Curcumin-loaded silk fibroin scaffold promotes cartilage regeneration by inhibiting angiogenesis via Drp1/ROS-mediated mitochondrial regulation

Abstract

Articular cartilage injury often leads to vascular endothelial cell (VEC) infiltration, disrupting the microenvironment between cartilage and subchondral bone, thereby compromising cartilage repair quality. Curcumin (Cur) is a natural polyphenol with anti-inflammatory and anti-angiogenic properties that holds promise for therapeutic applications. However, its clinical utility is limited due to poor solubility and instability. To address these challenges, we developed a curcumin-loaded silk fibroin nanoparticle (Cur-SN) delivery system to inhibit VEC infiltration and promote cartilage regeneration. Cur-SNs were prepared and characterised to evaluate their physicochemical properties. The effects of Cur-SN on VEC apoptosis and senescence were assessed, and the underlying mechanism by which Cur-SN regulates mitochondrial homeostasis via the Drp1/ROS pathway was investigated. Additionally, a rat knee cartilage defect model was established, in which Cur-SN combined with a BMSC-loaded hydrogel was implanted. Cartilage differentiation and VEC infiltration levels in newly formed tissues were subsequently analysed. In vitro experiments demonstrated that Cur-SN upregulated Drp1 and ROS levels, leading to mitochondrial homeostasis disruption. This, in turn, induced VEC apoptosis and senescence while significantly inhibited VEC infiltration. Furthermore, Cur-SN effectively counteracted the inhibitory effects of VEC activation on BMSC chondrogenic differentiation. In vivo experiments revealed that Cur-SN reduced VEC infiltration and angiogenesis in newly formed tissues, thereby promoting hyaline cartilage regeneration at the defect site. Cur-SN enhances cartilage repair by upregulating Drp1 expression and ROS levels, thereby disrupting mitochondrial homeostasis, inducing VEC apoptosis and senescence, and inhibiting VEC infiltration.

Introduction

Cartilage defects are globally prevalent and challenging-to-treat conditions that significantly impair patients’ quality of life [1]. Due to the avascular nature of cartilage, its nutrient supply relies entirely on synovial fluid diffusion, which limits its self-repair capacity [2, 3]. Traditional treatments, such as autologous and allogeneic cartilage transplantation and stem cell therapy, provide some relief but often fail to regenerate high-quality hyaline cartilage with adequate biomechanical properties [4]. Tissue engineering has emerged as a promising approach, combining bone marrow mesenchymal stem cells (BMSCs) with biological scaffolds and growth factors to significantly improve repair success [5–7]. Angiogenesis plays a crucial role in the early stages of cartilage repair by supplying essential nutrients, oxygen, and progenitor cells necessary for tissue regeneration. However, while angiogenesis is beneficial in promoting early healing, excessive or prolonged vascular invasion into the non-calcified cartilage can disrupt the hypoxic microenvironment, which is essential for chondrocyte function. This disruption impedes the formation of hyaline cartilage and leads to the development of fibrocartilage, characterized by inferior mechanical properties that undermine the long-term effectiveness of the repair [8–12]. Strategies aimed at inhibiting excessive vascular endothelial cell (VEC) proliferation, regulating pathological angiogenesis, and reducing vascular invasion have shown potential for enhancing cartilage repair quality.

Curcumin (Cur), a natural polyphenolic compound extracted from turmeric rhizome, possesses a wide range of bioactivities, including anti-inflammatory, antioxidant, and anti-angiogenic properties [13, 14]. The pleiotropic bioactivities of curcumin are attributed to its interaction with multiple cellular targets and signaling pathways. Its potent anti-oxidative effect is partly mediated through the activation of the Nrf2/ARE pathway, leading to the upregulation of downstream cytoprotective enzymes such as heme oxygenase-1 (HO-1) [15, 16]. As a classic anti-inflammatory agent, curcumin effectively suppresses the NF-κB signaling pathway by inhibiting IκB kinase (IKK), thereby preventing the nuclear translocation of NF-κB and the subsequent expression of pro-inflammatory cytokines (e.g., TNF-α, IL-1β, IL-6) [17]. Furthermore, its well-documented anti-angiogenic property involves the direct downregulation of key pro-angiogenic factors like Vascular Endothelial Growth Factor (VEGF) and its receptor VEGFR2 [18, 19]. The potential of curcumin to protect chondrocytes and promote the chondrogenic differentiation of BMSCs further underscores its promise for cartilage repair applications [20–22]. However, despite this compelling mechanistic rationale and therapeutic potential, the clinical translation of curcumin is severely hampered by its inherent pharmacokinetic drawbacks, including poor aqueous solubility, rapid systemic metabolism, and chemical instability in physiological environments [23]. Moreover, its potential role in regulating mitochondrial dynamics—a crucial process governing cellular fate—remains less explored, particularly in the context of cartilage repair and its associated anti-angiogenic effects. To overcome this limitation, we propose the use of a nano-delivery system to enhance the bioavailability and stability of Curcumin in vivo.

Silk fibroin (SF), a natural protein derived from silk fibres, has shown significant promise for biomedical applications due to its excellent biocompatibility, tunable degradation rate, and exceptional drug-loading capacity [24–27]. These properties make silk fibroin an ideal material for providing a stable drug-release environment, thereby enhancing Curcumin’s bioavailability and addressing the demands of an effective drug-delivery system.

In summary, while the anti-angiogenic properties of curcumin are well-recognized, the precise molecular mechanisms by which it orchestrates mitochondrial dynamics in vascular endothelial cells to inhibit angiogenesis, particularly in the context of cartilage repair, remain largely unexplored. Therefore, this study was designed to investigate the hypothesis that a novel curcumin-loaded silk fibroin nanoparticle (Cur-SN) delivery system enhances cartilage regeneration by disrupting mitochondrial homeostasis in VECs via the Drp1/ROS pathway, thereby inhibiting vascular invasion (Fig. 1). We systematically evaluated the efficacy of this Cur-SN system both in vitro and in a rat cartilage defect model to elucidate this potential mechanism and assess its therapeutic potential.

Fig. 1.

Schematic diagram of the preparation process, functions, and potential mechanisms of the curcumin-loaded silk fibroin nanoparticle hydrogel scaffold.(A) Construction of the Curcumin-loaded Silk Fibroin Nanoparticle Hydrogel Scaffold. (B) After implanting the Cur-SFNP-GM containing 3D-cultured BMSCs into cartilage defects, Cur-SFNP promotes the expression of Drp1 and the production of ROS in endothelial cells, thereby affecting mitochondrial homeostasis, inducing endothelial cell apoptosis and senescence, and ultimately enhancing cartilage repair

Materials and methods

Preparation of the Cur-SN-GM drug delivery system

Synthesis of Cur-SN

A total of 125 mg of SF was dissolved in 5 mL of ultrapure water and vortexed every 5 min to ensure uniform dispersion, serving as the aqueous phase. Precisely 300 µL of Span 80 was mixed with 30 mL of soybean oil and stirred in a water bath with a magnetic stirrer for 5 min at 45–50℃ to achieve uniform dispersion, forming the oil phase. The aqueous phase was then added to the oil phase and emulsified by stirring at 350 rpm in a water bath maintained at 45–50℃. The resulting emulsion was cooled to 4℃, filtered, and washed with cold acetone. The microspheres were then dispersed in a glutaraldehyde-acetone solution (0.01% v/v): water solution (3:1) at approximately 2 mg/mL and incubated overnight in the dark. After incubation, the solution was centrifuged to collect the microspheres, which were subsequently redispersed in a 10 mg/mL glycine solution, washed with ultrapure water, centrifuged again, and freeze-dried to obtain silk fibroin nano-microspheres (SNs).

To load Curcumin, 100 mg of SNs was immersed in 400 µL of a 5 mg/mL curcumin solution in ethanol at 4 °C for 12 h. The excess curcumin solution was removed, and the product was dried overnight at 4 °C, yielding curcumin-loaded silk fibroin microspheres (Cur-SN).

Characterisation of Cur-SN

The microstructure of the microspheres was observed using scanning electron microscopy (SEM; HITACHI Regulus 8100, Japan). The particle size distribution was determined via dynamic light scattering (DLS) using a Zetasizer Nano ZS (Malvern Instruments, USA).

To analyse the encapsulation efficiency and drug-loading capacity, 5 mg of Cur-SN was dispersed in 10 mL methanol, vortexed, ultrasonicated until fully dissolved, and centrifuged to collect the supernatant. The curcumin content in the supernatant was quantified using UV-Vis spectrophotometry at 425 nm, with curcumin standards serving as a reference. The encapsulation and drug-loading efficiencies were calculated using the following equations:

For the drug release study, 10 mg of Cur-SN was suspended in 25 mL of phosphate-buffered saline (PBS) and incubated at 37 °C in the dark. Samples were taken on days 1, 3, 5, 7, 9, 11, 13, and 15, and fresh PBS was replenished as needed. The curcumin content was measured relative to a blank SN leachate control, and the cumulative release rate was calculated.

Preparation of Cur/SN-GM

Gelatin methacryloyl (GelMA; EFL-GM-90, Suzhou, China) was purchased from Suzhou Yongqinquan Intelligent Equipment Co., Ltd. A 10% (w/v) GelMA solution was prepared by dissolving it in 0.25% lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP; EFL, Suzhou). Next, 5 mg of SN or Cur/SN was premixed with 1 mL of GelMA solution and cured under a 405 nm blue light source within a custom silicone mould for 2 min. This process resulted in a transparent cylindrical scaffold measuring 2 mm in diameter and 3 mm in length.

Characterisation of Cur-SN-GM

The gross morphology of the prepared hydrogels was observed under a microscope (Leica, Germany). Young’s modulus of the hydrogel scaffold was determined under axial compression using a biomechanical tester (MTS, USA). To measure the swelling rate (SR), the initial mass of the hydrogel (M0) was recorded, and the hydrogel was then incubated in 5 mL of PBS at 37 °C with agitation at 100 rpm. At predetermined time points, surface water was removed with filter paper, and the weight (M1) was recorded. The SR was calculated using the following formula:

To determine the degradation rate (DR), the hydrogel was weighed (M0) and incubated in 5 mL of PBS containing 10 µg/mL of type II collagenase at 37 °C under 100 rpm agitation. At designated time points, the surface water was removed with filter paper, and the weight (M1) was measured. The DR was calculated as follows:

In a sterile environment, the BMSCs were mixed with preheated hydrogel at 37 °C to achieve a final cell concentration of 1 × 10⁷ cells/mL. The mixture was cured using a 405 nm blue light source and a curing ring (EFL-SCR-3D, Suzhou) for 2 min and subsequently cultured in a 12-well plate. Cell Counting Kit-8 (CCK-8) assays were performed on days 1 and 3 to evaluate hydrogel cytocompatibility.

Cell culture

VECs were purchased from Fenghui Biotechnology and cultured in high-glucose Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% foetal bovine serum (FBS) and 1% antibiotics. Cells were incubated at 37 °C in a 5% CO₂ atmosphere. CD31 monoclonal antibody immunostaining was performed to observe VEC morphology. First, cells were fixed with 4% paraformaldehyde (PFA), permeabilised with 0.5% Triton X-100, and blocked with 5% bovine serum albumin (BSA) at room temperature. Subsequently, cells were incubated overnight at 4 °C with CD31 antibody (Proteintech, Nanjing, 1:100). Afterwards, they were treated with Goat anti-rabbit IgG H&L (Alexa Fluor^® 594; Abcam, USA) for 2 h at room temperature. Finally, nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) in preparation for fluorescence microscopy.

To investigate molecular mechanisms, VECs were pretreated with either Mdivi-1 (2.5 µM; Abmole, USA), a Drp1 signalling inhibitor or N-acetylcysteine (5 µM; Abmole, USA), a reactive oxygen species (ROS) scavenger. Following pretreatment, cells were analysed as outlined in the experimental schedule.

BMSCs were isolated from Sprague-Dawley rats using the whole-bone marrow adherence method. Six-week-old rats were euthanised, and their femurs and tibias were harvested under sterile conditions. The bone marrow cavity was flushed to collect marrow cells suspended in DMEM and cultured at 37 °C with 5% CO₂. After 48 h, tightly adherent cells were considered passage 1 BMSCs, while passage 3 BMSCs were used for this study. To investigate molecular mechanisms, a chondrogenic differentiation medium was prepared, containing 10 mg/mL transforming growth factor-beta (TGF-β; Abbkine, USA), 100 nM dexamethasone, 50 µg/mL Vitamin C, 40 µg/ml proline, and a 1:40 dilution of ITS cell culture additive (Yuanye, China).

CCK-8 assay

The CCK-8 assay (Vazyme, Jiangsu) was performed according to the manufacturer’s instructions to evaluate cytotoxicity, cell viability, and biocompatibility of Cur, SN, and Cur-SN. VECs were seeded into 96-well plates at a density of approximately 3 × 10 cells/well and incubated at 37 °C in a 5% CO₂ atmosphere for 12 h. The cells were then treated with increasing concentrations of Cur (0, 0.1, 0.5, 1.0, 2.5, 5, 10, 20, 30, 50 µM), SN (122 µg/mL), and Cur-SN (with Cur concentrations of 10, 20, 30, 40, 50 µM). After 48 h of treatment, the medium was discarded, and 100 µL of CCK-8 reagent (CCK8: DMEM = 1:10) was added to each well. The cells were incubated for an additional 3 h, and the absorbance was measured at 450 nm using a microplate reader (Tanon, Shanghai) to calculate cell viability.

Cell migration assay

For the cell migration assay, 4 × 10⁴ VECs were seeded into the upper chamber of a 24-well Transwell plate, and 700 µL of supplemented DMEM (DMEM containing 10% FBS and 1% penicillin-streptomycin) was added to the lower chamber. The experimental groups were as follows: Control (C) (supplemented DMEM), Cur20 (supplemented DMEM + 20 µM Cur-SN), Cur30 (supplemented DMEM + 30 µM Cur-SN), and Cur40 (supplemented DMEM + 40 µM Cur-SN). After 48 h, the cells in the Transwell chamber were fixed with 4% PFA for 10 min and stained with crystal violet (Beyotime, Shanghai) for 15 min at room temperature. Five random fields were selected for quantitative analysis of cell migration.

Wound healing assay

VECs were seeded in six-well plates and allowed to reach confluence. A scratch was created at the bottom of each well using a 200 µL pipette tip. After gently rinsing the wells with PBS, treatments were applied according to the experimental groups (C, Cur20, Cur30, and Cur40), using intervention concentrations identical to those in the cell migration assay, the supplemented DMEM was replaced with serum-free DMEM to specifically evaluate cell migration,. Images were captured immediately after scratching (0 h) to document the initial wound area. After 48 h, the plates were rinsed again with PBS, and final images were taken to calculate the migration rate of each group.

Apoptosis detection

VECs were collected and seeded into six-well plates. Treatments were applied according to the following groups: control (C), Cur20, Cur30, and Cur40 (with intervention concentrations consistent with those used in the cell migration assay). After 48 h, cells from each group were collected and adjusted to a concentration of 1 × 10⁶ cells/mL. Apoptosis was assessed using the Annexin V Alexa Fluor 647/PI Apoptosis Detection Kit (Fcmacs, Nanjing), and apoptotic status was analysed using flow cytometry (FlowSight, USA) with IDEAS software.

β-Galactosidase staining

To investigate the molecular mechanisms, VECs were categorised into the following treatment groups:

VECs were seeded into six-well plates, and the respective treatments were applied. After 48 h, cells were fixed with 4% PFA and assessed for senescence using the Cell Senescence β-Galactosidase Staining Kit (Yeasen, Shanghai).

Mitochondrial morphology

VECs were collected, seeded into six-well plates, and assigned to the following treatment groups:

After 48 h, cell clusters were fixed with 2.5% glutaraldehyde. The mitochondrial ultrastructure was examined using a transmission electron microscope (HITACHI HT7700, Japan). Additionally, cells were stained with MitoTracker^® Red CM-H2Xros (Yeasen, Shanghai) and imaged using an inverted fluorescence microscope to analyse mitochondrial morphology.

Mitochondrial function detection

VECs were collected, seeded into six-well plates, and treated according to the following groups:

(Intervention concentrations were consistent with those used in the cell migration assay and β-Galactosidase Staining). After 48 h, the cells were harvested, and the mitochondrial membrane potential was measured using the JC-1 Mitochondrial Membrane Potential Assay Kit (Yeasen, Shanghai).

ROS measurement

VECs were collected and seeded into six-well plates and treated according to the following groups:

(intervention concentrations remained identical to those in the cell migration assay and β-Galactosidase Staining). After 48 h, cells were collected, and intracellular ROS levels were quantified using the CheKine™ ROS Detection Fluorometric Assay Kit (Abbkine, USA) following the manufacturer’s instructions. ROS levels were analysed using fluorescence microscopy and flow cytometry.

In vitro culture of VECs and BMSCs with conditioned medium

Establishment of a conditioned medium culture system

The culture supernatant of VECs was collected and used as a conditioned medium to culture BMSCs.The following experimental groups were prepared:

• Ctrl-Med (Control Medium): supplemented DMEM only.

• VEC-CM (VEC Conditioned Medium): from untreated VECs.

• VEC-Cur-CM (VEC Curcumin Conditioned Medium): from VECs treated with 30 µM curcumin.

• VEC-CurSN-CM (VEC Curcumin-Loaded Silk Nanoparticle Conditioned Medium): from VECs treated with Cur-SN leachate.

Chondrogenic factors were added to the culture medium of each BMSC group. After 14 days, the chondrogenic differentiation status of BMSC was assessed.

Chondrogenic culture of BMSC

After fixation with 4% PFA, each group of BMSCs was stained with toluidine blue solution (Yuanye, Shanghai, China) and alizarin red solution (Beyotime) to assess chondrogenic and osteogenic differentiation. Additionally, total RNA was extracted from each group of BMSCs to evaluate the mRNA expression levels of phenotype-related genes.

Real-time quantitative PCR

Total RNA was extracted from the VECs and BMSCs using the FastPure Cell/Tissue Total RNA Isolation Kit (Vazyme, Nanjing, China). RNA was reverse-transcribed into complementary DNA (cDNA) using the All-In-One 5× RT MasterMix (Abm, Canada). Real-time PCR was performed with EvaGreen 2× qPCR MasterMix (Abm, Canada) on an automated PCR analyser (ABI QuantStudio 3, USA) under the following conditions: initial denaturation at 95 °C for 10 min, denaturation at 95 °C for 15 s, and annealing/extension at 60 °C for 60 s. This cycle was repeated for a total of 40 cycles. GAPDH was used as the internal reference gene. The relative expression levels of target genes were determined using the 2-∆∆Ct method. Primer sequences are listed in Table 1.

Table 1.

Target genes primer sequences used in qRT-PCR

Gene	Primer Direction	Sequence (5’ → 3’)
Hif-1α	Forward	GTCTGAGGGGACAGGAGGAT
	Reverse	CTCCTCAGGTGGCTTGTCAG
Vegfa	Forward	GCACGTTGGCTCACTTCCAG
	Reverse	TGGTCGGAACCAGAATCTTTATCTC
VegfR	Forward	GGCCACCACTCAGGACTACT
	Reverse	GGCGCTTCCAAATCTCTAAC
Dll4	Forward	TGCGGATAACCAACGACG
	Reverse	GCCCACAAAGCCATAAGGAC
TGF-β	Forward	CAAAGACATCACACACAGTA
	Reverse	GGTGTTGAGCCCTTTCCAGG
Aggrecan	Forward	CAGTGCGATGCAGGCTGGCT
	Reverse	CCTCCGGCACTCGTTGGCTG
COL1A1	Forward	GGCAGCAGCACTATGACCCAA
	Reverse	ACAGGCCTACCCAAACGTGAGTCC
COL2A1	Forward	CTCATCCAGGGCTCCAATGAT
	Reverse	TCTGTGATCGGTACTCGATGA
Sox9	Forward	AGTACCCGCATCTGCACAAC
	Reverse	ACTTGTAATCGGGGTGGTCT
DRP1	Forward	TGTGCTGGTGTTGAGGATG
	Reverse	GCTGTTGACTTGTGATGAGG
FIS1	Forward	GAGGAGTGTGATGGCAGAAA
	Reverse	CGACAGCAGGTCTTCAGGA

Western blot assay

Total protein from VECs was extracted using RIPA lysis buffer and quantified with a BCA assay kit (Yuanye, Shanghai). Sample buffer was added, and proteins were denatured in a boiling water bath for 10 min. Proteins were separated via sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto membranes. The membranes were blocked with 5% non-fat milk at room temperature for 2 h, followed by overnight incubation at 4 °C with primary antibodies. The antibodies included GAPDH (Proteintech, Nanjing), BAX (Proteintech, Nanjing), BCL-2 (Proteintech, Nanjing), Caspase-9 (SAB, USA), P16 (Santa, USA), P21 (Proteintech, Nanjing), P53 (Proteintech, Nanjing), Drp1 (Proteintech, Nanjing), P-Drp1 (CST, USA), and FIS1 (Proteintech, Nanjing), each diluted at 1:1000. After primary antibody incubation, membranes were incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG (YIFEIXUE Bio Tech, Nanjing, China) at a 1:10,000 dilution for 2 h at room temperature. An ECL chemiluminescent reagent (Yeasen, Shanghai) was added, and the bands were visualised using a gel imaging system (Tanon, Shanghai). The band intensity was quantified using ImageJ software.

Evaluation of chondrogenic differentiation of BMSC in 3D scaffold culture

Under sterile conditions, BMSCs were mixed with preheated GM hydrogel at 37 °C to achieve a final concentration of 0.5 × 10⁷ cells/mL for the BMSCs.The following experimental groups were established: Negative Control (N), consisting of blank GM + supplemented DMEM; Induction group (C), consisting of blank GM + supplemented DMEM + TGF-β (1 ng/mL); SN group, consisting of SN-GM + supplemented DMEM + TGF-β (1 ng/mL); and CUR-SN-GM group, consisting of Cur-SN-GM + supplemented DMEM + TGF-β (1 ng/mL). The hydrogel mixtures were polymerised under a 405 nm blue light source for 2 min using a curing ring, then cultured in 12-well plates and co-cultured for 4 weeks. After fixation in 4% PFA and dehydration using a graded ethanol series, the samples were embedded in an optimal cutting temperature (OCT) compound overnight. Cryosections were prepared using a cryostat (Leica) and stained with toluidine blue to assess the chondrogenic differentiation of BMSCs in vitro.

In vivo animal study

Animal model

All animal experiments followed the Guiding Principles for the Care and Use of Animals and were approved by the Animal Experimental Ethics Committee of Nanjing University of Chinese Medicine (Approval No. 202206A042). 75 male Sprague-Dawley (SD) rats weighing 250 ± 25 g were purchased from Jiangsu Huachuang Sino Pharmaceutical Technology Co., Ltd. The rats were randomly assigned into four groups: sham operation (SH), no treatment (NT), blank scaffold control (SN), and drug-loaded scaffold (Cur-SN).

Anaesthesia was induced with isoflurane using a gas anaesthesia machine (RWD R500, Shenzhen). The bilateral joint cavities were exposed under sterile conditions. With the exception of the SH group, all other groups underwent drilling of the articular cartilage surface of the knee joint (2 mm in diameter and 3 mm in depth, reaching the full thickness of the cartilage). The NT group received no treatment for the cartilage defects, the SN group received the SN-GM scaffold implanted at the defect site, and the Cur-SN group received the Cur-SN-GM composite implanted at the defect site. The rats were sacrificed 12 weeks post-surgery, and knee joint samples were collected for further analysis.

Gross morphology evaluation

The morphology of each knee-joint sample was evaluated based on cartilage defect filling, surface smoothness, tissue regeneration, texture, and structural integrity.

Histological and immunohistochemical evaluation

Knee joint repair tissue samples were fixed in 4% PFA for 2 days and decalcified in 19% ethylenediaminetetraacetic acid (EDTA) for 14 days. The samples were then dehydrated using a graded ethanol series, embedded in paraffin and sectioned into 7 μm slices. Sections were stained with haematoxylin and eosin (H&E) and toluidine blue (Yuanye, Shanghai) and examined under a microscope (Leica, Germany).

The sections were stained with hematoxylin and eosin (HE), toluidine blue, and type II collagen (Yuanye, Shanghai) immunohistochemistry, observed under a Leica microscope, and then collagen fiber fraction was measured using a digital image analysis system (ImageJ). The new tissue structure and level of cartilage differentiation in each group were evaluated.

Evaluation of vascular infiltration by immunofluorescent staining

Paraffin-embedded sections of the repaired cartilage tissue and synovium were prepared. Sections were de-paraffinized with xylene, rehydrated through a graded ethanol series, and subjected to antigen retrieval using sodium citrate buffer in a boiling water bath. After permeabilization with 0.5% Triton X-100 and blocking with 5% BSA at room temperature, the sections were incubated overnight at 4 °C with a primary antibody against CD31 (Proteintech, Nanjing; 1:100 dilution). The sections were then incubated with Goat Anti-Rabbit IgG H&L (Alexa Fluor^® 594; Abcam, USA; 1:500 dilution ) for 2 h at room temperature, counterstained with DAPI, and mounted with anti-fade mounting medium (Yuanye, Shanghai). Observations were performed using a fluorescence microscope (Leica, Germany).

Statistical analysis

Experimental data were expressed as means ± SD. Differences between groups were analysed using SPSS 20.0 software. The normality of data distribution was verified using Q-Q plots and Shapiro-Wilk tests. Parametric tests were employed as the data met the necessary assumptions, and these tests are robust to minor deviations from normality given our sample sizes. Statistical comparisons were performed using one-way ANOVA or a two-tailed Student’s t-test. A P-value < 0.05 was considered statistically significant.

Results

Characterisation of Cur-SN

Observations revealed that SN appeared as a white powder, whereas Cur-SN exhibited a yellow colour, attributable to the intrinsic pigmentation of Curcumin (Fig. 2A). Scanning electron microscopy (SEM) analysis demonstrated that SN and Cur-SN microspheres were approximately spherical and uniformly distributed. While SN displayed a relatively smooth and even surface, Cur-SN exhibited a slightly rougher texture, possibly due to curcumin loading on the surface or within the microsphere matrix (Fig. 2B). The particle size distribution for both microsphere types was narrow, indicating uniform particle sizes measuring less than 1000 nm, with peak sizes ranging between 200 and 300 nm. This classification places them within the nanoparticle range, making them suitable for drug delivery applications (Fig. 2C).

Fig. 2.

Characterisation and biocompatibility of Cur-SN and SN. (A) Camera Images of Cur-SN and SN. (B) Scanning electron microscopy of Cur-SN and SN. (C) Cur-SN and SN particle size distribution. (D) Cur-SN drug loading and encapsulation efficiency. (E) Curcumin impact on VECs viability detected by CCK8. (F) Cumulative release profile of Curcumin in Cur-SN. (G) Cur-SN and SN impact on VECs viability detected by CCK8. (H) Dose–response relationship of Cur-SN in VECs. Compare with Control group or SN group, *P < 0.05, **P < 0.01, ***P < 0.001

Cur-SN exhibited a drug loading capacity of 15.08 ± 0.62% and an encapsulation efficiency of 73.98 ± 3.4%. The drug release profile showed that Cur-SN demonstrated a rapid release rate in neutral PBS buffer at 37 °C during the first three days, with approximately 80% of the drug released by the third day. Thereafter, the release rate gradually slowed, reaching nearly 100% cumulative release by day 15, indicating the sustained-release characteristics of the system (Fig. 2F).

Curcumin was first applied to the cells to determine the effective microsphere dosage for VECs and to assess its concentration-dependent effects. The CCK-8 assay revealed that curcumin concentrations of 20 µM, 30 µM, and 50 µM inhibited VEC proliferation after 24 h, with more pronounced inhibition at 48 and 72 h, where cell viability dropped below 50% (Fig. 2E).

For subsequent experiments, Cur-SN and SN were applied to cells at drug concentrations of 10 µM, 20 µM, 30 µM, 40 µM, and 50 µM for 48 h, followed by a cell viability assessment. The CCK-8 assay showed that SN-treated cells maintained viability close to 100%, with no significant difference from the control group, suggesting that the microspheres were non-toxic and biocompatible. In contrast, Cur-SN at 20 µM, 30 µM, 40 µM, and 50 µM significantly inhibited VEC proliferation (p < 0.05, p < 0.001) (Fig. 2G). Notably, at 50 µM, Cur-SN reduced cell viability to below 50%. To ensure cell viability in subsequent experiments, Cur-SN concentrations of 20 µM, 30 µM, and 40 µM with an exposure time of 48 h were selected.

Effects of Cur-SN on apoptosis and senescence of endothelial cells in vitro

CD31 fluorescence staining of the VECs was positive, indicating that the cells were in a healthy state (Fig. 3A). Scratch wound healing assays and Transwell migration experiments were conducted to explore the effects of Cur-SN on VEC migration and invasion. VECs were divided into a control group (Group C) and experimental groups treated with different curcumin concentrations (20, 30, and 40 µM) (Groups Cur20, Cur30, Cur40). The results (Fig. 3B–C) demonstrated significant dose-dependent inhibition of migration at 48 h post-treatment.

Fig. 3.

Cur-SN inhibited the migration, invasion, apoptosis and senescence of VECs in vitro. VECs were treated with different concentrations of Cur-SN (0, 20, 30, 40µM) (A) Representative images of VECs under a light microscope (400×). CD31 fluorescence staining (red) and stained for nucler (blue). (B-1) Representative photomicrographs of the wound edge in the scratch assay after 48 h treatment in each group. (B-2) Results of quantitative analysis of cell migration rate. (C-1) Representative images of the transwell migration of VECs was determined by transwell migration assay after 48 h treatment in each group. (C-2) Results of quantitative analysis of VECs transwell migration assay. (D) Representative photomicrographs of senescence-β-galactosidase staining was performed and green staining indicated the senescence-β-galactosidase staining positivity. (E) Apoptotic rates of different groups were determined using a flow apoptosis kit. Quadrants drawn on histograms allow to define the repartition of VECs populations among apoptotic (Annexin V+/PI+), necrotic (Annexin V-/PI+), pre‐apoptotic (Annexin V+/PI‐), and normal (Annexin V‐/PI‐) cells. (F) Representative images and quantitative analysis of BCL2, BAX, Caspase 9, P21, P53, GAPDH in VECs of WB analysis. All data are expressed as mean ± SD, n = 3. *p < 0.05,**p < 0.01 and ***p < 0.001 as compared with the control group

To evaluate cellular senescence, β-galactosidase staining was performed (Fig. 3D). The results showed that Cur-SNs promoted the VEC senescence, with the most significant effect observed at a medium concentration. VEC apoptosis across all groups was assessed using an Annexin V/PI apoptosis assay kit (Fig. 3E). The findings indicated that Cur-SNs significantly promoted apoptosis in VECs, with the apoptosis rate being significantly higher in the medium-dose group compared to the control group.

The levels of apoptosis-related proteins BAX, BCL2, and Caspase-9 were quantified using Western blotting (Fig. 3F). The results indicated that compared to the control group (Group C), the pro-apoptotic proteins BAX and Caspase-9 were upregulated, whereas the apoptosis-inhibitory protein BCL2 was downregulated across all dose groups. The most significant effect was observed in the Cur30 group (p < 0.01). Furthermore, the expression of senescence-associated proteins P21 and P53 was analysed (Fig. 3F). Western blotting results revealed that relative to Group C, P21 expression was significantly increased in all dose groups, with the highest expression detected in the Cur40 group (p < 0.001). P53 expression increased in the Cur20 and Cur30 groups (p < 0.05) but decreased in the Cur40 group (p > 0.05).

Regulation of mitochondrial morphology and function by Cur-SN

To investigate the effects of Cur-SN on mitochondrial function in VECs, we examined the mitochondrial structure using transmission electron microscopy. The images revealed that medium-dose Cur-SN treatment resulted in a higher number of mitochondria with reduced size (Fig. 4A). MitoTracker Red probe staining was used to assess mitochondrial mass, and a significant increase in red fluorescence intensity was observed following medium-dose Cur-SN treatment (Fig. 4B).

Fig. 4.

Cur-SN regulates mitochondrial morphology and function of VECs in vitro. (A) Mitochondrial morphology observed using transmission electron microscopy after treatment of 0, 30µM Cur-SN. (B1-2) The number of mitochondria was detected by Mito-Tracker Red staining in VECs treated with Cur-SN at 0 and 30µM concentrations. (C1-2) Representative images of JC-1 staining, and quantification of JC-1 fluorescence intensities. (D1) Representative images of DCFH-DA fluorescent staining. Green-stained DCFH-DA positive cells represent intracellular ROS. (D2) Quantification of DCFH-DA fluorescence intensity. (E1-2) Representative images of MitoSOX Red staining, and quantification of MitoSOX Red fluorescence intensities. (F) Mean fluorescence intensity (MFI) of DCFH-DA measured by flow cytometry. (G) RT-qPCR analysis of mRNA expression for two mitochondrial fission-related genes in VECs. (H) Representative images and quantitative analysis of DRP1, p-DRP1 and FIS1 in VECs of WB analysis. All data are expressed as mean ± SD, n = 3. *p < 0.05,**p < 0.01 and ***p < 0.001 as compared with the control group

JC-1 staining was performed to detect mitochondrial membrane potential (Fig. 4C-1/2). The results demonstrated a progressive reduction in red fluorescence intensity (p < 0.001), indicating that Cur-SNs induced a concentration-dependent decrease in mitochondrial membrane potential. A DCFH-DA fluorescent probe was used to measure intracellular ROS levels. Quantitative analysis of fluorescence intensity (Fig. 4D-1/2) demonstrated that compared to the control group (C), ROS levels were significantly elevated in the Cur20, Cur30, and Cur40 groups (p < 0.01), with the highest ROS levels observed in the Cur30 group.We used MitoSOX™ Red, a mitochondrial superoxide indicator, to assess mitochondrial superoxide production. The results clearly show a significant, dose-dependent increase in mitochondrial superoxide levels following Cur-SN treatment compared to the control group (p < 0.01), with the most pronounced effect observed at the 30 µM concentration (Fig. 4E).Flow cytometry measured the mean fluorescence intensity (MFI) of DCFH-DA staining (Fig. 4F). In the Cur30 and Cur40 groups, the MFI of intracellular ROS was significantly higher than in the control group (p < 0.001).

Western blot analysis (Fig. 4H) showed that p-Drp1/Drp1 expression was significantly upregulated in the Cur30 group (p < 0.05, whereas FIS1 protein levels were significantly higher in the Cur20 group compared to the control group.RT-qPCR analysis (Fig. 4G) showed that all dose groups had significantly higher levels of Drp1 and FIS1 mRNA than the control group (p < 0.05), with the highest expression observed in the Cur30 group (p < 0.01). Based on these findings, 30 µM Cur-SN was chosen for subsequent experiments.

Effects of Cur-SN on BMSC chondrogenic differentiation in vitro

To investigate whether Cur-SNs influence BMSC differentiation when co-cultured with a VEC-conditioned medium, toluidine blue and Alizarin Red staining were performed (Fig. 5A). Group VEC-CM showed weaker toluidine blue staining but stronger alizarin red staining than the Ctrl-Med group, suggesting that exposure to activated VEC supernatant reduced BMSC chondrogenic differentiation. Conversely, In the VEC-Cur-CM and VEC-CurSN-CM groups, toluidine blue staining was intensified, while Alizarin red staining was diminished relative to the VEC-CM group, indicating that both Cur and Cur-SN promote BMSCs differentiation into chondrogenic cells.

Fig. 5.

Effects of Cur-SN on chondrogenic differentiation of BMSCs in vitro. BMSCs were chondrogenically induced and cultured in conditioned media prepared from: (1) VEC supernatant (VEC-CM), (2) VEC supernatant supplemented with 30 µM free Cur (VEC-Cur-CM), and (3) VEC supernatant supplemented with 30 µM Cur-SNs (VEC-CurSN-CM). (A)Representative images stained with toluidine blue and alizarin red. (B)The mRNA expressions of 6 genes related to chondrogenic differentiation in BMSCs of RT-qPCR analysis. All data are expressed as mean ± SD, n = 6, Compared with the Ctrl-Med group, *P < 0.05,**P < 0.01, ***P < 0.001; Compared with the VEC-CM group, ^#p < 0.05,^##p < 0.01 and ^###p < 0.001; Compared with the VEC-Cur-CM group, ⁺p < 0.05, ⁺⁺p < 0.01 and ⁺⁺⁺p < 0.001

The expression of mRNAs related to BMSC chondrogenic and osteogenic differentiation was also quantified using RT-qPCR (Fig. 5B). Compared to the Ctrl-Med group, the VEC-CM group exhibited a significant reduction in the mRNA expression of TGF-β, Aggrecan, SOX-9, COL2A1, COL1A1, and Osterix (P < 0.001, P < 0.001, P < 0.001, P < 0.001, P < 0.001, P < 0.01). In contrast, compared to the VEC-CM group, the VEC-Cur-CM group showed downregulation of TGF-β, SOX-9, COL1A1, and Osterix (P > 0.05), whereas Aggrecan and COL2A1 were upregulated (P > 0.05, P < 0.01). In the VEC-CurSN-CM group, the mRNA expression of TGF-β, Aggrecan, SOX-9, COL2A1, COL1A1, and Osterix were significantly increased compared to the VEC-CM group (P < 0.001, P < 0.001, P < 0.001, P > 0.5, P < 0.001, P < 0.05).

Effect of Drp1 blocking and ROS scavengers on Inhibition of angiogenesis by Cur-SN

To examine the effect of Mdivi-1 and N-Acetylcysteine on Cur-SN-mediated inhibition of angiogenesis, RT-qPCR was performed to quantify the mRNA expression of DLL4, HIF-1α, VEGFA, and VEGFR. Based on the results (Fig. 6A), compared to the C group, the J group exhibited downregulation of DLL4, HIF-1α, VEGFA, and VEGFR (P<0.01, P<0.05, P>0.05, P<0.05). Conversely, mRNA expression levels in the Mdivi1 and NAC groups were significantly upregulated following treatment with Mdivi-1 and N-acetylcysteine (P<0.01). In the Cur-Mdivi1 and CurSN-NAC groups, mRNA expression levels were significantly higher than in the J group (P < 0.05), but no significant differences were observed between the Mdivi1 and CurSN-Mdivi1 groups or between the NAC and CurSN-NAC groups. These findings suggest that N-acetylcysteine and Mdivi-1 counteract the inhibitory effects of Cur-SN on angiogenesis.

Fig. 6.

Cur-SN inhibits angiogenesis through the mitochondrial DRP1/ROS pathway. C: VECs, CurSN: VECs + 30µM Cur-SN, Mdivi1: VECs + 2.5µM Mdivi-1, CurSN-Mdivi1: VECs + 2.5µM Mdivi-1 + 30µM Cur-SN, NAC: VECs + 5 µM N-Acetylcysteine, CurSN-NAC: VECs + 5 µM N-Acetylcysteine + 30µM Cur-SN. (A) The mRNA expressions of 4 genes of angiogenesis in BMSCs cells of RT-qPCR analysis. (B) Representative images and quantitative analysis of apoptosis and senescence-related proteins in VECs of WB analysis. (C) Representative photomicrographs of senescence-β-galactosidase staining was performed and green staining indicated the senescence-β-galactosidase staining positivity. (D1-2) The number of mitochondria was detected by Mito-Tracker Red staining. (E1-2) Representative images of JC-1 staining; the red images on the left represent JC-1 aggregates, and the green images represent JC-1 monomer. The blue images represent DAPI, and the right panel shows the merged images. Quantification of JC-1 fluorescence intensities. (F) Mean fluorescence intensity (MFI) of DCFH-DA measured by flow cytometry. (G) Representative images and quantitative analysis of DRP1, p-DRP1 and FIS1 in VECs of WB analysis. All data are expressed as mean ± SD, n = 3. *p < 0.05,**p < 0.01 and ***p < 0.001

Western blot analysis was conducted to assess the expression of apoptosis- and senescence-related proteins. The results (Fig. 6B) showed that Mdivi-1 significantly inhibited BAX and P16 protein expression in the M group compared to the C group (P < 0.05, P < 0.001). Similarly, N-acetylcysteine significantly suppressed the expression of BAX, Caspase-9 and P16 in the NAC group compared to the Cur group (P < 0.01, P < 0.001, P < 0.001). The Cur group displayed significantly higher expression of BAX, Caspase-9, P53, and P16 than the Cur group (P < 0.05, P < 0.001, P < 0.001, and P < 0.05, respectively). In the CurSN-Mdivi1 and CurSN-NAC groups, the expression of these proteins was significantly lower than in the Cur group (P < 0.05), with no significant differences observed between the Mdivi1 and CurSN-Mdivi1 groups or the NAC and CurSN-NAC groups. These findings indicate that Mdivi-1 and N-acetylcysteine mitigate Cur-S-induced apoptosis and senescence in VECs. Cell senescence (Fig. 6C) was further evaluated by β-galactosidase staining, and the results were consistent with the Western blot analysis trend for senescence-related proteins.

We also examined the impact of Mdivi-1 and N-acetylcysteine on Cur-SN-induced mitochondrial fission. Results from MitoTracker Red-Probeto (Fig. 6D1-2) showed that relative fluorescence intensity was significantly lower in the J and M groups than in the C group (P<0.001, P<0.05). No significant difference was observed between the NAC and C groups. However, fluorescence intensity in the CurSN-Mdivi1 and CurSN-NAC groups was significantly higher than in the Cur group (P<0.001), with no significant differences detected between the Mdivi1 and CurSN-Mdivi1 groups or the NAC and CurSN-NAC groups. Fluorescence microscopy with JC-1 staining was used to assess mitochondrial membrane potential. The results (Fig. 6E1-2) showed that the red-to-green fluorescence intensity ratio was significantly lower in the J group than in the C group (P < 0.001). The Mdivi1 and NAC groups also exhibited significantly lower membrane potential than the C group (P < 0.001). No significant difference was observed between the CurSN-Mdivi1 and Cur group, although membrane potential in the CurSN-NAC group was higher than in the Cur group (P < 0.05). No significant difference was observed between the Mdivi1 and CurSN-MDIVI1 groups or the NAC and CurSN-NAC groups.

Intracellular ROS levels were measured using a DCFH-DA fluorescent probe, and MFI was quantified using flow cytometry (Fig. 6F). MFI was significantly higher in the J group compared to the C group (P < 0.001), whereas MFI values in the Mdivi1 and NAC groups were lower than in the C group (P < 0.001). MFI values in the CurSN-Mdivi1 and CurSN-NAC groups were significantly lower than in the Cur group (P < 0.001), with no significant differences observed between the Mdivi1 and CurSN-Mdivi1 groups or the NAC and CurSN-NAC groups.

Furthermore, mitochondrial fission-related proteins, including FIS1, DRP1, and p-Drp1, were detected using Western blotting (Fig. 6G). The Cur group exhibited significantly increased expression of FIS1 and p-Drp1/2-DRP1 compared to the C group (p < 0.01, P < 0.001, respectively). However, the Mdivi1 and NAC groups showed no significant differences in expression levels compared to the C group. In the CurSN- Mdivi1 group, FIS1 and p-Drp1/DRP1 expression were significantly lower than in the J group (p < 0.001), but no significant difference was observed between the Mdivi1 and CurSN- Mdivi1 groups. Similarly, In the CurSN-NAC group, FIS1 expression was significantly lower than in the Cur group (p<0.001), yet no significant difference was found between the NAC and CurSN-NAC groups. These findings suggest that Mdivi-1 and N-Acetylcysteine counteract Cur-SN-induced mitochondrial fission in VECs. This supports the hypothesis that Cur-SN may promote mitochondrial fission via the Drp1/ROS signalling pathways.

Cur-SN-GM scaffold characterisation and effect on BMSC chondrodifferentiation in vitro

To simulate the cartilage repair process induced by Cur-SN in vitro, GelMA hydrogel was used for 3D BMSC culture. The constructed Cur-SN-GM hydrogel formed an opaque cylindrical structure with a diameter of 2 mm and a length of 3 mm (Fig. 7A). The CCK-8 assay results indicated no significant differences in BMSC proliferation among the GM, SN-GM, and Cur-SN-GM hydrogels after 1 and 3 days of culture, demonstrating good biocompatibility (Fig. 7B). Mechanical testing revealed that Young’s modulus of GM and Cur-SN-GM was approximately 0.5 MPa, with no significant differences (P > 0.05). This value falls within the reported range for natural human articular cartilage (0.46–1.43 MPa) (Fig. 7C) [28]. The Cur-SN-GM scaffold was degraded by type II collagenase in vitro, demonstrating good degradability (Fig. 7D). Both Cur-SN and SN demonstrated minimal swelling and maintained their structural integrity, indicating mechanical stability suitable for cartilage tissue. This stability is essential, as it minimises mechanical stress on the surrounding native articular cartilage following scaffold implantation (Fig. 7E). Toluidine blue staining revealed that, compared to the N group, TGF-β stimulation (C group) resulted in deeper staining and enhanced chondrogenic differentiation. Notably, in the presence of Cur-SN-GM (Cur-SN group), toluidine blue staining was more intense than in the C group, indicating that Cur-SN-GM effectively promoted BMSC chondrogenic differentiation in vitro (Fig. 7F).

Fig. 7.

Characterisation of Cur-SN-GM scaffold and its effect on in vitro chondrogenic differentiation of BMSCs.(A) Camera imaging of the Cur-SN-GM scaffold. (B) BMSCs viability in 3D cultures with Cur-SN-GM, measured by CCK-8 assay. (C) Young’s modulus of Cur-SN-GM, measured using a tensile tester. (D) Degradation profile of SN-GM-GM.(E) Swelling behavior of SN-GM-GM. (F) Representative toluidine blue staining images of BMSCs following chondrogenic induction in conditioned media derived from different sources.Scale bar = 200 μm

Effects of Cur-SN-GM on new cartilage formation and VECs infiltration in vivo

The therapeutic effect of Cur-SN-GM on cartilage repair was evaluated in vivo using a rat model of femoral trochlear cartilage defects (Fig. 8). After a 12-week recovery period, significant differences were observed among the groups in terms of gross morphology, histological architecture, and the expression of specific molecular markers.

Fig. 8.

Effects of Cur-SN-GM on a rat model the femoral trochlear cartilage defect in vivo. (A)Images of repairing cartilage injury in knee joint of rats. (B) HE staining images (C) Toluidine blue staining images. (D) collagen-II immumohistochemical staining images. (E) Representative fluorescent images of CD31 in synovium of joint. (F) relative quantitative analysis of toluidine blue average optical density. (G) Collagen type II average optical density (H)CD31 integrated flourescence density. (I) ICRS scor. Scale bar = 200 μm, *p < 0.05,**p < 0.01 and ***p < 0.001

Gross morphological examination (Fig. 8A) revealed that the defects in the NT (non-treated) group and the SN-GM (blank scaffold) group were only partially repaired, with irregular and uneven surfaces. In contrast, the Cur-SN-GM group demonstrated markedly superior repair outcomes: the defects were completely filled with newly formed tissue, where the boundary between the neocartilage and the surrounding native cartilage was indistinct. The regenerated cartilage was seamlessly integrated with the adjacent tissue, exhibiting a smooth surface and a color and texture similar to those of normal articular cartilage.

Histological analysis provided further insight into the microstructural quality of the repaired tissue (Fig. 8B and H). Toluidine blue staining and type II collagen immunohistochemistry indicated that the optical density in both the NT and SN-GM groups was significantly lower than that in the SH (sham-operated) group (Fig. 8C, D, F, G), confirming that the defects in these groups were primarily filled with disorganized fibrous or fibrocartilaginous tissue. Notably, although the overall tissue volume in the SN-GM group was slightly lower than that in the fibrotic NT group, the mean fluorescence intensity of CD31 immunofluorescence was significantly reduced compared to the NT group (Fig. 8E, H). This suggests that the GelMA scaffold alone, despite lacking bioactivity, provided a biocompatible matrix that effectively suppressed uncontrolled fibrovascular invasion, thereby creating a more favorable microenvironment for tissue regeneration. However, likely due to the absence of bioactive cues, the repair process in this group appeared slower and remained incomplete at the observation time point.

In stark contrast, the Cur-SN-GM group exhibited characteristic features of high-quality hyaline-like cartilage regeneration. The staining intensity for both toluidine blue and type II collagen was significantly higher in this group than in the NT and SN-GM groups (Fig. 8C, D, F, G), indicating abundant proteoglycans and the synthesis of a specific extracellular matrix. The newly formed cartilage seamlessly integrated with the surrounding host tissue, demonstrating robust chondrogenic differentiation capability.

Angiogenesis was assessed via CD31 immunofluorescence staining (Fig. 8E, H). A significant increase in fluorescence intensity was observed in the NT group compared to the SH group, indicating pronounced pathological angiogenesis following defect creation. Conversely, the CD31 fluorescence intensity in the Cur-SN-GM group was significantly lower than that in the NT group, demonstrating that Cur-SN-GM effectively suppressed excessive vascularization upon implantation, which may contribute to maintaining a stable microenvironment conducive to cartilage regeneration.

Discussion

In articular cartilage, the distribution of blood vessels is defined by the tidemark: the upper layer consists of avascular hyaline cartilage, whereas the lower calcified zone contains sparse blood vessels [29]. During cartilage injury, the barrier that prevents angiogenesis is disrupted, leading to vascular infiltration into the non-calcified cartilage layer above the tidemark [12]. Angiogenesis plays a vital role in the early stages of cartilage repair by supplying essential nutrients, oxygen, and progenitor cells required for tissue regeneration [30]. However, when angiogenesis becomes excessive or prolonged, it can disrupt the cartilage microenvironment. Specifically, vascular invasion, particularly into the non-calcified cartilage layer, impedes chondrocyte function through the release of cytokines from newly formed vascular endothelial cells (VECs), ultimately compromising the quality of cartilage repair [11, 31, 32]. While angiogenesis is necessary for initial healing, excessive or sustained angiogenesis leads to the formation of fibrocartilage and impairs the regeneration of high-quality hyaline cartilage. Given the detrimental effects of VEC infiltration and activation on cartilage repair, we adopted a strategy to inhibit VEC infiltration to promote high-quality hyaline cartilage formation.

Curcumin, a natural polyphenolic compound derived from turmeric and ginger, exhibits multiple biological activities, particularly anti-inflammatory properties. It promotes chondrocyte proliferation and improves the cartilage repair microenvironment [33, 34]. Therefore, it has been widely explored for the treatment of arthritis and cartilage injuries [35, 36]. Additionally, studies have demonstrated that Curcumin exerts significant effects on apoptosis induction and anti-angiogenic processes [37]. Specifically, it inhibits VEGF-induced proliferation, migration, and tube formation of VECs and promotes VEC apoptosis in a dose-dependent manner [13]. However, the precise mechanisms underlying curcumin-mediated inhibition of VEC activation remain unclear.

Mitochondrial homeostasis is crucial for cellular functions, including ATP production, metabolic regulation, signalling, and apoptosis [38, 39]. The dynamic balance between mitochondrial fusion, fission, mitophagy, and biogenesis maintains mitochondrial morphology, number, and function [40]. Drp1, a key regulator of mitochondrial fission, preserves mitochondrial morphology and distribution by controlling fission [41, 42]. However, overactivation of Drp1 induces excessive mitochondrial fragmentation, disrupting mitochondrial homeostasis and triggering apoptosis [43]. This excessive mitochondrial fragmentation reduces mitochondrial membrane potential (Δψm) and impairs electron transport chain efficiency, resulting in increased production of ROS [44]. ROS, a class of highly reactive oxygen-derived molecules or ions, plays a critical role in regulating cellular signalling. However, their excessive accumulation can damage cells by inducing apoptosis or accelerating senescence [45, 46]. Furthermore, ROS can promote Drp1 phosphorylation and mitochondrial localisation via oxidative modification or activation of upstream signalling pathways, such as p38 MAPK, thereby exacerbating mitochondrial fission and disrupting mitochondrial homeostasis [47].

We hypothesised that Curcumin induces VEC apoptosis and senescence by promoting Drp1 expression and ROS generation, inhibiting VEC infiltration. However, its clinical application remains limited due to poor water solubility, rapid metabolism, and low stability, all of which impair its bioavailability. To address these challenges, we developed a silk fibroin nanoparticle (SFNP) delivery system (Cur-SN) with high biocompatibility and drug-loading capacity of up to 150 µg/mg, enabling sustained release of Curcumin under neutral PBS conditions. Cur-SN significantly inhibited VEC proliferation and migration. Following Cur-SN treatment, expression of the anti-apoptotic protein BCL2 in VECs decreased, whereas expression of the pro-apoptotic proteins BAX and Caspase-9 increased. Additionally, the senescence markers p21 and p53 were significantly upregulated, indicating that Cur-SN induces apoptosis and senescence in VECs.

The induction of VEC apoptosis by Cur-SN exhibited a clear dose-dependency, yet a closer examination of the cell death dynamics revealed a notable nuance. Interestingly, while our data demonstrate a dose-dependent induction of VEC apoptosis by Cur-SN up to 30 µM, a relative decrease in the proportion of Annexin V+/PI- and Annexin V+/PI + cells was observed at the highest concentration (40 µM), despite a continued decline in overall cell viability (Figs. 2G and 3E). This apparent discrepancy suggests a potential shift in the dominant mode of cell death under conditions of extreme metabolic stress. The profound mitochondrial dysfunction and massive ROS burst induced by high-dose Cur-SN, as evidenced in Fig. 4, may overwhelm the classical apoptotic machinery, potentially triggering alternative, non-apoptotic cell death pathways such as ferroptosis, which is characterized by lipid peroxidation and is not reliably detected by the Annexin V/PI assay. It is also plausible that rapid and extensive plasma membrane disruption under severe toxicity could compromise the binding of both Annexin V and PI. Elucidating the precise mechanism of death at this high concentration, potentially through assessment of lipid peroxidation or other necrosis markers, represents an important direction for future investigation.

Notwithstanding this high-concentration phenomenon, the primary and therapeutically relevant mechanism of Cur-SN action within the effective dose range (e.g., 30µM) is the Drp1/ROS-mediated disruption of mitochondrial homeostasis, leading to coordinated apoptosis and senescence. The results demonstrated that Cur-SN promotes Drp1 and FIS1 expression, leading to mitochondrial fission imbalance. Transmission electron microscopy revealed an increased number of smaller mitochondria in VECs treated with Cur-SN, which was associated with reduced mitochondrial membrane potential and increased ROS generation. When DRP1 inhibitors and ROS scavengers were applied, the effects of Cur-SN on mitochondrial structure, function, and apoptosis/senescence signalling in VECs were significantly attenuated, confirming that Cur-SN inhibits angiogenesis by promoting Drp1/ROS expression. Moreover, in the conditioned medium culture experiments, Cur-SN effectively modulated the secretory behavior of VECs, thereby alleviating the inhibitory effect of VEC supernatant on the chondrogenic differentiation of BMSCs. Collectively, these findings demonstrated that Cur-SN induces apoptosis and senescence in VECs under the dual pressures of high oxidative stress and mitochondrial imbalance, inhibits VEC activation to achieve anti-angiogenesis, and promotes high-quality cartilage defect repair.

Mechanistically, our study unveils the disruption of mitochondrial homeostasis via the Drp1/ROS axis as a pivotal event underpinning these phenotypic effects. This discovery provides a fresh perspective on curcumin’s action, suggesting that targeting mitochondrial dynamics may serve as an upstream integrator that orchestrates several of its well-documented effects. For instance, the elevated ROS levels we observed, while directly damaging to VECs, also function as signaling molecules that influence other pathways. Interestingly, excessive ROS is a known activator of the NF-κB pathway [48, 49]. It is therefore plausible that curcumin-induced ROS accumulation might initially activate NF-κB, while the well-established direct inhibition of IKK by curcumin ultimately supersedes this to yield a net suppressive effect in a time- or context-dependent manner [17]. This complex interplay warrants further investigation. Similarly, mitochondrial dysfunction is a potent inducer of the Nrf2-mediated antioxidant response [50]. The potent anti-angiogenic outcome we observed could thus be interpreted as the net result of a dynamic interplay between pro-death signals (Drp1/ROS-mediated apoptosis/senescence) and potential compensatory survival signals (e.g., Nrf2 activation). The clear dominance of the pro-apoptotic phenotype in our data suggests that Cur-SN treatment effectively pushes VECs beyond a salvageable threshold. Therefore, while curcumin’s roles in NF-κB inhibition and Nrf2 activation are crucial in other contexts, our findings position the mitochondrial fission pathway as the primary and initiating mechanism in its anti-angiogenic action within our cartilage repair model.

To evaluate the in vivo effects, we established a rat knee cartilage defect model and implanted a GM hydrogel loaded with BMSCs and Cur-SN into the defect site. Three months post-implantation, the Cur-SN group exhibited dense, well-structured hyaline cartilage-like tissue with good integration into the surrounding tissues. Consistent with the in vitro findings, the NT group showed significant VEC (CD31+) activation in the newly formed tissue, whereas Cur-SN treatment significantly reduced VEC activation at the repair site. These results indicated that Cur-SN effectively inhibited vascular invasion at cartilage defect sites and enhanced cartilage repair quality. The consistency between the in vitro and in vivo results highlighted the biocompatibility and anti-angiogenic efficacy of the Cur-SN-GM scaffold, demonstrating its potential for clinical applications.

Conclusion

In summary, we successfully designed, prepared, and developed a curcumin-loaded silk fibroin nanoparticle (Cur-SN) delivery system. In vitro experiments elucidated the mechanism by which Cur-SN disrupts mitochondrial homeostasis and function, primarily through the upregulation of Drp1 and ROS. This upregulation, in turn, promoted VEC apoptosis and senescence while inhibiting VEC infiltration. Additionally, Cur-SN demonstrated superior efficacy in promoting the chondrogenic differentiation of BMSCs. In vivo experiments using a rat knee cartilage defect model further confirmed the effectiveness of Cur-SN in inhibiting VEC infiltration and activation, thereby enhancing cartilage repair. Collectively, our findings emphasise the potential of Cur-loaded SN composite hydrogels in suppressing VEC activation and promoting articular cartilage injury repair through regulating mitochondrial homeostasis via the Drp1/ROS pathway, which contributes to cartilage regeneration. Overall, this study provides valuable insights into future advancements in tissue engineering and regenerative medicine.

Author contributions

Zitong Zhao, Jingyue Fu, and Jun Han contributed equally to this work. Zitong Zhao was responsible for Conceptualization, Investigation, and Writing – Original Draft; Jingyue Fu performed Formal Analysis, Data Curation, and Visualization; Jun Han contributed to Resources and Funding Acquisition; Yalan Pan carried out Investigation and Validation; Baojun Xue contributed to Methodology and Resources; Xiaoxian Sun assisted in Investigation and Visualization; Yong Ma participated in Formal Analysis and Supervision; Bin Li was involved in Software and Validation; Kan Chen provided Resources; Zining Li helped in Investigation and Validation; Zhongqing Liang Funding and provided Supervision; Xin Zhou acquired Funding and provided Supervision; Yang Guo contributed to Conceptualization, Supervision, and Funding Acquisition; Pengcheng Tu led Conceptualization, Funding Acquisition, and Supervision. All authors reviewed and approved the final manuscript.

Data availability

The authors declare that all data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials.

Declarations

Competing interests

The authors declare no competing interests.