferroptosis-related signals within heterogeneous injured facial nerve tissues. Future studies using single-cell RNA sequencing, spatial transcriptomics, or cell-type-specific validation will be needed to further clarify whether Schwann cells are the predominant cellular population undergoing ferroptosis-related injury after FNI. Second, the H₂O₂-triggered release assay was performed as a simplified in vitro oxidative microenvironment model to verify the ROS-responsive release capability of F-MHC@PNRLs. It should be noted that the current study does not provide direct in vivo evidence demonstrating TK bond cleavage or F-MHC release at the injured facial nerve site. The proposed ROS-responsive release behavior in vivo is mainly inferred from the in vitro H₂O₂-triggered release profile, the elevated ROS accumulation observed in injured facial nerve tissues, and the improved therapeutic outcomes after local administration. Therefore, the H₂O₂-triggered release assay should be considered a simplified in vitro model to verify ROS-responsive release capability, rather than direct evidence of in vivo TK cleavage or intralesional F-MHC release. Future studies using dual-labeled, FRET-based, or chemically traceable systems will be needed to directly monitor TK bond cleavage and F-MHC release in the injured nerve microenvironment.The current imaging approach cannot distinguish intraneural distribution from perineural retention, and therefore the therapeutic effects may result from improved local exposure around the injured nerve, modulation of the surrounding inflammatory and oxidative microenvironment, or partial nerve-associated accumulation. Future studies using higher-resolution ex vivo fluorescence imaging, tissue-clearing-based 3D imaging, confocal co-localization with nerve fascicle markers, or intraneural section-level quantitative analysis will be required to define the precise spatial distribution of the formulation within the injured nerve region. Finally, the present in vivo evaluation mainly focused on early repair within 14 days after FNI. Longer observation periods, such as 28 or 42 days, together with long-term behavioral assessment, electrophysiological testing, axonal regeneration markers, remyelination analysis, muscle reinnervation evaluation, and extended systemic biosafety studies, will be necessary to determine the long-term therapeutic efficacy and safety profile of F-MHC@PNRLs.

Funding

This work was supported by the Natural Science Foundation of Zhejiang Province, China (No. MS25H090029),the Medical and Health Research Project of Zhejiang Province (No. 2024KY399), Wenzhou science and technology project of China (No. Y20220935), European Union's Research and Innovation Program under the Marie Skłodowska Curie grant agreement (No. 101064861), and Young Scientists Program of Hangzhou Natural Science Foundation (No. 2025SZRJJ1086).

CRediT authorship contribution statement

Qiang Zhou: Data curation, Formal analysis, Investigation, Methodology. Xiaolong Xu: Data curation, Formal analysis, Investigation, Methodology. Shuyi Lin: Data curation, Formal analysis, Investigation, Methodology. Yida Wang: Data curation, Formal analysis, Investigation, Methodology. Yuqi Han: Data curation, Formal analysis. Wenxin Liang: Data curation, Formal analysis. Ran Mo: Data curation, Formal analysis. Xianlong Wang: Formal analysis. Yiyang Cheng: Formal analysis. Chengxuan Tang: Formal analysis. Lingxiao Zhang: Conceptualization, Funding acquisition, Supervision, Writing – review & editing. Yiheng Yang: Conceptualization, Funding acquisition, Project administration, Writing – review & editing. Xianzhen Chen: Conceptualization, Funding acquisition, Project administration, Writing – review & editing. Liangle Liu: Conceptualization, Funding acquisition, Project administration, Writing – review & editing.

Declaration of competing interest

The authors declare the following financial interests/personal relationshipswhich may be considered as potential competing interests: Prof. Lingxiao Zhang is aguest editor for this journal. To ensure a fair and impartial review process, Prof. Zhanghad no involvement in the peer review of this article and had no access toinformation regarding its review. The editorial handling and decision-making for thisarticle were delegated to another journal editor. All other authors declare that theyhave no known competing financial interests or personal relationships that couldhave appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is/are the supplementary data to this article.

Data availability

Data will be made available on request.