A dual-functional hydrogel integrating adhesive and lubricating interfaces for mitochondrial protection–Driven cartilage regeneration

TL;DR

A new hydrogel with adhesive and lubricating properties helps protect mitochondria and regenerate cartilage in osteoarthritis.

Contribution

A dual-functional hydrogel combining adhesion, lubrication, and mitochondrial protection for cartilage regeneration is developed.

Findings

The hydrogel showed excellent adhesion and lubrication under wet conditions.

It preserved chondrocyte viability, reduced oxidative stress, and improved cartilage and bone structure in a rat OA model.

Transcriptomic analysis revealed suppression of inflammatory pathways like IL-17A and TNF-α.

Abstract

Osteoarthritis (OA) progression is driven by chronic inflammation, oxidative stress, and mitochondrial dysfunction, which together disrupt cartilage homeostasis and hinder regeneration. Here, we developed a paeoniflorin-loaded multifunctional hydrogel (AdHy@Pae) composed of 2-Hydroxyethyl methacrylate, sulfobetaine methacrylate, and sodium polyglutamate. The latter two components endowed the material with excellent adhesive properties, whereas the introduction of Pae provided abundant hydroxyl groups to decrease interfacial energy, resulting in superior lubrication under wet condition. More importantly, with sustained release of Pae and scavenge of ROS, AdHy@Pae alleviated oxidative injury, restored mitochondrial membrane potential, and preserved cell viability of chondrocyte. It also rebalanced extracellular matrix (ECM) metabolism and maintained cartilage phenotypic stability. In a rat OA model, intra-articular administration of AdHy@Pae markedly restored cartilage and subchondral bone architecture, as evidenced by improved trabecular continuity, smooth cartilage surfaces, and significantly reduced OARSI and Mankin scores. Transcriptomic profiling and network analysis revealed that AdHy@Pae reprograms mitochondrial dynamics homeostasis and inflammatory gene networks, notably suppressing IL-17A and TNF-α signaling. Collectively, AdHy@Pae provides a robust and durable strategy for precise cartilage protection and repair in osteoarthritis.

Image 1

•Paeoniflorin was introduced to a photocrosslinked injectable hydrogel.•The prepared hydrogel exhibits excellent adhesion and lubrication properties, and can slowly release paeoniflorin.•The hydrogel could maintain cellular energy homeostasis and attenuating inflammatory feedback.•The hydrogel could significantly improved cartilage integrity and subchondral bone reconstruction in vivo.

Paeoniflorin was introduced to a photocrosslinked injectable hydrogel.

The prepared hydrogel exhibits excellent adhesion and lubrication properties, and can slowly release paeoniflorin.

The hydrogel could maintain cellular energy homeostasis and attenuating inflammatory feedback.

The hydrogel could significantly improved cartilage integrity and subchondral bone reconstruction in vivo.

Figures

Gelation and physicochemical characterization of hydrogels. (A) Photographs of the AdHy@Pae precursor solution before and after UV irradiation. (B) XRD patterns and (C) FTIR spectra of Hy, AdHy and AdHy@Pae hydrogels. (D) SEM images of freeze-dried Hy, AdHy and AdHy@Pae hydrogels at different magnifications and (E) corresponding average pore size. (F) Representative water contact angle images and (G) quantitative results of Hy, AdHy and AdHy@Pae hydrogels. (H) Quantitative results of fraction coefficient. μs represents static fraction coefficient, μd represents dynamic fraction coefficient. Ca

Adhesive, deformable and antioxidant performance of hydrogels. (A) Storage modulus and loss modulus of Hy, AdHy and AdHy@Pae. (B) Interfacial toughness, (C) Peel strength, and (D) lap shear strength of the three hydrogel. (E) Photographs showing adhesion of hydrogels to different substrates, deformability under various operations and stretchability. (F) *In vitro* adhesion of Hy, AdHy and AdHy@Pae on joints before and after soaking in PBS and after 500 flexion cycles. UV–vis spectra of different groups for (G) •OH scavenging, (H) ABTS•^+^ radical scavenging, and (I) superoxide anion (O_2_•^-^)

Regulation of extracellular matrix metabolism by AdHy@Pae in LPS-induced chondrocytes. (A) Representative immunofluorescence images showing the expression of COL2A1, SOX9, MMP9, and ADAMTS5 in different treatment groups and corresponding (B) quantitative analysis. Data are shown as mean ± SD (n = 3, ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001).Fig. 5

Expression analysis of cartilage-related proteins after treatment with AdHy@Pae. (A) Representative Western blot images showing ECM synthesis–related proteins (COL2A1, ACAN, COMP, and SOX9) and antioxidant markers (NRF2 and HO-1) in different groups. (B) Quantitative analysis of matrix-degrading enzymes (MMP13, ADAMTS5, ADAMTS1) and apoptosis-related proteins (BCL-2, Bax, and Caspase-3). Data are presented as mean ± SD (n = 3; ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001).Fig. 6

Transcriptomic analysis of AdHy@Pae–treated cartilage tissue. (A) Heatmap displaying DEGs between AdHy@Pae–treated and OA groups. (B) Volcano plot showing upregulated and downregulated DEGs identified between AHP and OA groups (|log_2_FC| ≥ 0.38, p < 0.05). (C) GSEA of DEGs indicating significantly enriched biological pathways. (D) WGCNA dendrogram constructed based on topological overlap. (E) Module–trait correlation heatmap showing relationships between co-expression modules and sample traits. (F) Scatter plot showing the correlation between module membership and gene significance within the

Identification and validation of the mechanism that AdHy@Pae alleviate OA. (A) Venn diagram illustrating the intersection among DEGs and WGCNA modules from OA and AdHy@Pae datasets. (B) Heatmap displaying the expression profiles of the five core genes in control, OA, and AHP groups. (C) Volcano plots of DEGs from OA vs. control and AHP vs. OA comparisons, with the five intersecting genes highlighted. (D) KEGG pathway enrichment analysis showing the distribution of the five core genes across inflammation- and metabolism-related signaling pathways. (E) Validation of the protein expression by Wes

Full text

Introduction

1

Osteoarthritis (OA) and related musculoskeletal disorders are chronic degenerative diseases characterized by inflammation-driven cartilage degradation, leading to persistent pain, joint stiffness, and restricted mobility [1,2]. The destruction of cartilage not only compromises the biomechanical stability of the joint [3] but also triggers subchondral bone remodeling [4] and synovial inflammation [5], ultimately resulting in irreversible functional impairment. Although pharmacological [6] and surgical interventions [7] can alleviate symptoms and improve joint mobility to some extent, most current treatments are limited to short-term pain relief and anti-inflammatory effects, failing to halt or reverse the progressive cartilage degeneration [8]. Therefore, developing novel therapeutic strategies capable of effectively delaying or even reversing cartilage degeneration has become an urgent clinical need.

At the molecular level, sustained inflammatory stimulation is recognized as a key driver of OA progression [9]. Proinflammatory cytokines such as TNF-α and IL-17A disrupt cellular redox homeostasis, leading to excessive accumulation of reactive oxygen species (ROS) [[10], [11], [12], [13]]. Elevated ROS levels cause mitochondrial dysfunction, ATP depletion, and activation of intrinsic apoptotic pathways, forming a vicious cycle between oxidative stress and inflammation [14]. This persistent cellular damage impairs chondrocyte viability and extracellular matrix (ECM) synthesis, accelerating tissue degeneration and joint destruction [15,16]. Hence, restoring mitochondrial function, modulating oxidative stress, and reestablishing cellular homeostasis are critical therapeutic approaches to mitigate inflammation-associated cartilage injury.

Despite the well-established role of oxidative stress in OA pathogenesis, current pharmacological therapies (e.g., nonsteroidal anti-inflammatory drugs [17], corticosteroids [18]) and intra-articular injections (e.g., hyaluronic acid [19]) mainly offer transient symptom relief without addressing the underlying pathology. These formulations exhibit poor tissue adhesion, short intra-articular retention, and limited anti-inflammatory persistence, thereby failing to sustain cellular modulation or promote cartilage regeneration [20].

Paeoniflorin (Pae), a bioactive monoterpene glycoside extracted from Paeonia lactiflora, has recently attracted considerable attention due to its multitarget pharmacological effects [[21], [22], [23], [24]]. Pae exerts potent anti-inflammatory and antioxidant activities by suppressing TNF-α and IL-17A expression and activating the NRF2/HO-1 signaling pathway, thereby alleviating oxidative stress, protecting mitochondrial function, and enhancing ECM synthesis [[25], [26], [27], [28], [29]]. However, its strong hydrophilicity, poor stability, and rapid clearance in physiological fluids greatly limit its long-term therapeutic efficacy in vivo [30]. Consequently, it is essential to design a suitable drug delivery platform capable of ensuring sustained release and biological protection of Pae within the inflammatory microenvironment, thereby achieving synergistic anti-inflammatory, antioxidative, and tissue-regenerative effects.

Injectable hydrogels have emerged as promising carriers for joint repair due to their excellent biocompatibility, high water content, and structural versatility [31,32]. Compared with traditional delivery systems, hydrogels form a three-dimensional polymeric network through physical or chemical crosslinking, enabling localized retention and controlled drug release [33,34]. However, conventional hydrogels often suffer from insufficient tissue adhesion and suboptimal lubrication, limiting their stability under the high-friction and dynamically loaded environment of the joint cavity [35]. An ideal hydrogel system for OA therapy should possess both strong adhesiveness and superior lubrication, allowing firm attachment to wet cartilage surfaces while effectively reducing friction and providing mechanical buffering, thus creating a favorable interface for controlled drug release and tissue regeneration [[36], [37], [38]].

Based on this rationale, we developed an Pae-loaded hydrogel with strong adhesion to bone while maintaining lubricity in wet environments (AdHy@Pae) for the integrated treatment of inflammation-associated joint degeneration (Scheme 1). The hydrogel network was formed via UV-initiated free radical polymerization of 2-hydroxyethyl methacrylate (HEMA), where the introduction of sodium polyglutamate (γ-PGA-Na) and sulfobetaine methacrylate (SMBA) endowed the hydrogel with moderate interfacial adhesion, while the incorporation of Pae further reinforced both adhesion and lubrication through enhanced interfacial hydration and molecular mobility. Benefiting from the synergy between mechanical reinforcement and pharmacological regulation, as well as ROS scavenging, AdHy@Pae significantly improved cartilage integrity via mitochondrial protection, offering a promising strategy for precise treatment of inflammatory osteoarthritic diseases.Scheme 1A schematic illustration of the preparation of AdHy@Pae and its cartilage repair via mitochondrial protection.Scheme 1

Experimental section

2

Preparation of AdHy@Pae

2.1

The AdHy hydrogels were prepared via UV-initiated free radical polymerization. Briefly, 30 wt% HEMA (Shanghai Yuanye, Y56248), 2 wt% γ-PGA-Na (Macklin, P762215, 92%), 30 wt% SMBA (Shanghai Yuanye, S70859), and 1 wt% 2959 (Bidepharm, BD17449) were sequentially dissolved in ultrapure water under stirring to obtain a homogeneous precursor solution. The mixture was then transferred into molds and exposed to UV light (365 nm) for 20–30 min to form the crosslinked hydrogel network. The obtained hydrogels were rinsed with PBS (Macklin, P917808) to remove unreacted residues and stored at 4 °C before use. Hy hydrogel was prepared only with HEMA. For AdHy@Pae preparation, Pae (Chengdu Herbpurify, 23180-57-6) was added to the precursor.

To determine the working concentration of Pae, LPS-stimulated chondrocytes were cultured with AdHy@Pae prepared with different Pae loadings, and cell viability was evaluated using a CCK-8 assay. Based on these results, the optimal non-cytotoxic concentration (150 μg/mL) was selected for subsequent experiments (Fig. S1).

Material characterization

2.2

To assess the adhesion, deformability, and stretchability of AdHy@Pae, the hydrogel was laminated onto glass, polypropylene (PP), wood, rubber, foam, and porcine skin, then subjected to bending, twisting, and 1 h water immersion to evaluate adhesion stability under mechanical deformation and wet conditions. Rheology was conducted at 37 °C on a rotational rheometer (TA Instruments, USA) using a 20 mm parallel-plate geometry, with oscillatory frequency sweeps from 0.1 to 100 rad s^−1^ to record storage (G′) and loss (G″) moduli. Crystalline features were analyzed by X-ray diffraction (Bruker D8 Advance, Germany), and freeze-dried samples were imaged by scanning electron microscopy (Hitachi SU-8010, Japan) to examine internal morphology and porosity. Tribological performance was measured on a universal micro-tribometer under dry and aqueous conditions with a normal load of 1 N and a sliding speed of 2 mm s^−1^. Surface wettability was determined by static contact angle (Dataphysics OCA25, Germany), and functional groups were identified by Fourier-transform infrared spectroscopy (Nicolet iS50, USA) over 4000–500 cm^−1^. Unless otherwise specified, all measurements were performed at room temperature under identical conditions. Adhesion performance including lap shear, peel strength and interfacial toughness was measured on a universal testing machine (Kesichuang. Co., Ltd.) with speed 50 mm/min.

In vitro evaluation of reactive oxygen species scavenging capacity

2.3

The ROS scavenging capacity of the hydrogels was evaluated in vitro using a series of colorimetric assays. For •OH scavenging, a Fenton reaction–methylene blue (MB; Aladdin, M196499) system was employed. Total antioxidant capacity was assessed using an ABTS•^+^ decolorization assay (Merck millipore, ES004). Superoxide anion scavenging was determined using pyrogallol (Selleck, S3885). Finally, H_2_O_2_ scavenging capacity was measured using a commercial hydrogen peroxide assay kit (Rhawn, R139829) according to the manufacturer's instructions.

In vitro adhesion and retention on rat knee joints

2.4

Sprague–Dawley (SD) rats aged 8–10 weeks were euthanized, and the knee joints were carefully dissected with the joint capsule intact. After removal of the surrounding muscles and soft tissues, the articular cartilage surfaces were rinsed with sterile saline and gently blotted dry. Precursor solutions of Hy, AdHy and AdHy@Pae containing blue dye solution for visualization were pipetted onto the femoral condyles (approximately 100 μL per joint) and exposed to UV to achieve in situ gelation. The joints were photographed immediately after gel formation. They were then immersed in PBS at 37 °C for 5 min, removed and imaged again. Finally, each knee joint was manually flexed and extended to about 90° for 500 cycles at a frequency of 1 Hz while kept moist with PBS, after which the integrity and retention of the hydrogels on the cartilage surface were recorded.

Pae release test

2.5

In vitro release of Pae from AdHy@Pae was assessed using a dynamic dialysis approach. Briefly, 0.2 g of Pae-loaded hydrogel was placed in a 5 mL centrifuge tube with 2 mL phosphate-buffered saline (PBS, pH 7.4,Solarbio, P1020) and incubated at 37 °C with orbital shaking at 60 rpm. At predetermined intervals (1, 4, 8, 12, and 24 h; and 2, 4, 7, 10, and 14 days), samples were centrifuged and the supernatants collected for analysis; an equal volume of fresh PBS was replenished to maintain a constant release volume. The concentration of released Pae was quantified at 230 nm on a UV–Vis spectrophotometer (Shimadzu, Japan), and cumulative release was calculated from a Pae calibration curve.

Cell culture

2.6

Primary rat chondrocytes (RCs, P0) were isolated from femoral head cartilage of 2-week-old Sprague–Dawley rats (Shanghai Slac Laboratory Animal Co., Ltd., China). Cartilage was aseptically dissected, minced to ∼1 mm^3^, digested with 0.25% trypsin (Gibco, 25200056) at 37 °C for 30 min, and then with 0.025% collagenase II (Solarbio, C8150) overnight. Released cells were collected, centrifuged, and cultured in high-glucose DMEM (Gibco, C11995500BT) supplemented with 10% fetal bovine serum (FBS, VivaCell, C04001) and 1% penicillin–streptomycin (Biosharp, BL505A). Cultures were maintained at 37 °C in a humidified 5% CO_2_ atmosphere, and passages 1–3 were used for subsequent experiments to ensure phenotypic stability.

Proliferation and apoptosis evaluation of chondrocytes

2.7

Rat chondrocytes (P1) were seeded in 24-well plates (5 × 10^4^ cells per well) and cultured overnight at 37 °C in 5% CO_2_. Inflammation was induced with LPS (10 μg mL^−1^, Biosharp, BS904) for 12 h, followed by 24 h treatment with hydrogel extracts (Hy, AdHy, or AdHy@Pae). Proliferation was assessed using an EdU fluorescence kit (Beyotime, C0078S) according to the manufacturer's instructions: cells were incubated with 10 μM EdU for 2 h, fixed with 4% paraformaldehyde (Beyotime, P0099), and sequentially stained with Apollo® 567 (Ribobio, C10310) and Hoechst 33342 (Beyotime, C1022); fluorescence images were acquired on a microscope (Olympus BX53, Japan), and EdU-positive percentages were calculated from five randomly selected fields. Apoptosis was evaluated by TUNEL staining (Beyotime, C1088) after identical pretreatment, with fixation and permeabilization followed by incubation with the TUNEL reaction mixture at 37 °C for 1 h in the dark and DAPI (Beyotime, P0131) counterstaining; images were collected on the same microscope and quantified in ImageJ. For quantitative apoptosis analysis by flow cytometry, chondrocytes were seeded in 6-well plates (1 × 10^6^ cells per well), subjected to the same LPS induction and hydrogel treatment, harvested, washed twice with cold PBS, resuspended in 1 × binding buffer, and stained with Annexin V-APC/7-AAD (Beyotime, C1063L) for 15 min at room temperature in the dark; apoptotic fractions were determined on a BD LSRFortessa X-20 (USA).

Cell cycle analysis

2.8

Rat chondrocytes (P1) were seeded in 6-well plates (1 × 10^6^ cells per well) and cultured at 37 °C in 5% CO_2_. After LPS stimulation (10 μg mL^−1^, 12 h, ServiceBio, GC205009) to induce an inflammatory phenotype, cells were treated with hydrogel extracts (Hy, AdHy, AdHy@Pae) for 24 h. Cells were then washed twice with PBS, harvested, and fixed overnight at 4 °C in 70% cold ethanol; after washing, they were stained with 500 μL cell-cycle staining solution (MULTI SCIENCES, CCS012) for 30 min in the dark. DNA content was acquired on a BD LSRFortessa X-20 flow cytometer (excitation 488 nm), and cell-cycle phases were quantified in FlowJo.

Mitochondrial membrane potential assay

2.9

Rat chondrocytes (P1) were seeded in 24-well plates (5 × 10^4^ cells per well) and cultured at 37 °C with 5% CO_2_. After LPS stimulation (10 μg mL^−1^, 12 h, ServiceBio, GC205009), cells were treated with hydrogel extracts (Hy, AdHy, AdHy@Pae) for 24 h, washed twice with PBS, and incubated with JC-1 working solution (Beyotime, C2006) at 37 °C for 30 min in the dark. Following three PBS washes, red (aggregates) and green (monomers) fluorescence were acquired on a confocal microscope (Nikon Eclipse C2), and the red/green intensity ratio was used to assess mitochondrial membrane potential (ΔΨm).

Intracellular ROS measurement

2.10

Rat chondrocytes (P1) were seeded in 6-well plates (1 × 10^6^ cells per well) and cultured at 37 °C with 5% CO_2_. After induction of an inflammatory phenotype by LPS (10 μg mL^−1^, 12 h, ServiceBio, GC205009), cells were treated with hydrogel extracts (Hy, AdHy, AdHy@Pae) for 24 h, washed twice with PBS, and incubated with ROS detection working solution (Servicebio, G1706) at 37 °C for 30 min. Following three gentle PBS washes, intracellular ROS was acquired on a BD LSRFortessa X-20 flow cytometer (excitation 488 nm, emission 525 nm) and quantified in FlowJo to assess oxidative stress across groups.

Immunofluorescence staining

2.11

Rat chondrocytes (P1) were seeded in 24-well plates (5 × 10^4^ cells per well) and cultured to ∼70–80% confluence, stimulated with LPS (10 μg mL^−1^, 12 h) to induce an inflammatory phenotype, and then treated with hydrogel extracts (Hy, AdHy, AdHy@Pae) for 24 h. Cells were washed three times with PBS, fixed in 4% paraformaldehyde for 15 min, permeabilized with 0.1% Triton X-100 (Aladdin, T434386) for 10 min, and blocked with 5% BSA (Solarbio, SW3015) at room temperature for 1 h. Primary antibodies were applied overnight at 4 °C: COL2A1 (Proteintech, 28459-1-AP, 1:200), SOX9 (HUABIO, HA723548, 1:1000), MMP9 (HUABIO, ET1704-69, 1:200), and ADAMTS5 (HUABIO, HA722011, 1:100). After three PBS washes, fluorescent secondary antibodies (Servicebio, Alexa Fluor 488, GB25303; Alexa Fluor 594, GB28303) were incubated for 1 h in the dark, nuclei were counterstained with DAPI (1 μg mL^−1^, 5 min, Servicebio, G1012), and images were acquired on a Nikon Eclipse C2 confocal microscope (Japan).

Western blot analysis

2.12

Following the above treatments (LPS 10 μg mL^−1^ for 12 h, then 24 h with Hy, AdHy, or AdHy@Pae extracts), rat chondrocytes (P2) were washed with ice-cold PBS and lysed on ice in RIPA buffer supplemented with protease inhibitors. Lysates were clarified by centrifugation at 12,000×g for 10 min at 4 °C, and protein was quantified by BCA (Beyotime, P0012S). Equal protein (20–30 μg per lane) was mixed with 5 × SDS sample buffer, boiled at 95 °C for 5 min, resolved by SDS–PAGE (5% stacking/10% resolving gel), and transferred to PVDF membranes (0.22 μm, Servicebio, G6015-1-0.22). Membranes were blocked with 5% BSA (Beyotime, ST2254) in TBST (Servicebio, G0004) for 1 h at room temperature and incubated overnight at 4 °C with primary antibodies diluted in TBST containing 1% BSA. After three TBST washes (3 × 10 min), membranes were incubated with HRP-conjugated secondary antibody (Affinity, S0001, 1:3000 in TBST–1% BSA) for 1 h at room temperature, washed again (3 × 10 min TBST), and developed with ECL (HUABIO, K1802). Chemiluminescence was captured on a digital imager, bands were quantified in ImageJ, and targets were normalized to β-actin. All blots were performed with at least three independent biological replicates.

qPCR analysis

2.13

Total RNA was isolated from treated rat chondrocytes (P2) with TRIzol (Invitrogen, 15596026CN) following the manufacturer's instructions, and purity/concentration were verified spectrophotometrically (A260/280 = 1.8–2.0). One microgram of RNA was reverse-transcribed to cDNA using a commercial RT kit (Gooniebio, 500). qPCR was performed on a StepOne Plus Real-Time PCR System (Thermo Fisher Scientific) with SYBR Green Master Mix in 20 μL reactions (10 μL SYBR mix, 0.4 μL forward primer, 0.4 μL reverse primer, 2 μL cDNA, nuclease-free water to volume, TransScript, AQ211-01). Cycling conditions were 95 °C for 30 s, followed by 40 cycles of 95 °C for 5 s and 60 °C for 30 s; a 65–95 °C melt curve confirmed specificity. GAPDH was used as the internal control, and relative expression was calculated by the 2^–ΔΔCt method. Each condition included ≥3 biological replicates with triplicate technical wells; primer sequences are provided in Table S1.

Animal experiments

2.14

All animal experiments were approved by the Animal Ethics Committee of Yunnan University of Chinese Medicine (YNUTCM-XMSS-G-20250064). 8-week-old male Sprague–Dawley rats were randomly assigned to experimental groups. OA was induced by surgically transecting the medial meniscotibial ligament (DMM surgery) of the right knee to establish a mechanically destabilized OA model. Sham-operated rats underwent joint exposure without ligament transection. Four weeks post-surgery, intra-articular therapeutic interventions were initiated. Each rat received weekly intra-articular injections (10 μL per joint) for four consecutive weeks via subpatellar insertion. The groups included Sham, PBS, Hy, AdHy, and AdHy@Pae. Body weight and general health conditions were monitored throughout the experiment. After the final injection, the rats were euthanized, and knee joints were collected for microstructural and histological analyses.

Histological and microstructural analysis

2.15

Micro-computed tomography (Micro-CT; NEMO NMC-200, PINGSENG Healthcare, China) was used to assess subchondral bone microarchitecture of rat knees, including bone volume fraction (BV/TV), bone mineral density (BMD), and trabecular number (Tb.N). Specimens were then fixed in 4% paraformaldehyde (Servicebio, G1101), decalcified, paraffin-embedded, and sectioned at 5 μm for histology and immunohistochemistry. Hematoxylin and eosin (H&E) (Servicebio, G1076), Safranin O–Fast Green (SO-FG) (Servicebio, G1053), and Toluidine Blue (TB) (Servicebio, G1032) staining evaluated cartilage morphology and matrix preservation, with histological grading performed using OARSI and Mankin scores. For immunohistochemistry, primary antibodies were COL2A1 (HUABIO, HA722733, 1:1000), SOX9 (Proteintech, 68350-1-Ig, 1:500), MMP3 (Servicebio, GB11132, 1:500), and ADAMTS1 (Affinity, DF13268, 1:100); sections were incubated with HRP-conjugated polymer secondary antibodies (Affinity, S0001), developed with DAB (Servicebio, G1212), and counterstained with hematoxylin. Whole-slide images were acquired on a digital scanner (Pannoramic MIDI, 3DHISTECH, Hungary), and quantification was performed in ImageJ.

TEM analysis

2.16

To evaluate mitochondrial ultrastructure in vivo, cartilage from rat knee joints (Sham, PBS, Hy, AdHy, AdHy@Pae) was collected at the study endpoint, immediately fixed in 1% glutaraldehyde (Servicebio, G1102) for 2 h at 4 °C, rinsed in PBS, and post-fixed in 1% osmium tetroxide for 1 h. Tissues were dehydrated through graded ethanol (30%, 50%, 70%, 90%, 100%), infiltrated and embedded in epoxy resin, sectioned at ∼70 nm on an ultramicrotome, double-stained with uranyl acetate and lead citrate, and examined by transmission electron microscopy (HITACHI HT7800, Japan) to assess mitochondrial morphology, cristae organization, and membrane integrity.

Transcriptomic and bioinformatic analysis

2.17

Total RNA was extracted from rat cartilage tissues of OA and AHP-treated groups for transcriptome sequencing. Raw sequencing data were processed using the Bioconductor package in R for background correction, normalization, and expression quantification. Differentially expressed genes (DEGs) between groups were identified via the limma package with a threshold of P < 0.05 and |log_2_FC| ≥ 0.38, and visualized by heatmap and volcano plots. Gene set enrichment analysis (GSEA) was performed using clusterProfiler to identify significantly enriched biological pathways (Hallmark gene sets).

To explore gene co-expression relationships, weighted gene co-expression network analysis (WGCNA) was conducted. Genes with expression variance above the 75th percentile were used to construct co-expression networks, determine the optimal soft-thresholding power (β), and define gene modules via dynamic tree cutting. Module eigengenes (MEs) were correlated with OA and AHP phenotypes to identify hub modules and key regulatory genes.

Mitochondria-associated genes were obtained from the MitoCarta 3.0 and GSEA databases. Overlapping genes among OA DEGs, AHP DEGs, WGCNA hub genes, and mitochondrial genes were identified using R/Perl scripts and visualized via Venny 2.1. Functional annotation of overlapping targets was performed using clusterProfiler, with Gene Ontology (GO) and KEGG enrichment analyses focusing on oxidative phosphorylation, mitochondrial metabolism, and inflammatory signaling pathways. Enrichment significance was evaluated by Fisher's exact and hypergeometric tests, and results were visualized using the pathview package.

Finally, key pathways identified from transcriptomic analysis were experimentally validated in vivo using intra-articular injection of IL-17A and TNF-α pathway antagonists, followed by Western blot analysis of cartilage tissue proteins to confirm pathway regulation.

All analyses were conducted in R (v4.3.2), and statistical significance was set at P < 0.05.

Statistical analysis

2.18

All data analysis was performed with GraphPad Prism software. Data are presented as mean ± standard deviation (SD). For comparisons involving more than two groups, one- or two-way ANOVA with appropriate multiple-comparisons correction was used. Two-group differences were tested with Student's t-test, or with Welch's t-test when variances were unequal. ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001 were considered to be significant.

Results and discussion

3

Synthesis and characterization of paeoniflorin-loaded adhesive and lubricating hydrogel (AdHy@Pae)

3.1

OA is sustained by persistent synovial inflammation and oxidative stress. Herein, we designed a hydrogel aimed to couple mechanical protection with mitochondria-targeted anti-inflammatory/antioxidant modulation for OA therapy. As shown in Fig. 1A, In the presence of the photoinitiator, the precursor solution composed of HEMA, SMBA, γ-PGA-Na, and Pae rapidly converted from a free-flowing liquid into a self-supporting 3D network upon UV irradiation, enabling rapid network formation during in vitro material fabrication. XRD spectra is displayed in Fig. 1B. The continuous leftward shift of the diffraction peaks in the AdHy and AdHy@Pae groups indicates an increase in intermolecular spacing [39], suggesting that the introduction of SMBA and γ-PGA-Na reduces the crosslinking density and thereby enhances the energy-dissipation capacity of the hydrogel network [40]. For FTIR spectra (Fig. 1C), a characteristic band at ∼1720 cm^−1^ was assigned to the C=O stretching of HEMA ester groups, the band at 1550–1600 cm^−1^ to the asymmetric stretching of COO^−^ from γ-PGA-Na, the peak at 1640–1660 cm^−1^ to the amide I band of SMBA and γ-PGA-Na, and the signals at 1050–1100 cm^−1^ to the C–O–C glycosidic bonds of Pae. In addition, the broadened O–H band in the 3200–3600 cm^−1^ region in AdHy@Pae indicated the presence of multiple hydroxyl groups in Pae and their hydrogen-bonding interactions within the polymer matrix. SEM observations revealed that all three hydrogels exhibited interconnected porous structures, whereas AdHy@Pae displayed a loosest architecture with largest pores (Fig. 1D). Quantitative analysis confirmed that its average pore size exceeded that of Hy and AdHy (Fig. 1E), a morphology that is favorable for energy dissipation as well as nutrient and drug transport. Compared to Hy and AdHy, the introduction of Pae decreased the contact angle, which indicated the contribution of abundant hydroxyl groups to decrease interfacial energy (Fig. 1F and Fig. S2).Fig. 1. Gelation and physicochemical characterization of hydrogels. (A) Photographs of the AdHy@Pae precursor solution before and after UV irradiation. (B) XRD patterns and (C) FTIR spectra of Hy, AdHy and AdHy@Pae hydrogels. (D) SEM images of freeze-dried Hy, AdHy and AdHy@Pae hydrogels at different magnifications and (E) corresponding average pore size. (F) Representative water contact angle images and (G) quantitative results of Hy, AdHy and AdHy@Pae hydrogels. (H) Quantitative results of fraction coefficient. μs represents static fraction coefficient, μd represents dynamic fraction coefficient. Catri represents cartilage. (I) Quantitative results of fraction force.Fig. 1

Decreased interfacial energy under hydrated environments may promote lubrication. To verify this hypothesis, we carried out adhesion and lubrication experiments. Data in Fig. 1H indicated that introduction of SMBA and γ-PGA-Na increased the static friction coefficient and dynamic friction coefficient. This increase can be further enhanced via addition of Paeoniflorin (Pae), which could be contributed to the abundant hydroxyl that generate hydrogen bonding interaction with interface. As we all know, the joint cavity is a special cavity tissue rich in lubricating fluid and lubricating fluid redistribute at the tissue interface with joint movement. Thus, we also simulated the frictional force in the presence of lubricating fluid. In the presence of PBS, the adhesion performance of AdHy@Pae rapidly decrease, resulting in a rapid decrease in friction force and reaching its lowest point (0.05 N) compared to that without PBS (∼0.25 N) (Fig. 1I). This huge contrast can be attributed to the rapid hydrated ability of abundant hydroxyl groups from Pae. Based on this dynamic hydration, this hydrogel can generate robust adhesion and lubrication simultaneously.

The performance of hydrogel under joint motion also depends on the viscoelastic response and shear stability of the hydrogel network. Therefore, rheological measurements was conducted. As displayed in Fig. 2a, the base Hy hydrogel had a relatively low storage modulus (G′) of only a few hundred pascals. Incorporation of the zwitterionic monomer modestly increased the viscoelasticity, while further addition of Pae led to a pronounced increase in G′ across the entire frequency range, approaching the modulus range reported for native cartilage [41]. These results indicate that Pae plays an important role in stabilizing the network structure and enhancing its ability to dissipate mechanical energy. To further ensure durable intra-articular retention, we next quantified the adhesive performance of AdHy@Pae. The results of lap shear, peel strength and interfacial toughness indicated that Pae can enhance the adhesion strength compared to Hy and AdHy (Fig. 2B–D). At the macroscopic level, AdHy@Pae firmly adhered to diverse substrates, including glass, polypropylene, wood, rubber and foam, and remained tightly attached after bending, twisting, rinsing and immersion in water for 1 h (Fig. 2E). Under tensile strains of 50% and 70%, the hydrogel maintained an intact continuous shape and largely recovered its original length upon unloading, highlighting its outstanding elasticity and resilience. These features suggest that AdHy@Pae can accommodate the complex motion and deformation of the articular cavity. In vitro experiments using SD rat knee joints further confirmed its tissue-level performance (Fig. 2F), Hy, AdHy and AdHy@Pae all formed continuous coatings on the articular cartilage surface, whereas AdHy@Pae maintained superior coverage and integrity after soaking in PBS followed by 500 cycles of flexion, indicating more durable adhesion and lubricating protection under conditions that mimic the intra-articular environment.Fig. 2. Adhesive, deformable and antioxidant performance of hydrogels. (A) Storage modulus and loss modulus of Hy, AdHy and AdHy@Pae. (B) Interfacial toughness, (C) Peel strength, and (D) lap shear strength of the three hydrogel. (E) Photographs showing adhesion of hydrogels to different substrates, deformability under various operations and stretchability. (F) In vitro adhesion of Hy, AdHy and AdHy@Pae on joints before and after soaking in PBS and after 500 flexion cycles. UV–vis spectra of different groups for (G) •OH scavenging, (H) ABTS•^+^ radical scavenging, and (I) superoxide anion (O_2_•^-^) scavenging. (J) H_2_O_2_ scavenging rate.Fig. 2

Given that excessive accumulation of ROS during OA progression aggravates chondrocyte damage and matrix degradation, we next evaluated the radical-scavenging ability of the hydrogels. In the MB–Fenton system used to assess •OH scavenging, AdHy@Pae showed the highest residual absorbance at 650–670 nm (Fig. 2G), indicating the strongest protection of the probe and thus the most effective •OH elimination. When ABTS•^+^ was employed as a probe (Fig. 2H), AdHy@Pae again produced the greatest decrease in absorbance, suggesting superior scavenging of this radical species. Using a specific probe for superoxide anion (O_2_^•-^), the AdHy@Pae group displayed a marked reduction in absorbance in the 250–400 nm range (Fig. 2I), further demonstrating broad-spectrum antioxidant activity. Consistently, H_2_O_2_ scavenging assays showed that AdHy@Pae achieved the highest removal rate among all groups (Fig. 2J), highlighting its potent capacity to eliminate peroxides. Collectively, these data indicate that the introduction of Pae to the network not only strengthens the mechanical and adhesive properties of the hydrogel but also endows it with multi-target ROS-scavenging functions, which are expected to alleviate oxidative stress and protect chondrocytes and their extracellular matrix within the inflamed niche.

Moreover, the release profile of Pae from AdHy@Pae was slow and sustained, with approximately 85% cumulative release achieved over 10 days (Fig. S2). This controlled release guarantees prolonged pharmacological stimulation while preserving the mechanical integrity of the hydrogel network. Taken together, AdHy@Pae integrates UV-triggered in situ gelation, cartilage-matched mechanics, strong adhesion to multiple substrates and tissues, favorable wettability and lubrication, broad-spectrum antioxidant activity and sustained drug delivery, thereby providing a comprehensive biomaterial platform for inflammation-associated cartilage repair.

In vitro biocompatibility and cytoprotective performance of AdHy@Pae

3.2

Maintaining chondrocyte viability within an inflammatory microenvironment is a key prerequisite for effective cartilage regeneration. During the progression of OA, oxidative stress and pro-inflammatory cytokines disrupt mitochondrial function, induce apoptosis, and impair extracellular matrix synthesis, thereby accelerating cartilage degeneration [42]. Developing a biocompatible material capable of protecting chondrocytes from inflammatory stress while sustaining proliferation is therefore essential for joint repair. To evaluate the cytocompatibility and protective properties of AdHy@Pae, an in vitro LPS-induced oxidative injury model was established.

The CCK-8 assay revealed that AdHy@Pae exhibited excellent biocompatibility and negligible cytotoxicity during 1, 3, and 7 days of incubation (Fig. S3). Live/dead staining further confirmed these results, showing that cells in the AdHy@Pae group maintained intact membranes and metabolic activity, while those exposed to LPS displayed extensive red fluorescence, reflecting severe cell death (Fig. S4). Quantitative analysis showed that the proportion of dead cells in the AdHy@Pae group was less than half that of the LPS group, demonstrating a remarkable cytoprotective capacity under inflammatory conditions (Fig. S5).

Cell proliferation and apoptosis were then assessed using EdU incorporation (Fig. 3A and B) and TUNEL staining (Fig. 3C and D), respectively. LPS exposure led to a sharp reduction in EdU-positive nuclei, indicating inhibition of DNA synthesis. Treatment with AdHy@Pae effectively restored proliferative activity, with a significantly higher proportion of proliferating cells than in both the LPS and AdHy groups. TUNEL analysis revealed extensive nuclear fragmentation after LPS induction, while AdHy@Pae treatment markedly reduced apoptotic nuclei. Quantitative evaluation confirmed that the number of apoptotic cells decreased by more than 70% compared with the LPS group, suggesting that AdHy@Pae efficiently prevents inflammation-induced apoptosis and supports cellular regeneration. Flow cytometry further validated these findings (Fig. 3E and Fig. S6). Upon LPS stimulation, the proportion of viable cells dropped to approximately 44%, while late apoptotic and necrotic cells exceeded 50%. Treatment with AdHy partially improved viability, whereas AdHy@Pae preserved over 85% viable cells and reduced late apoptosis to around 10% (Fig. 3F). These results were consistent with TUNEL analysis, confirming that AdHy@Pae mitigates inflammation-induced injury by maintaining mitochondrial stability and suppressing apoptotic signaling cascades. Collectively, these findings demonstrate that AdHy@Pae possesses excellent cytocompatibility and strong cytoprotective performance, effectively attenuating oxidative damage, reducing apoptosis, and promoting chondrocyte proliferation.Fig. 3In vitro biocompatibility and cytoprotective evaluation of AdHy@Pae. (A) EdU staining images of chondrocytes in different groups. (B) Quantification of EdU-positive cells. (C) TUNEL staining images showing apoptotic cells. (D) Quantitative analysis of TUNEL-positive cells. (E) Flow cytometry analysis of apoptosis. (F) Statistical results of apoptotic cell proportions. Data are shown as mean ± SD (n = 3).Fig. 3

In vitro antioxidant and mitochondrial homeostasis regulatory effects of AdHy@Pae

3.3

Inflammation-induced oxidative stress disrupts intracellular redox balance, impairs mitochondrial function, and interferes with the progression of the chondrocyte cell cycle [43]. To evaluate whether AdHy@Pae could restore cellular homeostasis under such stress conditions, cell cycle analysis was conducted. Flow cytometric results revealed that LPS stimulation markedly arrested cells in the G0/G1 phase, accompanied by a substantial reduction in the S and G2/M phases (Fig. 4A–D and Fig. S7), indicating severe inhibition of DNA synthesis and mitotic activity under inflammatory stress. Treatment with AdHy slightly alleviated this arrest, whereas AdHy@Pae treatment nearly normalized cell cycle distribution, significantly increasing the S-phase population and restoring mitotic activity. These findings suggest that AdHy@Pae effectively counteracts inflammation-induced cell cycle arrest and preserves proliferative potential. Meanwhile, given that oxidative stress is often accompanied by mitochondrial dysfunction, JC-1 staining was used to assess changes in ΔΨm (Fig. 4E and F). The cells in the control group displayed few green monomer fluorescence, reflecting well-polarized mitochondria, whereas LPS treatment caused a sharp increase in green monomer emission, indicative of mitochondrial depolarization. In contrast, AdHy@Pae treatment significantly decreased green fluorescence compared to LPS group, suggesting recovery of ΔΨm and preservation of mitochondrial integrity and energy homeostasis. This finding parallels the restoration of cell cycle activity, indicating that AdHy@Pae protects mitochondrial bioenergetics from inflammatory damage.Fig. 4In vitro antioxidant and mitochondrial homeostasis regulatory effects of AdHy@Pae. (A) Flow cytometric analysis of the cell cycle. (B–D) Quantitative statistics of the percentage of cells in G0/G1, S, and G2/M phases. (E) JC-1 fluorescence staining. (F) Quantitative analysis of JC-1 red/green fluorescence ratio (ΔΨm). (G) Flow cytometric detection of intracellular ROS using DCFH-DA probe. Data are shown as mean ± SD (n = 3, ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001).Fig. 4

Because mitochondrial impairment often leads to excessive ROS accumulation, intracellular ROS levels were quantified using the DCFH-DA fluorescent probe (Fig. 4G and Fig. S8). LPS exposure markedly elevated ROS fluorescence intensity, indicating pronounced oxidative stress, whereas AdHy@Pae treatment substantially reduced ROS signals to near-baseline levels. This demonstrates the strong free-radical scavenging and antioxidative regulatory capability of AdHy@Pae. Collectively, our data indicate that AdHy@Pae preserves chondrocyte viability under inflammatory stress by coordinating cell-cycle recovery, stabilizing the mitochondrial membrane potential, and limiting ROS accumulation, thereby sustaining cellular homeostasis. These findings align with prior evidence that oxidative stress and mitochondrial dysfunction are central drivers of osteoarthritic degeneration and that restoring mitochondrial bioenergetics/ΔΨm mitigates chondrocyte injury [[44], [45], [46]].

Chondroprotective effects of AdHy@Pae

3.4

Inflammation-induced oxidative stress disrupts metabolic homeostasis within cartilage tissue, leading to ECM degradation, loss of the chondrocyte phenotype, and ultimately cartilage degeneration [47]. Therefore, we elucidate the chondroprotective effect of AdHy@Pae. Immunofluorescence staining (Fig. 5A) revealed intense COL2A1 and SOX9 fluorescence signals in the control group, indicating a typical chondrocytic phenotype and active ECM synthesis. In contrast, LPS stimulation markedly attenuated these signals, suggesting pronounced dedifferentiation and impaired cartilage anabolism. Meanwhile, the red fluorescence intensity of matrix-degrading enzymes MMP9 and ADAMTS5 was strongly enhanced, reflecting elevated ECM degradation under inflammatory stress [48]. After treatment with AdHy or AdHy@Pae, the cellular phenotype was gradually restored; notably, the AdHy@Pae group exhibited COL2A1 and SOX9 levels comparable to the control, accompanied by a substantial reduction in MMP9 and ADAMTS5. These findings indicate that AdHy@Pae effectively restores the dynamic balance between matrix synthesis and degradation, thereby maintaining the phenotypic stability of chondrocytes under inflammatory conditions [49]. Consistently, quantitative analysis (Fig. 5B) confirmed this trend, showing that AdHy@Pae markedly upregulated COL2A1 and SOX9 mRNA expression while suppressing the elevated transcription of MMP9 and ADAMTS5 observed in the LPS group.Fig. 5. Regulation of extracellular matrix metabolism by AdHy@Pae in LPS-induced chondrocytes. (A) Representative immunofluorescence images showing the expression of COL2A1, SOX9, MMP9, and ADAMTS5 in different treatment groups and corresponding (B) quantitative analysis. Data are shown as mean ± SD (n = 3, ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001).Fig. 5

We next profiled key cartilage-related genes and proteins to further substantiate the chondroprotective role of AdHy@Pae in maintaining cartilage homeostasis. Because OA-associated inflammatory stress typically elicits oxidative and metabolic disturbances that suppress ECM anabolism while promoting catabolic degradation and cell death, we quantified representative anabolic, catabolic, antioxidant, and apoptotic markers. Accordingly, qPCR (Fig. S9) and Western blot analyses (Fig. 6) were performed to comprehensively evaluate these molecular alterations.Fig. 6. Expression analysis of cartilage-related proteins after treatment with AdHy@Pae. (A) Representative Western blot images showing ECM synthesis–related proteins (COL2A1, ACAN, COMP, and SOX9) and antioxidant markers (NRF2 and HO-1) in different groups. (B) Quantitative analysis of matrix-degrading enzymes (MMP13, ADAMTS5, ADAMTS1) and apoptosis-related proteins (BCL-2, Bax, and Caspase-3). Data are presented as mean ± SD (n = 3; ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001).Fig. 6

Under LPS stimulation, the expression of cartilage-specific structural proteins COL2A1, ACAN, and COMP, together with the transcription factor SOX9, was markedly downregulated, indicating impaired ECM synthesis and phenotypic instability of chondrocytes [50]. In contrast, AdHy@Pae treatment effectively restored their expression, particularly COL2A1 and ACAN, to levels comparable with the control, suggesting recovery of anabolic activity under inflammatory stress. Meanwhile, LPS stimulation induced a significant upregulation of matrix-degrading enzymes MMP13, ADAMTS5, and ADAMTS1, consistent with previous findings that excessive expression of these proteases accelerates OA-related cartilage degradation [51]. Notably, AdHy@Pae administration significantly attenuated this overexpression, thereby maintaining the balance between ECM synthesis and degradation. In addition, both mRNA and protein analyses demonstrated that AdHy@Pae restored the expression of antioxidant markers NRF2 and HO-1, which were markedly reduced in the LPS group. This observation agrees with earlier results showing its ability to mitigate oxidative damage and preserve mitochondrial stability [52]. Furthermore, the expression pattern of apoptotic markers exhibited a consistent protective tendency, with AdHy@Pae treatment leading to an increase in BCL-2 and a concomitant decrease in Bax and Caspase-3, suggesting an attenuation of inflammation-induced apoptotic activity [52]. These results confirm that AdHy@Pae effectively restores the expression of key anabolic, catabolic, antioxidant, and apoptotic regulators at both gene and protein levels, thereby maintaining cartilage metabolic homeostasis under inflammatory conditions. This comprehensive expression validation provides strong experimental evidence supporting the chondroprotective efficacy of AdHy@Pae.

In vivo evaluation of AdHy@Pae for cartilage protection and OA regeneration

3.5

In the cellular-level studies, AdHy@Pae was confirmed to restore cartilage homeostasis through multi-pathway coordination, significantly alleviating inflammation-induced metabolic imbalance and apoptosis. To further verify whether these effects were preserved in vivo, we established a destabilization of the DMM model in mice to mimic OA-related cartilage degeneration and evaluate the chondroprotective and osteoregenerative potential of AdHy@Pae (Fig. 7A). Notably, the hydrogel was cryogenically milled into a powder using liquid nitrogen and subsequently partial rehydrated with PBS prior to use in vivo. The rehydrated powder exhibited good injectability and rapidly re-formed a cohesive bulk hydrogel within 10 min (Fig. S10). Upon partial rehydration, it remains in a highly water-absorbing, swelling-driven state. When injected and brought into contact with wet cartilage, the material can rapidly wick interfacial fluid via capillary imbibition through its porous network and further generate osmotic swelling pressure, which together reduce the lubricating fluid film thickness at the cartilage interface. This dehydration-assisted interfacial consolidation enables conformal contact and mechanical interlocking with cartilage surface microtexture. Four weeks after surgery, mice received intra-articular injections of PBS, Hy, AdHy, or AdHy@Pae once weekly for two and four consecutive weeks, followed by histological and imaging assessments.Fig. 7In vivo evaluation of the therapeutic effects of AdHy@Pae on osteoarthritic cartilage repair after therapy for 4 weeks. (A) Schematic illustration of the DMM-induced OA model and intra-articular administration. (B) Representative Micro-CT images of knee joints from different groups. (C) Quantitative analysis of bone microarchitecture parameters, including Tb.N, BMD, and BV/TV. (D) Histological staining of articular cartilage with H&E, SO-FG, and TB to evaluate cartilage integrity and proteoglycan content. (E) Quantitative assessment of cartilage degeneration using OARSI and Mankin scores. Data are presented as mean ± SD (n = 3, ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001).Fig. 7

After 2 weeks of treatment, Micro-CT results found that both Hy and AdHy treatments moderately improved bone integrity, with less disruption of trabecular continuity (Figs. S11A-B). In contrast, AdHy@Pae treatment markedly preserved joint architecture and maintained dense, well-organized trabeculae, indicating a clear advantage in suppressing early bone resorption and promoting bone remodeling at this early 2-week postoperative stage. HE staining further confirmed the protective effects of AdHy@Pae on cartilage morphology and extracellular matrix integrity (Fig. S11C). At 4 weeks of treatment, severe subchondral bone destruction and trabecular rarefaction were observed in the PBS group (Fig. 7B), displaying typical osteoarthritic bone resorption. Both Hy and AdHy treatments moderately improved bone integrity but failed to restore normal trabecular continuity. In contrast, AdHy@Pae treatment markedly preserved the joint structure and maintained compact and well-aligned trabeculae, suggesting a pronounced advantage in inhibiting bone resorption and promoting bone remodeling. Quantitative analysis (Fig. 7C) revealed that AdHy@Pae significantly increased trabecular number (Tb.N), bone mineral density (BMD), and bone volume fraction (BV/TV) to near-sham levels, confirming its potent capability to reconstruct the osteochondral unit in vivo. Histological staining further validated the protective effects of AdHy@Pae on cartilage morphology and ECM integrity (Fig. 7D). H&E staining showed that the PBS group exhibited rough cartilage surfaces, disorganized chondrocyte alignment, and deep fissures. Although Hy and AdHy groups showed partial recovery, the zonal structure remained incomplete. In sharp contrast, the AdHy@Pae group displayed a smooth articular surface, well-preserved tidemark, and columnar chondrocyte arrangement, resembling the sham control. Safranin O–Fast Green (SO–FG) and Toluidine Blue (TB) staining results were consistent, showing that AdHy@Pae markedly restored proteoglycan content and maintained ECM continuity. Quantitative scoring (Fig. 7E) revealed lowest OARSI and Mankin scores in the AdHy@Pae group (>60% reduction), further confirming its remarkable in vivo chondroprotective and reparative efficacy.

To further verify the in vivo cartilage-protective effects of AdHy@Pae at the tissue level, IHC staining was performed. In the PBS group, cartilage anabolic markers COL2A1 and SOX9 were markedly reduced (Fig. 8A), while catabolic enzymes MMP3 and ADAMTS1 were highly upregulated, indicating suppressed ECM synthesis and enhanced degradation. Treatment with Hy or AdHy partially restored these imbalances, whereas the AdHy@Pae group exhibited intense COL2A1 and SOX9 staining along with attenuated MMP3 and ADAMTS1 signals, approaching the levels of the sham group. The quantitative IHC analysis (Fig. 8B) confirmed these findings, demonstrating that AdHy@Pae effectively maintained ECM homeostasis and suppressed matrix catabolism in vivo.Fig. 8In vivo validation of the chondroprotective and mitochondrial regulatory effects of AdHy@Pae after therapy for 4 weeks. (A) Representative immunohistochemical staining of cartilage sections for COL2A1, SOX9, MMP3, and ADAMTS1 in different groups (Sham, PBS, Hy, AdHy, and AdHy@Pae). (B) Quantitative analysis of the immunohistochemical staining intensity showing relative expression levels of anabolic (COL2A1, SOX9) and catabolic (MMP3, ADAMTS1) markers. (C) TEM images of chondrocytes showing mitochondrial ultrastructure in each group, with high-magnification views (bottom panels) indicating cristae integrity and membrane morphology. Data are presented as mean ± SD (n = 3); ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗P < 0.0001.Fig. 8

Given the central role of mitochondrial integrity in inflammation-associated cartilage degeneration, we further examined the ultrastructural morphology of chondrocytes using TEM (Fig. 8C). In the PBS group, mitochondria exhibited pronounced swelling, disrupted cristae, and extensive vacuolization, indicative of depolarization and metabolic dysfunction. Hy and AdHy groups showed partial recovery, yet mitochondrial swelling and cristae disorganization persisted. Remarkably, chondrocytes in the AdHy@Pae group displayed intact mitochondria with dense and well-defined cristae and continuous membranes, closely resembling the sham group. These observations are consistent with the JC-1 and ROS analyses in vitro (Fig. 4), providing further evidence that AdHy@Pae preserves mitochondrial homeostasis and mitigates energy stress at the organelle level, thereby preventing inflammation-induced chondrocyte degeneration.

Transcriptomic profiling confirms mitochondrial-targeted anti-inflammatory regulation by AdHy@Pae in OA cartilage

3.6

Building upon the histological and biochemical evidence of cartilage regeneration and mitochondrial protection in vivo, we next sought to delineate the transcriptional landscape underlying the therapeutic effects of AdHy@Pae. A comprehensive transcriptomic strategy integrating differential gene expression profiling, GSEA, and WGCNA was employed to uncover the molecular pathways and hub genes mediating the observed phenotypic recovery.

Differential expression analysis between osteoarthritic and healthy cartilage identified 44 upregulated and 6 downregulated genes under the threshold of |log_2_FC| ≥ 0.38 and P < 0.05. These genes exhibited distinct transcriptional alterations, as visualized by the heatmap and volcano plots (Fig. S12A–B). Gene set enrichment analysis revealed that OA-related DEGs were prominently enriched in inflammation and immune response, including IL-17, TNF, and NF-κB signaling pathways, as well as rheumatoid arthritis and chemokine signaling pathway (Fig. S12C), reflecting the hyperactivation of inflammatory and catabolic cascades characteristic of cartilage degeneration. To further identify the functional gene clusters related to OA pathogenesis, we constructed a weighted gene co-expression network using the WGCNA algorithm. The optimal soft thresholding power (β = 5, correlation >0.9) was determined to ensure a scale-free topology (Fig. S13A). Hierarchical clustering confirmed high sample consistency, and subsequent dynamic tree cutting partitioned the transcriptome into 35 distinct co-expression modules (Fig. S13B). Among these, the MEmagenta module showed a strong positive correlation with OA progression, while the MEbrown4 module exhibited a significant negative correlation (|r| > 0.8, P < 0.05) (Figs. S14A and B). These two modules contained 1512 putative hub genes, representing coordinated expression patterns associated with cartilage inflammation and degradation. Their distribution across modules was visualized through a topological overlap matrix heatmap (Fig. S14C).

To verify the transcriptional remodeling induced by AdHy@Pae treatment, RNA sequencing was performed on cartilage from the AdHy@Pae-treated group. Compared with untreated OA tissues, 6 genes were significantly upregulated and 6 were downregulated, indicating a partial reversal of the OA transcriptomic profile (Fig. 9A and B). GSEA showed that AdHy@Pae modulated multiple signaling axes involved in tissue repair, with significant enrichment in ErbB signaling, DNA replication, and metabolic recovery pathways (Fig. 9C). The clustering analysis showed that the samples clustered tightly with no apparent outliers, indicating that the dataset is suitable for subsequent analyses (Fig. 9D). The subsequent WGCNA (β = 24) yielded 21 distinct gene modules (Fig. 9E). Among them, the MEwhite module displayed a strong negative correlation with OA (|r| > 0.9, P < 0.05) (Fig. 9F), comprising 197 hub genes potentially responsible for the anti-inflammatory and mitochondrial-protective properties of AdHy@Pae. A clustering heatmap of these modules clearly depicted the expression reorganization driven by treatment (Fig. S15).Fig. 9. Transcriptomic analysis of AdHy@Pae–treated cartilage tissue. (A) Heatmap displaying DEGs between AdHy@Pae–treated and OA groups. (B) Volcano plot showing upregulated and downregulated DEGs identified between AHP and OA groups (|log_2_FC| ≥ 0.38, p < 0.05). (C) GSEA of DEGs indicating significantly enriched biological pathways. (D) WGCNA dendrogram constructed based on topological overlap. (E) Module–trait correlation heatmap showing relationships between co-expression modules and sample traits. (F) Scatter plot showing the correlation between module membership and gene significance within the MEwhite module.Fig. 9

To establish the molecular intersection between OA pathology and AdHy@Pae therapy, we integrated four gene datasets, including OA DEGs, AHP DEGs, OA WGCNA hub genes, and mitochondrial gene sets. This comprehensive intersection yielded five core targets (Tgm2, IL6, Fgr, Mmp9, and IL1β) representing the convergence of inflammatory, metabolic, and mitochondrial regulation (Fig. 10A). These hub genes were further validated using expression matrices from the GEO database, where heatmap and volcano plots confirmed consistent expression patterns (Fig. 10B and C). Functional enrichment revealed their strong involvement in IL-17, TNF signaling pathways, and antifolate resistance (Fig. 10D), consistent with the inflammatory suppression and matrix reconstruction observed histologically. Therefore, the mechanistic focus of this study centered on the IL-17A/TNF-α inflammatory axis identified by transcriptomic analyses and its downstream effects, including oxidative stress, mitochondrial dysfunction, and chondrocyte fate regulation.Fig. 10. Identification and validation of the mechanism that AdHy@Pae alleviate OA. (A) Venn diagram illustrating the intersection among DEGs and WGCNA modules from OA and AdHy@Pae datasets. (B) Heatmap displaying the expression profiles of the five core genes in control, OA, and AHP groups. (C) Volcano plots of DEGs from OA vs. control and AHP vs. OA comparisons, with the five intersecting genes highlighted. (D) KEGG pathway enrichment analysis showing the distribution of the five core genes across inflammation- and metabolism-related signaling pathways. (E) Validation of the protein expression by Western blot analyses in different groups. (F) Schematic illustration of the involvement of IL-17A and TNF-α signaling pathways in mitochondrial function and extracellular matrix regulation under AdHy@Pae treatment.Fig. 10

To experimentally validate the transcriptomic predictions, we employed siRNA knockdown and overexpression strategies targeting IL-17A/TNF-α. qPCR and Western blot assays confirmed that AdHy@Pae significantly regulated key components of the IL-17A and TNF-α pathways, key components of mitochondrial fusion/fission (MFN1, MFN2, FIS1, Dnm1L), and the along with downstream effectors including COL2A1, and MMP9 (Fig. 10E and Figs. S16-17). Consistent with these molecular changes, mitochondrial functional assays, including mitochondrial membrane potential, mitochondrial ATP, and oxygen consumption rate, further substantiated the involvement of mitochondrial restoration in the therapeutic effects of AdHy@Pae (Fig. S18-S21). These outcomes collectively indicate that AdHy@Pae attenuates IL-17A and TNF-α–driven inflammatory cascades, regulated mitochondrial dynamics homeostasis (MFN1, MFN2, FIS1, DNM1L), thereby regulated expression of matrix-degrading enzymes (MMP9) and cartilage homeostatic matrix (COL2A1).

Concurrently, AdHy@Pae integrates sustained local Pae delivery with effective control of cellular stress, simultaneously alleviating inflammatory/oxidative insults and maintaining therapeutic availability within the joint (Fig. 10F). This favorable microenvironment supports mitochondrial quality control, as evidenced by the rebalanced fusion–fission machinery. Through the coordinated regulation of inflammation, mitochondrial function, and ECM metabolism, AdHy@Pae effectively mitigates cartilage degradation and promotes structural repair under chronic inflammatory stress.

Conclusion

4

In this work, we report an interface-engineered, paeoniflorin-loaded multifunctional hydrogel (AdHy@Pae). AdHy@Pae exhibits robust interfacial adhesion and excellent lubrication, enabling durable retention and reduced frictional stress within the synovial environment. More importantly, the hydrogel effectively alleviates oxidative and inflammatory damage, reconstructs cartilage integrity and restores the physiological balance of the joint microenvironment. Mechanistically, AdHy@Pae re-established cellular redox homeostasis and mitochondrial integrity via IL-17A and TNF-α signaling pathways, recovering mitochondrial membrane potential, and rescuing oxidative phosphorylation–associated bioenergetics. This mitochondrial stabilization, in turn, preserved chondrocyte viability and phenotypic commitment, while concurrently suppressing catabolic remodeling. Collectively, these findings highlight mitochondrial dynamics as an actionable therapeutic node that can be effectively targeted by an interface-engineered hydrogel platform, bridging materials chemistry with regenerative biology and offering a mechanistically grounded strategy for OA therapy.

Nevertheless, several limitations warrant mention. We did not directly interrogate load-dependent mechanotransduction beyond the IL-17/TNF-centered transcriptomic findings, and the in vivo exposure and contribution of Pae relative to the hydrogel matrix were not quantified using pharmacokinetic measurements or an appropriate free-drug control. A time-resolved correlation between Pae release kinetics and sustained biological effects was also not established. In addition, validation was performed mainly in the DMM post-traumatic OA model without evaluation in age-related or metabolic OA settings or large-animal models, and with limited long-term assessment of joint function, in situ degradation, and synovial immunoresponses, which will be important to address in future studies to further strengthen the translational robustness of this platform.

CRediT authorship contribution statement

Zugui Wu: Writing – original draft, Investigation, Funding acquisition, Data curation. Gen Wu: Writing – original draft, Funding acquisition. Yue Zhu: Investigation, Data curation. Gaoquan Zheng: Data curation. Tao Wang: Investigation. Jian Liu: Investigation. Zhen Shen: Investigation, Funding acquisition. Xiaoying Wang: Software. Xingqiang Bei: Investigation. Yanfei Xu: Software. Jiao Li: Funding acquisition, Data curation. Rong Yuan: Investigation. Zhiwei Wu: Software. Ying Guo: Project administration, Funding acquisition. Yu Zhang: Writing – review & editing, Supervision. Feng Peng: Writing – review & editing, Supervision, Conceptualization.

Ethics approval and consent to participate

All animal experiments were approved by the Animal Ethics Committee of Yunnan University of Chinese Medicine (YNUTCM-XMSS-G-20250056).

Declaration of competing interest

The authors declare no conflict of interest.