A continuous DNA repairing system for alleviating intervertebral disc degeneration

Abstract

DNA damage triggering senescence-associated secretory phenotype (SASP) is a key pathway of Intervertebral Disc Degeneration (IVDD). Current therapeutic strategies for IVDD face significant challenges in effectively alleviating DNA damage-mediated senescence of nucleus pulposus cells (NPCs) and IVDD repair. Here, we propose a strategy for the continuous regulation of DNA repair: starting with the scavenging of the key upstream trigger, ROS, proceeding to the repair of the DNA damage, and culminating in the promotion of DNA replication and extracellular matrix synthesis. Through single-cell sequencing of clinical samples and in vitro modeling, we elucidated how oxidative stress drives DNA damage, leading to SASP and inflammatory microenvironment within the intervertebral disc. Subsequently, we developed a continuous DNA repairing system. In this system, bimetallic-curcumin nanozymes (AuCu-Cur) scavenge ROS in NPCs and repair damaged DNA via the increased DNA glycosylase NEIL3. Additionally, the combined human umbilical cord-derived vesicles (hUCMSC-EVs) significantly enhance the bioavailability of the AuCu-Cur, promoting DNA replication, and promoting tissue repair via p-JNK pathway. In a rat tail needle puncture model, this system markedly alleviated IVDD. This study presents a promising approach to reverse the DNA damage-induced cellular senescence and IVDD, providing a repair strategy for addressing diseases associated with DNA damage and cellular senescence.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04615-8.

Introduction

Low back pain (LBP) is a prevalent chronic/age-related disease worldwide. Among spinal degenerative diseases [1–4], IVDD is a primary cause of LBP and imposes a substantial socioeconomic burden [1, 5]. Oxidative stress is a major pathogenic factor in IVDD [6–11]. Reactive oxygen species (ROS), which encompass oxygen-derived free radicals and peroxides generated during oxygen metabolism, are central destructive agents that disrupt the REDOX homeostasis within the intervertebral disc, thereby accelerating IVDD progression [6, 9, 11, 12]. Cellular senescence of NPCs is a critical hallmark of IVDD advancement [3–16]. REDOX imbalance and ROS accumulation lead to oxidative DNA damage [17, 18], and the accumulation of such damage, along with the excessive DNA damage signaling, are key inducers in the initiation and maintenance of NPCs senescence [9, 13, 18–21]. The inflammatory microenvironment within the disc is a definitive manifestation of IVDD pathology. Senescent NPCs adopt SASP [18, 22, 23], characterized by increased secretion of inflammatory cytokines, angiogenic factors, and matrix metalloproteinases (MMPs) [19–21, 24, 25]. This SASP, in turn, recruits inflammatory macrophages into the disc and accelerates the degradation of the extracellular matrix [26–28], ultimately remodeling the disc microenvironment and driving IVDD. Although numerous studies have focused on developing antioxidant biomaterials to treat IVDD [6, 11], their primary aims are to scavenge ROS and suppress oxidative stress-induced inflammation to alleviate the pain [8, 29–39]. However, the critical role of ROS-induced DNA damage is ignored which still damages IVDD tissue. Besides, repairing degenerated disc tissue resulting from inflammatory senescence is indispensable. Moreover, the DNA repair functionality of nanozymes remains largely unexplored.

Nanozymes are artificial enzymes with catalytic properties that can mimic the structures of natural antioxidant enzymes, enabling them to exhibit superoxide dismutase (SOD) and catalase (CAT) activities [40–45]. However, the antioxidant efficiency and bioavailability of nanozymes require significant improvement [41–45].

Therefore, we have developed a continuous DNA repair system (AuCu-Cur-EVs) designed to treat IVDD through a sequential process: restoring REDOX balance, repairing DNA damage, and reversing the senescent phenotype of NPCs, promoting intervertebral disc repair. For NPCs, the system first scavenges ROS via the AuCu-Cur, preventing further genomic damage. Transcriptomic analysis further revealed that curcumin facilitates the excision of oxidized DNA bases through enzymes such as NEIL3, thereby delaying cellular senescence and the SASP, which successfully restores the disc’s immune microenvironment. This helps reverse the senescent state of NPCs and subsequently ameliorates the inflammatory immune microenvironment within the intervertebral disc. Furthermore, hUCMSC-EVs are enriched with various tissue regeneration-related RNAs and proteins [46–49]. They not only significantly improve the bioavailability of curcumin and the nanozyme but also promote the synthesis of extracellular matrix via chondrogenic differentiation mechanisms. Additionally, beyond inducing stem cell differentiation, hUCMSC-EVs activate DNA replication-related pathways in NPCs, further mitigating cellular senescence. In a rat tail needle puncture model, we encapsulated this system in an alginate hydrogel and injected it into degenerated discs, successfully achieving therapeutic reversal of IVDD. In summary, we have constructed a novel platform for the continuous regulation of DNA repair and demonstrated its efficacy in promoting the regeneration of degenerated intervertebral discs (Fig. 1).

Fig. 1.

DNA Repair-Mediated Strategy for Alleviating Intervertebral Disc Degeneration. a, Oxidative stress is a major pathogenic factor in IVDD. ROS accumulation and redox imbalance cause DNA damage, triggering a DNA damage response and promoting NPCs senescence. Senescent cells release chemokines and inflammatory factors. Chemokines recruit immune cells, forming an inflammatory immune microenvironment, while inflammatory factors accelerate extracellular matrix degradation, ultimately leading to intervertebral disc degeneration. b, The system is constructed by loading a Au-Cu bimetallic nanozyme and curcumin onto the surface of hUCMSC-EVs, encapsulating them in an alginate hydrogel, and injecting them into the degenerated disc of rats for sustained release. The bimetallic nanozyme combined with curcumin scavenges ROS and restores redox balance. Additionally, curcumin can repair DNA damage and inhibit SASP via the DNA glycosylase NEIL3. AuCu and curcumin promote the polarization of M1 macrophages to the M2 phenotype through ROS scavenging and DNA repair. hUCMSC-EVs contain various tissue-repair proteins and RNAs, promoting the synthesis of disc extracellular matrix and cellular regeneration; ultimately, this system reverses the inflammatory immune microenvironment and alleviates IVDD progression

Results

Senescence of NPCs in IVDD: oxidative stress-induced DNA damage triggers SASP

NPCs senescence is a major driver of IVDD [13, 19–21, 24–26]. To characterize the mechanisms of NPCs senescence and its relationship with IVDD progression, we re-analyzed single-cell transcriptomic data from the publicly available GSE244889 database [24] (Fig. 2a). Subpopulation analysis identified 8 distinct cell clusters (Fig. 2b). NPCs constituted the majority within nucleus pulposus (NP) tissue, alongside the presence of immune cell populations dominated by macrophages, indicating an inflammatory immune microenvironment in the degenerated disc. Further sub-clustering of NPCs revealed 6 distinct subpopulations (Fig. 2c). Among these, stress-responsive NPCs (SR-NPs) [24], marked by CP expression, were widely distributed across all disease stages. Given that CP exhibits glutathione peroxidase-like activity [50], this suggests the persistent presence of oxidative stress during IVDD. In contrast, the proportions of progenitor NPCs (pro-NPs, marked by UBE2C) and integral regulatory NPCs (IR-NPs, marked by CHI3L2) gradually decreased with advancing degeneration [24], indicating a progressive decline in the disc’s repair capacity (Fig. 2d, e). Samples were stratified into moderate degeneration (MDD) and severe degeneration (SDD) groups based on Pfirrmann grade [24]. Gene ontology (GO) enrichment analysis of genes upregulated in SDD NPCs compared to MDD revealed significant enrichment in pathways related to oxidative stress, DNA damage, inflammatory regulation, and myeloid cell differentiation (Fig. 2f). The enrichment of the myeloid cell differentiation pathway correlates with the observed macrophage infiltration, suggesting that the SASP from senescent NPCs may help shape the inflammatory microenvironment by recruiting macrophages. Pseudotime trajectory analysis of selected genes showed that NPCs marker genes (SOX9, ACAN, COL2A1) and DNA repair genes (NEIL3, BRCA2, ERCC1) were expressed early in degeneration but gradually decreased. Concurrently, the expression of inflammatory (NFKB1, MMP9, IL6) and senescence-related genes (TP53, STAT3, CDKN2B) increased over time (Fig. 2g, genes clustered into 2 groups). Gene Set Enrichment Analysis (GSEA) further confirmed significant enrichment of the inflammatory response pathway (Fig. 2h). We therefore hypothesize that oxidative stress in NPCs promotes DNA damage, induces cellular senescence and SASP, and subsequently recruits macrophages to form an inflammatory microenvironment, ultimately driving IVDD. Furthermore, we verified an overall increase in the proportion of macrophages within the degenerated nucleus pulposus using flow cytometry. Specifically, the proportion of M1 macrophages was notably elevated, whereas the proportion of M2 macrophages was significantly decreased(Supplementary Fig. 13).

Fig. 2.

Oxidative Stress-Induced DNA Damage Promotes NPCs Senescence in IVDD. a, Schematic diagram of single-cell RNA sequencing for NP tissue from the database. b, UMAP plot of human NP tissue, clustered into eight cell types (NPCs, endothelial cells, neutrophils, T cells, B cells, macrophages, smooth muscle cells, erythrocytes) by reference markers, with distinct colors for each cluster. c, UMAP plot showing further subclustering of NPCs. Pro-NPCs: progenitor NPCs, marked with ACAN, SOX9, UBE2C, and TOP2A; Fibro-NPCs: fibrotic NPCs, marked with ACAN, SOX9, FBLN1; IR-NPCs: integral regulatory NPCs, marked with ACAN, SOX9, CH3L2; Met-NPCs: metabolic NPCs, marked with ACAN, SOX9, DKK1; Adh-NPCs: adhesive NPCs, marked with ACAN, SOX9, and MSMO1; and SR-NPCs: stress-responsive NPCs, marked with ACAN, SOX9, and CP. d, Single-cell atlas revealing NP cell distribution in mild degenerative disc (MDD) and severe degenerative disc (SDD) groups based on Pfirrmann grading. e, Dot plot of marker genes for six cell types. f, DEGs between SDD and MDD were identified using the FindMarkers function in Seurat (min.pct = 0.1). Upregulated DEGs (log2FC > 0.25) were subjected to GO enrichment analysis. Fisher’s exact test was applied with p-values adjusted for multiple comparisons (FDR < 0.05). g, Heatmap of scaled gene expression along the pseudotime trajectory. h, Representative GSEA result from scRNA-seq (control: MDD; experimental: SDD). Displayed pathway: GO:0050727. i, Human NP tissue subjected to histology and RNA-seq after 3-day LPS induction (10 µmol/L) to promote senescence. j, Heatmap of selected transcriptomic DEGs (p < 0.05, |log2FC| > 1). k, Representative GSEA result of RNA-seq (control: Control; experimental: LPS). Displayed pathway: hsa04623. l, T2-weighted MRI of human intervertebral discs from MDD and SDD groups. m, Immunofluorescence and immunohistochemical staining of γ-H2AX in MDD and SDD groups. Scale bar, 50 μm. Representative images from three independent experiments. n, Schematic mechanism of IVDD: oxidative stress induces DNA damage in NPCs, leading to cellular senescence and SASP. Abbreviations: NP, nucleus pulposus; MDD, mild degenerative disc; SDD, severe degenerative disc; UMAP, uniform manifold approximation and projection; FC, Foldchange

To model NPCs senescence in IVDD in vitro, we characterized the transcriptomic profiles of young and senescent human NPCs. Isolated human NPCs were treated with lipopolysaccharide(LPS) to induce senescence, followed by RNA sequencing (Fig. 2i). A heatmap demonstrated significantly increased expression of key genes involved in inflammation, senescence, and extracellular matrix degradation in LPS-induced NPCs (Fig. 2j). GSEA confirmed the presence of a DNA damage response in these cells (Fig. 2k). These results indicate that the LPS-induced NPCs senescence model effectively recapitulates the state of NPCs in IVDD. To further validate the single-cell transcriptomics findings, we performed histological analysis on human NP tissues of different Pfirrmann grades (Fig. 2l). Expression of the DNA damage marker γ-H2AX was significantly elevated in severely degenerated discs (Fig. 2m).

In summary, these findings elucidate a key pathological sequence in IVDD: oxidative stress in NPCs promotes DNA damage, which in turn induces cellular senescence and SASP, thereby driving disease progression (Fig. 2n).

Characterization of AuCu-Cur-EVs with antioxidant properties

As shown in Figure. 3b, it can be seen from the SEM (Scanning Electron Microscope) and HRTEM (High-Resolution Transmission Electron Microscope) images that the AuCu nanoclusters loaded with curcumin (AuCu-Cur) present a uniform nanoscale cluster structure with regular morphology and concentrated size distribution. The elemental composition of AuCu-Cur includes O, C, S, Au, and Cu (Supplementary Fig. 1). The uniform distribution of O, Cu, and Au elements in the element mapping (Figure. 3c) intuitively proves that curcumin is successfully and uniformly loaded on the surface of the AuCu nanoclusters. As shown in Figure. 3d, in the subsequent FTIR (Fourier Transform Infrared Spectroscopy) spectrum, AuCu-Cur shows a significant change in the position and intensity of the C = O and C-O stretching vibration peaks of the characteristic functional groups of curcumin compared with the AuCu nanoclusters (AuCu), indicating that curcumin is successfully loaded onto the surface of the nanoclusters through chemical bonding or intermolecular interactions. The full XPS (X-ray Photoelectron Spectroscopy) spectrum (Figure. 3e) and O 1s peak (Figure. 3f) further reveals from the chemical state level that there are various oxygen-containing functional groups such as -OH, C = O, Cu-O, etc. on the surface of AuCu-Cur. These functional groups not only participate in the loading process of curcumin, but also provide active sites for subsequent antioxidant reactions. In addition, the binding energy of O1s has decreased, indicating that curcumin has been successfully loaded on the surface of AuCu. The release curve of Cur (Figure. 3g) shows a time-dependent controlled release behavior. As time goes by, the amount of curcumin released gradually increases and tends to be stable. This controlled release characteristic provides a basis for the application of AuCu-Cur in organisms. In the antioxidant activity test, the DPPH experiment (Figure. 3h, i) showed that the characteristic absorption peak intensity of DPPH free radicals decreased significantly with the action time and concentration after AuCu-Cur treatment, indicating that AuCu-Cur can efficiently scavenge DPPH free radicals and has direct antioxidant capacity; and the SOD (Figure. 3j) and CAT (Figure. 3k) nanozyme activity test results showed that the enzyme activity of the AuCu-Cur group was significantly higher than that of the control group and the single components AuCu and Cur, indicating that AuCu nanoclusters and curcumin work synergistically to enhance the ability of simulated SOD to scavenge superoxide anions (·O²⁻) and simulated CAT to decompose H₂O₂. This synergistic effect stems from the electronic structure regulation of the nanoclusters and the complementarity of the active sites of curcumin. Theoretical calculations (Figure. 3L–q) further explain the synergistic antioxidant mechanism at the microscopic level: by constructing an atomic structure model (Figure. 3L), it was found that the Cu atoms on the surface of the AuCu nanoclusters can form coordination bonds with the carbonyl/hydroxyl groups of curcumin. The density of states (Figure. 3m) analysis shows that after AuCu and Cur are combined, the electric field near the Fermi level is significantly increased. The density of states is significantly enhanced, and the d orbital of Cu and the π orbital of Cur are hybridized, which promotes the transfer of electrons between the nanoclusters and curcumin, providing an electronic basis for the adsorption and activation of ROS. The charge density difference (Figure. 3n) intuitively shows the charge transfer at the interface. The AuCu nanoclusters transfer charge to curcumin, regulating the electron cloud distribution of the active center of curcumin, making it easier to interact with ROS. For the adsorption behavior of ROS (Figure. 3o), the adsorption energies (Eads) of ·OH, ·O₂⁻, and H₂O₂ on the AuCu-Cur surface are calculated to be -4.33 eV, -2.39 eV, and − 1.21 eV, all negative and with large absolute values, indicating that these ROS can spontaneously and stably adsorb on the material surface, providing a thermodynamic prerequisite for subsequent scavenging reactions. Reaction path kinetic calculations (Figure. 3p-q) show that the energy barrier for ·O₂⁻ conversion in CAT and SOD simulations also dropped from 1.5 eV to 0.8 eV. This barrier reduction is attributed to the electronic structure regulation of the nanoclusters and the stabilization of the transition state by the curcumin conjugated system, which together accelerate the antioxidant reaction process.

Fig. 3.

Structural Characterization of the Material and Theoretical Calculation of Antioxidant Properties. a, Schematic illustration of the material preparation process. b, SEM (left) and HRTEM (right) images of AuCu-Cur. Scale bar, 200 and 100 μm. c, O, Cu, and Au element mapping distribution of AuCu-Cur. d, FTIR spectra of AuCu and AuCu-Cur. e, XPS full spectra of AuCu, Cur, and AuCu-Cur. f, XPS peak splitting of O 1s in AuCu-Cur. g, Time-dependent release curve of Cur. h, i, Spectral changes of DPPH free radical scavenging (h) and inter-group comparison (i). j, k, Quantitative analysis of nanoenzyme activities of SOD (j) and CAT (k). Data are expressed as mean ± SD, **p < 0.01, ***p < 0.001, ****p < 0.0001. l–n, Atomic structure models calculated theoretically, adsorption energy (l, o), density of states (m), and charge density difference (n). p, q, Free energy path diagrams of simulated reactions of CAT (p) and SOD (q)

ROS scavenging and inflammation modulation in vitro

Oxidative stress is a key inducer of NPCs senescence and intervertebral disc degeneration [31]. Under pathological conditions, excessive ROS accumulation disrupts REDOX homeostasis, exacerbating oxidative stress within the disc. It is important to note that the material exhibited no significant toxicity within a specific concentration range (Supplementary Fig. 2). While increasing material concentrations can potentially reduce cell viability (Supplementary Fig. 3). Furthermore, after being loaded into hUCMSC-EVs, the tolerable concentration of the AuCu and Cur increased significantly, likely attributable to the vesicles’ ability to improve biocompatibility, with an optimal effect observed at a concentration of 100 ppm after 24 h of stimulation. To evaluate the ROS-scavenging capacity of the composite material and its individual components, NPCs were first stimulated with 10 µM LPS for 3 days, followed by co-culture with the material or its components for 1 day. Intracellular ROS levels were assessed using the fluorescent probe DCFH-DA. Confocal microscopy revealed that AuCu, Cur, AuCu-Cur, and AuCu-Cur-EVs all demonstrated significant ROS-scavenging ability (Fig. 4a-b), whereas the vesicles alone could not scavenge ROS. This conclusion was corroborated by flow cytometry (FCM) results (Fig. 4c). Mitochondria are a major source of intracellular ROS, and their functional status can reflect cellular ROS levels. To further investigate mitochondrial function, JC-1 staining was used to assess changes in mitochondrial membrane potential (ΔΨm) (Fig. 4d-g). LPS stimulation disrupted the ΔΨm in NPCs, leading to mitochondrial depolarization, as evidenced by the accumulation of JC-1 monomers. Treatment with AuCu, Cur, AuCu-Cur, and AuCu-Cur-EVs improved mitochondrial functional morphology, promoting the increase of JC-1 aggregates. The cellular experimental results are highly consistent with the physicochemical data derived from our extensive nanozyme characterization and theoretical calculations. These findings demonstrate that AuCu-Cur possesses a robust ROS-scavenging capacity. Notably, compared to AuCu alone, AuCu-Cur exhibits significantly enhanced ROS-scavenging performance. In summary, compared to other components, AuCu-Cur-EVs possessed the strongest capacity for ROS clearance.

Fig. 4.

In Vitro ROS-Scavenging and Anti-inflammatory Ability. a, hBMSCs’ immunofluorescence and statistic analysis of DCFH-DA after 24 h of culture under different conditions. scale bar: 20 μm; three independent experiments were repeated. b, Statistical result of DCFH-DA. Data are expressed as mean ± SD, ***p < 0.001, ****p < 0.0001. c, FCM analysis of DCFH-DA. d, hBMSCs’ immunofluorescence and statistic analysis of JC-1 after 24 h of culture under different conditions. scale bar: 20 μm; three independent experiments were repeated. e, Statistical result of JC-1 aggregates. f, Statistical result of JC-1 monomers. Data are expressed as mean ± SD, *p < 0.05, ***p < 0.001, ****p < 0.0001. g, FCM analysis of JC-1. h, NPs’ immunofluorescence and statistic analysis of IL-1β after 24 h of culture under different conditions. scale bar: 20 μm; three independent experiments were repeated. i, Statistical result of IL-1β. Data are expressed as mean ± SD, **p < 0.01, ****p < 0.0001. j-l, qRT-PCR results of IL-6, IL-1β, IL-8 in NPCs after different treatments, n = 3 independent experiments per group. Data are expressed as mean ± SD, **p < 0.01, ****p < 0.0001. All of the culture conditions: culture medium (Group Control), LPS containing medium (Group LPS), EVs containing medium (Group EVs), AuCu containing medium (Group AuCu), Cur containing medium (Group Cur), AuCu-Cur containing medium (Group AuCu-Cur), AuCu-Cur-EVs containing medium (Group AuCu-Cur-EVs). All of statistical analyses were performed using one-way ANOVA followed by Tukey’s multiple comparison test (b, e-f, i-l)

Since senescent NPCs and macrophages can accelerate disc degeneration by inducing inflammation [13, 14, 19–21, 25–28, 39], we investigated whether the material could concurrently reduce inflammation alongside ROS. After treating the LPS-induced senescent NP cell model, the intracellular level of IL-1β was measured by immunofluorescence. AuCu, Cur, AuCu-Cur, and AuCu-Cur-EVs all significantly suppressed IL-1β levels (Fig. 4h-i), with the full composite material (AuCu-Cur-EVs) exhibiting the most potent effect. The anti-inflammatory capacity was further confirmed by quantitative real-time PCR (qRT-PCR), which showed that the material significantly suppressed the expression of IL-6, IL-8, and IL-1β (Fig. 4j-l). Collectively, these results indicate that the composite material possesses superior ROS-scavenging and anti-inflammatory capabilities compared to its individual components.

DNA repair and immune modulation in vitro

To evaluate the material’s potential in addressing DNA damage response (DDR), we assessed its DNA repair capability in vitro. In the LPS-induced NP cell senescence model, treatment with the material significantly reduced the level of the DNA damage marker γ-H2AX, as shown by confocal microscopy (Fig. 5a-b). This effect was primarily attributed to the curcumin component. Transcriptomic analysis revealed significant enrichment of DNA repair-related pathways and a marked increase in NEIL3 expression (Fig. 6h-i). A heatmap of differentially expressed genes (DEGs) demonstrated a substantial reduction in SASP factors in the AuCu-Cur-EVs treatment group (Supplementary Fig. 5). This conclusion is further supported by the previous qRT-PCR results showing downregulation of IL-6, IL-8, and IL-1β. Collectively, these findings indicate, for the first time, that curcumin mitigates oxidative stress-induced DNA damage and alleviates the SASP in NPCs, likely through the DNA glycosylase NEIL3. We established a DNA damage model using etoposide and verified the DNA repair capacity and underlying mechanism of Cur via the comet assay. The results support that Cur repairs damaged DNA through the NEIL3 pathway( Supplementary Fig. 12).

Fig. 5.

In Vitro DNA Repair and Immunomodulatory Ability. a, hBMSCs’ immunofluorescence and statistic analysis of γ-H2AX after 24 h of culture under different conditions. scale bar: 20 μm; three independent experiments were repeated. b, Statistical result of γ-H2AX. Data are expressed as mean ± SD, *p < 0.05, ****p < 0.0001. c, The immunofluorescence of polarization in THP1 and statistic analysis of CD86 and CD206 after 24 h of culture under different conditions. scale bar: 30 μm; three independent experiments were repeated. d, Statistical result of CD86. e, Statistical result of CD206. Data are expressed as mean ± SD, *p < 0.05, **p < 0.01, ****p < 0.0001. f, FCM analysis of polarization in THP1; three independent experiments were repeated. g, Statistical result of the propotion of CD86 + CD206- cells (%). h, Statistical result of the proportion of CD86-CD206 + cells (%). Data are expressed as mean ± SD, *p < 0.05, **p < 0.01, ****p < 0.0001. i, Graphic illustration of AuCu-Cur-EVs in immunomodulatory ability. All of the culture conditions: culture medium (Group Control), LPS containing medium (Group LPS), LPS + γ-IFN containing medium (Group M1 Positive), IL-4 + IL-13 containing medium (Group M2 Positive), EVs containing medium (Group EVs), AuCu containing medium (Group AuCu), Cur containing medium (Group Cur), AuCu-Cur containing medium (Group AuCu-Cur), AuCu-Cur-EVs containing medium (Group AuCu-Cur-EVs). All of statistical analyses were performed using one-way ANOVA followed by Tukey’s multiple comparison test (b, d-e, g-h)

Fig. 6.

In vitro chondrogenic potential and promotion of DNA replication. a, Representative Alcian Blue and Safranin O staining images of hBMSCs under different culture conditions, scale bar: 500 µm, three independent experiments were repeated. b, Statistical result of Alcian Blue intensity. c, Statistical result of Safranin O Intensity. Data are expressed as mean ± SD, *p < 0.05, **p < 0.01, ****p < 0.0001. d, hBMSCs’ immunofluorescence and statistic analysis of Collagen II after 24 days of culture under different conditions. scale bar: 50 µm, three independent experiments were repeated. e, Statistical result of Collagen II. Data are expressed as mean ± SD, *p < 0.05, ***p < 0.001, ****p < 0.0001. f, Representative SA-β-Gal staining images of NPs under different culture conditions, scale bar: 100 µm, three independent experiments were repeated. g, Statistical result of SA-β-Gal+ cells (%). Data are expressed as mean ± SD, *p < 0.05, ****p < 0.0001. h, DEGs between LPS and LPS + AuCu-Cur-EVs (|log2FC|> 1, p < 0.05) were subjected to GO Enrichment analysis, Fisher’s exact test was applied with p-values adjusted for multiple comparisons (FDR < 0.05). i, Volcano plot of RNA-seq, selected genes were presented in the plot. j, WB result of p-JNK, JNK, GAPDH, n = 3 independent experiments per group. k, Representative GSEA result from RNA-seq (control: Control; experimental: LPS + AuCu-Cur-EVs). l, Graphic illustration of AuCu-Cur-EVs in chondrogenic and promotion of DNA replication. Displayed pathway: GO:0006310, GO:0006260, GO:0006261, GO:0045005, GO:0006268, GO:0051052, GO:0051050, GO:0006275, GO:0140097. All of the culture conditions: culture medium (Group Control), complete chondrogenic media (Group Positive Control), LPS containing medium (Group LPS), AuCu containing medium (Group AuCu), Cur containing medium (Group Cur), AuCu-Cur containing medium (Group AuCu-Cur), EVs containing medium (Group EVs), AuCu-Cur-EVs containing medium (Group AuCu-Cur-EVs). All of statistical analyses were performed using one-way ANOVA followed by Tukey’ s multiple comparison test(b-c, e, g)

Given that disc degeneration is often accompanied by the infiltration of inflammatory cells (primarily M1 macrophages), and that senescent NPCs recruit macrophages via SASP factors, a vicious cycle ensues: M1 macrophages secrete pro-inflammatory cytokines that further accelerate NPCs senescence and sustain the inflammatory microenvironment [24, 27, 31, 51]. Therefore, we investigated whether the material could polarize M1 macrophages towards the M2 phenotype by reducing their DNA damage and ROS levels, thereby promoting disc repair. After differentiating THP-1 cells into M0 macrophages using Phorbol 12-myristate 13-acetate (PMA), we polarized them into M1 macrophages (using 100 ng/ml LPS and 20 ng/ml IFN-γ) or M2 macrophages (using 20 ng/ml IL-4 and 20 ng/ml IL-13). The material was then applied to the M1 macrophages. Confocal microscopy analysis of the polarization markers CD86 (M1) and CD206 (M2) showed that the material induced a phenotypic shift from M1 towards M2 (Fig. 5c-e). While the individual components Cur and AuCu exhibited similar polarizing ability, the full composite material was the most effective. Flow cytometry confirmed the material’s capacity to promote this M1-to-M2 transition, although the effect did not reach the level of the dedicated M2-positive control (Fig. 5f-h).

In summary, these experimental results demonstrate that the composite material promotes the polarization of M1 macrophages to the M2 phenotype. This effect is achieved through a dual mechanism: the DNA repair capacity of curcumin, combined with the ROS-scavenging activity of both the AuCu-Cur (Fig. 5i).

Chondrogenic potential and proliferative promotion in vitro

To evaluate the material’s potential in promoting disc repair and NPCs regeneration, its chondrogenic capacity was assessed in vitro. Human bone marrow mesenchymal stem cells (hBMSCs) were co-cultured with the composite material or its individual components, with a chondrogenic induction medium used as a positive control. After 21 days, Alcian Blue and Safranin O staining were performed (Fig. 6a, Supplementary Fig. 6). Staining intensity was more pronounced in the AuCu-Cur-EVs, EVs-only, and positive control groups compared to the AuCu, Cur, and AuCu-Cur groups. Notably, the AuCu-Cur-EVs group showed no statistically significant difference from the positive control (Fig. 6b-c), indicating its strong ability to enhance the chondrogenic potential of hBMSCs. To further characterize chondrogenic differentiation, the expression level of Collagen II, a key protein in cartilage formation, was detected via immunofluorescence. The EVs-only and AuCu-Cur-EVs groups exhibited significantly higher Collagen II expression than other groups (Fig. 6d-e), with the highest level observed in the AuCu-Cur-EVs group, potentially attributable to its effective suppression of oxidative stress. Furthermore, we evaluated the expression of aggrecan (ACAN), a pivotal marker of nucleus pulposus cells. The observed trend was consistent with the results for collagen type II ( Supplementary Fig. 14).

Given the crucial role of DNA damage in cellular senescence and the potent anti-senescence capacity of hUCMSC-EVs [46–48], senescence-associated β-galactosidase (SA-β-Gal) staining was used to assess senescence levels. Results showed a significant reduction in SA-β-Gal-positive cells in the material-treated group (Fig. 6f), an effect also observed in the EVs-only group. To elucidate the mechanism by which AuCu-Cur-EVs delay NPCs senescence, transcriptomic profiling was performed on senescent NPCs treated with the material. Principal component analysis (PCA) revealed distinct mRNA expression profiles between senescent NPCs and those treated with AuCu-Cur-EVs (Supplementary Fig. 7). GO enrichment analysis of upregulated DEGs identified significant enrichment in pathways related to extracellular matrix synthesis, MAPK signaling, and chondrogenesis in the treated group compared to senescent NPCs (Fig. 6h). A volcano plot demonstrated significant upregulation of DNA repair-related genes [17] (e.g., NEIL3, EME1, TRAIP, AUNIP, DTL, E2F) and downregulation of inflammation-related genes [22] (e.g., IL1B, TLR2, CXCL8, IL8), consistent with the material’s functions in promoting DNA repair and suppressing SASP (Fig. 6i). Since the MAPK pathway is often activated during chondrogenesis [52–54], Western blot analysis was used to assess JNK and phospho-JNK levels during material-induced chondrogenic differentiation of hBMSCs (Fig. 6j). A significant increase in JNK phosphorylation was observed, suggesting a potential molecular mechanism for the material’s chondrogenic effect. Furthermore, GSEA confirmed that AuCu-Cur-EVs significantly promoted pathways associated with DNA replication (Fig. 6k), providing strong evidence for its ability to mitigate NPCs senescence.

In conclusion, AuCu-Cur-EVs demonstrate excellent therapeutic efficacy in promoting chondrogenesis, facilitating tissue repair, and delaying cellular senescence.

Therapeutic efficacy of AuCu-Cur-EVs@SA in a rat tail needle puncture model

To minimize potential side effects from systemic exposure, a localized delivery strategy was employed. AuCu-Cur-EVs were encapsulated in an alginate (SA) hydrogel, confining their bioactivity to the lesion site and enabling sustained release within the intervertebral disc (IVD) to enhance therapeutic efficacy. The treatment potential of AuCu-Cur-EVs@SA was subsequently evaluated in a puncture-induced IVDD rat model (Fig. 7a). Blood samples collected two weeks post-treatment showed no significant differences in complete blood count or liver/kidney function parameters compared to the control group (Supplementary Fig. 8). Histological examination of major organs (heart, liver, spleen, lung, kidney) after four weeks revealed no pathological lesions (Supplementary Fig. 9), indicating minimal systemic toxicity. EVs@SA and AuCu-Cur-EVs@SA were injected into rat caudal IVDs using a 25G needle. The negative control group received no puncture, while the positive control group underwent puncture without injection. IVD status was analyzed using multiple imaging modalities. MRI, assessing water content via T2-weighted signal intensity for Pfirrmann grading, showed reduced signal in the positive control and EVs@SA groups at week 2. Although the AuCu-Cur-EVs@SA group did not fully recover to the level of the negative control, its signal intensity was significantly improved compared to the positive control and EVs@SA groups (Fig. 7b). By week 4, both EVs@SA and AuCu-Cur-EVs@SA groups showed some improvement, but the degeneration grade in the AuCu-Cur-EVs@SA group was significantly lower than in the EVs@SA group (Fig. 7c). X-ray and CT assessments of the Disc Height Index (DHI) yielded consistent results. At week 2, DHI was significantly reduced in the positive control, EVs@SA, and AuCu-Cur-EVs@SA groups (Fig. 7b). By week 4, disc height recovered in the treatment groups (Fig. 7d); however, the positive control and EVs@SA groups failed to reach the level of the negative control, whereas the AuCu-Cur-EVs@SA group showed no statistically significant difference in DHI from the negative control.

Fig. 7.

Therapeutic Efficacy of In Vivo AuCu-Cur-EVs@SA Injection for IVDD. a, Flow chart of animal experiment. b, Representative MRI, X-ray, Micro-CT images of rat tail intervertebral discs in each group at different time points. c, Quantitative analysis of Pfirrmann grade from MRI. d, Quantitative analysis of intervertebral DHI from micro-CT, n = 5 independent rats. Data are expressed as mean ± SD, *p < 0.05, **p < 0.01, ****p < 0.0001. e-f, Representative H&E staining and SO&FG staining images of rat tail intervertebral discs in each group at 8 weeks. g, Quantitative analysis of Histological score from H&E and SO&FG, scale bar: 100 μm and 1 mm, n = 5 independent rats. Data are expressed as mean ± SD, ****p < 0.0001. h-k, Representative immunofluorescence images of IL-1β and ACAN, representative immunohistochemical images of γ-H2AX. i, Statistical result of IL-1β. l, Statistical result of ACAN. scale bar: 150 μm and 1.5 mm, three independent experiments were repeated. Data are expressed as mean ± SD, ***p < 0.001, ****p < 0.0001. All of statistical analyses were performed using one-way ANOVA followed by Tukey’s multiple comparison test (c-d, g, i, l)

The therapeutic effect of AuCu-Cur-EVs@SA was further evaluated histologically. H&E staining revealed a significant reduction in NP area in the positive control, EVs@SA, and AuCu-Cur-EVs@SA groups compared to the negative control. However, tissue loss was substantially less in the AuCu-Cur-EVs@SA group than in the other two groups (Fig. 7e). Furthermore, while annulus fibrosus (AF) continuity was disrupted, laminar structure disordered, and the NP-AF border blurred at the puncture site in the positive control, EVs@SA, and AuCu-Cur-EVs@SA groups, the AuCu-Cur-EVs@SA group exhibited significantly better preservation. Safranin O & Fast Green (SO&FG) staining assessed proteoglycan levels. The positive control and EVs@SA groups showed significantly lower proteoglycan levels and more severe cartilage endplate degeneration than the negative control (Fig. 7f). The AuCu-Cur-EVs@SA group displayed intermediate proteoglycan levels between the EVs@SA and negative control groups. Histological scores confirmed this trend (Fig. 7g), with EVs@SA showing negligible therapeutic effect, while AuCu-Cur-EVs@SA significantly reduced the degeneration score compared to both the positive control and EVs@SA groups. Immunofluorescence and immunohistochemistry were used to assess the ability of AuCu-Cur-EVs@SA to scavenge ROS, repair DNA damage, and promote tissue regeneration. IL-1β staining showed that fluorescence intensity in the negative control and AuCu-Cur-EVs@SA groups was significantly lower than in the positive control and EVs@SA groups (Fig. 7h). The EVs@SA group showed no significant anti-inflammatory effect (Fig. 7i). Immunohistochemistry for γ-H2AX revealed that the proportion of γ-H2AX-positive cells in the negative control and AuCu-Cur-EVs@SA groups was significantly lower than in the positive control and EVs@SA groups (Fig. 7j). Aggrecan (ACAN), a chondrogenic marker, showed significantly higher fluorescence intensity in the negative control and AuCu-Cur-EVs@SA groups compared to the positive control and EVs@SA groups (Fig. 7k-l). Similarly, we examined the expression of Collagen II in sections of rat caudal intervertebral discs (Supplementary Fig. 10). The fluorescence intensity showed a similar trend to that of ACAN. The EVs@SA group failed to promote effective chondrogenesis, likely because upstream oxidative stress and DNA damage remained unaddressed. Finally, immunohistochemical analysis of p16 in the disc tissues demonstrated a significant reduction in the proportion of p16-positive cells in the material-treated group (Supplementary Fig. 11), confirming that the material alleviates senescence in NPCs and delays the progression of IVDD.

These results demonstrate that AuCu-Cur-EVs@SA alleviates IVDD by scavenging ROS, repairing DNA damage to reverse cellular senescence and SASP, and promoting extracellular matrix regeneration.

Discussion

The pivotal role of oxidative stress-induced DNA damage in both cellular senescence and the inflammatory microenvironment makes IVDD treatment particularly challenging. We successfully utilized AuCu-Cur to scavenge ROS and alleviate oxidative stress, thereby preventing further DNA damage. However, the bioavailability of the AuCu-Cur requires further enhancement, and crucially, the ROS-induced DNA damage itself remained unaddressed. To overcome these limitations, we employed hUCMSC-EVs to improve bioavailability. Within this system, curcumin promotes the expression of NEIL3 to repair damaged DNA. NEIL3 is a multifunctional DNA glycosylase, belonging to the helix–two-turn–helix (H2TH) superfamily, which performs base excision repair (BER) by excising oxidized bases and exhibiting DNA lyase activity [55–59]. Finally, the EVs themselves are enriched with various tissue-repair-related proteins and RNAs, which promote DNA replication in NPCs, reverse the senescent state, and stimulate the synthesis of extracellular matrix. In a rat tail needle puncture-induced IVDD model, we confirmed the strong antioxidant and DNA repair capabilities of this material, along with its efficacy in delaying cellular senescence, successfully alleviating IVDD progression. However, it is important to note the limitations of our animal model. The structure of the rat caudal disc differs significantly from the human lumbar disc. Furthermore, given the multifactorial complexity of human IVDD—including age-related degeneration, mechanical loading, inflammatory processes, and genetic factors—the needle puncture model in rat tails does not fully recapitulate the human disease process [60]. In conclusion, we have developed a continuous DNA repair system designed for the sequential treatment of IVDD, offering a novel DNA repair-based strategy for tackling aging-related diseases.

Methods

Single-cell transcriptome analysis

UMAP: Generated using the RunUMAP function in the Seurat package, with reduction set to “harmony”.

Cell Proportions: The cellular proportions of subpopulations within each group were calculated and visualized using ggplot 2.

DotPlot: Created using the DotPlot function in the Seurat package.

Cell culture

Human NP tissue samples were obtained from patients undergoing surgical treatment for lumbar vertebral fractures, idiopathic scoliosis, lumbar disc herniation, or lumbar spondylolisthesis. hBMSCs were isolated from bone marrow aspirates of healthy donors. All participants were recruited through the Department of Orthopedic Surgery at Wuhan Union Medical College Hospital, with informed consent and ethical approval granted by the Institutional Review Board of Tongji Medical College, Huazhong University of Science and Technology (No. S341). For hBMSC isolation, bone marrow aspirates were diluted with sterile PBS, layered over a density gradient lymphocyte separation medium, and centrifuged to isolate mononuclear cells. The resulting hBMSCs and NPCs were expanded in DMEM/F12 culture medium (Thermo Fisher Scientific, USA), supplemented with 10% fetal bovine serum (FBS; Gibco). All cell cultures were maintained at 37 °C in a humidified incubator with 5% CO₂.

The human monocyte leukemia cell line (THP-1) was obtained from the American Type Culture Collection (ATCC, USA). THP-1 cells were cultured in RPMI 1640 medium supplemented with 10% FBS (Gibco, USA) and 0.05 mM β-mercaptoethanol (Sigma, USA) at a density of 5 × 10⁵ cells/mL at 37 °C under 5% CO₂. The culture medium was replaced every two days. For all experiments, THP-1 monocytes were seeded in 6-well plates and treated with 100 ng/mL PMA (Abclonal, China) for 24 h to differentiate into adherent macrophages.

Experimental groups were designated as follows:

Control: PBS.

LPS: 10 µM LPS.

AuCu: 10 µM LPS + 5 ppm AuCu for 1 day.

Cur: 10 µM LPS + 5 ppm Cur for 1 day.

AuCu-Cur: 10 µM LPS + 5 ppm AuCu-Cur for 1 day.

EVs: 10 µM LPS + 100 ppm EVs for 1 day.

AuCu-Cur-EVs: 10 µM LPS + 100 ppm AuCu-Cur-EVs for 1 day.

For macrophage polarization, cytokines were administered at concentrations of 100 ng/mL LPS (Abclonal, China) with 20 ng/mL IFN-γ (Abclonal, China), or 20 ng/mL IL-4 (Abclonal, China) with 20 ng/mL IL-13 (Abclonal, China) to induce M1 and M2 phenotypes, respectively.

RNA sequencing

NPs were allocated into three experimental cohorts:

Control group: Cultured in basal medium (DMEM/F12 + 10% FBS).

LPS group: Treated with 10 µM LPS (L2630, Sigma-Aldrich) in basal medium.

AuCu-Cur-EVs group: LPS group + exposed to 100 ppm AuCu-Cur-EVs in basal medium.

Total RNA was extracted using TRIzol™ Reagent (15596026, Invitrogen, USA) following the manufacturer’s protocol. Then sequenced on an Illumina NovaSeq 6000 platform (Illumina, USA), generating ~ 40 million paired-end reads per sample. Raw reads were processed using a standardized pipeline with the assistance of Shanghai Bioprofile Technology Co., Ltd. (Shanghai, China):

DEGs: DESeq2 (v1.34.0) identified DEGs (|log2FC| > 1, adjusted p < 0.05).

Functional Enrichment:

GO: Biological processes, molecular functions, and cellular components were analyzed via clusterProfiler (v4.4.4).

GSEA: Pre-ranked GSEA (v4.3.2, Broad Institute) was performed to evaluate enrichment of hallmark gene sets from the Molecular Signatures Database (MSigDB v2023.1). Genes were ranked by log2 (fold change), and significance was determined using 1,000 permutations with a normalized enrichment score (NES) threshold of |NES| > 1.0 and false discovery rate (FDR) < 0.25.

Volcano Plot: DEGs were visualized using volcano plots.

PCA Analysis: PCA was performed using the prcomp package in R. The PC1 and PC2 scores from three groups were extracted and visualized using the ggplot2 package.

Venn Diagram: Differentially expressed genes from pairwise comparisons among three groups were identified, and their overlaps were visualized using the VennDiagram function in R.

Preparation of AuCu, AuCu-Cur and AuCu-Cur-EVs

Preparation of AuCu: First, 10 mL of ultrapure water was added to a 25 mL beaker. Then, 39.4 mg of chloroauric acid (HAuCl₄·3 H₂O, 0.1 mmol) and 17.0 mg of copper(II) chloride (CuCl₂·2 H₂O, 0.1 mmol) were dissolved in the water and magnetically stirred at 500 rpm for 5 min to form a blue-green transparent solution. Next, 153.6 mg of reduced glutathione (GSH, 0.5 mmol) was added, and the stirring was continued for 30 min until the solution turned light yellow. The mixture was subsequently injected dropwise into 1 mL of freshly prepared 0.1 M sodium borohydride solution. A hydrothermal reaction was then conducted for 3 h to obtain the AuCu nanocluster solution.

Preparation of AuCu-Cur: To prepare a 10 mM stock solution, 3.68 mg of high-purity curcumin was dissolved in 1 mL of anhydrous ethanol and mixed via ultrasonic treatment for 10 min. Subsequently, 5 mL of the synthesized AuCu nanocluster solution was transferred to a dark brown bottle. Under continuous magnetic stirring (600 rpm), 0.5 mL of the curcumin ethanol stock solution was slowly added dropwise at a rate of 0.1 mL/min using a microinjection pump. Following the addition, the mixture was stirred at 25 °C in the dark for 24 h to reach adsorption equilibrium. The mixture was then transferred into a 1000 Da dialysis bag and dialyzed for 12 h to remove free curcumin. The final AuCu-Cur solution was stored at 4 °C in the dark.

Preparation of AuCu-Cur-EVs complex: To isolate extracellular vesicles (EVs), the cell culture supernatant was sequentially centrifuged at 300 × g for 10 min to remove dead cells, at 1000 × g for 20 min to eliminate cellular debris, and at 10,000 × g for 30 min to deplete large vesicles. The resulting supernatant was then passed through a 0.22 μm PES filter membrane. The filtrate was transferred to an ultracentrifuge tube and centrifuged at 110,000 × g for 30 min at 4 °C to pellet the exosomes, which were subsequently resuspended in precooled 1× PBS. A 1 mL aliquot of the AuCu-Cur complex solution was equilibrated at 4 °C. While maintaining the complex solution in an ice bath under magnetic stirring (300 rpm), 200 µL of the exosome suspension was gradually added at a rate of 0.5 µL/s. After incubation, the mixed solution was transferred to an ultracentrifuge tube and centrifuged at 140,000 × g for 90 min at 4 °C. The supernatant was discarded, and the pellet was gently resuspended in 2 mL of precooled HEPES buffer. This centrifugation and resuspension process was repeated twice to completely remove unbound complexes. Finally, the precipitate was resuspended in 200 µL of HEPES buffer to yield the AuCu-Cur-EVs complex suspension.

CCK8 assay

NPCs and hBMSCs were pre-seeded in 96-well plates and cultured in complete medium. After allowing the cells to adhere for 24 h to ensure stabilization, experimental treatments were administered. The experimental design comprised six groups: blank control group, Group AuCu, Cur, AuCu-Cur, EVs, and AuCu-Cur-EVs. Each group was tested at 6 concentrations (0, 25, 50, 75, 100, and 150 ppm) to determine the optimal treatment dose. Cell viability was assessed at 24, 48, and 72 h post-treatment. To mitigate potential colorimetric interference from the test materials, the original medium was replaced with fresh serum-free medium prior to adding CCK-8 reagent (RM02823, ABclonal, China) at its working concentration. Following a 2-hour incubation at 37 °C, 100 µL of supernatant from each well was transferred to a new 96-well plate. Absorbance at 450 nm was quantified using a microplate reader. Cell viability was calculated using the following formula:

cell viability (%)=[(As-Ab)/(Ac-Ab)]x100% (1).

Where:

As : Absorbance of experimental wells (containing cells, medium, CCK-8, and test compounds).

Ab: Absorbance of blank wells (containing medium and CCK-8 only).

Ac: Absorbance of control wells (containing cells, medium, and CCK-8).

Hemolysis test

Whole blood (3 mL) was collected from a rat and subjected to centrifugation at 3000 rpm for 15 min to separate cellular components. The serum fraction was carefully aspirated, retaining the pelleted blood cells for subsequent processing. The cellular fraction was washed three times with PBS using centrifugation at 200×g for 15 min per cycle, with complete supernatant removal after each centrifugation step. Following the final wash, erythrocytes were resuspended in saline to prepare a standardized suspension. Test samples (AuCu, Cur, AuCu-Cur, EVs, AuCu-Cur-EVs) were introduced into the erythrocyte suspension to attain a final working concentration of 100 µg/mL. Experimental controls were established as follows: (1) A positive control consisting of 20 µL ultrapure water (ddH2O) added to an equivalent erythrocyte suspension aliquot; (2) A negative control comprising untreated erythrocyte suspension; (3) A background control containing AuCu-Cur-EVs solution with matched components to account for potential spectral interference from nanomaterial coloration during spectrophotometric analysis. All experimental EP tubes were maintained at 37 °C for 2 h under static incubation conditions. Post-incubation, samples were centrifuged at 200 × g for 15 min to pellet cellular debris. Aliquots (200 µL) of the clarified supernatant were collected for spectrophotometric quantification of hemoglobin release, with absorbance measurements recorded at 540 nm using an appropriate blank correction. The percentage of hemolysis was calculated using the following formula:

Hemolysis(%)=(OD_Samples-OD_Background)/(OD_{PositiveControl}-OD_{Negative Control})×100% (2).

DCFH-DA staining

Following the respective treatments, cells were incubated with 10 µM DCFH-DA (Beyotime, S0033S), prepared according to the manufacturer’s instructions, for 20 min at 37 °C. The DCFH-DA solution was then removed, and the cells were washed three times with PBS before observation under a confocal microscope. For FC analysis, cells were incubated with DCFH-DA following the same procedure, collected into flow cytometry tubes, washed three times with PBS, and then analyzed.

JC-1 staining

After the respective treatments, cells were stained using the JC-1 Assay Kit (Solarbio, M8650) by incubating for 30 min at 37 °C. The JC-1 staining solution was then removed, and the cells were washed three times with PBS before imaging with a confocal microscope. For flow cytometry, cells were processed identically, collected into flow cytometry tubes, washed three times with PBS, and subsequently analyzed.

Flow cytometry

Adherent cells were detached using trypsin and neutralized with complete RPMI-1640 medium. The cell suspension was centrifuged at 1000 rpm for 5 min, and the pellet was resuspended in 200 µL of PBS. For each sample, 200,000 cells were aliquoted into a tube, fixed, and blocked at 4 °C for 30 min. After three washes, cells were incubated in the dark with PE-Cy7-conjugated anti-CD86 antibody (BD Pharmingen, USA) for 40 min. Following another wash cycle, cells were permeabilized using BD Cytofix/Cytoperm solution (BD Pharmingen, USA) and incubated in the dark for 30 min. Cells were washed again and then stained in the dark with APC-conjugated anti-CD206 antibody (BD Pharmingen, USA) for 40 min. After a final series of washes, cells were resuspended in 200 µL of PBS for analysis.

All flow cytometry data were acquired on a BD FACSymphony A1 flow cytometer (BD Biosciences, USA), recording signals from the appropriate channels. Subsequent data analysis was performed using FlowJo software (BD Biosciences, USA).

Immunofluorescence

NPCs and hBMSCs were cultured in 12-well plates until reaching 70–80% confluency. After aspirating the culture medium, cells were fixed with 4% paraformaldehyde (PFA) for 30 min at room temperature, followed by permeabilization with 0.5% Triton X-100 in PBS for 20 min. Samples were washed three times with PBS (5 min per wash) and blocked with immunofluorescence blocking buffer (P0102, Beyotime, China) for 20 min at room temperature. Primary antibodies (diluted in TBST containing 1% BSA) were applied to the cells and incubated overnight at 4 °C in a humidified chamber. The next day, unbound antibodies were removed by three 5-minute washes with 0.1% PBST on an orbital shaker. Species-matched fluorescent secondary antibodies were incubated with the cells for 60 min at room temperature, protected from light. For cytoskeletal visualization, cells were stained with 200 µL of TRITC-conjugated phalloidin working solution for 30 min in the dark, followed by three 5-minute PBST washes. Nuclei were counterstained with DAPI (C1002, Beyotime, China) for 10 min at room temperature. Coverslips were mounted onto glass slides using anti-fade mounting medium. Fluorescence signals were initially screened using an inverted fluorescence microscope (Leica DMi8). High-resolution z-stack images were acquired with a confocal laser scanning microscope (FV3000, Olympus) using standardized excitation/emission parameters for each fluorophore. Image processing and colocalization analyses were performed using FV31S-SW (Olympus) and ImageJ (NIH, USA).

qPCR

Total RNA was extracted using the RNAeasy Kit (Beyotime, R0026) according to the manufacturer’s protocol. Complementary DNA (cDNA) was synthesized using the HiScript III RT SuperMix for qPCR (Vazyme, China). Quantitative PCR (qPCR) was performed using the AceQ qPCR SYBR Green Master Mix (Vazyme, China) on a CFX96 Touch Real-Time PCR Detection System (Bio-Rad, USA). The relative fold change in gene expression was calculated using GAPDH as the internal reference gene.

WB

Total cellular proteins were extracted using RIPA lysis buffer (BL504A, Biosharp, China) supplemented with a protease and phosphatase inhibitor cocktail (BL615A/BL612A, Biosharp, China). Protein lysates were resolved via sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and subsequently transferred onto polyvinylidene fluoride (PVDF) membranes using a semi-dry transfer system. Membranes were blocked for 1 h at room temperature with 5% bovine serum albumin (BSA; BS114, Biosharp, China) dissolved in 0.1% TBST. Primary antibodies (detailed in Supplementary Table S1) were diluted in blocking buffer and incubated with membranes overnight at 4 °C. Following three 10-minute washes with 0.1% TBST, membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 1 h at room temperature. Unbound antibodies were removed by three additional TBST washes (10 min each). Protein bands were detected using an enhanced chemiluminescence (ECL) substrate and visualized using a chemiluminescence imaging system (Tanon 5200, Tanon, China). Band intensity was quantified using ImageJ software (NIH, USA) with normalization to loading controls.

Alcian blue staining

hBMSCs were cultured in 6-well plates for 21 days under 7 experimental conditions:

(1) Basal medium(Negative control): Standard growth medium (DMEM/F12 supplemented with 10% FBS).

(2) Chondrogenic differentiation medium(Positive control): Basal medium supplemented with chondrogenic induction factors (TGF-β3, Dexamethasone, Sodium pyruvate, ITS, Ascorbic acid).

(3) AuCu treatment: Basal medium supplemented with 5 ppm AuCu.

(4) Cur treatment: Basal medium supplemented with 5 ppm Cur.

(5) AuCu-Cur treatment: Basal medium supplemented with 5 ppm AuCu-Cur.

(6) EVs treatment: Basal medium supplemented with 100 ppm EVs.

(7) AuCu-Cur-EVs treatment: Basal medium supplemented with 100 ppm AuCu-Cur-EVs.

Cells maintained in basal medium served as the control group. Following the 21 day differentiation period, cells were washed twice with PBS and fixed with 4% PFA for 30 min at room temperature. Each well was incubated with 100 µL of Alcian Blue staining solution (pH 2.5; Beyotime, C0155S) in a humidified chamber for 1 h at room temperature. Stain solution was aspirated, and cells were gently rinsed with distilled water (three 10-second washes). Plates were air-dried prior to microscopic imaging. Glycosaminoglycan (GAG) deposition, indicative of chondrogenic differentiation, was visualized and documented using a phase-contrast microscope (Leica DMi8).

Safranine O and fast green cartilage staining

Cells maintained in basal medium served as the control group. Post-culture, cells were washed with PBS and fixed in 4% PFA for 30 min at room temperature. Each well was incubated with 100 µL of 0.1% Fast Green staining solution (Beyotime, C0621S) for 5 min. Excess stain was removed by briefly rinsing with running tap water until cartilaginous matrix appeared colorless. Then cells were immersed in 1% acetic acid solution for 30 s to enhance staining specificity. Rinsed with running tap water for 10–30 s. Treated with 100 µL of 0.1% Safranin O solution (Beyotime, C0621S) for 8 min to visualize proteoglycan-rich matrix. Rapidly dehydrated through three sequential 30-second washes in absolute ethanol. Air-dried plates were imaged using an inverted phase-contrast microscope (Leica DMi8).

Senescence β-Galactosidase Staining

Following the respective treatments, cells were stained using the Senescence β-Galactosidase Staining Kit (Beyotime, C0602). After overnight incubation at 37 °C, cells were observed and imaged under an optical microscope (Olympus, Japan).

Animal experiment

Sprague-Dawley (SD) rats purchased from BIONT were selected as the experimental animal model. All rats were male, weighing between 180 and 220 g. All experiments were approved by the Animal Research Committee of Huazhong University of Science and Technology (NO.4721) and conducted in accordance with international animal welfare standards.

The SD rats were housed under standardized conditions in a specific pathogen-free (SPF) environment, with a constant temperature of 21–24 °C and a 12-hour light/dark cycle (1:1). Animals were randomly assigned to groups using a blinded method for in vivo experiments. Rats were first anesthetized via intraperitoneal injection of 3% pentobarbital sodium (0.3 mg/kg). After confirming the target coccygeal intervertebral disc segment, surgery was performed on the rat tail. The IVD was punctured using a needle with a diameter of 0.5 mm (25G). For the EVs@SA group and the AuCu-Cur-EVs@SA group, the respective hydrogel (pre-mixed 5 min prior) was injected into the puncture site using a syringe. The concentrations of EVs and AuCu-Cur-EVs were consistent with those used in the in vitro cell experiments (100 µg/mL). 28 days post-treatment, the corresponding coccygeal IVD segments were harvested for histological analysis, including Safranin O-Fast Green staining, H&E staining, IHC, and IF. Additionally, the heart, liver, spleen, and kidneys were collected for H&E staining. Complete blood count (CBC) and assessments of liver and kidney function were also performed on day 14 post-treatment.

Hematological analysis

Peripheral blood samples were collected via retro-orbital venous plexus puncture from rats under inhaled anesthesia (isoflurane) on postoperative day 14. Whole blood anticoagulated with K2EDTA was analyzed using a veterinary hematology analyzer (BC-2800vet, Mindray, China) to quantify erythrocyte, leukocyte, and platelet parameters.

Fresh blood samples were collected in anticoagulant tubes and mixed thoroughly. After centrifugation at 3000 × g for 10 min, the upper pale yellow supernatant was collected for subsequent analysis. Liver function parameters (ALT, AST, ALP), glucose (GLU), and renal function parameters (Cr, UA) were measured.

Radiological analysis

Following sacrifice, rat caudal vertebrae were collected and subjected to X-ray, micro-CT, and MRI analyses.

X-ray: tissues were imaged using a DRX Ascend System (Carestream, Canada). We use Bruker MI SE for images visualization.

Micro-CT: tissues were imaged using a high-resolution micro-CT system (SkyScan 1176, Bruker, Belgium) at an isotropic voxel size of 18 μm, with acquisition parameters set to 50 kVp and 500 µA. Three-dimensional reconstructions were generated using CTVox software (v3.3.0, Bruker).

MRI: tissues were imaged by a fast spin echo sequence, 0.5 mm slice thickness, 3000 ms repetition time, 70 ms echo time and 7.0 T magnetic field intensity using a MRI Scanning System (Bruker, Germany). We use MRIcroN to achieve visualization of the images.

Histopathological evaluation

Tissue specimens (excluding cardiac, hepatic, splenic, lung and renal samples) were decalcified in 10% EDTA (pH 7.4) for 21 days, paraffin-embedded, and sectioned at 5 μm thickness. Serial sections were stained with:

Hematoxylin & Eosin (H&E): For general histoarchitectural assessment.

Safranin O & Fast Green (SO&FG): For general histoarchitectural assessment.

Stained slides were mounted with neutral resin and imaged under a brightfield microscope (Olympus, Japan).

DHI: Measurements were performed on reconstructed CT images.

Comet assay

After the respective treatments, cells were harvested and the comet assay was performed according to the manufacturer’s instructions using a Comet Assay Kit (C2041S, Beyotime) and specially treated glass slides (FSL061, Beyotime). Finally, the cells were observed under a fluorescence microscope, and quantitative analysis was conducted using the CASP software (CaspLab, Comet Assay Software Project Lab).

Olive Tail Moment=Tail DNA (%) × Tail Moment Length (3).

Statistical analysis

Continuous data are presented as mean ± standard deviation (SD) from ≥ 3 independent biological replicates. Intergroup comparisons were analyzed using two-way ANOVA with Bonferroni post hoc correction for multi-group datasets, or one-way ANOVA followed by Tukey’s multiple comparison test. Significance thresholds were defined as: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns (non-significant): p ≥ 0.05. All analyses were performed using GraphPad Prism 9.0 (USA) with experimenters blinded to group assignments during data acquisition and processing. No outliers were excluded. All experimental procedures, including animal group allocation and cell treatments, were conducted with strict randomization. Data collection and outcome analyses were performed by investigators blinded to the treatment conditions. Additionally, all key experiments were repeated independently using different batches of synthesized vesicles to eliminate potential batch effects, thereby validating the reliability and reproducibility of the therapeutic outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

K.W., C.Y., S.P., Z.Z. and L.T. designed the experiments. S.P., Y.D., R.W. and M.Z. performed most of the experiments and analyzed the data with the assistance of X.Zhang., J.L., and D.Z., B.T., X.Zhou., H.L., H. Xu., D.W., and H.W. assisted in the synthesis of materials. Z.O., J.W., Z.D., X.L., R.S., S.Z., and Y.C. helped to perform the animal surgery. K.W., C.Y., Y.L., and H.L. collected the clinical specimens. S. P., K.W., and C.Y. wrote this manuscript.

Funding

This work was supported by the National Science Foundation of China (No. 82130072, No. 82072505, No. 82472511) and ShenZhen Science and Technology Program (SGDX20230116093544006, JCYJ20240813153421028, JCYJ20250604190927038).

Data availability

The datasets generated and/or analyzed during the current study have been deposited in the China National Center for Bioinformation (CNCB) database under the accession number HRA017878. The data will be made available upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.