To synthesize the amphiphilic polymer P₁M₅, we performed acylation between the anhydride groups of PMHC₁₈ [poly(maleic anhydride-alt-1-octadecene)] and mPEG-NH₂ (average Mw = 5 kDa). Briefly, PMHC₁₈ (70.2 mg, 0.2 mmol; Sigma-Aldrich, USA) was dissolved in dichloromethane (DCM), followed by the addition of 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (75.4 mg, 0.4 mmol). The mixture was stirred for 30 min to activate carboxyl groups. Subsequently, mPEG-NH₂ (200 mg, 0.04 mmol) and triethylamine (TEA, 0.8 mmol; Macklin, China) were introduced, and the reaction proceeded for 24 h under stirring. The resulting product was dried under a nitrogen stream to yield a white solid, which was redissolved in dimethyl sulfoxide (DMSO) with stirring until fully dissolved. The solution was then diluted with ultrapure water, transferred to a dialysis membrane (MWCO = 14 kDa), and dialyzed against distilled water for 24 h. The final product was lyophilized and preserved at -20 °C for subsequent use.

Idebenone (MedChemExpress, USA) was encapsulated with amphiphilic polymer P₁M₅ by the self-assembly method. In short, 10 mg of Idebenone and 10, 20, 40, 80, 120, 160, and 200 mg of P₁M₅ were dissolved in dimethyl sulfoxide (DMSO), respectively. Idebenone solution was added to the P₁M₅ solution at different mass ratios (P₁M₅: Idebenone) and mixed thoroughly. Then, it was added dropwise to a round-bottom flask filled with double-distilled water, and the P@I solution was obtained by ultrasonic reaction for 30 min.

SP@I was synthesized by conjugating SS31(DGpeptides, China) to P@I via an amide reaction. In brief, carboxyl groups on the P@I solution (containing 160 mg P₁M₅, equivalent to 0.188 mmol COOH) were activated by reacting with EDC (145.2 mg, 0.752 mmol) and Sulfo-NHS (159.7 mg, 0.830 mmol) for 0.5 h. SS31 (15.1 mg, 0.0236 mmol) was dissolved in 200 µL of double-distilled water and then slowly added to the activated P@I solution. The reaction was allowed to proceed for 12 h under light-protected conditions. For fluorescent labeling, Cy5.5 (10 mg/mL; Aladdin, China) was added during the P@I preparation to obtain P@I-Cy5.5. SP@I-Cy5.5 was then synthesized using the same conjugation method as described above, starting from P@I-Cy5.5. The final products (SP@I and SP@I-Cy5.5) were purified by centrifugal filtration using an Amicon Ultra ultrafiltration tube (MWCO = 10 kDa) and washed three times with double-distilled water until no drug residue was detected in the filtrate.

The successful PEGylation and SS31 conjugation were confirmed by ¹H NMR spectroscopy (Bruker DRX 600, USA) and analyzed using MestReNova software (version 11.0.4). The hydrodynamic diameter (Dh) and zeta potential were measured using a Zetasizer Nano ZS instrument (Malvern, UK) at 25 °C. Each sample was equilibrated for 120 s before measurement, and data were collected from six repeated runs. The morphology and size of the nanoparticles were further characterized by transmission electron microscopy (TEM; Hitachi HT7800, Japan). The drug loading efficiency was determined by high-performance liquid chromatography (HPLC).

HT22 cells, a mouse hippocampal neuron cell line, were provided by the Central Laboratory of the Third Affiliated Hospital of Sun Yat-sen University. Cells were plated in 96-well plates at a density of 1 × 10⁴ cells per well and cultured for 24 h. Upon reaching confluence, the medium was exchanged with 100 µL of fresh DMEM (Thermo Fisher Scientific, USA) containing graded concentrations of RSL3 (Selleckchem, USA), and cells were incubated for an additional 12 h. CCK-8 reagent (TargetMol, China) was prepared following the manufacturer’s protocol. Subsequently, 100 µL of the CCK-8 working solution was added to each well, and incubation proceeded for 2 h. Absorbance at 450 nm was recorded using a microplate reader to determine the IC₅₀ value of RSL3.

HT22 cells were seeded into 96-well plates at 7 × 10³ cells per well and incubated for 24 h. The cells were then treated with 100 µL of different concentrations of SP@I, all containing a fixed concentration of RSL3 (2 µM), for 12 h. After treatment, the medium was discarded, and the CCK-8 working solution was added. Following 2 h of incubation, the OD at 450 nm was measured using a microplate reader to calculate cell viability.

The mitochondrial targeting of the therapeutics was assessed using laser scanning confocal microscopy. HT22 cells (2.8 × 10⁴ cells/well) were seeded and cultured for 24 h, followed by incubation with 1 µM SP@I-Cy5.5, P@I-Cy5.5, or free Cy5.5 for 2, 6, or 10 h. Cells were subsequently stained with MitoLite™ Green (2 mL in DMEM) for 1 h to label mitochondria. Confocal imaging was performed with excitation/emission settings at 498/520 nm for mitochondria (green) and 640/710 nm for therapeutics (red).

HT22 cells were seeded in 96-well plates at 10 × 10³ cells per well and incubated for 24 h. The cells were then treated with 100 µL of 1 µM SP@I or P@I, each containing 2 µM RSL3, for 12 h. After treatment, cell viability was evaluated using the CCK-8 assay as described above.

HT22 cells were seeded in 6-well plates at 1.6 × 10⁵ cells per well and incubated for 24 h. The cells were then divided into four groups: (1) untreated Control, which received 2 mL of fresh medium alone; (2) RSL3-only group, which received 2 mL of medium containing 2 µM RSL3; (3) SP@I group, which received 2 mL of medium containing 1 µM SP@I and 2 µM RSL3; and (4) P@I group, which received 2 mL of medium containing 1 µM P@I and 2 µM RSL3. All groups were incubated for 12 h. Following treatment, the cells were incubated with 1 µM MitoSOX Red reagent for 30 min in the dark. After being washed three times with phosphate buffered saline (PBS), the cells were collected and analyzed by flow cytometry to quantify mitochondrial superoxide levels.

All animal experiments were approved by the Laboratory Animal Welfare and Ethics Committee of Jinan University (approval no. GZJLAWE-20231124-03). Female C57BL/6J mice (9-10 weeks, 20-22 g) were purchased from Beijing HFK Bioscience Co., Ltd. A total of 33 mice were used in this study. Specifically: For the pharmacokinetic study, 9 mice were randomly assigned to three groups (n = 3 per group): free Cy5.5, P@I-Cy5.5, and SP@I-Cy5.5. For in vivo imaging and biodistribution, 12 mice were randomly assigned to four groups (n = 3 per group): PBS, free Cy5.5, P@I-Cy5.5, and SP@I-Cy5.5. Mice were imaged at 1, 3, 6, 12, and 24 h post-injection and then euthanized at 24 h for ex vivo imaging of the spinal cord and brain. For the therapeutic study, 12 mice were randomly assigned to four groups (n = 3 per group): Sham, PBS, P@I, and SP@I. Mice were anesthetized with 4% sodium pentobarbital (50 mg/kg, i.p.)^[47] for all surgical procedures. A laminectomy was performed at T10 to expose the spinal cord, followed by contusive injury at the same segment using a Louisville Injury System impactor (0.6 mm drop height, 0.6 s dwell time). Post-operatively, gentamicin (10 mg/kg; Guangzhou Tianxin Pharmaceutical, China) was given subcutaneously for 3 days. Mice were housed under standard conditions (22-24 °C, 12 h light/dark cycle), and bladders were manually expressed twice daily until spontaneous voiding resumed. Sham-operated mice and PBS controls received 100 µL of PBS via the tail vein. Treatment groups (SP@I, P@I) received equivalent idebenone doses (5 mg/kg) via tail vein once daily for 7 consecutive days.

The basso mouse scale (BMS) consists of 0 to 9 points, where 0 indicates no movement of any kind observed and 9 indicates normal. Using a single-blind method, the average score of the two hindlimbs recorded by the two observers was considered as the BMS of the sample. BMS scores of each treatment group were assessed on the 1st, 3rd, 5th, and 7th day, and once a week until 8 weeks after model establishment, while videos were recorded on day 1, 7, 14, 28, and 56 post-operation.

For the pharmacokinetic study (n = 9), mice were administered free Cy5.5, P@I-Cy5.5, or SP@I-Cy5.5 via tail vein injection post-operatively. For P@I-Cy5.5 and SP@I-Cy5.5, the idebenone-equivalent dose was 5 mg/kg. The free Cy5.5 group received an equivalent Cy5.5 dose matching the Cy5.5 content in the SP@I-Cy5.5 formulation. Blood samples were collected via the tail vein or retro-orbital plexus into heparinized tubes at pre-dose, 5 min, and 1, 3, 6, 12, 24, and 48 h post-injection. After centrifugation, plasma fluorescence intensity was measured at 640/710 nm (Ex/Em) using a multifunctional microplate reader.

For the in vivo imaging and biodistribution study (n = 12), mice were administered PBS, free Cy5.5, P@I-Cy5.5, or SP@I-Cy5.5. The P@I-Cy5.5 and SP@I-Cy5.5 groups received an idebenone-equivalent dose of 5 mg/kg, while the free Cy5.5 group received a Cy5.5 dose equivalent to that of the SP@I-Cy5.5 group. Mice were anesthetized and imaged at 1, 3, 6, 12, and 24 h post-injection using a small-animal in vivo imaging system (excitation: 640 nm). At 24 h, these mice were euthanized, and the spinal cord and brain were harvested for ex vivo imaging.

The gait of each group of mice was analyzed using the CatWalk XT at day 56. Deficits in hindlimb motor function were assessed by the animal’s ability to walk on a 50 cm-long runway. Footprints were recorded and transferred to a computer via a connected camera. Footprints were analyzed using XT version 9.0.

On day 56, the spinal cord segment centered on the injury site was harvested and fixed in 4% paraformaldehyde for 24 h. After dehydration and paraffin embedding, the tissue was sectioned into 4 µm thick slices. The sections were deparaffinized, rehydrated, and stained with an aniline blue solution for 10 min. Finally, the sections were dehydrated, cleared, and mounted. Images were captured using an upright light microscope.

Spinal cord paraffin sections adjacent to those used for Nissl staining were prepared. After deparaffinization and antigen retrieval, the tissue sections were encircled with a hydrophobic barrier pen and blocked with 2% bovine serum albumin for 30 min. The sections were then incubated overnight at 4 °C with a mixture of primary antibodies: rabbit anti-NeuN (GB11096, Servicebio, China) and goat anti-GFAP (GB13138-1, Servicebio, China). After washing, the sections were incubated for 50 min at room temperature in the dark with a mixture of appropriate fluorescent secondary antibodies. Finally, the sections were mounted with an anti-fade mounting medium. Images were captured using a fluorescence microscope. The fluorescence intensity of NeuN-positive and glial fibrillary acidic protein (GFAP)-positive areas was quantified using ImageJ software.

In vitro cytotoxicity: The cytotoxicity of SP@I and P@I was evaluated on HT22 cells. Cells were incubated with each formulation at concentrations of 0.1, 0.5, 1, 5, 10, and 15 µM for 24 h. Subsequently, cell viability was assessed using the CCK-8 assay according to the manufacturer’s protocol. After incubating with the CCK-8 solution for 2 h, the optical density (OD) at 450 nm was measured with a microplate reader, and the cell viability rate was calculated.

Body weight monitoring: The body weights of the mice were recorded before anesthesia (baseline) and then monitored weekly until 8 weeks after surgery.

Histopathological analysis: On day 56, mice were deeply anesthetized with 4% sodium pentobarbital (50 mg/kg, i.p.) and perfused via the left ventricle with 50 mL of PBS followed by 50 mL of 4% paraformaldehyde. Major organs (heart, liver, spleen, lungs, kidneys, and brain) were harvested, fixed in 4% paraformaldehyde for 24 h, dehydrated, paraffin-embedded, and sectioned at 4 µm. Sections were stained with hematoxylin and eosin (H&E), mounted, and examined under an upright light microscope (Nikon, Japan) for morphological evaluation.

Data are expressed as mean ± standard deviation (SD) (n = 3 unless otherwise stated). Two-way repeated-measures ANOVA followed by Bonferroni's post-hoc test was used to evaluate group differences over time in BMS scores, body weights, and plasma fluorescence intensity in the pharmacokinetic study. One-way analysis of variance (ANOVA) with Tukey's multiple comparison test was applied for other comparisons. Pharmacokinetic parameters, including Tmax, terminal half-life (t₁/₂), and mean residence time (MRT), were calculated by non-compartmental analysis using Phoenix WinNonlin. The terminal half-life was calculated as ln(2)/λz, where λz is the terminal elimination rate constant determined by linear regression of the terminal phase of the log-transformed plasma concentration–time curve. Fluorescence intensity from in vitro confocal imaging and ex vivo tissue imaging was quantified using ImageJ and expressed as mean fluorescence intensity (a.u.). A significance level of P < 0.05 was adopted.

P₁M₅ was synthesized via an acylation reaction between anhydride groups on poly (maleic anhydride-alt-1-octadecene)(PMHC₁₈) and amino groups on 5 k mPEG-NH₂^[48]. The characteristic peaks of PMHC₁₈ and PEG were detected at chemical shifts of 1.1-1.4 ppm and 3.5-3.9 ppm, respectively [Supplementary Figure 1]. The PEG conjugation efficiency on PMHC₁₈ was calculated to be 12.43% based on ¹H nuclear magnetic resonance (¹H NMR). To improve the bioavailability of poorly water-soluble idebenone, the drug was subsequently encapsulated with P₁M₅ at a mass ratio of 1:16 (Idebenone: P₁M₅) to fabricate polymeric P₁M₅@Idebenone (P@I) micelles [Supplementary Figures 2 and 3]. Both P@I and SP@I solutions showed no significant changes in size during incubation for 7 days at room temperature [Supplementary Figure 4], confirming remarkable stability for poorly water-soluble idebenone after well-designed formulation, which is conducive to further in vivo experiments.

The synthetic route for SP@I micelles is illustrated in Figure 1A. To confer mitochondrial targeting to P@I micelles, they were modified with the mitochondrial-targeting SS31 peptide by conjugating it to carboxyl groups on P@I via EDC/NHS-activated amidation, yielding SP@I micelles [Figure 1A]. Dynamic light scattering (DLS) measurements showed that the average hydrodynamic diameter increased from approximately 9 nm for P@I to 16 nm for SP@I [Figure 1B], which was attributed to successful surface conjugation of SS31 peptides. The zeta potential of SP@I micelles was nearly electrically neutral (-0.2 mV), much higher than that of P@I (-8.3 mV) [Figure 1C], further verifying the successful modification with the positively charged SS31 peptide. Transmission electron microscopy (TEM) images showed that both P@I and SP@I micelles possessed uniform, spherical morphologies [Figure 1D]. Fourier-transform infrared spectroscopy (FT-IR) spectra of SP@I displayed new characteristic stretching signals at 1,245 cm^-1, 1,559 cm^-1, and 1,654 cm^-1, corresponding to amide bonds formed during SS31 peptide conjugation with P₁M₅@Idebenone^[49,50] [Figure 1E]. Moreover, characteristic ¹H NMR peaks of SS31 were observed at approximately 5.4-5.8 ppm and 6.1-6.6 ppm, confirming successful conjugation of SS31 onto P@I micelles. The conjugation efficiency of the SS31 peptide to P@I was quantified to be 7.56% by 1H NMR [Figure 1F].

Ferroptosis accelerates neuronal death and functional loss in SCI^[12], and this pathological process is dominated by mitochondrial ROS overproduction^[35]. Accordingly, normalizing ROS in neuronal mitochondria would inhibit ferroptosis and rescue neurons. We first validated the mitochondrial targeting capability of SS31-modified micelles. HT22 cells were co-incubated with MitoLite^TM mitochondrial tracker dyes and Cy5.5-labeled P@I or Cy5.5-labeled SP@I. Confocal fluorescence images were then collected at 2, 6, and 10 h post-incubation, respectively [Figure 2A]. Strong overlap between green fluorescence from MitoLite^TM and red fluorescence from Cy5.5-labeled SP@I was observed at 6 h after co-incubation, verifying the efficient mitochondrial accumulation of SP@I micelles. In comparison, the P@I group exhibited only faint intracellular red fluorescence with a significantly lower fluorescence intensity, indicating negligible mitochondrial uptake of non-targeted P@I micelles. Collectively, these results confirm the precise mitochondrial targeting capacity conferred by the surface-grafted SS31 peptide.

An in vitro ferroptosis model was established by incubating HT22 cells with ferroptosis inducer RSL3 for 12 h. The half-maximal inhibitory concentration (IC50) of RSL3 was determined to be 1.95 μΜ [Figure 2B]. Cytotoxicity assays demonstrated that P@I and SP@I micelles exhibited no obvious toxic effects on HT22 cells within the concentration range of 0-15 μM [Figure 2C], confirming satisfactory biocompatibility. To further evaluate the therapeutic effect of SP@I, the cell viability was assessed after HT22 cells were co-treated with 2 μM RSL3 and different concentrations of SP@I for 12 h. SP@I exerted a dose-dependent anti-ferroptosis effect at concentrations below 1 μM [Figure 2D], and the cell viability reached 96.23% at 1 μM SP@I, signifying the prominent neuroprotective function. In contrast, cell viability was only 70.94% after treatment with the same concentrations of RSL3 and P@I [Figure 2E]. This suggests that the mitochondrial-targeting peptide SS31 could improve ferroptosis resistance, probably by facilitating uptake of idebenone formulations by mitochondria.

Mitochondrial superoxide homeostasis plays a vital role in normal cellular physiological functions, and excessive mitochondrial superoxide can mediate ferroptosis^[35,51]. We further studied whether SP@I could reduce mitochondrial superoxide in an in vitro ferroptosis model, which features excess superoxide. Mitochondrial superoxide was detected and quantified after HT22 cells were incubated with RSL3+SP@I, RSL3 +P@I, and RSL3 for 12 h, respectively. MitoSOX in HT22 cells was quantified by flow cytometry [Figure 2F and G]. HT22 cells treated with RSL3 showed the highest mitochondrial superoxide levels. The P@I treatment reduced mitochondrial superoxide levels by approximately 49%. Notably, the SP@I group exhibited mitochondrial superoxide content comparable to that of normal untreated cells. These results verify that the synergistic effect of SS31-mediated mitochondrial targeting and idebenone-based antioxidant activity enables SP@I to efficiently eliminate mitochondrial superoxide and inhibit neuronal ferroptosis.

Having validated the favorable mitochondrial targeting and anti-ferroptosis capacities of SP@I in vitro, we then investigated the in vivo pharmacokinetics and tissue accumulation of SP@I micelles. SCI mice were modeled by injury at the thoracic level 10 (T10). SCI mice were intravenously administered with SP@I-Cy5.5, P@I-Cy5.5, and free Cy5.5. Strong fluorescence was observed at the injured spinal cord and brain at 1, 3, 6, 12, and 24 h post-injection with SP@I-Cy5.5, indicating effective accumulation of SP@I-Cy5.5 [Figure 3A]. In contrast, markedly lower fluorescence signals were detected at the injured spinal cord and brain in the control or P@I-Cy5.5 treatment groups.

Subsequent blood pharmacokinetic analysis based on temporal blood fluorescence intensity changes revealed distinct circulation profiles among groups [Figure 3B]. The blood half-lives of SP@I-Cy5.5, P@I-Cy5.5, and free Cy5.5 were approximately 49.24 ± 5.32, 24.52 ± 2.58, and 18.72 ± 0.55 h, respectively, demonstrating that SS31 modification significantly prolonged the systemic circulation time of micellar nanoparticles. The prolonged circulation time of SP@I prompted us to explore its potential enrichment in the injured spinal cord and brain. Mice were sacrificed at 24 h post-injection, and the spinal cord and brain were harvested for ex vivo fluorescence quantification. Consistent with in vivo observations, ex vivo imaging and quantitative fluorescence analysis confirmed remarkably enhanced accumulation of SP@I in the injured spinal cord and brain [Figure 3C-E]. In the spinal cord, the mean fluorescence intensities were 12.61 × 10⁷ (a.u.) for SP@I-Cy5.5, 6.55 × 10⁷ (a.u.) for P@I-Cy5.5, 3.73 × 10⁷ (a.u.) for free Cy5.5, and 0.53 × 10⁷ (a.u.) for PBS [Figure 3D]. In the brain, the corresponding values were 19.34 × 10⁷ (a.u.) for SP@I-Cy5.5, 8.01 × 10⁷ (a.u.) for P@I-Cy5.5, 7.40 × 10⁷ (a.u.) for free Cy5.5, and 1.28 × 10⁷ (a.u.) for PBS [Figure 3E]. These findings confirmed that modification of P@I-Cy5.5 with the SS31 peptide could prolong pharmacokinetics with advantageous neuronal tissue targeting and accumulation, providing a favorable prerequisite for subsequent SCI therapeutic intervention.

In the following, we evaluated the therapeutic outcome of SCI mice with each treatment. Standard contusive SCI mice were established and randomly grouped. SCI mice were intravenously administered SP@I, P@I, and PBS for 7 consecutive days, while sham-operated mice without spinal cord injury were set as the positive control. The Basso mouse scale (BMS) test was used to evaluate the motor function of mice in each group during the 56-day observation period^[52]. BMS scores in SP@I and P@I showed varying degrees of recovery [Figure 4A]. On post-operative day 7, the average BMS score of the SP@I group reached 2.67, which was significantly higher than that of the P@I and PBS groups, indicating early functional recovery. By day 56, the SP@I group reached a BMS score of 6, representing nearly complete motor functional restoration. In contrast, the PBS-treated SCI group only obtained a final BMS score of 0.66, and the P@I group exhibited limited functional improvement with a score of merely 3. Moreover, assessment of the mice’s gait and locomotion patterns provided reliable insights into their motor function recovery after SCI^[53]. The mice in the SCI group walked with dragging hindlimbs even on day 56. Their heels could not even effectively touch the ground, showing barely visible footprints. No obvious hindlimb footprints were found in the P@I treatment group on day 56, and only weak average contact area and support strength of the hindlimb were detected. Notably, the SP@I group exhibited remarkably enhanced hindlimb contact area and mechanical support, as well as coordinated forelimb and hindlimb movements with regular gait patterns [Figure 4B-D].

Subsequently, mice were sacrificed for histological evaluation of spinal cord tissue. The spinal cord tissues from the SP@I treatment group showed more intact and regular tissue morphology [Figure 4E]. Consistently, Nissl staining results further confirmed integrated spinal cords with negligible cystic cavities in the SP@I treatment group [Figure 4F]. In comparison, the P@I and PBS groups displayed incomplete tissue structures and prominent lesion cavities. Collectively, these results indicated that SCI mice treated with SP@I could regain integrated spinal cord structures and effectively accelerate their motor function recovery.

To further verify neuronal regeneration following different treatments, spinal cord tissues were immunofluorescently stained with anti-NeuN and anti-GFAP antibodies, labeling neurons and astrocytes, respectively. The SP@I group exhibited robust NeuN fluorescence intensity comparable to the sham group and markedly higher than the P@I and PBS groups, indicating substantial neuronal regeneration and effective spinal cord repair. Meanwhile, the fluorescence of GFAP in P@I and PBS treatment groups was significantly higher than that of SP@I treatment and the sham group [Figure 5A-C], suggesting excessive astrocyte activation and glial scar formation that impedes endogenous neuronal regeneration after injury. In addition, weight was monitored during the treatment period. Mice in the SP@I group weighed more than mice in the P@I and PBS groups, also confirming better functional recovery [Figure 5D]. No abnormal pathology was observed in the heart, liver, spleen, lung, kidney, and brain of H&E-stained tissues in all treatment groups [Figure 5E]. These results suggested favorable biocompatibility and better recovery after SP@I treatment.

The modest sample size and use of only female mice may limit generalizability. Moreover, the detailed molecular pathway of SP@I-mediated ferroptosis inhibition was not fully elucidated. In addition, long-term efficacy and safety beyond 56 days remain to be determined. Furthermore, the mechanism of blood-spinal cord barrier penetration requires further characterization. Nonetheless, our findings provide proof of concept for mitochondria-targeted idebenone delivery in SCI therapy.