Cell-inspired dual-tandem catalytic nanoreactors drive targeted hearing protection

Abstract

Sensorineural hearing loss is driven by disruption of cochlear homeostasis, where dysregulated redox, ionic, and metabolic factors lead to hair cell degeneration and synaptic injury. Current strategies lack the capacity to restore the enzymatic clearance of reactive oxygen species or penetrate the blood-labyrinth barrier. Here, we report a cell-inspired dual-tandem catalytic nanoreactor (HSC@Se) integrating GPx-mimetic diselenide with superoxide dismutase and catalase to reconstruct the detoxification cascade. We show that the albumin scaffold stabilizes the nanoarchitecture and enables active blood-labyrinth barrier penetration of neutrophil-adhesive transport and megalin-mediated uptake, facilitating cochlear delivery to hair cells and spiral ganglion neurons. In noise-induced mouse models, we demonstrate that HSC@Se restores redox equilibrium, suppresses ferroptosis, normalizes iron handling, preserves hair-cell integrity, maintains ribbon synapses, and reduces auditory threshold shifts. We perform transcriptomics to reveal activation of antioxidative and metabolic repair pathways. This work presents cell-inspired nanoreactors that emulate natural antioxidant networks to rebalance cochlear redox homeostasis and drive hearing protection.

Here, the authors present a biomimetic nanoreactor that mimics natural antioxidant enzymes, rebuilds cochlear redox homeostasis and protects against noise‑induced hearing loss by preserving hair cells and synaptic function.

Introduction

Sensorineural hearing loss (SNHL) is a highly prevalent and debilitating sensory disorder that imposes major clinical and socioeconomic burdens. In addition to impairing communication, SNHL is strongly associated with accelerated cognitive decline and contributes substantially to dementia risk^1,2. Its incidence continues to increase with population aging, cumulative noise exposure, ototoxic drug use, and age-related cochlear degeneration³. Current clinical interventions, primarily limited to hearing aids or cochlear implants, restore auditory perception only mechanically and do not prevent the progressive loss of cochlear sensory and neural structures, highlighting the need for therapies that directly protect and restore function⁴.

The integrity of cochlear function relies on the coordinated regulation of ion gradients, cellular metabolism, and redox homeostasis^5–7. Oxidative stress and apoptosis are well‑established mechanisms driving sensory hair cell damage in noise‑induced hearing loss. Acoustic overexposure and ototoxic injury rapidly elevate reactive oxygen species (ROS; O₂^•-, H₂O₂, •OH) levels⁸, and trigger interconnected pathological cascades involving excitotoxicity⁹, neuroinflammation¹⁰, and mitochondrial dysfunction⁸. In addition, emerging evidence suggests that ferroptosis, a lipid peroxidation-driven cell death modality, may also contribute to hair cell and synaptic degeneration¹¹. Redox stability is normally maintained by a hierarchical enzymatic network comprising superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GPx)¹². SOD and CAT eliminate superoxide and hydrogen peroxide, whereas GPx uniquely reduces lipid hydroperoxides and suppresses ferroptotic injury. Impairment of these antioxidant systems substantially increases vulnerability to noise- and age-related hearing loss^13–16. Despite extensive efforts to modulate oxidative stress, therapeutic translation remains limited. Small-molecule antioxidants exhibit poor stability, short circulation time, and extremely low cochlear bioavailability due to the restrictive blood- labyrinth barrier (BLB)^17–19. Moreover, most single-agent antioxidants cannot recapitulate the multistep detoxification architecture required to interrupt ROS cascades effectively^20–24. Nanozymes offer catalytic robustness and tunability²⁵, yet many lack sufficient selectivity, biocompatibility, or the capacity to mimic the compartmentalized, cascade-type enzymatic reactions characteristic of physiological antioxidant systems²⁶. These challenges underscore the need for a multifunctional catalytic platform capable of (i) replicating endogenous antioxidant hierarchy, (ii) functioning stably in the cochlear microenvironment, and (iii) traversing the BLB for targeted intervention.

To address these challenges, we engineered a cell-inspired dual-tandem catalytic nanoreactor (HSC@Se) that reconstructs key features of natural antioxidant networks (Fig. 1a). Human serum albumin (HSA) was used as a structural scaffold to co-load SOD and CAT, forming hybrid enzymatic nanoparticles (HSC) that reestablish sequential ROS detoxification. Introduction of diselenide (Se-Se) linkers endowed the platform with GPx-like catalytic activity, yielding HSC@Se with dual SOD-CAT and SOD-GPx-mimetic pathways (Fig. 1b). This configuration enables broad-spectrum ROS elimination, lipid peroxidation control, and ferroptosis inhibition in noise-induced hearing loss (NIHL) (Fig. 1c). Compared to other common biocompatible nanoreactors (e.g., silica nanoreactors), HSC@Se is synthesized through a mild self-assembly and dual-crosslinking strategy, therefore avoiding complex synthesis processes. Structurally, it combines high stability with H₂O₂ responsiveness, addressing the common issue of insufficient physiological stimulus sensitivity found in other nanoreactors. Functionally, HSC@Se leverages the inherent biocompatibility of HSA to achieve efficient cellular targeting and uptake, a key advantage of the HSC@Se system²⁷. HSA facilitates two endogenous transport mechanisms, neutrophil-adhesive-mediated chemotaxis and megalin-mediated uptake at the vascular plexus, allowing active BLB traversal and precise delivery to sensory hair cells and spiral ganglion neurons. Protein-protein docking further supports robust HSA-megalin interaction, validating the targeting strategy²⁸ (Fig. 1d). Transcriptomic profiling reveals the molecular pathways through which HSC@Se regulates redox repair, ferroptotic suppression, and metabolic recovery. In vivo, the nanoreactor effectively penetrates the BLB and significantly improves auditory thresholds in NIHL models. By uniting biomimetic catalytic architecture with dual-tandem redox regulation and endogenous transport-mediated cochlear targeting, HSC@Se offers a powerful and versatile strategy for treating oxidative-stress-driven cochlear degeneration.

Fig. 1. Schematic illustration of the dual-tandem catalytic mechanism of the HSC@Se nanoreactor for treating noise-induced hearing loss (NIHL).

a HSC@Se is fabricated by crosslinking HSA to co-encapsulate SOD and CAT, followed by the incorporation of diselenide (Se-Se) linkers to impart GPx-mimetic activity, forming a biomimetic hierarchical antioxidant architecture. b The nanoreactor reproduces the sequential enzymatic detoxification cascade, SOD-mediated conversion of O₂^•- to H₂O₂, CAT- and GPx-like reduction of H₂O₂ and lipid hydroperoxides, achieving efficient, compartmentalized ROS elimination. c This dual-tandem catalytic synergy interrupts key mechanistic targets in NIHL, including excessive ROS accumulation, oxidative damage, excitotoxic stress, and synaptic injury. d By blocking lipid-peroxidation-driven ferroptosis and restoring cochlear redox equilibrium, HSC@Se protects sensory hair cells and spiral ganglion neurons from noise-induced degeneration. Created in BioRender. Xu, B. (2026) https://BioRender.com/twp8wku.

Results

Fabrication and characterization of HSC@Se

HSC@Se nanoreactors were constructed through the co-assembly of HSA, SOD, and CAT, followed by covalent stabilization using a selenide-containing diselenide linker (Fig. 2a). The introduction of the Se-Se crosslinker produced a compact protein-based architecture in which the enzymes were confined within a stable, nanoscale matrix. The linker not only reinforced structural integrity but also endowed the hybrid with selenium-dependent catalytic capabilities. The successful synthesis of the diselenide linker was confirmed by ¹H NMR (Supplementary Fig. 1). Transmission electron microscopy (TEM) showed that HSC@Se forms uniform spherical nanoparticles with diameters of approximately 80–100 nm. The well-defined morphology indicates that the protein crosslinking process was well-controlled, preventing the formation of large aggregates or irregular structures. Dynamic light scattering (DLS) results showed a hydrodynamic size of approximately 169 nm, with uniform dispersion. In contrast, HSC prepared without the diselenide crosslinker formed irregular aggregates with heterogeneous sizes, and their hydrodynamic size distribution also exhibited polydispersity (Fig. 2b, and Supplementary Fig. 2), clearly highlighting the crucial role of the diselenide crosslinker in forming a stable structure.

Fig. 2. Characterization of HSC@Se.

a Schematic illustration of HSC@Se, a hybrid protein nanoreactor integrating SOD and CAT via glutaraldehyde crosslinking and further modified with diselenide (Se-Se) linkers to establish dual-tandem catalytic activities (SOD-CAT and SOD-GPx-like) for efficient ROS detoxification. Created with BioRender (www.biorender.com). b TEM image of HSC@Se. c EDX elemental mapping confirming Se incorporation. The experiment was repeated three times independently, and the results were consistent. d XPS survey spectrum of HSC@Se. e High-resolution Se 3d X-ray photoelectron spectrum of HSC@Se. f UV-vis-NIR spectra of HC@Se-FL, HS@Se-R6G and HSC@Se. g Hydrodynamic size of freshly prepared HSC@Se and its hydrodynamic size after storage at 4 °C for 30 days. The illustration shows digital photographs of HSC@Se dispersed in ddH₂O, PBS, and culture medium. h High-resolution Se 3d X-ray photoelectron spectrum of HSC@Se after 30 days of storage at 4 °C. i Zeta potential of different nanoparticles (n = 3 for each group). Data are shown as mean ± SD. Source data are provided as a Source Data file.

The homogeneous distribution of selenium observed in energy-dispersive spectroscopy (EDS) mapping (Fig. 2c) suggests efficient integration of the diselenide linker throughout the protein network. X-ray photoelectron spectroscopy (XPS) analysis provided further confirmation of selenium incorporation, with clear Se 3d signatures (Fig. 2d). The characteristic Se 3d_5/2 and Se 3d_3/2 doublets at 55.48 and 56.18 eV correspond to Se-Se bonds (Fig. 2e), demonstrating preservation of the chemical motifs required for GPx-like activity²⁹. Fourier-transform infrared spectroscopy (FTIR) spectra revealed distinct vibrational bands associated with Se-Se (≈550 cm⁻¹) and C-O (≈1126 cm⁻¹)³⁰ (Supplementary Fig. 3), reinforcing this conclusion.

To confirm the co-assembly of the dual-enzyme nanoreactor, SOD and CAT were fluorescently labeled with BODIPY R6G-NHS and BODIPY FL-NHS, respectively, producing HS@Se-R6G and HC@Se-FL, respectively. UV-vis spectroscopy showed characteristic peaks of ~530 nm and ~509 nm for HS@Se-R6G and HC@Se-FL. The final hybrid, HSC@Se, displayed both peaks, indicating successful integration of the two enzymes within the glutaraldehyde- and diselenide cross-linked framework (Fig. 2f). This co-localization is crucial for enabling cooperative catalytic activity; segregated or weakly bound enzymes would be expected to exhibit diminished cascade efficiency.

HSC@Se exhibits good dispersibility under physiological environments. The nanoparticles were uniformly dispersed in ddH₂O, PBS, and culture medium, with a persistent Tyndall effect (Fig. 2g, and Supplementary Fig. 4). Notably, after storage at 4 °C for 30 days, TEM image demonstrated that the nanoparticles maintained a uniform spherical morphology without deformation or degradation. This observation was consistent with DLS measurements, which showed that HSC@Se remained stable with a narrow size distribution (Fig. 2g, and Supplementary Fig. 5). Moreover, XPS analysis verified the sustained presence of selenium in HSC@Se after storage (Fig. 2h), demonstrating a high level of structural stability.

Zeta potential measurements further clarified the sequential construction and redox responsiveness of the nanoreactor (Fig. 2i). Native HSA exhibited a surface charge of −14.20 ± 1.73 mV, whereas introduction of the diselenide linker produced a markedly more negative potential in H@Se (−18.91 ± 1.01 mV), indicating successful incorporation of the selenium-containing crosslinker and altered surface chemistry. Enzyme loading had only a modest effect, with HSC showing a potential of −15.04 ± 0.95 mV, whereas subsequent Se-Se reinforcement again shifted the charge toward a more negative value in HSC@Se (−18.73 ± 0.46 mV). Upon exposure to H₂O₂, the surface potential returned to −14.83 ± 0.37 mV, closely matching the uncrosslinked HSC intermediate, demonstrating cleavage of the Se-Se bonds. This reversible shift underscores the redox-adaptive nature of the nanoreactor, suggesting that oxidative conditions promote structural relaxation that may facilitate catalytic accessibility and enhance reactivity toward ROS. Notably, the use of two types of crosslinkers endows HSC@Se with ROS responsiveness while maintaining a stable core structure. Following co-incubation with H₂O₂, both TEM imaging and DLS analysis confirmed that cleavage of the Se-Se bonds does not compromise the core structural integrity of HSC@Se (Supplementary Fig. 6), thereby establishing a solid foundation for its dual cascading catalytic function.

Catalytic performance evaluation

To dissect the contribution of each catalytic component and assess the cooperative behavior of the dual-tandem system, we prepared a graded series of nanostructures, including H@Se (selenium-crosslinked scaffold), HS@Se (single-enzyme nanoreactor), and HSC@Se (dual-enzyme nanoreactor), allowing direct comparison across the catalytic hierarchy. HSC@Se integrates SOD and CAT within a diselenide-crosslinked protein matrix that also imparts GPx-like functionality, forming a multi-enzyme architecture capable of sequential ROS detoxification. Superoxide anion (O₂^•-) was first quantified using the riboflavin-NBT assay. HSC@Se displayed strong O₂^•- elimination in a concentration-dependent manner and achieved removal efficiencies of > 60% at higher doses, demonstrating potent antioxidant capacity (Fig. 3a). Its activity closely matched that of free SOD (Supplementary Fig. 7), indicating that enzymatic function was retained after nanoassembly. Furthermore, after trypsin treatment, HSC@Se retained higher SOD activity than free SOD, indicating that the nanoreactor formulation confers resistance to proteolytic degradation and helps preserve catalytic function under enzyme-rich conditions (Supplementary Fig. 8). CAT activity was next assessed by monitoring O₂ evolution during H₂O₂ decomposition. Using a portable dissolved oxygen meter, we observed that at a fixed HSC@Se concentration (120 μg mL⁻¹), increasing H₂O₂ from 0.125 to 4 mM produced a nearly four-fold rise in oxygen generation. The reaction gradually plateaued after 300 s, consistent with enzyme-driven kinetic saturation (Fig. 3b).

Fig. 3. Synergistic dual-tandem enzymatic catalysis of HSC@Se for enhanced ROS scavenging.

a SOD activity of HSC@Se (n = 4 for each group). b CAT activity of HSC@Se. c Confocal image of HSC@Se showing BODIPY-FL-labeled CAT (green) and BODIPY-R6G labeled SOD (red), confirming dual-enzyme co-assembly. d Schematic of Förster Resonance Energy Transfer (FRET) between the two enzyme components. Created in BioRender. Xu, B. (2026) https://BioRender.com/8c9plym. e Absorption and emission spectra of HC@Se-FL and HS@Se-R6G. f Fluorescence spectra of HC@Se + HS@Se mixture and HSC@Se, illustrating energy transfer and FL/R6G ratio. g Dual enzyme-driven cascade catalytic performance assay. h GPx-like enzyme activity of HSC@Se. i Comparative antioxidant activity of different materials (n = 4, Independent biological replicate experiments). j Quantification of total antioxidant capacity of HSC@Se at varying concentrations (n = 4, Independent biological replicate experiments). k Schematic summarizing the three enzymatic activities of HSC@Se (SOD, CAT, GPx-like) and their cooperative ROS-scavenging mechanism. Created in BioRender. Xu, B. (2026) https://BioRender.com/tynsquk. Data are shown as mean ± SD. P values, indicated on the bars, were calculated by one-way ANOVA for comparisons versus their respective controls. n.s., not significant. Source data are provided as a Source Data file.

To evaluate nanoscale enzyme co-localization, a prerequisite for efficient cascade catalysis, Förster Resonance Energy Transfer (FRET) analysis was performed. Confocal imaging showed substantial overlap between BODIPY-FL-labeled CAT and BODIPY-R6G-labeled SOD within HSC@Se (Fig. 3c, d). Spectral pairing (CAT-FL: 495/530 nm; SOD-R6G: 520/570 nm) and decreased donor emission with concurrent acceptor enhancement upon 459 nm excitation confirmed effective donor-acceptor coupling (Fig. 3e, f), indicative of close enzyme proximity. Cascade activity was further validated using the Amplex Red assays, which demonstrated rapid conversion of SOD-generated H₂O₂ by downstream catalytic sites in HSC@Se (Fig. 3g). The NADPH-coupled GSH oxidation assay indicates that H@Se leads to a significant decrease in absorbance at 340 nm, while the PBS control group remains unchanged, suggesting that H@Se may possess GPx-like activity (Fig. 3h, and Supplementary Fig. 9).

These combined results demonstrate that HSC@Se performs sequential detoxification of O₂^•-, H₂O₂, and lipid hydroperoxides, thereby preventing the accumulation of reactive intermediates. Notably, the free SOD + CAT mixture exhibited limited performance due to the absence of spatial confinement, while the HSC nanoparticles (dual-enzyme but selenium-free) showed markedly improved activity, reflecting the benefit of nanoscale co-localization. Incorporation of the diselenide linker produced an additional catalytic tier, yielding the highest overall activity in HSC@Se. Quantitatively, HSC@Se exhibited a total antioxidant capacity that was 6–8 times higher than that of the free enzyme mixture and ~1.3-fold greater than that of HSC (Fig. 3i, j). Furthermore, both the standalone Se and the simple mixed HSC+Se group exhibit some antioxidant activity; however, their antioxidant effects are substantially weaker than that of the HSC@Se group (Supplementary Fig. 10). Given the superior cascade performance of HSC@Se and the comparatively poor activity of the enzyme mixture, subsequent cellular experiments focused on nanoparticle-based systems. Collectively, HSC@Se functions as a multi-enzyme, dual-tandem nanoreactor capable of eliminating multiple ROS species through integrated SOD, CAT, and GPx-mimetic pathways. The HSA-based framework enables catalytic synergy far exceeding that of free enzymes, providing a robust platform for restoring redox balance under pathological oxidative stress in cochlear environments (Fig. 3k).

Anti-apoptotic and anti-oxidative stress effects of HSC@Se

We next examined whether the multi-enzyme HSC@Se nanoreactor could protect cochlear HEI-OC1 cells from oxidative injury. HSC@Se displayed excellent biocompatibility, with cell viability remaining above 95% at concentrations up to 120 μg mL⁻¹ (Fig. 4a, and Supplementary Figs. 11 and 12). These results indicate that incorporation of SOD, CAT, and the GPx-mimetic Se-Se linker does not introduce detectable cytotoxicity, a critical requirement for applications within the inner ear.

Fig. 4. Anti-apoptotic and anti-oxidative stress effects of HSC@Se.

a Cell viability of HEI-OC1 cells incubated with different concentrations of HSC@Se for 48 h (n = 5 for each group). b Cellular uptake of HSC@Se after different durations of uptake by confocal laser scanning microscopy (CLSM). c Cell viability of HEI-OC1 cells pre-incubated with 120 µg mL^-1 of different nanoreactors (H@Se, HS@Se and HSC@Se) for 3 h, followed by treatment with 500 µM H₂O₂ for 24 h (n = 5 for each group). d Flow cytometry showing the percentage of apoptotic cells after different treatments (n = 3 for each group). e Staining of intracellular ROS levels in different groups (n = 3 for each group, the experiment was repeated three times independently, and the results were consistent). f Flow cytometry analysis of ROS by DCFH-DA staining for each treatment group (n = 3 for each group). Data are shown as mean ± SD. P values, indicated on the bars, were calculated by one-way ANOVA for comparisons versus their respective controls. Source data are provided as a Source Data file.

Cellular internalization of BODIPY-FL-labeled HSC@Se was monitored to investigate uptake kinetics. After 2 h, fluorescence appeared mainly as discrete puncta near the cell periphery, consistent with early membrane association and initial internalization. Over time, the signal intensified and redistributed within the cytoplasm, with increased perinuclear accumulation observed by 8 h (Fig. 4b). These findings indicate that HSC@Se is efficiently internalized and trafficked intracellularly, providing the opportunity for the nanoreactor to exert its antioxidant and ferroptosis-modulating effects within target cells.

To evaluate cytoprotection, HEI-OC1 cells were exposed to increasing concentrations of H₂O₂ (200–600 μM), resulting in a steep decline in viability, with approximately 50% reduction at 500 μM (Supplementary Fig. 13). At this level of injury, neither H@Se nor HS@Se provided substantial protection, with residual viabilities of 37.15 ± 1.46% and 34.74 ± 3.48%, respectively. In contrast, HSC@Se significantly improved cell survival to 66.14 ± 4.61% (Fig. 4c). This marked enhancement reflects the integrated catalytic sequence within HSC@Se, in which SOD, CAT, and the GPx-mimetic Se linker collectively target superoxide, hydrogen peroxide, and lipid hydroperoxides. The inability of the single-enzyme or selenium-only controls to achieve similar protection underscores the importance of combining all three catalytic components within one nanosystem.

Flow cytometric analysis using Annexin V-FITC/PI staining corroborated these findings. H₂O₂ exposure increased the proportion of apoptotic cells to 88.7%, whereas untreated controls displayed only 2.55% apoptosis. Treatment with HSC@Se (120 μg mL^-1) dramatically lowered apoptosis to 2.32%, effectively restoring cells to near-baseline conditions (Fig. 4d, and Supplementary Fig. 14). This pronounced reduction indicates that HSC@Se efficiently interrupts the cascade linking oxidative stress to programmed cell death. Intracellular ROS levels were further quantified using DCFH-DA staining. Cells exposed to H₂O₂ demonstrated intense fluorescence, while those pretreated with HSC@Se showed markedly weaker signals. Flow cytometry revealed that ROS-positive cells accounted for 56.5% of the H₂O₂ group, a value that was reduced to 31.8% by HS@Se and 21.8% by HSC. In comparison, HSC@Se lowered this fraction to only 7.46% (Fig. 4e, f, and Supplementary Fig. 15). This corresponds to an 87.7% reduction in ROS accumulation, demonstrating that combined SOD-CAT catalysis with GPx-mimetic activity generates a broad and highly effective antioxidant response.

Ferroptosis suppression via dual-tandem catalysis of HSC@Se

Ferroptosis is a lipid-peroxidation-driven form of cell death marked by dysregulated Fe²⁺ handling, excessive ROS accumulation, and mitochondrial oxidative stress, features that are highly relevant to cochlear degeneration under noise exposure³¹. To model this process, HEI-OC1 cells were treated with Erastin, a System Xc⁻ inhibitor that depletes intracellular cysteine³². Erastin induced a pronounced, dose-dependent decline in viability (Fig. 5a), with survival falling to 32.49 ± 1.90% at 15 µM, indicating robust ferroptotic stress. Under the same conditions, HSC@Se and H@Se substantially improved cell viability, whereas HSC alone produced only modest protection (Fig. 5b). This differential response underscores the functional advantage of the dual-tandem catalytic configuration, in which SOD, CAT, and GPx-mimetic Se-Se linkers operate cooperatively to suppress superoxide, hydrogen peroxide, and lipid hydroperoxide accumulation, mirroring the multi-tiered antioxidant network.

Fig. 5. Ferroptosis suppression via dual-tandem catalysis of HSC@Se.

a Cell viability assays by treating HEI-OC1 cells with different concentrations of Erastin for 24 h (n = 6 for each group). b Cell viability of HEI-OC1 cells pre-incubated with 120 µg mL^-1 of different nanoreactors (H@Se, HSC, and HSC@Se) for 3 h, followed by treatment with 15 µM Erastin for 24 h (n = 6 for each group). c, d FeRhoNox-1 fluorescent probe was used to detect the intracellular Fe²⁺ level of HEI-OC1 cells after different treatments and its statistical analysis (n = 3 for each group). e, f CLSM image of mitochondrial superoxide levels by MitoSOX staining (red) and its statistical analysis (n = 3 for each group). g, h Intracellular superoxide levels of HEI-OC1 cells after different treatments, as detected by DHE fluorescent probe method (n = 3 for each group). i The mitochondrial membrane potential of HEI-OC1 cells after different treatments (n = 3 for each group). Data are shown as mean ± SD. P values, indicated on the bars, were calculated by one-way ANOVA for comparisons versus their respective controls. Source data are provided as a Source Data file.

Because ferroptosis is tightly linked to iron dysregulation, intracellular Fe²⁺ levels were assessed using FeRhoNox-1. Erastin elicited a marked increase in labile Fe²⁺, whereas HSC@Se and H@Se significantly reduced these signals, a capability not observed with selenium-free HSC (Fig. 5c, d). This reduction is consistent with the GPx-mimetic activity of the Se linker, which limits lipid hydroperoxide propagation and thereby prevents iron-dependent ROS amplification. The efficient intracellular distribution of HSC@Se and H@Se, aided by the HSA scaffold, further supports this effect by positioning the catalytic modules in proximity to iron-rich subcellular regions. Mitochondrial oxidative stress, probed with MitoSOX Red, exhibited a similar pattern. Control cells showed minimal fluorescence, whereas Erastin triggered strong mitochondrial oxidation (Fig. 5e, f). Neither H@Se nor HS@Se mitigated this response, while HSC@Se pretreatment nearly abolished mitochondrial ROS accumulation, preserving organelle-specific redox balance. This protection aligns with the cell-inspired catalytic layout of HSC@Se, which enables rapid and coordinated detoxification at mitochondrial and perinuclear sites where ferroptotic signaling converges.

Total cellular O₂^•- levels were evaluated using dihydroethidium (DHE) staining. HS@Se produced partial suppression, but HSC@Se reduced O₂^•- fluorescence to nearly baseline levels (Fig. 5g, h), demonstrating efficient elimination of upstream oxidants before their conversion into lipid-reactive intermediates. Such effectiveness reflects synergistic cascade catalysis rather than simple additive effects of individual enzymes. Mitochondrial membrane potential was subsequently examined using JC-1. Healthy cells displayed predominantly red JC-1 aggregates, whereas Erastin induced a shift to green monomers in over half of the population, signifying mitochondrial depolarization. HSC@Se treatment restored red aggregate fluorescence in > 95% of cells (Fig. 5i), indicating robust preservation of mitochondrial bioenergetics. This level of protection is consistent with the intracellular targeting and catalytic efficiency provided by the HSA-based nanosystem.

Transcriptome sequencing analysis

To elucidate the molecular mechanisms underlying HSC@Se-mediated protection, transcriptome analysis was systematically conducted. Differential gene expression (DEGs) were identified using a fold-change threshold of >2.0 or <0.5 and a p-value < 0.05, generating a robust dataset for downstream analysis. The volcano plot revealed extensive transcriptional reprogramming upon H₂O₂ exposure, while Venn diagram comparisons showed a substantial subset of stress-induced genes that were reversed following HSC@Se treatment. In total, 8285 DEGs were detected across comparisons, with many genes showing a transition from marked upregulation under oxidative stress to partial or full normalization after HSC@Se intervention (Fig. 6a–d, and Supplementary Fig. 16).

Fig. 6. Transcriptome detection for analyzing the underlying mechanism of HSC@Se treatment for NIHL.

a Principal component analysis of Control, H₂O₂ and HSC@Se. b Volcano plots showing differentially expressed genes between HSC@Se-treated and Control groups. Red, blue, and gray dots indicate up-regulated, down-regulated, and no differentially expressed genes in the HSC@Se-treated group compared to the Control group, respectively. c Wayne plot demonstrating DEGs up-regulated after H₂O₂-induced effect and down-regulated after HSC@Se treatment. d Differential gene heatmap. e GO enrichment analysis of DEGs in biological process, cellular component, and molecular function. f GO pathway enrichment analysis of DEGs. The “rich factor” represents the ratio of the number of differentially expressed genes annotated to a given term or pathway to the total number of genes annotated to that term or pathway. Pathways are prioritized the selection of terms that meet the predefined cutoff criteria (corrected p-value < 0.05) and have the highest significance and biological relevance. g Chord diagram showing the association between differentially expressed genes and enriched KEGG signaling pathways. Colors are used solely to distinguish different pathways and do not represent changes in expression. h KEGG pathway enrichment analysis of differentially expressed genes. i MDA expression level under different treatment conditions. j GPX4 expression level under different treatment conditions. Enrichment analysis was performed using two-sided Fisher’s exact hypergeometric test, and raw P-values were corrected for multiple comparisons by Benjamini–Hochberg FDR adjustment. The experiment was repeated three times independently, and the results were consistent. Source data are provided as a Source Data file.

GO analysis provided insight into the functional categories enriched among these DEGs. Terms related to signal transduction, response to stimulus, cytokine regulation, and ion transport were prominently represented (Fig. 6e, f). These categories mirror biological processes implicated in noise-related cochlear pathology, suggesting that HSC@Se acts by modulating pathways central to oxidative injury, inflammatory activation, and stress-induced cell death. Chord diagram analysis highlighted several regulatory nodes, including Akt1, Jak1, Mapk3, GPx1, GPx8, Mapk11, and Src, that were substantially altered by HSC@Se (Fig. 6g). These genes represent key components of antioxidant defense, stress-kinase regulation, and ferroptosis control. Their coordinated normalization indicates that the nanoreactor influences multiple tiers of intracellular signaling rather than acting through a single pathway. KEGG pathway enrichment further supported this interpretation. HSC@Se treatment significantly affected signaling modules linked to cellular survival and inflammatory modulation, including the PI3K-Akt pathway, JAK-STAT signaling, and arachidonic acid metabolism (Fig. 6h). These pathways are tightly associated with oxidative damage propagation, immune dysregulation, and lipid-peroxidation dynamics in NIHL, underscoring the biological relevance of HSC@Se-mediated transcriptional rescue.

To validate these transcriptomic findings at the protein level, we examined biomarkers of oxidative stress and ferroptosis via Western blotting (Fig. 6i, j, and Supplementary Fig. 17). H₂O₂ exposure markedly increased malondialdehyde (MDA) accumulation and reduced GPx4 levels relative to untreated controls, consistent with enhanced lipid peroxidation and ferroptotic signaling. HSC@Se treatment partially restored GPx4 expression and reduced MDA abundance (Supplementary Fig. 18), confirming potent antioxidant and ferroptosis-suppressive activity. These molecular signatures align with the dual-tandem catalytic functions of HSC@Se and further support its ability to rebalance redox homeostasis in HEI-OC1 cells.

In vivo bio-distribution and barrier penetration of HSC@Se

To determine whether HSC@Se could effectively reach cochlear targets in vivo in the acutely injured cochlea, Cy5.5-labeled HSC@Se was intravenously administered to noise-exposed mice immediately, with free Cy5.5 used as a control. Fluorescence imaging 24 h after injection showed stronger HSC@Se-Cy5.5 signals in the cochlea, approximately twice that of the free dye, indicating enhanced cochlear delivery of HSC@Se-Cy5.5 compared to the free dye (Fig. 7a, b). These results provide direct evidence that HSC@Se is capable of penetrating the BLB following noise trauma, a delivery challenge highlighted in the introduction as a major obstacle to the entry of antioxidants and nanomaterials into cochlear tissues.

Fig. 7. In vivo biodistribution and barrier penetration of HSC@Se.

a IVIS images of the mouse cochlea and major organs 24 h after injection. b Quantitative analysis of fluorescence intensity in the cochlea (n = 4 for each group). c Cochlear anatomy diagram. Created in BioRender. Xu, B. (2026) https://BioRender.com/qdup4cj. d ZDOCK results between HSA and megalin. e Confocal images reveal the uptake of the nanoreactors within both auditory hair cells and SGNs. Organ of Corti: Scale bar = 25 μm, SGNs: Scale bar = 100 μm. f, g Statistical analysis of the average fluorescence intensity in the cochlea, organ of Corti, and SGN (n = 3 for each group). Data are shown as mean ± SD. P values, indicated on the bars, were calculated by one-way ANOVA for comparisons versus their respective controls. Source data are provided as a Source Data file.

As illustrated schematically in Fig. 7c, the enhanced cochlear distribution of HSC@Se may be explained by two endogenous transport routes. One possible mechanism is neutrophil-associated transport, as albumin-based nanomaterials have been reported to bind neutrophils via Fcγ receptor interactions under inflammatory conditions^28,33,34. Second, megalin, an endocytic receptor abundantly expressed at the vascular plexus, binds albumin with high affinity³⁵. Protein-protein docking (ZDOCK) simulations showed that HSA forms multiple stable hydrogen bonds with megalin (Fig. 7d), indicating strong ligand-receptor complementarity. This suggests that megalin-mediated internalization may function in parallel with neutrophil adhesion to facilitate translocation of HSC@Se across the BLB. These dual transport pathways resonate with the introduction’s emphasis on leveraging endogenous receptor and immune-cell interactions to overcome the intrinsic impermeability of inner-ear barriers.

After confirming BLB penetration, we evaluated whether HSC@Se can reach the cochlear cells most susceptible to oxidative stress. Ex vivo confocal imaging of cochlear frozen sections revealed minimal autofluorescence in untreated animals. In contrast, mice treated with HSC@Se exhibited strong Cy5.5 signals. In noise-exposed mice, HSC@Se showed broad cochlear distribution, with prominent signals observed in hair cells, SGNs, and the lateral wall/stria vascularis, whereas little signal was detected in Reissner’s membrane (Fig. 7e, and Supplementary Fig. 19). These high-energy regions are more fragile and susceptible to damage, allowing HSC@Se to better target these affected areas through the neutrophil adhesion capability of HSA. Subsequent analyzes focused on these structures to assess therapeutic efficacy. The CY5.5-labeled HSC@Se co-localizes with Myo7a-positive outer and inner hair cells, as well as SGNs within the modiolus. In orthogonal views, the green fluorescence from HSC@Se overlapped with the red Myo7a signal, generating yellow merged signals within the hair-cell layer. Quantitative co-localization analysis further supported the close association of Cy5.5-labeled HSC@Se with Myo7a-positive hair cells (Supplementary Fig. 20, and Supplementary Table 1). Quantitative analysis further confirmed significantly elevated fluorescence throughout the organ of Corti and SGN regions relative to controls (Fig. 7f, g). This biodistribution pattern carries important mechanistic implications. Hair cells and SGNs, as emphasized in the introduction, possess limited intrinsic redox buffering capacity and are among the earliest cellular targets of noise-induced oxidative and ferroptotic damage. Efficient delivery of an antioxidant nanoreactor to these sites is therefore essential for restoring redox homeostasis and preventing irreversible functional decline. The ability of HSC@Se to accumulate specifically in these compartments suggests that HSA-mediated chemotaxis and receptor-assisted uptake collectively position the catalytic nanoreactor precisely at the loci of pathological ROS generation.

In vivo protective effect of HSC@Se for noise-induced hearing impairment

Excessive acoustic stimulation triggers oxidative injury, ferroptosis, synaptopathy, and neuronal degeneration within the cochlea, ultimately producing irreversible hearing loss. Given the potent ROS-scavenging and ferroptosis-suppressive properties of HSC@Se, we evaluated whether systemic administration of the nanoreactor could mitigate noise-induced hearing impairment in vivo. C57BL/6J mice (4 weeks old) were exposed to broadband noise (2–20 kHz, 114 dB SPL, 2 h) and treated before and after noise exposure with HSC@Se, five times in total over 16 days (Supplementary Fig. 21). Auditory function was assessed using auditory brainstem response (ABR) measurements, with wave I serving as a readout of cochlear output and auditory-nerve synchrony (Supplementary Fig. 22). Four-week-old mice were randomly assigned to three groups: Control (no noise exposure or treatment), NE+Saline (noise-exposed mice treated with saline), and NE + HSC@Se (noise-exposed mice treated with HSC@Se). Hearing function was confirmed to be normal before noise exposure and no significant differences were observed among groups at baseline (Supplementary Fig. 23). In saline-treated mice, noise exposure resulted in a sustained increase in ABR thresholds across all frequencies, exceeding 20 decibels on the first day post-exposure, with no recovery observed by day 14 (Supplementary Fig. 24). One day after noise exposure, the ABR threshold shifts in the NE + HSC@Se group were comparable to those in the NE+Saline group. However, during the subsequent recovery period, mice treated with HSC@Se showed improved restoration of hearing function. By day 14 post-noise exposure, threshold shifts in the NE + HSC@Se group were significantly lower than those in the NE+Saline group, remaining below 10 dB at 8, 16, 24, and 32 kHz (Fig. 8a, and Supplementary Fig. 25). These findings indicate substantial functional preservation across the frequency regions most vulnerable to acoustic trauma.

Fig. 8. Protective effect of HSC@Se on noise-induced hearing loss.

a ABR recording of mice on day 1 and day 14 (n = 4 for each group). Baseline: pre‑noise thresholds. b, c Amplitude of Wave I from ABR waveforms and latency for 60–90 dB SPL stick stimulus at 16 kHz in each group (n = 3, Independent biological replicate experiments). d Representative immunofluorescence images of hair cells in basal regions of Corti’s organ. White circles indicate missing OHCs. e H&E staining of SGNs (n = 3 for each group, the experiment was repeated three times independently, and the results were consistent). f CLSM images of ribbon synapses in inner hair cells. The whole-mount cochlear preparation was double stained for CtBP2 (red) and Homer1 (green). IHCs were labeled with Vglut3 (blue), and white arrow points to synaptic loss (n = 3 for each group). g Statistical analysis of outer hair cells (n = 3 for each group). h Quantitative analysis of SGN densities (n = 3, Independent biological replicate experiments). i Average number of ribbon synapses per IHC (n = 3, Independent biological replicate experiments). Data are shown as mean ± SD. P values, indicated on the bars, were calculated by one-way ANOVA for comparisons versus their respective controls. Source data are provided as a Source Data file.

Morphological analysis of wave I further supported this improvement. Noise exposure caused pronounced amplitude reductions and latency delays at 16 kHz, which are hallmarks of impaired neural synchrony and diminished cochlear output. HSC@Se treatment restored both parameters to near-baseline levels (Fig. 8b, c), demonstrating that the nanoreactor maintains auditory-nerve function in the wake of acoustic injury. This functional rescue aligns with the introduction’s emphasis on the necessity of restoring redox homeostasis within hair cells and SGNs to sustain neural signaling integrity. To evaluate structural preservation, cochlear whole mounts were examined by immunofluorescence. Consistent with known patterns of NIHL pathology, noise exposure resulted in substantial OHC loss in mid- to high-frequency regions^36,37. HSC@Se treatment markedly preserved OHC survival across these domains (Fig. 8d, g). H&E staining showed that SGN degeneration after noise exposure was region-dependent, with more pronounced neuronal loss in the basal turn than in the middle or apical regions. Treatment with HSC@Se attenuated this loss across the affected cochlear regions (Fig. 8e, h, and Supplementary Fig. 26).

Given that cochlear synaptopathy is an early and often irreversible component of NIHL, we examined presynaptic CtBP2 and postsynaptic Homer1 colocalization to quantify functional ribbon synapses. Noise exposure dramatically reduced the number of CtBP2/Homer1-positive puncta, particularly at 16–32 kHz. HSC@Se robustly mitigated this synapse loss, with the strongest rescue at 16 kHz and substantial protection at 24 and 32 kHz (Fig. 8f, i). This preservation of synaptic integrity is consistent with the nanoreactor’s ability to suppress ROS, lipid peroxidation, and iron-dependent injury, key drivers of synaptic degeneration highlighted in the introduction. Collectively, these results demonstrate that HSC@Se provides comprehensive protection against noise-induced auditory injury by preserving hair cell viability, maintaining ribbon synapses, rescuing SGN survival, and sustaining neural synchrony. These multilayered protective effects reflect the ability of the dual-tandem catalytic nanoreactor to counteract oxidative stress and ferroptotic signaling within the BLB-shielded cochlea, thereby targeting key pathological mechanisms underlying noise-induced hearing loss.

Molecular mechanisms underlying HSC@Se-mediated hearing protection

Oxidative stress is a well‑established central mechanism underlying NIHL, while ferroptosis, an emerging form of regulated cell death tightly linked to oxidative stress and hearing impairment, also plays a critical role in noise‑induced cochlear degeneration, as excessive ROS production and iron-catalyzed lipid peroxidation collapse membrane integrity and trigger sensory-neuronal injury^11,38,39. To determine whether HSC@Se mitigates these molecular events in vivo, we examined oxidative and ferroptotic markers in cochlear tissues following acoustic trauma.

DHE staining was used to assess ROS accumulation (Fig. 9a). Control cochleae exhibited only faint, diffuse fluorescence within the basilar membrane and spiral ganglion, consistent with low physiological ROS levels. These signals showed minimal overlap with DAPI, indicating limited nuclear oxidative stress. In sharp contrast, noise-exposed mice displayed intense DHE fluorescence, particularly within spiral ganglion neurons, a population known to exhibit heightened vulnerability to oxidative injury. The strong co-localization of DHE and DAPI suggested not only cytosolic but also nuclear ROS overload and oxidative DNA damage. Systemic HSC@Se administration attenuated ROS-associated fluorescence signals in multiple cochlear regions. Quantitative analysis showed significant suppression of superoxide signals in the spiral ganglion and stria vascularis, whereas the fluorescence intensity in the basilar membrane/organ of Corti exhibited only a downward trend and did not reach statistical significance (Fig. 9b), consistent with the dual-tandem catalytic activity described earlier. Ferroptosis also plays a key role in noise-induced cochlear degeneration, with noise exposure significantly increasing the levels of Fe²⁺ in the cochlea (Fig. 9c). In addition, the distribution of Fe²⁺ in different cochlear turns following noise exposure and HSC@Se treatment is clearly illustrated. Interestingly, noise exposure resulted in an increased distribution of Fe²⁺ in the basal turn, while the middle and apical turns were less affected. Meanwhile, HSC@Se effectively suppressed the distribution of Fe²⁺ in both the basal and apical turns, with minimal impact on the middle turn. These findings provide more comprehensive and compelling evidence for the role of ferroptosis in NIHL (Supplementary Fig. 27). HSC@Se treatment reduced Fe²⁺ signals by approximately 40–60% across these regions, with the strongest reduction observed in the spiral ganglion (Fig. 9d). This preferential impact on neuronal iron pools suggests that the nanoreactor not only neutralizes ROS but also suppresses the iron-dependent feedback amplification that fuels ferroptotic progression.

Fig. 9. Molecular mechanisms underlying HSC@Se-mediated hearing protection.

a Detection of oxidative stress levels in the cochlea by DHE staining. b Statistical analysis of integrated optical density (IOD)/area of DHE staining (n = 3, Independent biological replicate experiments). c FeRhoNox-1 staining detects cochlear Fe²⁺ levels. d Statistical analysis of IOD/area of FeRhoNox-1 staining (n = 3, Independent biological replicate experiments). e, f 4-HNE immunofluorescence staining and mean quantitative analysis of cochlear sections from each group (n = 3, Independent biological replicate experiments). g Schematic diagram of HSC@Se rescuing hair cells. Created in BioRender. Xu, B. (2026) https://BioRender.com/3jpuaab. Data are shown as mean ± SD. P values, indicated on the bars, were calculated by one-way ANOVA for comparisons versus their respective controls. Source data are provided as a Source Data file.

To further evaluate ferroptosis-related lipid damage, cochlear sections were immunostained for 4-HNE, a downstream product of polyunsaturated lipid peroxidation. Noise exposure caused substantial accumulation of 4-HNE in both SGNs and the basilar membrane (Fig. 9e), echoing the severe oxidative burden observed in DHE and FeRhoNox-1 assays. Notably, HSC@Se significantly reduced 4-HNE fluorescence intensity in the examined cochlear regions, including the stria vascularis (Fig. 9f), which indicates attenuation of lipid peroxidation associated with noise-induced cochlear injury indicating effective interception of lipid-peroxidation cascades and suppression of ferroptotic signaling. Together, these findings delineate a coherent molecular mechanism for HSC@Se-mediated otoprotection (Fig. 9g). The nanoreactor simultaneously eliminates excessive ROS through coordinated SOD-CAT-GPx-like tandem catalysis, thereby preventing oxidative stress from propagating into lipid peroxidation and oxidative DNA injury. In parallel, it reduces labile Fe²⁺ pools, disrupting the iron-dependent chain reactions that initiate and sustain ferroptosis. This dual regulation of redox balance and iron homeostasis directly targets the core pathological drivers of NIHL described in the introduction. By restoring these two interconnected axes, HSC@Se suppresses cochlear lipid peroxidation, preserves cellular viability, and interrupts the molecular cascade leading to irreversible auditory dysfunction.

Biosafety evaluation of HSC@Se

A rigorous assessment of biosafety is essential for the translational development of any nanotherapeutic. We therefore examined the auditory, hematological, and systemic tolerance of HSC@Se using in vivo assays (Supplementary Fig. 28). Longitudinal ABR measurements showed that HSC@Se administration did not alter hearing thresholds at any tested frequency relative to baseline, indicating an absence of ototoxicity at therapeutic doses (Fig. 10a). Concordantly, body-weight trajectories of treated mice remained comparable to those of untreated controls throughout the 3-week evaluation period, suggesting no impact on general health or metabolic status (Fig. 10b). Hemocompatibility was examined by in vitro hemolysis analysis, which revealed a hemolysis rate below the commonly accepted safety threshold of 1%, demonstrating excellent compatibility with red blood cells (Fig. 10c). To further assess systemic tolerance, comprehensive hematological and serum biochemical analyzes were performed. Key biochemical markers, including ALT, AST, BUN, and creatinine, remained within normal ranges, indicating preserved hepatic and renal function. Hematological indices (RBC, HGB, HCT, MCV, MCH, MCHC, PLT, MPV, PDW, WBC, LYM, and Neu) showed no abnormalities, suggesting that HSC@Se does not induce inflammation, infection, or hematologic dysregulation (Fig. 10d, e).

Fig. 10. Biocompatibility evaluation of HSC@Se.

a ABR detection of mice in HSC@Se group (n = 4 for each group). b Weight changes of mice in Saline and HSC@Se groups (n = 3 for each group). c Hemolysis evaluation of HSC@Se (n = 5 for each group). d Changes in ALT and AST in each group at 30 days post-treatment, and toxic effects of blood BUN and CREA (n = 4, Independent biological replicate experiments). e Changes in RBC, HGB, HCT, MCV, MCH, MCHC, PLT, MPV, PDW, WBC, LYM and Neu, for each group 3 weeks after treatment (n = 3 for each group). f H&E staining images of the brain, heart, liver, spleen, lungs, kidneys, and other organs on day 30 after different treatments for each group (n = 3 for each group). Data are shown as mean ± SD. Source data are provided as a Source Data file.

Histopathological examination of major organs, including brain, heart, liver, spleen, lung, and kidney, further supported the systemic safety of HSC@Se. H&E staining revealed no detectable structural abnormalities, inflammatory lesions, or tissue damage following treatment (Fig. 10f). Together, these findings demonstrate that HSC@Se possesses a favorable biosafety profile, with no evidence of ototoxicity, hemolysis, systemic toxicity, or organ-level pathology. The excellent biocompatibility and physiological tolerance observed across multiple evaluation modalities support the suitability of HSC@Se for continued preclinical development as a redox-modulating and ferroptosis-inhibiting nanoplatform for hearing preservation.

Discussion

In summary, we developed a biomimetic cascade nanoreactor (HSC@Se) that integrates natural antioxidant enzymes with a GPx-mimetic diselenide crosslinker, creating a cell-inspired platform capable of reconstructing endogenous antioxidant networks and restoring cochlear redox homeostasis. Through coordinated SOD-CAT-GPx-like tandem catalysis, HSC@Se efficiently eliminated O₂^•-, detoxified H₂O₂, and suppressed downstream lipid peroxidation, thereby preserving mitochondrial integrity and preventing oxidative damage at multiple levels.

In parallel, the nanoreactor attenuated ferroptotic signaling by limiting Fe²⁺ accumulation, reducing lipid peroxidation biomarkers (4-HNE, MDA), and maintaining mitochondrial membrane potential. These dual actions on redox balance and iron homeostasis translated into robust cytoprotection in vitro and comprehensive otoprotection in vivo. In vivo results convincingly demonstrated the protective effects of HSC@Se on auditory function, hair cells, synapses, and SGNs, and limited ABR threshold shifts to <10 dB across noise-vulnerable frequencies, demonstrating potent functional preservation following acoustic trauma. However, without direct in vivo comparisons with control groups (such as systems containing only albumin, a single enzyme construct, or only selenium), we cannot completely rule out the potential contributions of the nanoparticle scaffolds, individual components, or general delivery effects. Further research is needed to fully elucidate the specific in vivo contributions of the integrated SOD–CAT–GPx mimetic architecture. The quantitative analyzes in this study, including measurements of fluorescence intensity, are inherently semi-quantitative and may be influenced by tissue variability and regional sampling. Future studies will require a larger sample size and automated image analysis to further validate these findings.

This work establishes a dual-pathway therapeutic paradigm for NIHL that couples enzyme-based ROS detoxification with ferroptosis suppression. More broadly, it provides a generalizable strategy for constructing safe, biomimetic cascade nanoreactors capable of addressing complex oxidative stress-driven pathologies. By bridging nanobiotechnology, redox biology, and auditory medicine, HSC@Se represents a promising nanoplatform with wide potential for treating diseases rooted in dysregulated redox and ferroptotic mechanisms.

Methods

Materials and reagents

N-hydroxysuccinimide (NHS), Catalase (CAT), and 1-ethyl-(3-dimethylaminopropyl) carbodiimide (EDC) were obtained from Shanghai Yuanye Bio-Technology Co., Ltd (Shanghai, China). Superoxide dismutase (SOD), anhydrous sodium sulfate, human serum albumin (HSA), and sodium chloride were obtained from Macklin (Shanghai, China). Selenium granules were acquired from Sigma-Aldrich (Shanghai, China). Sodium borohydride and ethyl acetate were procured from Titan Scientific Co., Ltd (Shanghai, China). 3-Bromopropanoic acid was supplied by Sinopharm Chemical Reagent Co., Ltd (Shanghai, China). Hydrogen peroxide (H₂O₂) was purchased from Sinopharm Chemical Reagent Co., Ltd (Shanghai, China). Dulbecco’s modified Eagle medium (DMEM) and Fetal bovine serum (FBS) were purchased from Gibco (GrandIsland, USA). Phosphate buffer solution (PBS) was purchased from Lanjike Technology Co., Ltd (Beijing, China). L-methionine was obtained from Sigma-Aldrich (Shanghai, China). CCK-8 cell proliferation reagent was purchased from YoBiBiotech Co., Ltd (Shanghai, China). 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonate) (ABTS), riboflavin, and nitro blue tetrazolium (Nitor-TB) were obtained from Macklin (Shanghai, China). Reactive Oxygen Species Assay Kit, enhanced mitochondrial membrane potential assay kit with JC-1, Annexin V-FITC Apoptosis Detection Kit were purchased from Beyotime Biotechnology (Shanghai, China).

Synthesis of 3,3-diselenodipropionic acid di(N-succinimidyl) ester

Initially, sodium borohydride (2.8 g) and selenium (2.8 g) were placed in a cold three-neck flask equipped with a condenser, gas inlet adapter, and a drop funnel under a nitrogen atmosphere. While stirring, 150 mL of water was added. After approximately 15 min, the ice bath was removed, and a second portion of selenium (2.8 g) was added to the colorless mixture. The mixture was stirred for an additional 3 h. The resulting brown-red aqueous solution of Na₂Se₂ was cooled to room temperature, and 10 g of 3-bromopropanol was added to 50 mL of H₂O. The resulting yellow solution was stirred for 12 h. Ethyl acetate was used to extract the aqueous layer, followed by back extraction with sodium chloride. The organic layer was dried, and the solvent was concentrated under vacuum to yield the precursor powder of the diselenide crosslinker.In the second phase, the precursor powder (1 g) was dissolved in 15 mL of anhydrous tetrahydrofuran (THF) in a dual-neck flask under a nitrogen atmosphere. Subsequently, 1.1 g of succinimide was added, and the mixture was stirred continuously. After 30 min, a solution of 1-ethyl-(3-dimethylaminopropyl) carbodiimide (1.65 g) in 15 mL of anhydrous THF was introduced gradually, drop by drop, while maintaining a temperature of 0 °C. The reaction mixture was then stirred at room temperature for an extended period, continuing overnight. The mixture was then filtered to remove impurities, and the tetrahydrofuran was removed using a rotary evaporator to obtain 3,3-diselenodipropionic acid di(N-succinimidyl) ester.

Preparation of HSC@Se

Following the established protocols in previous study³⁹, the synthesis of the selenium-complexed SOD@CAT cascade catalytic nanoplatform (HSC@Se) involves two primary steps. In the first step, 1 mg of CAT, 0.5 mg of SOD, and 7 mg of HSA are dissolved in 1 mL of ultrapure water. This mixture is thoroughly combined before adding 4 mL of a 25% wt glutaraldehyde solution and an additional 8 mL of ultrapure water. The solution is then stirred at room temperature at 1150 rpm for 12 h. During the second phase, 12 mg of the diselenide crosslinker (3,3-diselenodipropionic acid di(N-succinimidyl) ester) is incorporated into the stirred solution, which continues to be maintained at room temperature and stirred at 1150 rpm for another 12 h to yield HSC@Se. Following this, the mixture undergoes dialysis in phosphate-buffered saline (PBS) at room temperature for 24 h to eliminate any unreacted reagents. The final product is then stored at 4 °C for subsequent use.

Characterization of HSC@Se

The structure and chemical composition of the HSC@Se product were determined using a nuclear magnetic resonance spectrometer (AVANCE NEO 600M, Bruker, Switzerland). The size and surface charge of HSC@Se were evaluated using dynamic light scattering (DLS) with a Zetasizer Nano ZSE from Malvern Instruments (UK). Transmission electron microscopy (TEM) images, along with energy dispersive X-ray (EDX) data and elemental analysis, were obtained using a field emission TEM from JEOL (Japan). X-ray photoelectron spectroscopy (XPS) measurements were conducted with a Nexsa G2 system from Thermo Fisher Scientific (USA). The aqueous ultraviolet-visible-near-infrared (UV-vis-NIR) absorption spectra were recorded using a UV-3600 i Plus spectrometer from Shimadzu (Japan). Absorbance at the indicated wavelengths was measured using a Varioskan LUX multifunctional microplate reader (Thermo Fisher Scientific, USA).

Nuclear magnetic resonance (NMR)

The ¹H NMR spectrum of the HSC@Se was used to characterize the successful polymerization of the diselenide crosslinker. A 5 mg sample of the diselenide crosslinker was dissolved in 500 μL of deuterated DMSO for NMR analysis. The structure and chemical composition of the product were determined from the characteristic signals in the ¹H NMR spectrum.

SOD activity of HSC@Se

The activity of SOD in catalyzing the disproportionation of superoxide anions into H₂O₂ and oxygen (O₂) was assessed using the Total Superoxide Dismutase Assay Kit with Nitor-BT (Beyotime Biotechnology, S0109). In this assay, superoxide anions (O₂•⁻) were generated through a reaction involving xanthine and xanthine oxidase. This reaction facilitates the reduction of nitro blue tetrazolium to formazan, which exhibits strong absorption at 560 nm. The reagent kit was employed to interact with various materials, and changes in absorbance at 560 nm were recorded and analyzed using an enzyme labeling instrument.

O₂^•− scavenging activity of HSC@Se in vitro

The efficiency of HSC@Se in clearing superoxide anions (O₂^•−) was assessed by calculating the photoreduction inhibition rate of nitroblue tetrazolium (Nitor-TB). To begin, a mixture was prepared by combining 300 μL of 20 μM riboflavin, 313 μL of 125 mM L-methionine, and 375 μL of 75 μM Nitor-TB in 1800 μL of phosphate-buffered saline (PBS), ensuring thorough mixing. Next, 300 μL of HSC@Se at various concentrations (0, 150, 300, 600, and 1200 μg mL⁻¹) was added to the mixture. The resulting solution, totaling 3 mL, was transferred to a container and subjected to 15 min of exposure under an LED lamp. Following this, an equal volume of chloroform was introduced, facilitating phase separation into distinct layers. The absorbance of the organic phase was measured at 560 nm using a microplate reader. This process was repeated three times. The inhibition percentage was calculated using the following formula: [100-(A_s-A_b)/(A_c-A_b)×100]%, where A_s, A_b, and A_c represent the absorbance of the sample, the blank, and the control group, respectively.

CAT activity of HSC@Se

The activity of catalase (CAT) in HSC@Se was evaluated using a portable dissolved oxygen meter (JPBJ-608) to measure the oxygen (O₂) generated from the decomposition of hydrogen peroxide (H₂O₂). Specifically, 1 mL of HSC@Se at a concentration of 120 μg mL⁻¹ was introduced into 9 mL of H₂O₂ at varying concentrations (0.125, 0.25, 0.5, 1, 2, 4, and 8 mM). O₂ production was then observed and recorded for each concentration of H₂O₂.

GPX activity of HSC@Se

The activity of GPx in the samples was measured using the NADPH-coupled colorimetric method. The total volume of the reaction system was 200 μL, which included 120 μg/mL H@Se (40 μL), PBS buffer, 2 mM reduced glutathione (GSH), 1 mM NADPH, and 2 μg/mL glutathione reductase (GR). The reaction was carried out at a constant temperature of 25 °C, and then hydrogen peroxide (H₂O₂) was added to a final concentration of 50 μmol/L to initiate the reaction. The decrease in absorbance was continuously monitored at a wavelength of 340 nm using a microplate reader (or spectrophotometer), reading every 30 s for a duration of 10 min.

ABTS^•+ radical scavenging capacities

The detection of ABTS^•+ radicals was slightly modified based on the methodology described by Li et al.⁴⁰. To generate ABTS^•+ free radicals, a stock solution of 7.4 mM ABTS was prepared alongside a 2.6 mM stock solution of K₂S₂O₈. Equal volumes of these two stock solutions were mixed and allowed to oxidize in the dark for 12 h. When needed, this mixture was diluted 30-fold with absolute ethanol. For subsequent experiments, 100 μL of the diluted reagent was combined with 100 μL of different samples to be tested. The resulting mixture was thoroughly mixed and allowed to sit in the dark for 2 min. The absorbance at 734 nm was then measured using a multi-functional enzyme labeling instrument. This assay was performed in triplicate. Inhibition ratio = [100-(A_s-A_b)/(A_c-A_b) × 100] %, A_s, A_b and A_c are the absorbance of the sample, the absorbance of the blank, and the absorbance of the control group, respectively.

Cascade catalysis of HSC@Se

BODIPY FL-NHS fluorophore (Ex: 495; Em: 530) was conjugated to CAT, while BODIPY R6G-NHS fluorophore (Ex: 520; Em: 570) was conjugated to SOD. After synthesizing the HSC@Se as described above, co-localization was observed using a confocal microscope. Additionally, changes in the fluorescence emission of the two fluorophores were measured using a fluorescence spectrophotometer. The cascade catalytic activity of HSC@Se was determined using a hydrogen peroxide detection kit (Beijing Baiao Leibo Technology Co., Ltd, KFS378). Hydrogen peroxide, as an intermediate product in the cascade reaction, is closely related to the reaction process. In simple terms, Amplex Red acts as a hydrogen peroxide probe and serves as a fluorogenic substrate for peroxidase enzymes. In the presence of horseradish peroxidase (HRP), Amplex Red undergoes a reaction with hydrogen peroxide in a 1:1 stoichiometric relationship, resulting in the formation of resorufin, a compound that exhibits strong fluorescence. The fluorescence intensity is measured using a fluorescence spectrophotometer; higher concentrations of hydrogen peroxide correspond to increased fluorescence intensity, while lower concentrations result in decreased fluorescence.

Hydrolytic enzyme resistance test

To assess resistance to proteolytic degradation, free SOD and HSC@Se with matched SOD activity were incubated separately with trypsin (0.25%, Thermo Fisher Scientific, cat. no. 25200056) at room temperature for 6 h. The reaction was then terminated by addition of phenylmethanesulfonyl fluoride (PMSF), and the residual SOD activity was measured using a Total Superoxide Dismutase Assay Kit with NBT (Beyotime Biotechnology, cat. no. S0109) according to the manufacturer’s instructions.

Cell culture

The HEI-OC1 cell line was derived from the auditory tissues of the Immortomouse™, a transgenic model created by the House Ear Institute in Los Angeles, California (Purchased from SUNNCELL Biotechnology Co., Ltd, cat. no. SNL-728). These cells were cultured in Dulbecco’s Modified Eagle’s Medium (Gibco) supplemented with 10% (v/v) fetal bovine serum (FBS) and without the inclusion of antibiotics. The cells were maintained at a temperature of 33 °C in a humidity-controlled incubator with an atmosphere of 10% CO₂.

Cellular uptake

A total of 5 × 10⁴ HEI-OC1 cells were plated in culture dishes appropriate for confocal microscopy and incubated at 33 °C with 10% CO₂ for 24 h. After this period, the medium was replaced with DMEM containing FL-HSC@Se (120 µg mL⁻¹). The cells were then separated into three experimental groups based on different incubation durations (2 h, 4 h, and 8 h). Following the treatments, the cells were washed three times with PBS to remove any non-adherent HSC@Se particles. The cells were subsequently fixed using 4% paraformaldehyde (PFA) at room temperature for 30 min and washed three additional times with PBS. To prepare for imaging, a fluorescence quenching mounting medium containing DAPI was applied, and sample images were obtained using a laser scanning confocal microscope.

In vitro cytotoxicity assay

The cytotoxicity of HSC@Se at varying concentrations was evaluated using the CCK-8 assay. HEI-OC1 cells were plated at a density of 8000 cells per well in a 96-well plate and incubated overnight. After discarding the culture medium, fresh DMEM supplemented with different concentrations of HSC@Se (0, 15, 30, 60, and 120 μg mL^-1) was added to the wells, and the cells were further incubated for 24 h/48h. At the end of the incubation period, CCK-8 solution was added to the wells, and the cells were co-incubated for 30 min. Cell viability was assessed by measuring the absorbance at 450 nm using a microplate reader.

Cell viability assay

To assess cell viability under various concentrations of H₂O₂, a CCK-8 assay was conducted. Initially, an oxidative stress model was established by incubating the cells for 24 h with different levels of H₂O₂ (0, 200, 300, 400, 500, and 600 μM). The cells were treated with varying concentrations of HSC@Se (0, 15, 30, 60, and 120 μg mL⁻¹) for 3 h at 33 °C. Following this, they were co-incubated with 500 μM H₂O₂ for 24 h to evaluate the potential protective effect of the nanoreactors against oxidative stress induced by hydrogen peroxide. After two washes with PBS, 100 μL of the CCK-8 working solution was added to each well, and the cells were incubated for an additional 1 h at 37 °C. Finally, absorbance at 450 nm was measured using a multifunctional enzyme reader to assess cell viability. Similarly, the ability of Erastin-induced ferroptosis was assessed using the CCK-8 method, which included the establishment of a cellular ferroptosis model with varying concentrations of Erastin (0, 2.5, 5, 10, 15, 20, and 25 μM). Cell viability was assessed after incubation with 120 μg mL^-1 of various materials (H@Se, HS@Se and HSC@Se) at 33 °C for 3 h followed by 24 h of incubation in the presence of 500 μM H₂O₂ or 15 μM erastin, using the same methods as previously described. This assay was performed in triplicate. Inhibition ratio = [100-(A_s-A_b)/(A_c-A_b)×100] %, A_s, A_b and A_c are the absorbance of the sample, the absorbance of the blank, and the absorbance of the control group, respectively.

Intracellular ROS level

The levels of intracellular ROS were quantified using a reactive oxygen detection kit, which employed 2’,7’-dichlorofluorescein diacetate (DCFH-DA). This compound, which is initially non-fluorescent, is transformed into the fluorescent DCF form upon reaction with ROS inside the cells. In this assay, HEI-OC1 cells were plated in confocal dish imaging plates and cultured for 24 h. They were then treated with 120 μg mL⁻¹ of various materials (H@Se, HS@Se and HSC@Se) for 3 h at 33 °C, followed by a 24 h co-incubation with 500 μM H₂O₂. After treatment, the cells were stained with 1 mL of DCFH-DA working solution and incubated for 30 min at room temperature, shielded from light. After incubation, laser confocal microscopy was used to capture images, with an excitation wavelength set at 488 nm. In addition, HEI-OC1 cells were initially seeded in 6-well plates and allowed to incubate for 24 h. Following this, they were treated with various materials (H@Se, HS@Se and HSC@Se) at a concentration of 120 μg mL⁻¹ for 3 h at 33 °C, then 3 mM H₂O₂, for a duration of 12 h at 33 °C. After treatment, the cells were resuspended in 0.25% trypsin-EDTA, followed by centrifugation. The cell pellet was then resuspended in 1 mL of phosphate-buffered saline containing 10 μM DCFH and incubated for 30 min in the dark at room temperature. Subsequently, flow cytometry analysis of the samples was performed using Beckman Flow Cytometry.

Superoxide levels in mitochondria

Mitochondrial superoxide levels were assessed using the MitoSOX Red probe (MedChemExpress, CAS No. 1003197-00-9). MitoSOX Red is a cell-permeable, mitochondria-targeted fluorescent probe that is oxidized by superoxide to generate red fluorescence, thereby reflecting mitochondrial superoxide levels rather than mitochondrial morphology. HEI-OC1 cells were seeded in confocal dishes and cultured for 24 h. Cells were then treated with different materials (H@Se, HS@Se, and HSC@Se; 120 μg mL⁻¹) at 33 °C for 3 h, followed by co-incubation with 500 μM H₂O₂ for 24 h. After removal of the culture medium, cells were incubated with MitoSOX Red working solution for 30 min at room temperature, washed twice with culture medium, and imaged using a laser-scanning confocal microscope with excitation at 510 nm.

Intracellular superoxide levels

Cellular superoxide anion levels were detected using the DHE (CAS NO.: 104821-25-2) fluorescent probe. The probe freely enters cells, where it is oxidized by superoxide anions, and the oxidation product subsequently binds to RNA or DNA and produces red fluorescence. In this experiment, HEI‑OC1 cells were seeded in confocal culture dish imaging plates and cultured for 24 h. The cells were divided into five groups: control, H₂O₂(500 µM), H@Se, HS@Se, and HSC@Se. The concentration of each material was 120 µg/mL. After a 3‑h pretreatment, the cells were incubated with fresh DMEM containing 500 µM H₂O₂ for 24 h. Following treatment, the cells were stained with 5 µM DHE working solution and incubated for 30 min at room temperature in the dark. Fluorescence images were then captured using a laser scanning confocal microscope with an excitation wavelength of 355 nm, a scan speed of 2.0 µs/pixel, and a gain of 1×.

Cell apoptosis study

Cell apoptosis was assessed using flow cytometry with a FITC-Annexin V/PI double staining kit. HEI-OC1 cells (5 × 10⁵ cells per well) were cultured in 6-well plates for 24 h. The cells were then divided into five groups: Control, H₂O₂, H@Se, HS@Se and HSC@Se. After a 3-h pretreatment period, they were incubated with 6 mM H₂O₂ at 33 °C for an additional 6 h. Following this, the cells were detached from the plates using trypsin, centrifuged at 1500 rpm for 5 min, and resuspended in 200 μL of binding buffer containing 8 μL of PI and 5 μL of FITC-Annexin V. This mixture was incubated for 10 min at room temperature in the dark before being analyzed by flow cytometry. The data were subsequently processed using Flow Jo software.

Mitochondrial membrane potential (MMP) assay

Mitochondrial membrane potential kit using JC-1 dye to detect the change of mitochondrial membrane potential. HEI-OC1 cells were plated in confocal dish imaging plates and cultured for 24 h. They were then treated with 120 μg mL⁻¹ of various materials (H@Se, HS@Se and HSC@Se) for 3 h at 33 °C, followed by a 24 h incubation with 500 μM H₂O₂. After 24 h, wash twice with detection buffer and incubate for 20 min with the same volume of medium and JC-1 working solution to avoid light. Images were acquired by laser confocal microscopy (488 nm / 561 nm).

Transcriptome sequencing

HEI-OC1 cells were seeded in 6-well plates at a density of 5 × 10⁵ cells per well and cultured for 24 h. Cells were then assigned to three groups (Each group n = 3): (1) control, cultured under standard conditions; (2) H₂O₂, treated with H₂O₂ to induce oxidative stress; (3) HSC@Se + H₂O₂, pre-treated with HSC@Se followed by H₂O₂ exposure. At the end of treatment, total RNA was extracted, and cDNA libraries were prepared by LC-Bio Technologies Co., Ltd (Hangzhou, China) according to standard protocols. High-throughput sequencing was performed on the NovaSeq X Plus platform. RNA-seq reads were aligned to the reference genome using Hisat2⁴¹, differential expression analysis in this study was performed using DESeq2, which models count data from biological replicates using a negative binomial framework⁴². Multiple-testing correction was applied using the Benjamini–Hochberg method, and only genes with an adjusted P value < 0.05 were considered differentially expressed. GO enrichment analysis was conducted using OmicStudio tools (https://www.omicstudio.cn/tool). DEGs were annotated to the Gene Ontology database, covering three branches: biological process, cellular component, and molecular function. Only terms with a p‑value < 0.05 were considered significantly enriched. For visualization, the top‑ranked significantly enriched terms from each branch (by fold enrichment or gene count) were displayed.

Western blotting

Total protein was extracted from cell lysates using IP lysis buffer (Wuhan Servicebio Technology Co., Cat: G2038) supplemented with protease inhibitor (Wuhan Servicebio Technology Co., Cat: G2008). Protein concentration was determined using a BCA protein assay kit (Beyotime Biotechnology, Cat: P0012). Equal amounts of protein (20 μg per sample) were separated by 4–12% SDS-polyacrylamide gel electrophoresis and subsequently transferred onto polyvinylidene difluoride (PVDF) membranes (Wuhan Servicebio Technology Co., Cat: G6047). The membranes were blocked with 5% bovine serum albumin for 1 h at room temperature and then incubated with primary antibodies overnight at 4 °C (MDA, 1:1000, #ab243066, Abcam; GPX4, 1:1000, #11243, ABclonal). After washing with Tris-buffered saline containing Tween, the membranes were incubated with the appropriate secondary antibodies for 1 h at room temperature. The membranes were then washed three times with 0.05% TBS-Tween. Protein bands were visualized using enhanced chemiluminescence reagent (Wuhan Servicebio Technology Co., Cat: G2074). The sample size for all quantitative Western blot analyzes was n = 3. Band densitometric analysis was performed using ImageJ software. The uncropped and unprocessed scanned images are provided in the Source Data file.

Animals

The animal experiments were performed following the ethical guidelines set by the Ethics Committee of Shanghai University (ECSHU-2021-029). To create a noise-induced hearing loss (NIHL) model and evaluate the antioxidant effects of nanozymes, male C57BL/6J mice (4 weeks old) were obtained from Shanghai Legen Biotechnology Co., Ltd (Shanghai, China). Only male C57BL/6J mice were used in this study. This choice was made to reduce potential variability associated with sex-dependent physiological and hormonal fluctuations and to maintain consistency across experimental groups; accordingly, the findings of this study should be interpreted within the context of male mice⁴³. Prior to the experiment, the mice were housed in stainless steel ventilated cages under controlled conditions, with a temperature of 25 ± 2 °C, 60 ± 10% relative humidity, and a 12-h light-dark cycle, for a duration of 7 days. The mice were then randomly allocated to different experimental groups and provided with disinfected drinking water and granulated food ad libitum.

Acoustic overexposures

To create a mouse model for NIHL, mice were placed in individual compartments within a circular wire mesh enclosure. The sound intensity in each compartment was carefully controlled, ensuring it varied by no more than 1 decibel. For a duration of 2 h, broadband white noise at 114 dB, covering frequencies from 2 to 20 kHz, was delivered through a high-quality loudspeaker and amplifier from JBL, a well-known American brand. Prior to each exposure, the acoustic levels were meticulously calibrated to ensure accuracy.

In vivo HSC@Se treatment and noise stimulation

To assess the protective effects of HSC@Se against noise-induced hearing loss, a total of 15 mice were randomly assigned to three groups (n = 5 each). The control group was untreated and not subjected to noise exposure, and the other two received either saline or HSC@Se via tail vein injections and were subjected to noise exposure, as follows. The HSC@Se was administered at a concentration of 120 μg mL⁻¹, corresponding to a dosage of 0.8 mg/kg. The mice underwent five injections: one administered 1 day prior to noise exposure and another 1 h before exposure. Following the noise exposure, the mice received three additional injections over the subsequent 3 days. The mice were subjected to broadband noise exposure with a frequency range of 2–20 kHz at a sound pressure level of 114 dB SPL for a duration of 2 h.

Auditory brainstem responses (ABR) tests

Auditory function in the animals was assessed through ABR testing. For ABR measurements, mice were anesthetized by intraperitoneal injection of ketamine (60 mg kg⁻¹) and placed on a heating pad maintained at 37 °C in a sound-attenuating environment. For the recordings, three subcutaneously inserted electrodes were placed: one at the pinna (recording electrode), another at the vertex (reference electrode), and the third at the rump (ground electrode). Responses were recorded during ABR testing, where pure tone stimuli at frequencies of 4, 8, 12, 16, 24, and 32 kHz were presented at sound levels ranging from 0 to 90 dB, in 5 dB increments, using an open-field microphone. The acoustic signals were produced by a RZ6 signal processor and controlled through BioSig software (TDT, Alachua, FL, USA). The threshold for each frequency was determined by examining ABR wave II, defined as the lowest stimulus level that elicited a detectable response. Baseline auditory acuity was evaluated 2 days prior to noise exposure, with follow-up assessments conducted on the first, third, seventh, and fourteenth days after exposure.

Cross-blood labyrinth barrier tests

Mice were randomly assigned to three groups: (1) NE + saline, (2) NE + Cy5.5, and (3) NE + HSC@Se-Cy5.5. Immediately after acoustic exposure, each animal received the corresponding treatment via a single intravenous tail-vein injection (100 μL). Before injection, the fluorescence intensity of the free Cy5.5 solution was adjusted to match that of HSC@Se-Cy5.5 to minimize bias arising from differences in fluorophore input between groups. For in vivo imaging, animals were anesthetized with isoflurane, placed in a prone position with the head fixed at a standardized angle, and imaged using an IVIS Spectrum system. Fluorescence signals were quantified as mean fluorescence intensity using ImageJ.

Distribution of materials within the cochlea

Cochlear frozen sections, after permeabilization with Triton X−100, were incubated overnight at 4 °C with the primary antibody (myo7a, 1:500, #20720-1-AP, Proteintech), followed by incubation with the corresponding secondary antibody (Goat anti-Guinea Pig IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647, 1:1000, #A-21450) at room temperature for 1 h, and then washed for immunofluorescence imaging.

Hair cell count

Cochleae were carefully dissected and post-fixed overnight in 4% paraformaldehyde (PFA) at 4 °C. They were subsequently treated with 10% EDTA (pH 7.4) at room temperature for 5–7 days, with the decalcifying solution changed every 2 days. The careful dissection involved meticulously peeling away the surrounding tissues to expose the delicate structures of the cochlea, using microdissection scissors to remove small and intricate sections of bone and tissue in order to obtain the basilar membrane of the cochlea. To assess the survival rate of hair cells, the basilar membrane was stained with phalloidin, and the proportion of surviving OHCs at the base, mid-turn, and apex was determined, normalized to the length of the basilar membrane.

Fe²⁺ distribution detection

To assess the involvement of ferroptosis in NIHL, cochlear basilar membranes were stained with FeRhoNox‑1, a fluorescent probe specific for ferrous iron (Fe²⁺). Briefly, after fixation and permeabilization, the isolated basilar membranes were incubated with FeRhoNox‑1 (5 μM) in the dark at 37 °C for 30 min, followed by three washes with phosphate‑buffered saline. Fluorescence images were acquired using a confocal laser scanning microscope, using identical acquisition settings across all samples to ensure comparability. The mean fluorescence intensity was quantified using ImageJ software. For regional analysis, the cochlea was divided into three segments: basal, middle, and apical turns. The mean fluorescence intensity was measured in each segment, and data were collected from three independent biological replicates per group. Quantitative results are presented as mean±SD, and statistical comparisons were performed using two‑way ANOVA.

Immunofluorescence staining

Basal membrane specimens were microdissected into three curves (apical, middle and basal) for immunostaining. The specimens were permeabilized in 0.5% Triton X-100 solution for 1 h and then blocked with 5% goat serum for 1 h at room temperature, followed by incubation with primary antibodies: guinea pig (IgG) anti-Type 3 vesicular glutamate transporter (Vglut3) at 1:500 (Synaptic Systems, #135204), mouse (IgG1) anti-C-terminal Binding Protein (CtBP2) at 1:200 (BD Biosciences, #612044) and rabbit polyclonal (IgG) anti-Homer protein homolog 1 (Homer1) at 1:200 (GeneTex, # GTX103278) at 4 °C overnight. The specimens were then incubated with secondary antibodies from Invitrogen: goat anti-guinea pig IgG conjugated to AlexaFluor-647, goat anti-mouse IgG1 conjugated to Alexa Fluor 555, and goat anti-rabbit IgG conjugated to Alexa Fluor 488, all at a dilution of 1:200, at room temperature for 4 h. Confocal images were captured using a laser scanning confocal microscope (Zeiss Microsystems, Germany). For ribbon synapse analysis, CtBP2 and Homer1 immunofluorescent puncta were counted in the z-stack images and normalized by the number of IHCs. The resulting images were processed and analyzed using ImageJ software.

Immunohistochemistry

Cochlea samples were prepared following the previously described protocol⁴⁴. After undergoing fixation, decalcification, and a series of dehydration steps, the samples were embedded in paraffin. Mid-modiolar sections were cut to a thickness of 5-7 μm and subsequently stained using hematoxylin and eosin (H&E) to assess the density of spiral ganglion neurons (SGNs) or for immunostaining with 4-HNE (Abcam, #ab46545). The images obtained were analyzed using ImageJ software. The density of SGNs was calculated by determining the ratio of SGN count to the area of the SGN region. Additionally, the integrated optical density (IOD) of the 4-HNE staining within the SGN area was also measured using ImageJ software.

Hemolysis assay

In summary, red blood cells (RBCs) were isolated from fresh mouse blood and diluted to prepare a 5% RBC suspension in saline. A 0.5 mL aliquot of this suspension was mixed with: (1) 0.5 mL of saline as a negative control, (2) 0.5 mL of distilled water (ddH₂O) as a positive control, and (3) 0.5 mL of HSC@Se. The mixtures were incubated at room temperature for 3 h, followed by centrifugation at 8000 rpm for 5 min. Hemolysis was observed by capturing images, and the absorbance (A) of the supernatant was measured at 541 nm using a microplate reader. The hemolysis percentage was calculated as follows: Percent hemolysis (%) = (A_sample-A_negative) /(A_positive-A_negative)×100%, where A_sample is the absorbance of the RBC suspension mixed with HSC@Se solution.

In vivo biosafety assay

To evaluate the systemic safety of HSC@Se administration, we used an independent cohort of mice (n = 6). This group was not exposed to noise and was randomly divided into two subgroups: one received an intravenous injection of HSC@Se (200 μL, 120 μg mL⁻¹) via the tail vein, while the other received an equal volume of saline as a control. Subsequently, we monitored changes in body weight and conducted auditory brainstem response (ABR) tests over 3 weeks period to assess potential systemic and ototoxic effects. After 30 days, blood samples were collected for hematological and biochemical analysis, and the brain and major organs (heart, liver, spleen, lungs, and kidneys) were harvested for histopathological examination.

Image acquisition and processing

All fluorescence images were acquired using a confocal laser scanning microscope (Zeiss LSM 880 or Leica STELLARIS 8). To ensure comparability across experimental groups, identical acquisition settings (laser power, detector gain, offset, and exposure time) were maintained for all samples within each experiment. Images were processed using ImageJ. When adjustments to brightness or contrast were applied, they were performed uniformly across all images from the same experimental set to avoid introducing bias. Representative images are shown with consistent intensity scaling.

Statistical analysis

Statistical analysis was conducted using GraphPad Prism (GraphPad Software). Data from experiments are presented as mean ± standard deviation (SD). Ordinary one-way and two-way multiple comparison post hoc tests were appropriately applied during the analysis of single-factor variance (ANOVA) when necessary. A p-value of less than 0.05 was deemed statistically significant.

Statistics and Reproducibility

Unless otherwise stated, all experiments were performed with at least three independent biological replicate experiments. For image-based assays, including confocal microscopy, TEM, and histological/immunofluorescence analyzes, representative images are shown and similar results were obtained across independent repeats.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

B.X., W.F., Y.X., D.Y. and Y.C. conceived and designed the experiments. B.X., X.C., K.H., J.S., Y.H., and W.L. performed experiments. B.X., W.F., Y.X., and D.Y. contributed to the analysis and interpretation of the results and the writing of the manuscript. W.F., Y.X., D.Y. and Y.C. secured funding. Everyone reviewed and approved the manuscript. AI tools were used to improve the fluency of language and the standardization of academic expression. However, all research design, data analysis, interpretation of results and academic opinions were independently completed by the author team. The author is ultimately responsible for the entire content and academic integrity of the paper.

Peer review

Peer review information

Nature Communications thanks Anna Rita, Amit Kumar, Xiaodong Tan and the other anonymous, reviewer for their contribution to the peer review of this work. A peer review file is available.

Funding

We greatly acknowledge the financial support from National Key Research and Development Projects (2023YFC2306500 to Y.C), National Nature Science Foundation of China (52572305 to Y.C) and Shanghai Science and Technology Program (24520712900 to Y.C).

Data availability

The transcriptomics data generated in this study have been deposited in the Gene Expression Omnibus (GEO) database under accession number “GSE326834 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE326834]”. All other data supporting the findings of this study are available within the article and its supplementary files. Any additional requests for information can be directed to, and will be fulfilled by, the corresponding author. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76994-3.