ROS‐Responsive Wedelolactone Hydrogel Promotes Intervertebral Disc Repair by Disrupting the NF‐κB–LCN2 Inflammatory Feedback Loop
Zimei Wu, Yan Xu, Lang Qin, Zemin Ling, Yingjie Mai, Xiong Tian Guo, Jiajia Chen, Leping Yan, Lin Wang, Liming Bian, Fuxin Wei
TL;DR
A responsive hydrogel delivers wedelolactone to treat disc degeneration by targeting an inflammatory loop, restoring disc structure and function in rats.
Contribution
A ROS-responsive hydrogel is developed to specifically target the NF-κB–LCN2 inflammatory feedback loop for intervertebral disc repair.
Findings
WPG treatment suppressed NF-κB activation and LCN2 in macrophages and AF cells.
WPG preserved disc height and MRI signal while improving biomechanics in a rat model.
RNA-seq showed restoration of ECM genes and downregulation of inflammatory pathways.
Abstract
Intervertebral disc degeneration (IVDD) is driven by persistent inflammation–oxidative stress that disrupts annulus fibrosus (AF) homeostasis. Guided by network pharmacology and docking, we prioritized the NF‐κB–LCN2 axis as a druggable target of wedelolactone (WDL). To achieve targeted modulation, we engineered a dual‐network ROS‐responsive hydrogel (WPG) in which a phenylboronic‐ester/PVA redox‐cleavable network interpenetrates a covalently crosslinked GelMA–elastin matrix, enabling mechanically robust yet stimulus‐triggered WDL release. WDL suppressed NF‐κB activation and downregulated LCN2 in both macrophages and AF cells. Conditioned‐medium co‐culture demonstrated that WDL disrupts macrophage‐derived LCN2‐mediated paracrine amplification, breaking the self‐sustaining inflammatory loop. Bulk RNA‐seq across both cell types revealed coordinated downregulation of NF‐κB – driven…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
SCHEME 1
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7
FIGURE 8
FIGURE 9
FIGURE 10- —National Key Research and Development Program of China10.13039/501100012166
- —National Natural Science Foundation of China10.13039/501100001809
- —Shenzhen Key Laboratory of Bone Tissue Repair and Translational Research
- —Shenzhen Science and Technology Program10.13039/501100017610
- —Guangdong Basic and Applied Basic Research Fund
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsSpine and Intervertebral Disc Pathology · Cervical and Thoracic Myelopathy · Medical Imaging and Analysis
Introduction
1
Intervertebral disc degeneration (IVDD) is a leading cause of chronic low back pain and disability, imposing substantial medical and socioeconomic burdens [1, 2, 3]. A defining hallmark is the loss of extracellular matrix (ECM) homeostasis within a persistent inflammatory–oxidative niche, where catabolic enzymes and reactive oxygen species (ROS) synergistically deplete proteoglycans and collagens, disrupt disc architecture, and impair biomechanics [4, 5]. Despite advances in surgical decompression and conservative management, disease‐modifying strategies capable of reprogramming this hostile microenvironment remain lacking, particularly during in the early and intermediate stages of IVDD when biological rescue is still feasible [6, 7].
Inflammation‐driven crosstalk between infiltrating immune cells and disc‐resident stromal cells has emerged as a central driver of degeneration. Macrophages release cytokines and alarmins that enforce catabolism in annulus fibrosus cells (AFCs) and sustain oxidative stress [8]. Among inflammatory amplifiers, lipocalin‐2 (LCN2) is consistently upregulated and tightly coupled to nuclear factor‐κB (NF‐κB) signaling [9, 10]. NF‐κB directly induces LCN2 under inflammatory stress, while exogenous or exosome‐delivered LCN2 can reactivate NF‐κB in disc cells—amplifying cytokine release and cellular senescence—thus forming a self‐reinforcing NF‐κB–LCN2 inflammatory feedback loop [11]. Beyond transcriptional crosstalk, LCN2 binds matrix metalloproteinase 9 (MMP9) to stabilize its activity and promotes ECM proteolysis [12], thereby coupling inflammation to matrix breakdown. Whether this NF‐κB–LCN2 amplifier operates bidirectionally between macrophages and AFCs—and whether interruption can restore matrix homeostasis—remains unclear.
Translating mechanism insight into therapy is particularly challenging in the avascular, load‐bearing disc. Systemic small molecules often exhibit poor intradiscal bioavailability and off‐target toxicity, while existing responsive hydrogels rarely meet the dual demands of pathology‐specific release and mechanical robustness under compression [13, 14, 15]. ROS‐responsive systems are appealing because elevated ROS is a biochemical hallmark of degeneration [16]; and several studies have reported protective effects of ROS‐degradable depots in IVDD models [17, 18]. However, few formulations simultaneously achieve ROS‐triggered drug release and long‐term structural stability in situ [19]—two prerequisites for reliable therapeutic performance within this confined mechanical niche [20, 21].
Wedelolactone (WDL), is a coumarin‐type natural product that is a potent NF‐κB pathway inhibitor that limits IKK activation, suppresses P‐p65 nuclear translocation, and attenuates proinflammatory mediators such as IL‐6, TNF‐α, COX‐2, and iNOS [22, 23]. Concurrently, WDL exerts antioxidant effects through radical scavenging and upregulation of endogenous defense enzymes [24]. Nevertheless, its poor solubility and short residence time restrict local application, necessitating a delivery vehicle capable of releasing WDL in response to pathological cues while maintaining mechanical integrity [25]. To address this, we engineered an injectable dual‐network ROS‐responsive hydrogel (WPG) by integrating a phenylboronic‐ester/ poly(vinyl alcohol) (PT) redox‐labile within a covalently crosslinked GelMA–elastin (GAE) matrix [26, 27], achieving both ROS‐gated drug release and mechanical competence.
We hypothesized that WDL delivered via WPG disrupts a macrophage–AFC NF‐κB–LCN2 feedback circuit, alleviates oxidative and catabolic stress, and restores ECM homeostasis to promote disc repair. To test this hypothesis, we (i) applied network pharmacology and molecular docking to prioritize NF‐κB–LCN2 as a WDL‐responsive regulatory axis; (ii) established macrophage‐to‐AFC paracrine models incorporating recombinant LCN2 rescue and pharmacologic LCN2 inhibition, complemented by bulk RNA‐seq to profile transcriptomic remodeling; and (iii) evaluated structural (MRI T_2_ signal, disc height, histology), molecular (ECM and catabolic markers), and biomechanical outcomes in a rat annulus fibrosus defect model, alongside biosafety verification.
In summary, this study identifies a disc‐relevant NF‐κB–LCN2 inflammatory amplifier, delineates its contribution to matrix degeneration, and presents a mechanically robust, ROS‐gated WDL delivery system that concurrently interrupts inflammatory signaling and restores ECM integrity. By integrating mechanistic discovery with delivery logic, we provide a synergistic framework for stimuli‐responsive, disease‐modifying therapy in IVDD. The overall design rationale and proposed working mechanism of the ROS‐responsive WDL‐loaded hydrogel (WPG) for disrupting the NF‐κB–LCN2 feedback loop and promoting intervertebral disc repair are illustrated in Scheme 1.
Design rationale and mechanism of WDL‐loaded ROS‐responsive hydrogel (WPG) for intervertebral disc repair via disruption of the NF‐κB–LCN2 inflammatory feedback loop. The WPG hydrogel integrates a dynamic TSPBA–PVA (PT) redox‐sensitive network with a stable GelMA–elastin (GEA) framework, enabling both mechanical stability and ROS‐triggered drug release. Wedelolactone (WDL), a plant‐derived NF‐κB inhibitor, is released in response to oxidative stress to target p65, LCN2, and MMP9, thereby blocking the NF‐κB–LCN2 feedback loop between macrophages and AFCs. This dual inhibition alleviates oxidative and catabolic stress, promotes ECM remodeling, and facilitates structural and functional regeneration of the degenerative disc in vivo.
Results
2
Network Pharmacology and Molecular Docking Identify the NF‐κB–LCN2 Axis as a Key Actionable Target of WDL in IVDD
2.1
WDL, a plant‐derived NF‐κB inhibitor, was analyzed against IVDD through an integrated network‐pharmacology workflow. Intersecting WDL‐associated targets with IVDD‐related genes yielded 54 overlapping candidates (Figure 1A; drug–disease mapping in Figure S1). Gene Ontology (GO) enrichment revealed strong associations with inflammatory regulation, oxidative and hypoxic response, and ECM organization and catabolism. Cellular component (CC) terms were enriched in extracellular and membrane‐associated compartments, while molecular function (MF) terms emphasized chemokine receptor binding, metallopeptidase activity, and integrin binding (Figure 1B). KEGG pathway analysis highlighted the IL‐17/TNF/NOD‐like receptor/NF‐κB axis, together with PI3K–AKT/relaxin and adhesion–mechanotransduction pathways, as well as apoptotic at the efferocytic modules (Figure 1C). Although NF‐κB was not the nominally top single pathway, its position at the intersection of inflammatory, oxidative, and matrix‐remodeling cascades indicates its role as a central regulatory hub in IVDD.
Network pharmacology and molecular docking identify the NF‐κB–LCN2 axis as a key actionable target of WDL in IVDD. (A) Venn diagram showing 54 overlapping genes between IVDD‐related targets and WDL‐associated proteins. (B) GO enrichment (BP/CC/MF) indicating significant enrichment in inflammatory and oxidative responses, ECM organization/proteolysis, and adhesion‐related functions. Bubble size denotes gene count; color encodes −log10 (P‐value). (C) KEGG pathway analysis showing enrichment in IL‐17/TNF/NOD‐like receptor/NF‐κB signaling, PI3K–AKT/relaxin, adhesion–mechanotransduction, and cell‐fate modules (apoptosis/efferocytosis). (D) Simplified STRING PPI subnetwork connecting upstream cytokines (TNF/IL‐6) through MAPK14/PTK2/AKT1/KEAP1 to LCN2/MMP9 effectors. Node size ∝ degree centrality. Color code—red: inflammatory drivers; gold: LCN2; purple: MMP9; teal: adhesion; blue: redox; gray: cell‐fate regulators. (E–G) Molecular docking of WDL with p65/RELA, LCN2, and MMP9. WDL shown in yellow; key residues in green; interaction distances in Å. Enrichment analysis (B,C) was performed using over‐representation analysis with Benjamini–Hochberg FDR correction (FDR < 0.05). The PPI network (D) was built using STRING (confidence ≥ 0.4) and visualized in Cytoscape. Docking (E–G) was performed with AutoDock Vina, and binding poses were visualized in PyMOL (scores in kcal/mol).
To distill the dense interaction network into an interpretable structure, a high‐confidence PPI subnetwork was extracted from STRING, emphasizing actionable nodes (Figure 1D). This simplified network connects upstream cytokines (TNF, IL‐6) through MAPK14, PTK2, AKT1, and KEAP1 toward LCN2/MMP9 effectors, linking inflammatory, adhesion, and redox signaling to ECM degradation. The architecture aligning with a working model of the NF‐κB–LCN2 positive‐feedback loop, in which inflammation and proteolysis reinforce each other. The complete PPI map (Figure S2) places CASP3, EGFR, TNF, IL‐6, MMP9, and BCL2 among high‐centrality hubs, corroborating the enrichment results and supporting the curated backbone.
At the structural level, molecular docking validated potential direct interactions of WDL with multiple targets. WDL exhibited favorable binding energies with p65/RELA (near LEU‐32, ΔG ≈ −6.6 kcal mol^−1^), LCN2 (PRO‐112, ≈ −6.2 kcal mol^−1^), and MMP9 (ASN‐188, ≈−6.8 kcal mol^−1^) (Figure 1E–G), suggesting a multipronged mechanism involving: (i) blockade of NF‐κB transcriptional activation, (ii) suppression of LCN2‐mediated inflammatory amplification, and (iii) attenuation of MMP dependent ECM degradation. Together, these analyses prioritize the NF‐κB–LCN2 axis as a central, druggable regulatory node and establish a mechanistic foundation for targeting inflammatory and catabolic loops through the WDL/WPG strategy.
WDL Attenuates Macrophage Inflammation via NF‐κB Inhibition and Mitigates Oxidative Stress in Disc Cells
2.2
To establish safe and effective working concentrations, CCK‐8 assays showed that WDL was non‐cytotoxic up to 6.25 µg mL^−1^ and alleviated LPS‐induced viability loss in RAW 264.7 macrophages. Comparable safety and protective profiles were observed in AFCs exposed to oxidative stress (Figure S3A–D). Therefore, low‐dose WDL (1.5625–6.25 µg mL^−1^; LW1–LW3) was selected for subsequent experiments.
Macrophage NF‐κB Inhibition and Immunomodulation
2.2.1
LPS stimulation robustly induced nuclear translocation of phospho‐p65 (P‐p65), as revealed by immunofluorescence staining (Figure 2A). Quantitative analysis of nuclear P‐p65 fluorescence intensity (Figure S4) showed an approximately 3.5‐fold increase upon LPS challenge, which was dose‐dependently reversed by WDL. At the highest dose (WDL3), the nuclear P‐p65 level was reduced to about one‐third of that in the LPS group (p < 0.0001), indicating effective suppression of NF‐κB activation. Consistently, qPCR analysis confirmed that WDL significantly attenuated LPS‐induced upregulation of canonical NF‐κB‐dependent inflammatory genes, including IL‐6, IL‐1β, TNF‐α, Nos2 (iNOS), Ptgs2 (Cox‐2), and Vcam1 (Figure 2B).
*WDL suppresses NF‐κB–driven inflammation in macrophages and mitigates oxidative stress in disc cells. (A) Immunofluorescence staining showing LPS‐induced P‐p65 nuclear translocation and its dose‐dependent inhibition by WDL (1.5625–6.25 µg mL−1). (B) qPCR analysis of NF‐κB target genes (IL‐6, IL‐1β, TNF‐α, iNOS, Cox‐2, Vcam1). (C) Flow cytometry plots of CD86⁺CD11b⁺ (M1) and CD206⁺CD11b⁺ (M2) subsets; (D,E) corresponding quantifications. (F) Immunofluorescence analysis of IL‐10 and TNF‐α expression in macrophages under LPS ± WDL. (G) DCFH‐DA staining showing WDL‐mediated suppression of H2O2‐induced ROS in Raw264.7cells. (H,I) Antioxidant assays showing increased SOD activity (H) and GSH levels (I) in NPCs and AFCs following WDL treatment. (J) Schematic summarizing dual actions of WDL: NF‐κB inhibition and ROS scavenging. Created with BioRender.com. Scale Bar: (A) 30 µm; (F,G) 20 µm. Data are presented as mean ± SD. Sample size (n) represents independent biological replicates and is indicated in each panel (or in the corresponding quantification plot). Statistics: One‐way ANOVA with Tukey's multiple‐comparisons test was used for panels (B,D,E); two‐way ANOVA with Tukey's multiple‐comparisons test was used for panels (H,I). Significance: *p < 0.05, **p < 0.01, ***p < 0.001, ***p < 0.0001; ns, not significant.
Flow cytometric analysis revealed that LPS increased the proportion of CD86⁺CD11b⁺ (pro‐inflammatory) macrophages from approximately 3.4% to ∼60.3%, while reducing CD206⁺CD11b⁺ cells from 13.9% to 7.65%. WDL treatment dose‐dependently reversed this polarization, restoring the M1/M2 balance close to baseline (Figure 2C–E). Immunofluorescence staining further showed that LPS markedly elevated TNF‐α and suppressed IL‐10 expression, whereas WDL treatment dose‐dependently restored the anti‐inflammatory IL‐10/TNF‐α ratio (Figure 2F). Quantitative analysis (Figure S5) confirmed that WDL significantly reduced TNF‐α and increased IL‐10 in a concentration‐dependent manner (p < 0.0001). Collectively, these data demonstrate that WDL not only inhibits NF‐κB activation but also promotes macrophage polarization toward an anti‐inflammatory phenotype, thereby re‐establishing immune homeostasis under LPS challenge.
Redox Protection in Disc Cells
2.2.2
Given that oxidative stress functions downstream of the NF‐κB–LCN2 axis in IVDD, we next examined whether WDL confers direct antioxidant protection. In cell‐free radical–scavenging assays, WDL exhibited dose‐dependent scavenging activity toward superoxide (•O_2_ ^−^), hydroxyl (•OH), ABTS•⁺, and DPPH• radicals (Figure S6A–D), confirming its intrinsic antioxidant potential. Consistently, DCFH‐DA fluorescence imaging revealed that WDL markedly suppressed intracellular ROS accumulation induced by H_2_O_2_ (Figure 2G and Figure S7), with fluorescence intensity nearly returning to control levels at WDL3 concentration (*p *< 0.0001). In both nucleus pulposus cells (NPCs) and AFCs, WDL significantly enhanced SOD activity and restored GSH content in a concentration‐dependent manner (Figure 2H,I). These findings indicate that WDL reinforces endogenous antioxidant defenses while directly scavenging free radicals. A schematic model summarizes this dual mechanism—NF‐κB inhibition with immune modulation and redox homeostasis restoration—that together converges to limit ROS‐driven cellular injury (Figure 2J).
WDL Interrupts the NF‐κB–LCN2 Axis within AFCs and Disrupts the Macrophage–AFC Paracrine Loop
2.3
NF‐κB–LCN2 Signaling Suppression in Oxidatively Stressed AFCs
2.3.1
We next examined whether WDL suppresses inflammatory activation in AFCs by targeting the NF‐κB–LCN2 axis, and whether macrophage‐derived LCN2 amplifies this response in a paracrine manner. In H_2_O_2_‐challenged AFCs, immunofluorescence staining revealed robust P‐p65 nuclear translocation, which was dose‐dependently attenuated by WDL treatment (Figure 3A). Semi‐quantitative analysis showed that H_2_O_2_ increased the nuclear‐to‐cytoplasmic ratio of p65 by more than fourfold, whereas WDL reduced this ratio by approximately 35%, 55%, and 70% at LW1–LW3, respectively (Figure S8A–C). Consistent with this, WDL markedly downregulated canonical NF‐κB target genes, including TNF‐α, IL‐6, IL‐1β, and Lcn2 (Figure 3C–F).
*WDL interrupts the NF‐κB–LCN2 axis in AFCs and disrupts the macrophage–AFC paracrine amplification. (A) Immunofluorescence of p65/P‐p65 showing NF‐κB activation under H2O2 and its dose‐dependent suppression by WDL. (B) LCN2 immunofluorescence in AFCs under oxidative stress and its attenuation by WDL. (C–F) qPCR showing downregulation of TNF‐α, IL‐6, IL‐1β, and Lcn2 by WDL. (G) Pharmacologic blockade using NF‐κB inhibitor JSH‐23 (NF‐κB‐i) decreases LCN2 and MMP9; WDL+NF‐κB‐i further reduces both. (H) Schematic of the conditioned‐medium (CM) transfer from LPS‐stimulated RAW264.7 macrophages treated with WDL or LCN2 inhibitor (LCN2i). Created with BioRender.com. (I) ELISA of secreted LCN2 in CM. (J) Immunofluorescence of intracellular LCN2. (K–N) qPCR in AFCs after CM exposure: matrix anabolism (Acan, Col1a1) increases, catabolic enzymes (MMP13, Adamts5) decrease. (O,P) qPCR showing reduced IL‐6 and TNF‐αunder WDL/LCN2i‐conditioned CM. (Q–S) Antioxidant readouts (GSH, SOD) in AFCs exposed to CM. (T) Functional validation: rLCN2 enhances nuclear P‐p65, LCN2i reduces it; rLCN2+WDL partially rescues, while LCN2i+WDL produces maximal inhibition. Scale bars: 20 µm. Data are presented as mean ± SD. n is indicated in each panel. Statistics: one‐way ANOVA with Tukey's multiple‐comparisons test. *p < 0.05, **p < 0.01, ***p < 0.001, ***p < 0.0001; ns, not significant.
Immunofluorescence staining (Figure 3B) and semi‐quantitative analysis (Figure S9) revealed that LCN2 protein was strong induced (>2.5‐fold) by oxidative stress and progressively restored to near‐baseline levels by WDL in a concentration‐dependent manner. To clarify the signaling hierarchy, NF‐κB activity was pharmacologically blocked using JSH‐23 (denoted as NF‐κB‐i). NF‐κB inhibition alone reduced both LCN2 and its catabolic partner MMP9, while combined treatment with WDL and NF‐κB‐i further suppressed their expression (Figure 3G). Quantitative image analysis (Figure S10) confirmed that LCN2 and MMP9 fluorescence intensities were reduced by ∼ 60%–75% relative to H_2_O_2_ and that WDL+NF‐κB‐i achieved maximal inhibition. These results indicate that WDL effectively blocks NF‐κB activation and downstream LCN2/MMP9 signaling under oxidative stress, acting as a dual anti‐inflammatory and anti‐catabolic regulator in AFCs.
Disruption of Macrophage‐Derived LCN2 Paracrine Amplification
2.3.2
To examine whether WDL disrupts macrophages‐to‐AFC crosstalk, we modeled the paracrine loop using a conditioned‐medium (CM) system (Figure 3H). In LPS‐stimulated RAW264.7 macrophages, ELISA showed that secreted LCN2 levels increased dramatically, whereas WDL or an LCN2 inhibitor (LCN2i) significantly reduced LCN2 release (Figure 3I). Immunofluorescence staining (Figure 3J) and quantitative analysis (Figure S11) further confirmed that intracellular LCN2 fluorescence increased nearly threefold with LPS but was reduced by ∼55% with WDL and ∼70% with LCN2i. When AFCs were cultured with CM from WDL‐ or LCN2i‐treated macrophages, ECM anabolic gene (Acan, Col1a1) was restored, while catabolic enzymes (MMP13, Adamts5) were reduced (Figure 3K–N). Moreover, inflammatory cytokines (IL‐6, TNF‐α) were markedly suppressed (Figure 3O,P), and antioxidant capacity was enhanced, as evidenced by elevated intracellular GSH and SOD (Figure 3Q–S). These findings demonstrate that macrophage‐derived LCN2 functions as an amplifier of AFC inflammation and oxidative stress, and that WDL effectively interrupts this cross‐cellular feed‐forward loop.
Cross‐Compartment Validation Confirms NF‐κB–LCN2 Feedback Attenuation
2.3.3
Functional perturbation of LCN2 within AFCs confirmed the signaling directionality. Recombinant LCN2 (rLCN2) enhanced P‐p65 nuclear accumulation under LPS, whereas LCN2i suppressed it. Notably, rLCN2 partially reduced the inhibitory effect of WDL (LPS + rLCN2 + WDL), while LCN2i acted additively with WDL (LPS + LCN2i + WDL), producing the lowest P‐p65 signal (Figure 3T and Figure S12). Quantitative analysis (Figure S12A–C) showed that WDL reduced nuclear P‐p65 fluorescence by ∼45%, LCN2i by ∼60%, and the combined treatment by ∼80% relative to LPS, consistent with cooperative NF‐κB inhibition.
As an orthogonal validation in macrophages, pharmacologic inhibition of LCN2 similarly reduced P‐p65 nuclear intensity under LPS stimulation, and combined treatment with WDL+LCN2i exerted an additive effect (Figure S13). Quantitatively, LPS induced an approximately fourfold increase in nuclear P‐p65 compared with control, whereas WDL or LCN2i alone reduced this signal by ∼55% and ∼65%, respectively; their combination further lowered it nearly 80%. These results corroborate a cross‐compartment NF‐κB ‐LCN2 feed‐forward circuit and confirm that WDL cooperatively suppresses this inflammatory axis across macrophage and AFC populations.
Synthesis and Characterization of a ROS‐Responsive Dual‐Network Hydrogel for WDL Delivery
2.4
To enable cue‐responsive and temporally controlled delivery of WDL in the oxidative microenvironment of degenerated intervertebral discs, we developed a dual‐network hydrogel that coupled ROS sensitivity with mechanical robustness. The design integrates a ROS‐cleavable phenylboronic‐ester/PVA network (PT) with a covalently crosslinked gelatin–elastin framework (GAE), forming an interpenetrating hybrid (denoted PTGAE).
Synthesis and Chemical Verification
2.4.1
The boronic‐ester crosslinker TSPBA was synthesized via bromination and esterification reactions and confirmed by ^1^H NMR spectroscopy, displaying characteristic aromatic and linker proton resonances (Figure 4A‐I). In oxidative conditions, boronic‐ester bonds in the TSPBA–PVA network undergo H_2_O_2_‐mediated cleavage, producing benzenediol derivatives and leading to network disassembly (Figure 4A‐II). This process is consistent with the established oxidation mechanism of phenylboronic esters, ensuring a predictable ROS‐responsive degradation pathway [28]. The GAE framework interpenetrates the ROS‐labile PT network, providing long‐term structural support and elasticity even as the PT phase undergoes controlled degradation [26]—a strategy that decouples mechanical stability from stimulus‐responsiveness, a key advance over conventional single‐network systems.
Design, synthesis, and characterization of a ROS‐responsive PTGAE hydrogel system for WDL delivery. (A) (I) 1H NMR spectrum of TSPBA confirming successful synthesis. (II) schematic illustration of PVA–TSPBA network formation and ROS‐induced cleavage of boronic ester bonds under H2O2. (B) Schematic models and SEM images of GAE, PT, and PTGAE hydrogels showing distinct porous architectures with an interpenetrating morphology in PTGAE. Created with BioRender.com. (C) Photographs demonstrating the elasticity and stretchability of PTGAE. (D) Swelling ratios of GAE, PT, and PTGAE in PBS. (E) frequency‐dependent rheological profiles of G′ (storage) and G″ (loss) moduli, confirming viscoelastic reinforcement in PTGAE. (F) Macroscopic stability and degradation in PBS vs H2O2 over 7 days, showing ROS‐responsive disassembly. (G) Cumulative WDL release profiles under graded H2O2 concentrations, revealing ROS‐triggered, dose‐dependent release kinetics. Data are presented as mean ± SD (n = 3).
Microstructure and Mechanical Properties
2.4.2
SEM analysis revealed distinct porous morphologies across formulations (Figure 4B). While GAE exhibited a homogeneous, sponge‐like texture, and PT showed a looser structure, PTGAE displayed densely interconnected pores, consistent with the successful interpenetrating of the two networks. Macroscopic handling confirmed that PTGAE maintained high elasticity and stretch recoverability, enabling conformal fit and handing flexibility (Figure 4C). Swelling analysis demonstrated that the incorporation of the GAE framework moderately reduced equilibrium water uptake compared to GAE alone (Figure 4D), indicating enhanced structural integrity and crosslink density. Rheological measurements further validated mechanical reinforcement: the storage (G′) and loss (G″) moduli of PTGAE were substantially higher than those of either individual network, confirming improved viscoelastic performance and energy‐dissipation capacity (Figure 4E). Together, these data establish that the dual‐network architecture yields a mechanically competent yet responsive hydrogel.
ROS Responsiveness and Controlled WDL Release
2.4.3
The ROS‐responsiveness of PTGAE was assessed under graded oxidative conditions. Macroscopic stability tests showed that hydrogel samples remained intact in PBS but gradually disintegrated in H_2_O_2_ solutions, confirming selective oxidation‐triggered softening and erosion (Figure 4F). Correspondingly, in vitro WDL release profiles demonstrated dose‐dependent ROS‐triggered release kinetics: under 1 m H_2_O_2_, over 80% of WDL was released within 5 days, whereas cumulative release under physiological PBS remained below 20% over 14 days (Figure 4G). The in vitro release profiles (Figure 4G) demonstrated dose‐dependent responsiveness. We selected 100 µm H_2_O_2_ as the representative pathological concentration because established oxidative‐stress IVDD models have shown that H_2_O_2_ in the 100–200 µm range reliably induces senescence and apoptosis in disc cells without causing necrosis [29, 30]. Higher concentrations (0.1–1 M) were used as accelerated oxidative conditions to define the upper bounds of ROS sensitivity. Notably, the release behavior followed pseudo‐first‐order kinetics, showing a sustained, concentration‐dependent elution rate rather than a burst release. Under the pathological mimic (100 µM H_2_O_2_), WPG released approximately 45% of its payload over 21 days, whereas release in PBS stayed below 15%, confirming excellent stability in non‐oxidative environments.
This ROS‐responsive release behavior directly reflects the cleavage dynamics of boronic‐ester crosslinks within the PT network, validating the concept of a stimuli‐responsive, pathology‐triggered drug‐release mechanism. The integration of a stable GAE scaffold with a degradable PT phase enables sustained retention under basal conditions and accelerated drug release specifically within oxidative lesions—matching the pathological redox gradient of degenerated discs.
WPG Hydrogel Preserves Cell Viability, Promotes Adhesion and Migration of AFCs under Oxidative Stress
2.5
To evaluate the cytocompatibility and functional performance of WPG hydrogel under oxidative stress, AFCs were exposed to H_2_O_2_ and treated with WPG or control matrices. Live/Dead staining revealed that WPG maintained high cell viability comparable to untreated controls, whereas H_2_O_2_ exposure led to extensive cell death (Figure 5A). Quantitative CCK‐8 analysis confirmed that WPG preserved viability over 5 days and significantly outperforming the PG matrix without WDL (Figure 5B). Cell proliferation, assessed by EdU incorporation, was markedly restored by WPG treatment (Figure 5C). Quantitative volumetric analysis (Figure 5D) showed that H_2_O_2_ reduced EdU incorporation by >70%, whereas WPG recovered proliferation to ∼65% of control levels—substantially higher than PG (p < 0.001). Cytoskeletal organization and adhesion morphology were further evaluated by immunofluorescence staining of F‐actin (phalloidin) and focal adhesion protein vinculin. Oxidative stress caused cell shrinkage and disrupted actin filament alignment, while WPG preserved cytoskeletal continuity and enhanced vinculin‐rich adhesion contacts (Figure 5E). Quantitative analysis (Figure 5F) demonstrated that WPG restored vinculin volume to approximately 75% of control values (p < 0.0001), indicating recovery of adhesion strength and cellular spreading capacity. To examine functional motility, scratch wound assays were performed. WPG‐treated AFCs exhibited accelerated wound closure compared to both H_2_O_2_ and PG groups (Figure 5G). Quantification of the healing area (Figure 5H) showed that WPG reduced residual wound width to <20% of baseline within 24 h (*p *< 0.01), highlighting enhanced migration dynamics. Together, these findings demonstrate that the WPG hydrogel provides strong cytoprotective and mechano‐supportive cues, preserving AFC viability, proliferation, adhesion, and migration under oxidative stress. The results confirm that WPG not only mitigates ROS‐induced injury but also restores cytoskeletal architecture and motile functionality, underscoring its therapeutic potential for disc microenvironment repair.
*WPG hydrogel supports viability, proliferation, adhesion, and migration of AFCs under oxidative stress. (A) Live/Dead staining showing viability of AFCs cultured under H2O2 ± hydrogel treatments. (B) CCK‐8 assay quantifying cell viability over 5 days. (C,D) EdU assay and quantitative volumetric analysis of proliferation recovery. (E) Immunofluorescence staining of F‐actin (Phalloidin, red) and Vinculin (green) indicating cytoskeletal and adhesion remodeling. (F) Quantitative analysis of Vinculin volume showing enhanced adhesion in WPG‐treated AFCs. (G,H) Scratch wound assay and quantification of healing area at the indicated time points. Scale bars: (A) 100 µm; (C) 20 µm; (E) 50 µm. Data are presented as mean ± SD. n is indicated in each panel. Statistics: One‐way ANOVA with Tukey's multiple‐comparisons test was used for panels (B,F); two‐way ANOVA with Tukey's multiple‐comparisons test was used for panels (B,H). Significance: *p < 0.05, **p < 0.01, ***p < 0.001, ***p < 0.0001; ns, not significant.
WPG Mitigare Macrophage–AFC Inflammatory Crosstalk, Reduces Oxidative Stress, Limits Apoptosis, and Preserves ECM Integrity in a Co‐culture–Relevant Model
2.6
To recapitulate immune–stromal interactions in a disc‐like environment, we established a Transwell co‐culture in which LPS‐stimulated RAW264.7 macrophages were seeded in the upper insert together with hydrogel, while AFCs were cultured in the lower chamber (Figure 6A).
*WPG hydrogel mitigates macrophage–AFC inflammatory crosstalk, reduces oxidative stress and apoptosis, and preserves ECM integrity. (A) Transwell co‐culture schematic with RAW264.7 macrophages in the upper insert and AFCs in the lower chamber. Created with BioRender.com. (B) Immunofluorescence of TNF‐α (green) and IL‐10 (red) in RAW264.7 macrophages (CO‐Ctrl, CO‐LPS, CO‐LPS+PG, CO‐LPS+WPG). (C,D) DCFH‐DA staining and quantitative analysis of ROS levels in AFCs under H2O2 ± hydrogel treatments. (E) Annexin V‐FITC (AF488, green)/PI (red) double staining and corresponding fluorescence intensity profiles showing apoptosis modulation. (F) Immunofluorescence of ACAN (green) and ADAMTS5 (red) with F‐actin (phalloidin, yellow) illustrating ECM remodeling. (G,H) Quantification of ACAN and ADAMTS5 fluorescence volumes. Scale bars: (B) 40 µm; (C) 30 µm; (E) 100 µm; (F) 50 µm. Data are presented as mean ± SD. n is indicated in each panel. Statistics: one‐way ANOVA with Tukey's multiple‐comparisons test. *p < 0.05, **p < 0.01, ***p < 0.001, ***p < 0.0001; ns, not significant.
Macrophage Immunomodulation
2.6.1
Under LPS challenge, macrophages exhibited pronounced pro‐inflammatory activation, characterized by elevated TNF‐α and suppressed IL‐10 expression. Treatment with WPG (WDL‐loaded PTGAE) markedly rebalanced this cytokine profile—TNF‐α fluorescence decreased by ∼40%, while IL‐10 increased approximately twofold—indicating a shift toward an anti‐inflammatory phenotype (Figure 6B and Figure S14). In contrast, the blank PG hydrogel produced only partial modulation, underscoring the essential contribution of WDL in driving macrophage reprogramming.
Antioxidant and Cytoprotective Effects in AFCs
2.6.2
In parallel oxidative‐stress assays, intracellular ROS in AFCs—quantified by DCFH‐DA fluorescence—rose by ∼3.5‐fold under H_2_O_2_ exposure relative to control. PG provided limited protection (∼15% reduction), whereas WPG suppressed ROS accumulation by nearly 60%, bringing fluorescence intensity close to baseline levels (Figure 6C,D). Annexin V‐FITC/PI double staining revealed extensive early (Annexin V⁺/PI^−^) and late (Annexin V⁺/PI⁺) apoptosis following H_2_O_2_ insult. WPG substantially reduced both signals, with >50% decrease in overlapping fluorescence peaks, confirming mitigation of oxidative stress‐induced apoptosis and membrane damage (Figure 6E). PG offered only a minor rescue effect.
Matrix Preservation and Anti‐Catabolic Remodeling
2.6.3
To evaluate ECM integrity, immunofluorescence staining of aggrecan (ACAN) and ADAMTS5 was performed. H_2_O_2_ challenge caused severe ACAN loss and robust ADAMTS5 induction, while WPG treatment preserved ACAN deposition and markedly reduced ADAMTS5 expression (Figure 6F). Quantitative analysis showed that WPG increased ACAN volume by ∼3.2‐fold and lowered ADAMTS5 by ∼70% relative to H_2_O_2_ alone (Figure 6G,H), whereas PG achieved only partial recovery.
Collectively, these findings demonstrate that WPG hydrogel reprograms macrophage cytokine output, suppresses ROS‐driven oxidative injury, prevents apoptosis, and maintains matrix homeostasis. These multi‐level effects reflect the synergistic actions of WDL‐mediated NF‐κB inhibition and ROS scavenging, supporting WPG as an integrated immunomodulatory and cytoprotective platform for halting or reversing degenerative disc progression.
Transcriptomic Profiling Reveals That WPG Reprograms Inflammatory and Repair Pathways in AFCs and Macrophages
2.7
To elucidate the molecular basis underlying WPG‐mediated protection, bulk RNA sequencing (RNA‐seq) was performed on AFCs exposed to H_2_O_2_ with or without WPG and on LPS‐stimulated RAW264.7 macrophages with or without WPG treatment. Quality control metrics confirmed high data integrity and reproducibility (Figures S15–S19).
Transcriptomic Remodeling in AFCs
2.7.1
Under oxidative stress, AFCs exhibited transcriptomic signatures dominated by inflammatory, chemokine, and oxidative‐stress responses, accompanied by adhesion–cytoskeletal reorganization and ECM‐degradation pathways. WPG treatment attenuated these inflammatory and oxidative signatures while restoring expression of adhesion and matrix‐assembly programs (GO/KEGG, Figure 7A,B).
*Transcriptomic profiling shows that WPG reprograms inflammatory and repair pathways in AFCs and macrophages. (A,B) GO and KEGG analyses of DEGs between H2O2 and H2O2 + WPG in AFCs. (C) Sankey plot mapping AFC DEGs to KEGG pathways. (D–F) AFC DEGs associated with inflammation/oxidative stress (D), cell proliferation/adhesion (E), and ECM remodeling (F). (G–I) GO, KEGG, and Reactome analyses of macrophage DEGs (LPS ± WPG). (J–L) GSEA ridgeplots for GO, KEGG, and Reactome terms showing reciprocal enrichment of inflammatory versus ECM/adhesion modules. (M) Venn diagram of shared DEGs between LPS versus control and WPG versus LPS comparisons. (N) PPI network of the shared 115 core genes, highlighting reduced activity in chemokine/cell‐cycle hubs and increased reparative signaling nodes with WPG. Data are presented as mean ± SD (n = 3 biologically independent samples per group) for panels (D–F). Statistics for panels (D–F): one‐way ANOVA with Tukey's multiple‐comparisons test; *p < 0.05, **p < 0.01, ***p < 0.001, ***p < 0.0001; ns, not significant. Panels (A–C) and (G–L) show enrichment/GSEA results derived from RNA‐seq differential expression analysis (DEG cutoff: FDR < 0.05 and |log2FC| > 0.5; multiple testing correction: Benjamini–Hochberg). Panels (M,N) show overlap and PPI network visualization based on the DEG lists and STRING/Cytoscape.
Sankey mapping of differentially expressed genes (DEGs) revealed multi‐pathway regulators linking inflammatory, redox, and ECM processes (Figure 7C). DEG validation panels showed consistent trends: WPG suppressed pro‐inflammatory and oxidative genes (e.g., Lcn2, Tlr2, and C3) (Figure 7D), upregulated cell‐proliferation and migration‐related transcripts (e.g., Mki67, Itgb1, and Vcl) (Figure 7E), and restored ECM synthesis (e.g., Col1a1, Col3a1) while repressing catabolic enzymes (e.g., Adamts1 and Adamts5) (Figure 7F). These transcriptional changes align with the phenotypic outcomes described earlier—improved cytoskeletal integrity and adhesion (Figure 5E,F), reduced ROS and apoptosis (Figure 6C–E), and preserved Acan deposition (Figure 6F–H).
Transcriptomic Remodeling in Macrophages
2.7.2
In macrophages, LPS stimulation upregulated classical inflammatory cascades, including NF‐κB, TNF, and cytokine‐receptor pathways, as well as mitotic and ECM‐remodeling programs. WPG treatment markedly repressed these pro‐inflammatory networks and enhanced enrichment in ECM–receptor interaction, focal adhesion, and cytoskeletal reorganization (GO/KEGG/Reactome; Figure 7G–I).
Gene set enrichment analysis (GSEA) ridgeplots reinforced this shift in transcriptional polarity: LPS favored inflammatory and checkpoint‐associated gene sets, whereas WPG promoted ECM, adhesion, and metabolic remodeling programs (including glycolysis, pentose phosphate, and HIF‐1 signaling) (Figure 7J–L).
Intersection analysis identified 115 core genes shared between LPS versus control and WPG versus LPS comparisons (Figure 7M). Protein–protein interaction (PPI) mapping of this shared subset highlighted central chemokine and cell‐cycle regulators (Cxcl10, Psmb8, Cdc25a) that were downregulated by WPG, along with increased representation of reparative or regulatory genes (Tmem119, Clspn) (Figure 7N).
Collectively, the transcriptomic landscapes of AFCs and macrophages reveal a convergent therapeutic mechanism driven by WPG. Rather than targeting isolated pathways, WPG orchestrates synchronized cellular reprogramming: it simultaneously dismantles the pro‐inflammatory scaffold, characterized by the shared downregulation of NF‐κB and chemokine cascades (e.g., Cxcl10, Il1b), while reactivating the structural checkpoints required for repair, specifically the ECM‐receptor interaction and focal adhesion modules (Figure 7). This coordinated shift from a catabolic/inflammatory state to an anabolic/adhesive phenotype across both immune and stromal compartments provides the molecular basis for the structural restoration observed in vivo. These findings highlight WPG's coordinated regulation of inflammatory and reparative pathways across macrophages and AFCs.
WPG Hydrogel Restores Structural Integrity and Biomechanical Function in a Rat Model of Intervertebral Disc Degeneration
2.8
To evaluate the therapeutic efficacy of WPG in vivo, a rat annulus fibrosus defect model was established via percutaneous needle puncture, followed by intradiscal injection of the corresponding hydrogels (Figure 8A). Radiographic and MRI assessments were performed at 4 and 12 weeks post‐operation to monitor structural recovery and functional preservation.
*WPG hydrogel restores structural and biomechanical properties of intervertebral discs in a puncture‐induced degeneration model. (A) Experimental design outlining IVDD induction, hydrogel injection, and longitudinal imaging. Created with BioRender.com. (B) Representative X‐ray images showing progressive degeneration and repair at 1 and 3 months. (C) Quantification of disc height index (DHI). (D) T2‐weighted MRI scans at corresponding timepoints. (E) Quantitative optical density analysis confirming preservation of disc hydration. (F) Stress–strain curves of motion segments across groups. (G,H) Quantification of toe modulus (G) and linear modulus (H), demonstrating WPG‐mediated restoration of mechanical elasticity and damping. Data shown as mean ± SD (n = 6). Statistics: one‐way ANOVA with Tukey's post hoc test for ≥3 groups; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Quantification is shown in panels (C,E,G,H) as mean ± SD; n is indicated in each panel (biological replicates). Statistics for panels (C,E,G,H): one‐way ANOVA with Tukey's multiple‐comparisons test; *p < 0.05, **p < 0.01, ***p < 0.001, ***p < 0.0001; ns, not significant. Panel (F) shows stress–strain curves (descriptive).
X‐ray imaging revealed severe disc height loss and endplate irregularities in the puncture group, which were only partially alleviated by GAE or PG treatment. In contrast, WPG markedly preserved disc height and improved endplate morphology over time (Figure 8B). Quantitative disc height index (DHI) analysis confirmed that WPG maintained approximately 85% of the native disc height at 3 months—significantly higher than GAE (∼65%) and PG (∼70%) groups (Figure 8C). Consistent with the radiographic findings, T_2_‐weighted MRI demonstrated brighter and more homogeneous signals in the WPG group, indicative of improved nucleus pulposus hydration and proteoglycan retention (Figure 8D). Quantitative optical density analysis (Figure 8E) showed that WPG‐treated discs exhibited a ∼1.8‐fold increase compared with the puncture group and a ∼35%–40% improvement over PG, confirming superior matrix water retention and compositional integrity.
Functional restoration was further substantiated by ex vivo biomechanical testing. The punctured discs displayed a steep stress–strain curve characteristic of viscoelastic loss, whereas WPG‐treated discs regained a smooth sigmoidal profile comparable to sham controls (Figure 8F). Quantitatively, both toe modulus and linear modulus were significantly restored—recovering to ∼85% and ∼80% of sham levels, respectively (Figure 8G,H)—demonstrating enhanced load‐bearing capacity and elastic resilience.
Collectively, these results demonstrate that WPG hydrogel effectively preserves disc height and hydration, maintains nucleus pulposus composition, and restores near‐physiological viscoelastic mechanics. This integrated protection against structural and functional degeneration provides compelling preclinical evidence supporting the translational potential of WPG for intervertebral disc repair.
WPG Hydrogel Facilitates Histological Restoration of Annulus Fibrosus Structure in Degenerated Discs
2.9
Histological analyses were conducted to assess annulus fibrosus (AF) restoration following WPG treatment in vivo. Hematoxylin and eosin (H&E) staining at both 1 and 3 months revealed severe disorganization and fragmentation of AF lamellae in the puncture group, with discontinuous and irregular collagen alignment. In contrast, WPG‐treated discs exhibited well‐preserved lamellar architecture, maintaining continuous and parallel collagen layers similar to the sham group, whereas GAE and PG hydrogels resulted in only partial realignment with residual fissures (Figure 9A,B). Safranin O–Fast Green (SO/FG) staining further demonstrated substantial depletion of proteoglycans in punctured discs. WPG treatment effectively restored a proteoglycan‐rich matrix and preserved the lamellar organization, particularly at the inner annulus–nucleus interface. By comparison, GAE and PG groups exhibited only modest recovery with weak SO‐positive staining (Figure 9A,B). To evaluate collagen reorganization, Sirius Red–polarized light microscopy was performed. Puncture injury led to disrupted collagen orientation and weak birefringence, indicative of collagen degradation and fiber disarray. In contrast, WPG‐treated discs showed densely packed, uniformly oriented collagen bundles with strong birefringence, reflecting restored lamellar integrity and anisotropy (Figure 9C). These improvements were markedly greater than those observed in the GAE and PG groups.
WPG hydrogel restores annulus fibrosus morphology and matrix composition in degenerated discs. (A) One month post‐treatment: H&E (top) and Safranin O–Fast Green (bottom). Puncture induces AF disorganization and proteoglycan depletion; WPG maintains lamellar continuity and SO‐positive matrix compared with GAE and PG. (B) Three months post‐treatment: sustained preservation of AF lamellar structure and proteoglycan content in the WPG group. (C) Three months: Sirius Red staining under polarized light showing collagen fiber orientation and birefringence. WPG exhibits strong, uniformly aligned collagen bundles, consistent with structural restoration. Groups: Sham, Puncture, Puncture+GAE, Puncture+PG, Puncture+WPG. Scale bars: as indicated.
Collectively, these histological findings confirm that WPG hydrogel preserves annulus fibrosus architecture, replenishes matrix proteoglycans, and promotes ordered collagen remodeling. The observed lamellar restoration and compositional recovery provide morphological evidence consistent with the improved mechanical resilience demonstrated in Figure 8, highlighting WPG's capacity to drive multi‐level disc regeneration.
WPG Restores AF ECM Architecture and Dampens Inflammation In Vivo by Restraining the NF‐κB–LCN2 Axis
2.10
To determine whether WPG promotes ECM repair and mitigates inflammation in vivo, lumbar discs from the sham, puncture, puncture+GAE, puncture+PG, and puncture+WPG groups were examined by immunofluorescence staining.
ECM Remodeling
2.10.1
Puncture injury markedly increased the expression of the ECM‐degrading enzyme MMP13, accompanied by pronounced lamellar disorganization within the annulus fibrosus. WPG treatment substantially suppressed MMP13 immunoreactivity—by approximately 70% relative to the puncture group—and preserved the continuous chondroitin sulfate (Cs) boundaries characteristic of intact AF lamellae, whereas GAE and PG exhibited only modest reductions (Figures 10A and S20). Quantitative analysis confirmed that the MMP13‐positive volume decreased from ∼900 µm^3^ in punctured discs to ∼300 µm^3^ following WPG treatment, approaching near‐sham levels. Collagen I (Col1) staining revealed a similar restorative pattern. After puncture, Col1 fibers appeared sparse and fragmented, indicative of disrupted lamellar organization. WPG markedly enhanced collagen deposition and alignment, restoring a dense, continuous architecture comparable to the sham structure (Figure 10B). Semi‐quantitative assessment (Figure S21) showed a ∼2.3‐fold increase in Col1 fluorescence volume in the WPG group relative to the puncture group, confirming robust collagen remodeling and ECM repair. Together, these findings demonstrate that WPG effectively restrains matrix catabolism while promoting ECM reassembly in situ.
WPG hydrogel regulates ECM remodeling in degenerated intervertebral discs. (A) Immunofluorescence of chondroitin sulfate (CS, green) and MMP13 (red) showing ECM degradation after puncture and restoration following WPG treatment. (B) Double staining of (green) and collagen I (Col1, yellow) demonstrating aligned collagen lamellae recovery with WPG. (C) Cytokine expression (TNF‐α, green; IL‐10, red) showing reduced inflammation and enhanced anti‐inflammatory signaling with WPG. (D) P‐p65 (green) indicating suppressed NF‐κB activation. (E) LCN2 (green) reflecting downstream attenuation of the NF‐κB–LCN2 feedback loop. Scale bars: (A) 100 µm; (B–E) 50 µm.
Inflammatory Modulation
2.10.2
Double immunostaining revealed a cytokine imbalance in punctured discs, characterized by elevated TNF‐α and reduced IL‐10 expression. WPG treatment significantly decreased TNF‐α levels (by ∼60%) while restoring IL‐10 (approximately twofold increase), thereby shifting the local cytokine milieu toward an anti‐inflammatory and reparative profile (Figure 10C and Figure S22). This in vivo cytokine normalization parallels the macrophage and AFC reprogramming observed in vitro, underscoring the systemic consistency of WPG's immunomodulatory action.
NF‐κB–LCN2 Signaling Attenuation
2.10.3
Mechanistically, immunofluorescence staining demonstrated strong nuclear accumulation of P‐p65 after puncture, reflecting canonical NF‐κB activation. WPG treatment markedly reduced nuclear P‐p65 intensity by ∼45% relative to PG (Figure 10D and Figure S23). Correspondingly, the NF‐κB–responsive amplifier LCN2 was robustly induced in degenerated discs but suppressed by WPG by ∼55% (Figure 10E and Figure S24), approaching baseline expression levels observed in the sham group.
Collectively, these in vivo results demonstrate that WPG hydrogel restores annulus fibrosus ECM integrity, attenuates pro‐inflammatory cytokine production, and inhibits NF‐κB–LCN2 signaling, thereby preserving matrix homeostasis and mitigating degenerative progression.
In Vivo Biocompatibility of WPG Hydrogel
2.11
To evaluate systemic biosafety, major organs were examined 12 weeks after intradiscal administration of WPG. Hematoxylin and eosin (H&E) staining of the heart, liver, spleen, lungs, and kidneys revealed no histopathological abnormalities, such as inflammation, necrosis, or tissue fibrosis, in the WPG‐treated rats compared with the sham or vehicle controls (Figure S25A). Comprehensive hematological analysis further demonstrated no significant alterations in leukocyte, erythrocyte, or platelet parameters, including WBC, LYM, MON, NEU, RBC, HGB, HCT, MCV, MCH, MCHC, PLT, MPV, PDW, and PCT (Figure S25B). Collectively, these results confirm that WPG hydrogel exhibits excellent systemic biocompatibility and hematological safety during the 12‐week observation period, supporting its suitability for long‐term intradiscal application.
Discussion
3
IVDD progression is driven by feed‐forward inflammatory and oxidative cascades that progressively erode ECM homeostasis and biomechanical resilience, yet disease burden and unmet need remain substantial despite current care [31, 32, 33]. Our study introduces a closed‐loop therapeutic paradigm that couples a mechanistically informed molecular brake—WDL, which inhibits the NF‐κB–LCN2 inflammatory amplifier—with a ROS‐responsive depot (WPG) that enables pathology‐triggered, microenvironment‐adaptive drug release within oxidative niches disc microdomains. This approach aligns molecular target selection (the inflammatory–proteolytic NF‐κB–LCN2 circuit) with the pathology‐specific microenvironment for intervention, characterized by ROS‐rich and load‐bearing disc niches. It thus addresses intrinsic avascularity, poor molecular exchange, and delivery barriers that limit systemic small‐molecules in IVDD.
Our data delineate a bidirectionally coupled NF‐κB–LCN2 amplifier across macrophage–AFC crosstalk. In macrophages, LPS robustly activates NF‐κB and upregulates LCN2; WDL suppresses P‐p65 nuclear translocation and diminishes LCN2 release, thereby attenuating the paracrine drive to AFCs (Figures 2 and 3). This is consistent with findings in other inflammatory contexts where innate immune adaptors such as CARD9 activate NF‐κB to induce LCN2 in macrophages, and LCN2 deficiency reduces MMP activity in vivo, functionally linking inflammatory amplification to proteolysis [34]. In hepatic tissue, a LIFR–NF‐κB–LCN2 regulatory axis controls stress programs and cell fate, underscoring the conserved amplifier role of this pathway beyond a single organ context [35]. Moreover, proteostasis and stress‐response pathways modulate NF‐κB/LCN2 signaling in reactive glia, suggesting upstream regulatory layers that may likewise operate within the disc microenvironment [10]. Consistent with these cross‐system observations, both our conditioned‐medium and pharmacological intervention experiments confirm that macrophage‐derived LCN2 intensifies AFC inflammatory and catabolic outputs, whereas LCN2 inhibition or WDL treatment collapses this amplifier; RNA‐seq analysis further revealed coordinated downregulation of NF‐κB/chemokine modules together with restoration of ECM and adhesion programs (Figure 7). Complementarily, prior studies have shown exosomal LCN2 reactivates NF‐κB and accelerates disc‐cell senescence; our work provides a pharmacological brake to the same axis, delivered through a ROS‐gated hydrogel system [11].
Mechanistically, our results indicate that WDL suppresses Lcn2 expression primarily at the transcriptional level. This is evidenced by the synchronized downregulation of Lcn2 mRNA and protein (Figures 2B and 3F) and the observation that the specific NF‐κB inhibitor JSH‐23 mimics the suppressive effect of WDL (Figure 3G). Critically, WDL physically blocks the nuclear translocation of phospho‐p65 (Figures 2A and 3A), thereby depriving the nucleus of this essential transcription factor. While we did not directly assess p65–promoter binding in this study, our data demonstrate a functional NF‐κB–dependent suppression of Lcn2 that is consistent with prior chromatin immunoprecipitation (ChIP) and promoter‐reporter studies. These previous works have firmly established Lcn2 as a direct transcriptional target of NF‐κB, identifying functional κB response elements within its promoter [9, 35, 36, 37]. By preventing p65 from accessing these confirmed promoter sites, WDL effectively disrupts the inflammatory feed‐forward loop.
Although we identify LCN2 as a key amplifier of inflammation and matrix catabolism in our oxidative/inflammatory models, we acknowledge that LCN2 can exert context‐dependent functions in disc disease. Recent single‐cell atlases of human IVDD have reported LCN2‐high immune/stromal subsets with potential regulatory roles in certain early disease stages [38]. Our models instead capture an established yet still reparable oxidative–inflammatory microenvironment in which NF‐κB–driven LCN2 is strongly associated with ROS accumulation, MMP activation, and ECM degradation. In this context, WDL/WPG is designed to restrain pathological NF‐κB–LCN2 overactivation rather than to universally eliminate LCN2. The concomitant increase in CD206/IL‐10 and decrease in CD86/TNF‐α suggest a shift toward a more reparative macrophage phenotype, consistent with targeted modulation of a maladaptive LCN2 amplifier rather than global suppression of its physiological functions. Downstream of this transcriptional regulation, the LCN2 amplifier is functionally coupled to matrix degradation via protease stabilization. LCN2 stabilizes MMP9 by forming an NGAL/MMP9 complex, preserving gelatinolytic activity [12, 39, 40], consistent with our observed reduction in MMP13 and restoration of matrix markers in vivo (Figure 10). Together, these data position NF‐κB–LCN2 as a tractable, disc‐relevant amplifier whose disruption by WDL—delivered on demand by a ROS‐responsive depot—rebalances inflammation, redox, and ECM homeostasis. Lcn2 transcriptional regulation is multimodal, receiving inputs from parallel cytokine axes such as IL‐6/STAT3 and IL‐17A pathways, in addition to NF‐κB [41]. While these alternative pathways may contribute to the overall inflammatory milieu, our pharmacological blockade experiments (Figure 3G) demonstrate that specific inhibition of NF‐κB is sufficient to abrogate the majority of oxidative stress‐induced LCN2 upregulation in AFCs. This suggests that, in the specific pathological context of the degenerating disc, NF‐κB acts as the predominant driver. Therefore, we propose that WDL exerts its therapeutic effect primarily by dismantling the NF‐κB–dependent LCN2 induction, although concurrent modulation of auxiliary signaling pathways cannot be entirely excluded.
The WPG platform integrates a ROS‐labile phenylboronic‐ester/PVA network (PT) within a covalently crosslinked GelMA–elastin framework (GAE). The boronate‐diol chemistry enabling H_2_O_2_‐triggered cleavage and reversible covalency is well characterized in biomedical hydrogel systems [42]. Functionally, graded H_2_O_2_ accelerated WDL release, while release under PBS remained minimal; a pathology‐triggered profile that complements the interpenetrating, elastomeric GAE skeleton, which maintains structural integrity and moderates swelling under load—two disc‐specific constraints frequently overlooked in responsive gels [13].
SEM confirmed a porous interpenetrating microarchitecture, and rheology showed G′ > G″ demonstrated dominant elastic behavior, consistent with viscoelastic reinforcement by dual‐network integration [43, 44]. Moreover, this physicochemical reinforcement directly underpins the biological performance of WPG. The physicochemical reinforcement conferred by the dual‐network architecture directly underpins its biological performance. The enhanced storage modulus (G′, Figure 4E) provides viscoelastic stiffness to withstand intradiscal compression, thereby reducing the risk of mechanical washout or fragmentation of the depot under physiological load. At the same time, the reduced swelling ratio (Figure 4D) indicates a tighter and more hydrodynamically stable mesh, which limits passive solute diffusion and supports WDL release that is primarily governed by ROS‐mediated bond cleavage rather than nonspecific pore expansion. Together, these properties help create a mechanically stable microenvironment that secures the therapeutic payload and offers a consistent structural support for AFC adhesion and migration (Figure 5). Cytocompatibility was high by live/dead, CCK‐8, and EdU; phalloidin and vinculin staining confirmed cytoskeletal integrity and adhesion, and scratch assays supported preserved migratory capacity—key for AF repair. These biological results, together with the material's mechanical properties, demonstrate that the WPG hydrogel not only provides a structurally resilient scaffold but also promotes cellular behaviors essential for tissue regeneration. The enhanced mechanical stability, combined with the ROS‐triggered release mechanism, ensures that the hydrogel maintains its integrity and continues to support the therapeutic payload during critical stages of the repair process. This synergy between material properties and biological performance further emphasizes the potential of WPG for intervertebral disc regeneration.
Bulk RNA‐seq across macrophages and AFCs systems provided a cross‐compartment, unbiased readout, revealing down‐regulation of NF‐κB/chemokine and oxidant‐responsive modules, couples with recovery of ECM and adhesion‐related gene programs. Enrichment analysis consistently highlighted PI3K–AKT, IL‐17, and ECM–receptor signaling—pathways recurrently implicated in disc degeneration and matrix homeostasis in recent multi‐omics meta‐analyses [45, 46, 47]. The concordance between transcriptome shifts and experimental phenotypes (cytokine profiles, matrix gene expression, antioxidant indices, and macrophage polarization) reinforces the causal cascade from amplifier disruption to transcriptional reset and phenotypic rescue.
In a rat annulus fibrosus defect (needle‐puncture) model, intradiscal WPG preserved disc height and T_2_‐weighted MRI signal relative to controls. Histology demonstrated maintenance of annulus integrity, proteoglycan retention (SO‐Fast Green), and restored collagen alignment (Sirius Red). Biomechanics testing showed near‐native stress–strain behavior in WPG‐treated segments, indicating functional preservation rather than fibrotic stiffening. These outcomes parallel other ROS‐responsive hydrogel systems recently evaluated in IVDD rat models, yet WPG is distinguished by coupling a mechanism‐specific payload (NF‐κB‐targeting WDL) with a load‐bearing dual network scaffold, overcoming the classical trade‐off between responsiveness and mechanical robustness [48, 49]. Importantly, biosafety was supported by stable hematology and normal histology of major organs (heart, liver, spleen, lung, kidney), consistent with localized ROS‐biased release and minimal systemic exposure typical of rationally designed intradiscal depots [50].
Recent boronate‐ester hydrogels (e.g., HA‐PBA/PVA) have proven effective for ROS‐triggered cargo release and anti‐inflammatory/anti‐infective therapy in oxidatively stressed tissues, underscoring the responsiveness of PBA–diol chemistry [42, 51, 52, 53, 54]. Yet such single‐network formulations are generally tailored for cue‐responsiveness rather than mechanical endurance under sustained compression. In contrast, our dual‐network WPG interpenetrates a ROS‐labile PT subnetwork with a covalently crosslinked GelMA–elastin backbone, balancing triggered delivery with structural resilience appropriate for the load‐bearing disc microenvironment. Disc‐focused ROS‐addressable depots have likewise shown efficacy—for example, a PVA‐tsPBA injectable delivering SLC7A11‐modRNA in a rat IVDD model [48]—supporting the translational value of pathology‐gated release while highlighting WPG's distinct mechanics‐plus‐responsiveness profile. Pharmacologically, WDL is not merely an antioxidant—it intercepts NF‐κB at IKK/p65 and simultaneously scavenges free radicals, rendering it ideally suited to collapse the NF‐κB–LCN2 loop rather than buffer ROS alone [24, 55, 56]. Finally, our in vitro delivery design reflects intradiscal pharmacokinetic realities: low diffusion and rapid washout. A microenvironment‐adaptive, ROS‐triggered system enhances local drug exposure while minimizing systemic spillover [13].
The caudal rat differs from the human lumbar disc; therefore, future studies employing larger or upright‐loading animal models will be critical for translation. Long‐term clearance and repeat‐dose tolerance should be characterized, and intradiscal pharmacokinetic profiling, bioluminescence imaging, or mass spectrometry‐based exposure mapping could de‐risk future translation [57]. While our immune analyses centered on LCN2 and NF‐κB, deeper exploration of macrophage temporal plasticity(M1/M2 transitions via single‐cell RNA‐seq or spatial proteomics) and disc‐cell senescence programs would refine causal mapping [58]. Given the proteolytic partnership between LCN2–MMP9 [59], in vivo neutralization or interface‐blocking studies could further quantify its contribution to matrix degradation during IVDD.
By coupling a druggable inflammatory amplifier (NF‐κB–LCN2) with a ROS‐addressable depot, this study highlights that the timing and spatial precision of anti‐inflammatory intervention are as critical as its potency. The same logic is extensible beyond this system: WPG is payload‐flexible and could be adapted to deliver senomorphics, pro‐anabolic cues, or pro‐resolving mediators. Conversely, WDL could be paired with other ROS‐responsive scaffolds or bioadhesive AF patches to optimize retention under mechanical motion. Early‐to‐mid‐stage IVDD—where microenvironment reprogramming can still avert irreversible failure—represents the most compelling therapeutic window [60]. The WPG system incorporates design features that support potential clinical adaptation. Its injectable, photo‐crosslinkable formulation enables minimally invasive intradiscal delivery and conformal filling of annular fissures. The mechanically reinforced dual‐network backbone ensures initial structural stability, although long‐term integration under physiological multi‐axial loading will require further evaluation in large‐animal models. From a manufacturing standpoint, the modular synthesis of TSPBA and the established preparation of GelMA–elastin components are compatible with standardizable, scalable manufacturing workflows. These considerations outline a feasible translational trajectory while recognizing key steps needed for clinical progression.
Conclusion
4
We present a mechanism‐informed and material‐enabled therapeutic strategy for disease modification in IVDD. By delineating the NF‐κB–LCN2 axis as a disc‐relevant, druggable inflammatory amplifier and coupling it with a ROS‐responsive dual‐network hydrogel (WPG) capable of on‐demand wedelolactone (WDL) release, we achieve coordinated regulation of inflammation and oxidative stress, leading to matrix preservation and functional restoration. Through multi‐scale validation spanning cellular systems, transcriptomic profiling, and a rat annulus fibrosus defect model, WPG effectively collapses the NF‐κB–LCN2 feedback loop, suppresses cytokine and catabolic signaling, and preserves ECM architecture with high biocompatibility. The modular design—integrating a mechanically robust backbone with pathology‐triggered pharmacology—offers a blueprint for adapting diverse therapeutic payloads and incorporating mechanosensitive or immune‐modulatory modules. Looking forward, scalable synthesis, image‐guided intradiscal delivery with exposure tracking, and evaluation in large‐animal, load‐bearing models will advance translational readiness. Overall, this study bridges mechanism discovery with delivery design and highlights precision, pathology‐responsive biomaterials as a promising path toward disease‐modifying therapy for IVDD.
Experimental Section
5
Materials and Reagents
5.1
Poly (vinyl alcohol) (PVA, 31–145 kDa, ≥98% hydrolyzed, Sigma‐Aldrich), tris(4‐formylphenyl)boronic acid, gelatin methacryloyl (GelMA, Aladdin, cat. M299513), and Elastin‐PEG‐acrylate (Elastin‐PEG‐AC (modified elastin); provided by Yan. LP lab) were used as the polymeric components. Wedelolactone (WDL, ≥98%, MedChemExpress, cat. HY‐N0282) served as the payload. Hydrogen peroxide (H_2_O_2_, 30% w/w, Sigma), LPS (Sigma), DMEM, FBS, and penicillin‐streptomycin (Gibco) were used as received. All biochemical assay kits and fluorescent probes were purchased from Abbkine unless otherwise specified, including CCK‐8, EdU, DCFH‐DA, GSH, SOD, and Annexin V–FITC/PI kits. The LCN2 ELISA kit was obtained from Abcam (cat. ab199083). qPCR primers and antibodies are listed in Table S1 and Table S2 where noted.
All aqueous reagents were prepared with ultrapure water (18.2 MΩ cm). UV–vis measurements were performed on a Thermo Evolution 220 spectrophotometer (λ max = 332 nm). Image and statistical analysis used ImageJ (v1.53), Imaris (v9.8), and GraphPad Prism (v9.4). R (v4.3.0) was used for bioinformatic analyses.
Network Pharmacology and Molecular Docking
5.2
Wedelolactone (WDL) targets were predicted using Swiss Target Prediction, ChEMBL, and STITCH, while IVDD‐related genes were obtained from Gene Cards, OMIM, and DisGeNET. Datasets were filtered for Homo sapiens entries, standardized via UniProt, and intersected to identify overlapping genes. Protein– interaction (PPI) networks were constructed from STRING (confidence ≥ 0.7) and visualized in Cytoscape. Functional enrichment (GO, KEGG, Reactome) was conducted with cluster Profiler (Benjamini‐Hochberg FDR < 0.05). For molecular docking, crystal structures of representative targets were retrieved from RCSB PDB, and the ligand structure of WDL from PubChem. Docking was carried out with AutoDock Vina (default exhaustiveness = 8); receptor and ligand were prepared by adding hydrogens and Gasteiger charges; grid boxes centered on the reported binding pockets. Interaction poses were visualized using PyMOL and Discovery Studio.
Synthesis of the ROS‐Labile Crosslinker (TSPBA)
5.3
N ^1^‐(4‐boronobenzyl)‐N ^3^‐(4‐boronophenyl)‐N ^1^, N ^1^, N ^3^, N ^3^‐tetramethylpropane‐1,3‐diaminium (TSPBA) was synthesized based on a previously reported method [27]. Briefly, N, N, N′, N′‐tetramethyl‐1,3‐propanediamine and 4‐(bromomethyl)phenylboronic acid were reacted in DMF (60°C, 12–16 h). The mixture was precipitated into THF, filtered, washed (THF, 3×), and vacuum‐dried to afford TSPBA. Product identity was confirmed by 1H NMR.
Hydrogel Preparation and WDL Loading
5.4
PTGAE Dual‐Network Hydrogel
5.4.1
The ROS‐labile network (PT) was obtained by mixing PVA (5 wt.%) and TSPBA aqueous solutions (boronate: diol molar ratio = 0.5) at 25°C under continuous stirring to form dynamic boronic‐ester linkages. The elastic secondary network (GAE) was prepared by photocrosslinking GelMA (5 wt.%) and Elastin‐PEG‐AC (3 wt.%; degree of functionalization verified batch‐wise) under UV light (365 nm, 5 mW cm^−^ ^2^, 30 s). Equal volumes of the two prepolymer solutions were combined (1:1, v/v) and photo‐cured to obtain the interpenetrating dual‐network hydrogel (PTGAE). Product identity was confirmed by ^1^H NMR; full physicochemical characterization is provided in Figure 4A‐I
WDL Loading
5.4.2
Wedelolactone was dissolved in DMSO to prepare a 10 mm stock solution. The stock was added to the PTGAE pre‐gel mixture immediately before curing, yielding the drug‐loaded hydrogel (WPG). The total loading was adjusted to target a release window of ∼10–30 µm in citro (≈3–9 µg mL^−1^), matching the cytoprotective range (5–20 µm) determined by CCK‐8 assays in RAW 264.7 macrophages and AFCs (Figure S3). For each gel specimen (∼200 mg), the nominal loading was ∼270 µg WDL per specimen, supporting the targeted release profile over 48 h under basal and oxidative conditions. The final DMSO content in the pre‐gel solution was maintained below 0.5% v/v and matched in all vehicle controls. Sterility was maintained by 0.22 µm filtration of liquid components and brief UV exposure prior to crosslinking.
Morphological and Rheological Characterization
5.5
Morphology
5.5.1
The microstructures of freeze‐dried hydrogels (PT, GAE, and PTGAE) were examined using a field‐emission scanning electron microscope (FEI Nova NanoSem 450, Czech Republic) operated at 5 kV. Samples were frozen at −80°C for 12 h, lyophilized for 24 h, and sputter‐coated with a ≈10 nm gold layer. Cross‐sectional images were acquired, and pore diameters were analyzed in ImageJ from at least five randomly chosen fields.
Rheological Characterization
5.5.2
Dynamic oscillatory rheology was performed on a rotational rheometer (HAAKE MARS III, Thermo Scientific, Germany) with a 20 mm parallel‐plate geometry at 25°C; gap 1 mm. Amplitude sweeps (0.1%–100% strain, 1 Hz) were used to define the linear viscoelastic region (LVR), followed by frequency sweeps (0.1–10 Hz, 1% strain) within the LVR. The storage (G′) and loss (G″) moduli were recorded. Each sample was tested in triplicate (n = 3).
Swelling Behavior and In Vitro Drug Release
5.6
Swelling Test
5.6.1
Pre‐weighed cylindrical hydrogel disks (approximately Ø10 mm × 5 mm, ≈200 mg each) of PT, GAE, and PTGAE were immersed in PBS (pH 7.4, 37°C). At predetermined time intervals (0.5, 1, 3, 6, 12, and 24 h; 1, 3, 4, 7, and 14 days), samples were removed, gently blotted to remove surface water, and weighed (W t). The swelling ratio (SR) was calculated as
where W 0 is the initial dry mass. Each measurement was performed in triplicate.
In Vitro ROS‐Responsive Drug Release
5.6.2
Drug‐loaded hydrogels (WPG, ≈200 mg) were incubated in 10 mL PBS (pH 7.4, 37°C) containing different H_2_O_2_ concentrations (0, 10^−4^,10^−1^, and 1 m). At scheduled time points (0.5, 1, 3, 6, 12, and 24 h; 1–7, 14, and 21 days), 1 mL aliquots were withdrawn and replaced with equal volumes of fresh buffer. The released WDL was quantified by UV–vis spectroscopy at 332 nm using a pre‐established calibration curve (R ^2^ > 0.999; 5–50 µg mL^−1^).
Cell Culture and Treatments
5.7
RAW264.7 macrophages (ATCC, Manassas, USA) and primary s AFCs (isolated from rat caudal discs) were cultured in Dulbecco's Modified Eagle Medium (DMEM, Gibco) and DMEM/F12 supplemented with 10% FBS (Gibco) and 1% penicillin–streptomycin (Gibco) at 37°C in a humidified incubator with 5% CO_2_. Cells were passaged at 70%–80% confluence and used between passages 3–6 for all experiments.
Inflammatory Stimulation in Macrophages
5.7.1
To establish an inflammatory model, RAW264.7 cells were treated with 100 ng mL^−1^ lipopolysaccharide (LPS, E. coli O111:B4, Sigma) for 12 h, followed by incubation with WDL (1.56–6.25 µg mL^−1^; corresponding to LW1–LW3) or hydrogel extract (WPG leachate) for another 24 h. Vehicle controls (DMSO < 0.1%, v/v) were included in all groups. Macrophage polarization was analyzed by flow cytometry using CD86 (M1) and CD206 (M2) surface markers. Immunofluorescence staining was performed to detect TNF‐α, IL‐10, and P‐p65 localization. Intracellular ROS were visualized with DCFH‐DA probes, and fluorescence intensity was quantified using ImageJ.
Oxidative Stress Model in AFCs
5.7.2
AFCs were exposed to 200 µm hydrogen peroxide (H_2_O_2_) for 24 h to induce oxidative stress, and subsequently treated with WDL (1.56–6.25 µg mL^−1^) or WPG extracts for another 24 h. Gene expression levels of TNF‐α, IL‐6, IL‐1β, and Lcn2 were quantified by qPCR. Immunofluorescence staining was conducted to evaluate P‐p65 nuclear translocation, LCN2, and MMP9 expression.
Macrophage‐to‐AFC Conditioned‐Medium Experiment
5.7.3
To evaluate paracrine signaling through the NF‐κB–LCN2 axis, LPS‐stimulated RAW264.7 macrophages (Ctrl, LPS, LPS + WDL, LPS + LCN2 inhibitor) were cultured for 24 h. The conditioned medium (CM) was collected, centrifuged (3000 rpm, 10 min), and applied to AFCs for an additional 24 h. Expression of extracellular‐matrix genes (Acan, Col1a1, MMP13, Adamts5) and inflammatory cytokines (IL‐6, TNF‐α) in AFCs was analyzed by qPCR. Intracellular antioxidant activities were determined using GSH and SOD assay kits (Abbkine). The concentration of LCN2 in macrophage CM was quantified by ELISA (Abcam).
LCN2 Intervention
5.7.4
To verify the role of LCN2, recombinant LCN2 (rLCN2, 200 ng mL^−1^) or an LCN2 inhibitor (LCN2i, 10 µm, MedChemExpress, Cat# HY‐Q45780) was added to AFC or RAW264.7 cultures with or without WDL treatment for 24 h. LCN2 and P‐p65 localization were analyzed by immunofluorescence.
NF‐κB Inhibition
5.7.5
To confirm the involvement of the NF‐κB pathway, cells were pretreated with the NF‐κB inhibitor JSH‐23 (10 µm, MedChemExpress, Cat# HY‐13982) for 1 h before LPS stimulation, followed by WDL (6.25 µg mL^−1^) treatment for 24 h. NF‐κB activation was assessed by P‐p65 nuclear translocation, and downstream targets LCN2 and MMP9 were evaluated by immunofluorescence.
Cell Viability, Proliferation, Adhesion, and Migration Under Oxidative Stress
5.8
To evaluate the cytocompatibility and functional performance of the hydrogels under oxidative conditions, AFCs were exposed to 200 µm H_2_O_2_ for 24 h to induce oxidative stress and subsequently treated with WPG (or its extract) or PG (or its extract) for 24 h.
Live/Dead Assay
5.8.1
Cell viability was assessed using a Live/Dead Viability/Cytotoxicity Kit (Calcein‐AM/PI, Beyotime, China) according to the manufacturer's instructions. Fluorescence images were captured on a confocal microscope (Zeiss LSM 980), and the ratio of live (green) to dead (red) cells was quantified using ImageJ.
Cell Proliferation
5.8.2
Proliferative activity was determined by CCK‐8 assay (Abbkine) at 1, 3, and 5 days (A_450_), and by EdU incorporation assay (Abbkine) following the manufacturer's protocol. Nuclei were counterstained with Hoechst 33 342, and EdU‐positive cells were expressed as a percentage of total nuclei.
Cytoskeletal Organization and Adhesion
5.8.3
After treatments, AFCs were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X‐100. F‐actin filaments were stained with Phalloidin‐594 (Abbkine), and focal adhesion complexes were visualized using an anti‐Vinculin antibody (Abcam) followed by FITC‐conjugated secondary antibody. Images were acquired by confocal microscopy (Zeiss LSM 980) to evaluate cytoskeletal integrity and adhesion morphology.
Cell Migration
5.8.4
For wound‐healing analysis, confluent AFC monolayers were scratched with a sterile 200 µL pipette tip and washed with PBS to remove debris. Cells were then cultured with the indicated treatments under oxidative stress, and images were captured at 0 and 24 h using a phase‐contrast microscope. The migration area was quantified as % wound‐area closure in ImageJ.
Macrophage–AFC Coculture With Hydrogels
5.9
Coculture Setup
5.9.1
RAW264.7 macrophages (1 × 10⁵ per insert) were seeded in the upper Transwell chamber (0.4 µm pore size, Corning), and AFCs (2 × 10⁵ per well) were plated in the lower chamber. After 12 h, macrophages were stimulated with LPS (100 ng mL^−1^). WPG hydrogels, PG hydrogels, or WDL solution (6.25 µg mL^−1^) were added to the upper chamber at the time of LPS. Controls lacked LPS/hydrogel. Co‐cultures were maintained for 24 h at 37°C, 5% CO_2_.
Immunofluorescence Analysis
5.9.2
After coculture, macrophages were stained for TNF‐α and IL10; AFCs were stained for ACAN, ADAMTS5, and F‐actin (Phalloidin). Fluorescent images were acquired under identical parameters. Quantification of fluorescence intensity/volume was performed in Imaris (v9.8).
ROS and Apoptosis Assays
5.9.3
Intracellular ROS in AFCs were detected using DCFH‐DA. Apoptosis was assessed by Annexin V‐FITC/PI double staining according to the manufacturer's instructions. Fluorescence images were recorded by a confocal microscope, and mean fluorescence intensity or positive‐cell ratios were analyzed in ImageJ.
Bulk RNA‐Seq and Bioinformatic Analysis
5.10
Sample Preparation and RNA Extraction
5.10.1
AFCs and RAW 264.7 macrophages were treated as described. After 24 h, total RNA was extracted using TRIzol Reagent (Invitrogen). RNA integrity was confirmed (Agilent 2100 RIN ≥ 8); purity and concentration were measured with a NanoDrop 2000 (Thermo Scientific).
Library Construction and Sequencing
5.10.2
RNA‐seq libraries were generated with NEBNext Ultra RNA Library Prep Kit for Illumina (NEB) following the manufacturer's protocol and sequenced on an Illumina NovaSeq 6000 platform (PE 150). Each biological group included three independent replicates (n = 3).
Data Processing and Differential‐Expression Analysis
5.10.3
Raw reads were quality‐checked (FastQC) and trimmed (Trimmomatic). Clean reads were aligned to the appropriate reference genome (AFCs: Rnor_6.0; macrophages: GRCm39), and transcript abundance was estimated using featureCounts. Differentially expressed genes (DEGs) were identified with DESeq2 (v1.30.1) using |log_2_ fold change| ≥ 1 and adjusted p < 0.05. Principal‐component analysis and hierarchical clustering were performed in R.
Functional Enrichment and Pathway Analysis
5.10.4
GO and KEGG enrichment were conducted using clusterProfiler (v4.2.2) with the Benjamini–Hochberg FDR < 0.05. Reactome and GSEA were used for gene‐set level analysis. Networks were visualized in Cytoscape (v3.9.1), and PPI networks were constructed in STRING (confidence ≥ 0.7). Venn and Sankey‐plot analyses were generated in R.
In Vivo Rat Model of Annulus Fibrosus Defect and Treatment
5.11
Ethical Approval and Animal Care
5.11.1
All procedures complied with the Southern University of Science and Technology guidelines (Approval No. SUSTech‐JY202406106). Male Sprague–Dawley rats (6–8 weeks old, 200–250 g) were obtained from Jiangsu Jiyu Pharmaceutical Co., Ltd., China. Animals were housed in SPF conditions at 23 ± 2°C with a 12‐h light/dark cycle and ad libitum access to food and water. All assessments were performed under blinded allocation and outcome assessment.
Establishment of the AF Defect Model
5.11.2
A needle‐puncture–induced annulus fibrosus (AF) defect model was created under isoflurane anesthesia (1.5%–2%). The Co7/8 disc was exposed through a midline posterior tail incision (∼2 cm). An 18‐gauge needle was inserted perpendicular to a depth of 1.4 mm, generating a full‐layer AF defect (∼0.6 mm diameter) with minimal nucleus pulposus injury. Immediately after puncture, 5 µL of the indicated material was injected using a Hamilton microsyringe: (1) Sham (no puncture); (2) IVDD (puncture only); (3) IVDD + WDL (free WDL, 10 µg in 5 µL); (4) IVDD + WPG (WPG containing 10 µg WDL in 5 µL). In situ photo‐crosslinking was initiated using 405 nm light for ∼60 s to solidify the hydrogel. The incision was closed in layers and disinfected with 75% ethanol. Animals were allowed free movement postoperatively, and analgesics were administered as required.
Radiographic and MRI Evaluation
5.11.3
At 4 and 12 weeks post‐surgery, rats were anesthetized and subjected to imaging. Lateral micro‐computed tomography (micro‐CT) and MRI were used to assess disc structure and hydration. X‐ray micro‐CT imaging was performed using a high‐resolution system (SkyScan 1275, Bruker, Belgium) operated at 60 kV and 88 µA. The disc height index (DHI) was calculated as (disc height)/(mean height of adjacent vertebral bodies) on reconstructed sagittal images (CTAn, Bruker). T_2_‐weighted images were acquired on a Philips Ingenia Elition S 3.0 T scanner using a 3D T2 turbo spin‐echo sequence (TR = 1800 ms, TE = 122 ms, slice thickness = 1.2 mm). For each disc, signal intensity was measured within a mid‐sagittal ROI encompassing the nucleus pulposus and normalized to paraspinal muscle to reduce inter‐scan variability; values were averaged across three contiguous slices. For quantitative analysis, signal intensity was measured on mid‐sagittal slices using an ROI that strictly encompassed the nucleus pulposus while excluding the endplate margins to avoid partial‐volume artifacts. Signal values were normalized to the mean intensity of a paraspinal muscle ROI on the same slice to reduce inter‐scan variability, and the final value for each disc was obtained by averaging across three contiguous slices.
Histological and Immunohistochemical Analysis
5.11.4
At 4 and 12 weeks, Co6–Co8 motion segments were excised, fixed (4% paraformaldehyde), decalcified (10% EDTA), embedded in paraffin, and sectioned (5 µm). Sections were stained with H&E, Safranin O–Fast Green (SOFG), and Sirius Red (polarized light). Immunofluorescence used antibodies against Col I, Acan, MMP13, TNF‐α, IL‐10, P‐p65, and LCN2. Images were captured by confocal microscopy; quantitative analysis of positive area and mean intensity was performed in ImageJ.
Biomechanical Assessment
5.11.5
Remaining motion segments were subjected to axial compression testing using a 30 kN electronic universal testing machine (Instron 2367, USA). Samples were kept hydrated in PBS during testing. Compression was applied at 0.5 mm min^−1^ up to 50% strain, and stress–strain curves were recorded. The toe modulus was calculated as the slope of the initial low‐strain region (0%–10%), whereas the linear modulus was derived from the subsequent linear‐elastic region (typically 20%–40%) using least‐squares regression (R ^2^ > 0.95). Each measurement was performed in triplicate (n = 6).
Biocompatibility and Biosafety
5.11.6
At 12 weeks, major organs (heart, liver, spleen, lung, kidney) were harvested, fixed, paraffin‐embedded, sectioned (5 µm), and stained with H&E. Whole blood was collected by cardiac puncture under anesthesia and analyzed on an automated hematology analyzer.
Statistical Analysis
5.12
All quantitative data were presented as mean ± standard deviation (SD). Sample sizes (n), representing the number of independent biological replicates, were explicitly indicated in the figure legends. No specific pre‐processing methods for outlier removal were applied unless otherwise stated. Normality of data distribution was assessed using the Shapiro–Wilk test, and homogeneity of variance was verified using Levene's test. For comparisons between two groups, statistical significance was determined using a two‐tailed Student's *t‐*test (for parametric data) or the Mann–Whitney U‐test (for non‐parametric data). For comparisons involving three or more groups, a one‐way analysis of variance (ANOVA) was performed, followed by Tukey's post hoc test for multiple comparisons. For RNA‐seq functional enrichment, over‐representation analysis (hypergeometric test) was performed, and p‐values were adjusted by Benjamini–Hochberg FDR; adjusted *p *< 0.05 was considered significant. PPI networks were generated using STRING (minimum required interaction score ≥ 0.4) and visualized in Cytoscape. All statistical analyses were conducted using GraphPad Prism software (Version 9.4, GraphPad Software, San Diego, USA) for standard statistical tests. RNA‐seq differential expression was performed using DESeq2 in R (vendor pipeline), and GSEA was performed using GSEA software (Version 4.1.0). Enrichment analyses and GSEA were performed using the vendor pipeline with Benjamini–Hochberg FDR correction. A p‐value < 0.05 was considered statistically significant (**p *< 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).
Author Contributions
Z.M.W. and Y.X. contributed equally to this work. Z.M.W. dealt with validation, investigation, formal analysis, visualization, and wrote the original draft. Y.X. dealt with validation, investigation, and formal analysis. L.Q. dealt with validation and investigation. Z.M.L. dealt with validation and formal analysis. Y.J.M. dealt with validation and investigation. X.T.G. dealt with investigation and formal analysis. J.J.C. dealt with investigation and formal analysis. L.P.Y. dealt with formal analysis and visualization. L.W. dealt with validation, investigation, and formal analysis. L.M.B. dealt with formal analysis, visualization, and wrote the review and editing. F.X.W. dealt with conceptualization, investigation, visualization, supervision, project administration, funding acquisition, and wrote the review and editing.
Conflicts of Interest
The authors declare no conflict of interest.
Supporting information
Supporting File: advs73553‐sup‐0001‐SuppMat.docx.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1M. L. Ferreira , K. de Luca , L. M. Haile , et al., “Global, Regional, and National Burden of Low Back Pain, 1990–2020, Its Attributable Risk Factors, and Projections to 2050: a Systematic Analysis of the Global Burden of Disease Study 2021,” The Lancet Rheumatology 5, no. 6 (2023): e 316–e 329.37273833 10.1016/S 2665-9913(23)00098-XPMC 10234592 · doi ↗ · pubmed ↗
- 2V. M. Ravindra , S. S. Senglaub , A. Rattani , et al., “Degenerative Lumbar Spine Disease: Estimating Global Incidence and Worldwide Volume,” Global Spine Journal 8, no. 8 (2018): 784–794.30560029 10.1177/2192568218770769 PMC 6293435 · doi ↗ · pubmed ↗
- 3N. N. Knezevic , K. D. Candido , J. W. S. Vlaeyen , J. Van Zundert , and S. P. Cohen , “Low Back Pain,” The Lancet 398, no. 10294 (2021): 78–92.10.1016/S 0140-6736(21)00733-934115979 · doi ↗ · pubmed ↗
- 4Y. Wang , H. Cheng , T. Wang , K. Zhang , Y. Zhang , and X. Kang , “Oxidative Stress in Intervertebral Disc Degeneration: Molecular Mechanisms, Pathogenesis and Treatment,” Cell Proliferation 56, no. 9 (2023): 13448.10.1111/cpr.13448 PMC 1047253736915968 · doi ↗ · pubmed ↗
- 5X. Chen , A. Zhang , K. Zhao , et al., “The Role of Oxidative Stress in Intervertebral Disc Degeneration: Mechanisms and Therapeutic Implications,” Ageing Research Reviews 98 (2024): 102323.38734147 10.1016/j.arr.2024.102323 · doi ↗ · pubmed ↗
- 6W. Taylor and W. M. Erwin , “Intervertebral Disc Degeneration and Regeneration: New Molecular Mechanisms and Therapeutics: Obstacles and Potential Breakthrough Technologies,” Cells 13, no. 24 (2024): 2103.39768194 10.3390/cells 13242103 PMC 11674193 · doi ↗ · pubmed ↗
- 7D. Wu , G. Li , X. Zhou , et al., “Repair Strategies and Bioactive Functional Materials for Intervertebral Disc,” Advanced Functional Materials 32, no. 52 (2022): 2209471.
- 8Y. Dou , Y. Zhang , Y. Liu , et al., “Role of Macrophage in Intervertebral Disc Degeneration,” Bone Research 13, no. 1 (2025): 15.39848963 10.1038/s 41413-024-00397-7PMC 11758090 · doi ↗ · pubmed ↗