3D-printed bilayer hydrogel scaffolds incorporating HAP and KGN@Lip for osteochondral regeneration

Abstract

Osteochondral defects require the simultaneous regeneration of cartilage and subchondral bone, posing a major challenge for current biomaterial strategies. Here, we present a radially oriented 3D-printed bilayer hydrogel scaffold composed of a mechanically reinforced GelMA/HAP osteogenic layer and a compliant, KGN@Lip-loaded GelMA chondrogenic layer, to achieve coordinated osteochondral repair. Structural and physicochemical characterization confirmed the formation of a functionally graded architecture with uniform HAP and KGN@Lip distribution, while mechanical, rheological, swelling, and degradation analyses revealed a layer-dependent gradient that resembles certain structural features of the native osteochondral interface. In vitro, the GelMA/HAP layer promoted robust osteogenic differentiation through ion-mediated activation of osteogenic pathways, whereas the GelMA/KGN@Lip layer sustained chondrogenic stimulation and markedly enhanced SOX9, Col2a1, and Acan expression. The radially oriented microchannels provided an interconnected structure that may support nutrient transport and cell infiltration. In vivo implantation demonstrated substantial subchondral bone regeneration and the formation of cartilage-like tissue with improved matrix organization, along with enhanced integration at the osteochondral interface, confirming the scaffold's ability to provide spatially coordinated biochemical and structural cues. Overall, this anisotropic bilayer hydrogel scaffold offers a promising strategy for integrated and functional osteochondral regeneration.

---

1 Introduction

Osteochondral defects, resulting from trauma, osteoarthritis, or degenerative diseases, involve concurrent damage to the articular cartilage and subchondral bone.¹ Owing to the avascular nature of cartilage² and the structural characteristics of subchondral bone,³ these lesions exhibit extremely limited self-regeneration capacity. Disruption of the osteochondral interface further leads to mechanical imbalance and progressive joint degeneration.⁴ Conventional treatments, including microfracture,⁵ autograft/allograft transplantation,⁶ and chondrocyte implantation,⁷ can provide partial repair but usually fail to restore the integrated structure and long-term function of osteochondral tissue. Hence, the development of biomimetic scaffolds capable of supporting both cartilage and bone regeneration is of great importance.⁸ Ideal scaffolds should replicate the hierarchical architecture and mechanical anisotropy of native tissue, guide directional cell migration, and provide localized biochemical cues to induce osteogenic and chondrogenic differentiation.⁹ Three-dimensional (3D) printing enables precise control over scaffold composition and structure, offering a promising strategy for stratified and bioactive systems for efficient osteochondral repair.¹⁰

In recent years, tissue engineering scaffolds based on hydrogels have shown tremendous potential for osteochondral defect repair.¹¹ Hydrogels provide a hydrated, ECM-like microenvironment that supports cell adhesion, proliferation, and differentiation,¹² while enabling precise modulation of the local microenvironment through the incorporation of bioactive molecules, growth factors, or small-molecule drugs to guide specific cellular behaviors.¹³ The advent of 3D printing technology has further expanded the design capability of hydrogel scaffolds, allowing precise control over macrostructure, porosity, microarchitecture, and spatial composition.¹⁴ Such control is particularly crucial for mimicking the native osteochondral organization, which consists of a mechanically robust subchondral bone layer and a softer cartilage layer.¹⁵ Such designs allow different regions within a single scaffold to support bone- and cartilage-related tissue formation.¹⁶

Gelatin methacryloyl (GelMA) hydrogel has been widely applied in bone and cartilage tissue engineering owing to its excellent biocompatibility, tunable mechanical properties, and abundant cell adhesion motifs.¹⁷ GelMA can mimic the physical and biochemical characteristics of the native extracellular matrix (ECM), providing an ideal microenvironment for cell adhesion, migration, and differentiation.¹⁸ When combined with hydroxyapatite (HAP), the scaffold exhibits significantly enhanced osteoinductivity and mechanical strength. HAP can release Ca²⁺ and PO₄³⁻ ions, which are associated with osteogenic activity and bone matrix deposition,^19,20 while also improving protein adsorption and cell adhesion.²¹ In addition, its rigid structure contributes to mechanical stability and supports bone tissue formation.²²

In contrast, cartilage regeneration requires a softer microenvironment with strong biochemical cues.²³ Kartogenin (KGN) is a potent small-molecule chondrogenic agent (EC50 ≈ 100 nM) that promotes chondrogenesis by binding to filamin A and activating the CBFβ–RUNX1 signaling pathway.²⁴ KGN also enhances collagen synthesis and facilitates tendon–bone junction (TBJ) healing.²⁵ However, its hydrophobicity and rapid clearance in physiological environments limit its direct application in vivo.²⁶ To overcome these limitations, KGN was encapsulated into liposomes (KGN@Lip) to achieve controlled and sustained release.²⁷ Liposomal encapsulation improves the dispersibility and bio-stability of KGN while maintaining its bioactivity, enabling prolonged chondrogenic stimulation.²⁸ The combination of GelMA and KGN@Lip provides a suitable microenvironment for cartilage-related tissue formation.²⁹

Inspired by the hierarchical structure of the native osteochondral interface, a 3D printed bilayer hydrogel scaffold was developed. The lower GelMA/HAP layer serves as the bone-contacting region, providing mechanical support, while the upper GelMA/KGN@Lip layer serves as the cartilage-contacting region. The scaffold was designed with distinct structural characteristics in each layer to partially mimic native osteochondral architecture.³⁰ The lower GelMA/HAP layer was fabricated as a relatively compact structure to emulate the mineralized subchondral bone, providing mechanical support and serving as an osteoinductive environment that facilitates bone tissue ingrowth and integration.³¹ In contrast, a radially oriented channel structure was introduced into the upper GelMA/KGN@Lip layer, resembling the porous and hydrated nature of cartilage tissue. This anisotropic architecture provides an open and interconnected structure that may facilitate nutrient transport and cell infiltration from the surrounding cartilage and bone marrow toward the defect center.³² The radial channels provide an interconnected structure that may influence cell distribution at the osteochondral interface. In vitro and in vivo results showed that the scaffold supported cell viability and tissue formation. In a rabbit osteochondral defect model, CT and histological analyses indicated improved bone regeneration and tissue integration, suggesting its potential for osteochondral repair.

2 Materials and methods

2.1 Materials and reagents

Gelatin (95%) was purchased from Sigma (Shanghai, China). Methacrylic anhydride (98%) and hydroxyapatite (60–80 nm) were obtained from Macklin (Shanghai, China). Kartogenin (98.45%) was purchased from MCE. Hydrogenated soy phosphatidylcholine (HSPC) (98.5%) and cholesterol (CHO-HP) (95%) were obtained from AVT (Shanghai, China). Lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) was purchased from Shanghai YinChang New Materials Co., Ltd (Shanghai, China). Lemon yellow (purity ≥95%) was purchased from Aladdin (Shanghai, China). Collagenase type II was obtained from Sigma-Aldrich (Shanghai, China). Fetal bovine serum (FBS), penicillin, trypsin–EDTA (0.25%, with phenol red), and α-MEM were purchased from Gibco (Shanghai, China). Cell Counting Kit-8 (CCK-8) was obtained from Beyotime Biotechnology (Shanghai, China). Alizarin red S staining solution and hematoxylin–eosin (H&E) staining kit were purchased from Servicebio (Wuhan, China). TRAP/ALP Stain Kit was obtained from WAKO (Japan). Rhodamine-labeled phalloidin and cell viability (live/dead) assay kits were purchased from KeyGEN BioTECH (Jiangsu, China). Xylene, absolute ethanol, hydrochloric acid, ammonia solution, neutral balsam, and hydrogen peroxide were obtained from Sinopharm Chemical Reagent Co., Ltd (China). Antigen retrieval solutions including EDTA (pH 9.0), EDTA (pH 8.0), and citrate buffer (pH 6.0) were purchased from Servicebio (Wuhan, China). Bovine serum albumin (BSA) was obtained from Solarbio (Wuhan, China), and normal rabbit serum was purchased from Boster Biological Technology (Shanghai, China).

2.2 GelMA preparation

First, weigh 5.0 g of gelatin and dissolve it in 50 mL of PBS buffer solution under magnetic stirring at 50 °C in a water bath until completely transparent. Slowly add 3 mL of methacrylic anhydride to the gelatin solution dropwise, then stir magnetically at 700 rpm for 1 h while maintaining the reaction temperature at 50 °C. The reaction mixture will contain numerous oily droplets. After the reaction, transfer the solution to a cellulose dialysis bag with a molecular weight cut-off of 3500 Da. Dialyze in deionized water at 40 °C for 3 days to remove byproducts. Finally, collect the dialyzed solution, centrifuge at 5000 rpm for 10 minutes to remove the precipitate, collect the supernatant, and freeze-dry at −80 °C for 48 hours to obtain the final product, GelMA.

2.3 Preparation of Lip@KGN

Weigh 7.4 mg HSPC and 16.5 mg CHO-HP, dissolve in 1 mL ethanol, and disperse by ultrasonication. After complete dissolution, inject into 9 mL of vigorously stirred KGN ethanol solution (1 mg mL⁻¹). Stir for 15 min, then sonicate for 15 min (power 30%, 5 s on, 5 s off) to obtain the liposome solution.

2.4 Preparation of 3D-printed hydrogels

First, prepare the photoinitiator solution by dissolving 1 mg of LAP (photoinitiator) and 0.6 mg of lemon yellow (light absorber) in 1 mL of deionized water. Vortex for 30 seconds to ensure thorough mixing. Dissolve 5 mg HAP and 0.1 g GelMA in the above solution to obtain the first-layer ink. Take 1 mL of the liposome solution, add 1 mg LAP and 0.6 mg lemon yellow, vortex for 30 seconds, mix thoroughly, and obtain the photoinitiated liposome solution. Dissolve 0.1 g GelMA in the above solution to obtain the second layer ink. Then, first pour the prepared first layer bioink into the printer, select the print model and model size, and print the shape (light intensity: 17 mW cm⁻², exposure time: 20 s, base layer count: 1, base layer exposure time: 28 s). Continue printing the second layer of hydrogel onto this base layer. Pour the prepared second layer bioink into the printer, select the printing model and size, and print the structure (light intensity: 17 mW cm⁻², exposure time: 20 s, base layer count: 1, base layer exposure time: 27 s). Finally, wash the printed hydrogel 2–3 times with sterile PBS to obtain the hydrogel scaffold.

2.5 Chemical construction and morphology of hydrogel scaffolds

2.5.1 Chemical construction of GelMA. Fourier transform infrared (FT-IR) spectroscopy was employed to investigate the bonding characteristics of the synthesized GelMA. Spectra were recorded using an FTIR spectrometer (Nicolet iS20, Thermo Scientific, USA) within the range of 4000–400 cm⁻¹. Additionally, the chemical structure of GelMA was characterized by proton nuclear magnetic resonance (¹H NMR) spectroscopy using a BRUKER AVANCE 400 spectrometer (BRUKER Co. Ltd, Germany) at room temperature. The resulting spectra were analyzed using MestReNova software.

2.5.2 Dynamic light scattering (DLS) and transmission electron microscopy (TEM) observation. DLS was used to accurately measure the size and zeta potential difference of the liposomes (Zetasizer Nano ZS, Malvern, UK). The shape of the liposomes was observed using a TEM (H-800, Hitachi, Japan).

2.5.3 Chemical construction of Lip@KGN. X-ray diffraction (XRD; SmartLab SE, Rigaku, Japan) was applied to characterize the crystal phase composition of HAP, scanning from 10° to 80° using Cu-Kα radiation (λ = 1.5406 Å) at 40 kV/40 mA.

2.5.4 Scanning electron microscope (SEM) observation. The microstructures of the 3D printed hydrogels were characterized using a SEM (HITACHI SU8100, Hitachi, Japan). Prior to examination, the hydrogel scaffolds were dried, and their surfaces were coated with a thin layer of gold for 30 seconds.

2.5.5. Loading rate and encapsulation rate testing of KGN. To determine the drug loading parameters, the Lip@KGN liposome solution was centrifuged at 15 000 rpm for 30 min at 4 °C to ensure the complete separation of encapsulated liposomes from the free drug. The resulting supernatant, containing the unencapsulated (free) KGN, was carefully collected. The concentration of free KGN in the supernatant was then quantified by measuring the absorbance at 280 nm using a UV-Vis spectrophotometer, with reference to a standard calibration curve of KGN. The encapsulation efficiency (EE) and drug loading (DL) were calculated according to the following formulas:

The encapsulation rate (%) = (the amount of drug loaded/the amount of drug added) × 100%

Drug loading (%) = (total mass of loaded drugs/total mass of liposome) × 100%

2.6 Physical properties of hydrogel scaffolds

2.6.1 Rheology tests. The rheological behavior of the hydrogels was characterized on a rotational rheometer (Kinexus Pro+, Malvern, UK) using a parallel plate geometry (25 mm diameter). To determine the viscoelastic response, the storage modulus (G′) and loss modulus (G″) were monitored over an angular frequency range of 0.1–10 rad s⁻¹ at room temperature. These data were used to delineate the linear viscoelastic region and to evaluate the network stability of the hydrogels under oscillatory shear.

2.6.2 Compressive modulus. Hydrogel precursor solutions (600 µL) were poured into the wells of a 48-well plate, photo-crosslinked under UV irradiation, and demolded into uniform cylindrical specimens (diameter: 11 mm). The samples were then subjected to uniaxial compression using a universal testing machine. Before testing, the compression plates were carefully adjusted to ensure parallel contact with the sample surfaces. Compression was applied at a constant rate of 0.05 mm s⁻¹ until 70% strain, while the load and displacement were continuously recorded. The compressive stress was determined from the equation:

Compression stress = F/S (Pa)

where F is the recorded pressure value and S is the cross-sectional area of the hydrogel sample.

2.6.3 Swelling properties test. To assess the water absorption capacity, pre-weighed hydrogels (W₀) were immersed in 5 mL of deionized water at ambient temperature. At specified intervals, the samples were removed, gently wiped with filter paper to eliminate surface moisture, and weighed again (W_t). The degree of swelling was calculated according to the following equation:

2.6.4 Degradation properties test. To investigate degradation kinetics, freeze-dried hydrogel samples were weighed (W₁) and subsequently incubated in phosphate-buffered saline (PBS, pH 7.4) or a 2.5 U mL⁻¹ solution of collagenase II at 37 °C under gentle agitation (70 rpm). At predetermined time points (3, 7, 10, 14, 21, 28, 35, and 42 days or 6, 12, 24, 48, 72, and 96 hours), the hydrogels were retrieved, washed with ultrapure water to remove residual salts, and freeze-dried again. The remaining dry weight (W_t) was recorded, and the mass retention rate (W_r) was determined using the following formula:

W_r (%) = W_t/W₁ × 100%

2.6.5 In vitro release of KGN. To prepare KGN-loaded hydrogels, 9 mg of KGN was dissolved in 100 µL of ethanol to obtain a stock solution. Separately, 0.1 g of GelMA, 1 mg of LAP, and 0.6 mg of tartrazine were dissolved in 0.9 mL of deionized water. After complete dissolution, 0.1 mL of the KGN stock solution was added to the polymer precursor and mixed thoroughly. The resulting mixture was photocrosslinked to form KGN-containing hydrogel scaffolds, following the same procedure used for Lip@KGN-loaded constructs. Both types of scaffolds were immersed in 10 mL of sterile PBS and incubated at 37 °C. At scheduled intervals (1, 3, 7, 10, 14, 21, and 28 days), 1 mL of supernatant was withdrawn and replenished with an equal volume of fresh PBS. The amount of KGN released at each time point was quantified using UV-Vis spectrophotometry, and cumulative release profiles were subsequently plotted.

2.7 In vitro cytocompatibility evaluation

2.7.1 Cytotoxicity evaluation of hydrogel scaffolds. The sterilized hydrogel was extracted with α-MEM complete medium at a mass-to-volume ratio of 1 : 10 for 24 h. Bone marrow mesenchymal stem cells (BMSCs) were digested with 0.25% trypsin, resuspended, and seeded into 48-well plates at a density of 2 × 10⁴ cells per well. After 24 h of culture, the original medium was replaced with the hydrogel extract. Each experimental group included at least five replicates, and the medium was refreshed every 24 h. Cell viability was assessed at three time points: 24, 72, and 120 h. Cell survival was quantitatively evaluated using the CCK-8 assay. After culturing for 1, 3 and 5 days, 100 µL of CCK-8 working solution was added to each well, followed by incubation at 37 °C in a humidified incubator containing 5% CO₂ for 30 min. The absorbance at 450 nm was then measured using a microplate reader, and cell viability was calculated according to the following formula:

2.7.2 The live/dead fluorescence assay. BMSCs at passage 3 were digested with 0.05% trypsin, resuspended in α-MEM complete medium, and seeded into 48-well plates at a density of 5 × 10³ cells per well. Cells were cultured at 37 °C in a humidified incubator containing 5% CO₂ for 24 h to allow adherence. Subsequently, the wells were washed three times with PBS and incubated with hydrogel extract for 1, 3, and 5 days. At each time point, the culture medium was aspirated and discarded, and the cells were washed three times with PBS. Live/dead staining solution (2 µM calcein AM and 8 µM propidium iodide) was then added, and the cells were incubated at room temperature for 20 min. After a single PBS wash, an anti-fade reagent was added, and images were immediately captured using a fluorescence microscope to assess cell viability. Calcein AM labels live cells with green fluorescence, whereas propidium iodide (PI) labels dead cells with red fluorescence.

2.7.3 Cytoskeleton staining experiment. BMSCs (p3 generation) were digested with 0.05% trypsin, resuspended in α-MEM complete medium, seeded in 48-well plates at a density of 5000 per well/cell, and cultured in a 37 °C constant temperature carbon dioxide incubator (containing 5% CO₂) for 24 hours until the cells adhered. Subsequently, wash three times with PBS and continue to culture for 24 hours with the hydrogel extract. The culture medium was aspirated and discarded, washed three times with PBS, stained with rhodamine B labeled with piconidin, washed with PBS, and then images were captured with a fluorescence microscope.

2.8 Evaluation of osteogenic induction differentiation of BMSCs in vitro

2.8.1 Alizarin red staining experiment. The sterilized hydrogel was extracted in α-MEM complete medium at a mass ratio of 1 : 10 for 24 h to obtain hydrogel extracts. BMSCs (P3) were digested with 0.25% trypsin, resuspended, and seeded into 24-well plates at a density of 2 × 10⁴ cells per mL per well. After 24 h of incubation, the original medium was removed and replaced with the hydrogel extract, with medium changes performed every 24 h. After 21 days of culture, the medium was aspirated, and the wells were washed three times with PBS and fixed with 4% paraformaldehyde for 30 min. Subsequently, 2 mL of Alizarin red S solution was added to each well and incubated at room temperature for 5 min. The staining solution was discarded, and the wells were rinsed with PBS 2–3 times to remove excess dye. After complete air-drying, images of the Alizarin red S staining were captured.

2.8.2 ALP staining experiment. The cell culture procedure was identical to that described in Section 2.8.1. After 10 days of culture in the wells, the culture medium was aspirated and discarded. The wells were washed three times with PBS and fixed with 4% paraformaldehyde for 15 minutes, followed by removal of the fixative. After three additional PBS washes, ethanol/acetone permeabilization solution was added on ice for 1 minute. The solution was then aspirated, and the wells were washed three times with PBS. ALP working solution was added and incubated at 37 °C for 15–45 minutes. The excess solution was removed, and the samples were rinsed with distilled and deionized water. After air-drying, the staining results were observed under a microscope and imaged.

2.8.3 Osteogenic induction RT-qPCR gene detection. After 21 days of culture, the medium in the wells was aspirated and discarded. Then, PBS was added for washing three times. The RNA was lysed and extracted with Trizol to obtain the RNA. After detecting the concentration, the RNA was added to the gDNA adsorption column. The RNA without DNA was collected by centrifugation, denatured at 65 °C, cooled on ice and subjected to reverse transcription reaction to obtain cDNA. Set up the reaction program of the PCR instrument, combine the cDNA with the reaction solution, and then conduct on-machine detection to obtain the melting curve and amplification curve. Export and analyze the data. The primer sequences used for RT-qPCR are listed in Table 1.

Table 1 Primer sequences of q-PCR

Primer name	Forward primer	Reverse primer
Col-1	GCTCAAGTCGCTGAACAACC	AGTCTCCGCTCTTCCACTCT
OCN	CAAGCAGGAGGGCAGTAAGG	GTCCTGGAAGCCAATGTGGT
RUNX2	ATGGTACTTCGTCAGCGTCC	ATAGCGTGCTGCCATTCGAG
GAPDH	GCATCCTGGGCTACACTGAG	CCACCACCCTGTTGCTGTAG

---

2.9 Evaluation of chondrogenic induction and differentiation of BMSCs in vitro

2.9.1 Alcian blue staining experiment. BMSCs (P3) were seeded into 6-well plates pre-lined with sterilized cell slides and allowed to adhere overnight. After removal of the culture medium, hydrogel extracts from the respective groups were added and refreshed every 2–3 days for a 21 day induction period. At the end of induction, the cells were washed twice with PBS and fixed with 4% paraformaldehyde for 15 minutes. After discarding the fixative and washing twice with PBS, Alcian blue staining solution was added to fully cover the bottom of each well and incubated for 10–15 minutes. The staining solution was removed, and the wells were gently rinsed with running water. Subsequently, nuclear fast red solution was added and incubated for 3 minutes, followed by another rinse to remove excess dye. The samples were then dehydrated through graded ethanol solutions, air-dried, and examined under a microscope for imaging.

2.9.2 Quantitative detection of GAG. BMSCs (P3) were seeded into 6-well plates pre-lined with sterilized cell slides and allowed to adhere overnight. After removal of the medium, material extracts from the corresponding groups were added and refreshed every 2–3 days for a 21 day induction period. At the end of induction, the cell culture supernatant was collected and centrifuged for subsequent assays. Experimental procedures, including sample loading, incubation, reagent preparation, washing, enzyme addition, secondary incubation, color development, termination, and absorbance measurement at 450 nm, were performed according to the manufacturer's instructions. The obtained values were used for subsequent data analysis.

2.9.3 PCR genetic testing. After 21 days of culture, the medium in the wells was aspirated and discarded, followed by three washes with PBS. Total RNA was extracted using TRIzol reagent. After quantification, the RNA solution was applied to a gDNA removal column, and DNA-free RNA was collected by centrifugation. The purified RNA was denatured at 65 °C, cooled on ice, and subjected to reverse transcription to obtain cDNA.

The PCR program was set according to the instrument protocol, and cDNA was mixed with the reaction mixture for quantitative PCR analysis. Melting curves and amplification curves were recorded, and the resulting data were exported for subsequent analysis. The primer sequences used for RT-qPCR are listed in Table 2.

Table 2 Primer sequences of q-PCR

Primer name	Forward primer	Reverse primer
SOX9	GCCACCGAACAGACTCACAT	CTGTTTTGGGAGTGGTGGGT
Col2a1	GCTCAAGTCGCTGAACAACC	AGTCTCCGCTCTTCCACTCT
Acan	CCCAACAGTCTCCCTTGTGG	CCTGAACCACTGACGCTGAT
GAPDH	GCATCCTGGGCTACACTGAG	CCACCACCCTGTTGCTGTAG

---

2.10 Animal experiment

All animal experiments were reviewed and approved by the Ethics Committee of Hainan Medical University (Approval No. HYALL-2026-101). All procedures were performed in accordance with the guidelines for the care and use of laboratory animals and complied with relevant institutional and national regulations. All efforts were made to minimize animal suffering and reduce the number of animals used.

2.10.1 Modeling method for full-thickness femoral cartilage defects in New Zealand white rabbits. A total of 24 male New Zealand white rabbits (2.5–3.0 kg) were used in this study. The animals were randomly divided into 4 groups (n = 6), and anesthetics were administered by intravenous injection of 3% pentobarbital sodium through the auricular vein. The skin at the knee joint of the leg was prepared using a shaving machine. The skin was disinfected three times with alcohol and iodophor. Then, a 1.5–2 cm longitudinal incision was made at the knee joint of the leg with a sterile scalpel. The muscles and fascia were bluntly separated to expose the position of the femoral cartilage pulley. Using a sterilized mechanical electric drill, a circular defect with a diameter of 5 mm and a depth of 4 mm was drilled at the position of the cartilage pulley. After hemostasis, materials from different groups were filled into the injured area. Then, the muscles and skin were sutured in sequence. The surgical incision was disinfected. It was waited for the animal to wake up, and its condition was observed. Antibiotics should be injected into the animals within 1 week after the operation to prevent infection. The animals were euthanized at the designated time, and the samples were taken from the femurs and fixed in 4% paraformaldehyde fixative. To minimize potential bias, all downstream outcome assessments, including micro-CT imaging and histological analyses, were performed by independent investigators who were blinded to the specific group assignments.

2.10.2 Femoral CT scan. The dissected and fixed femurs of New Zealand white rabbits were placed in the MCT-Sharp fixator. The voltage of the scanner is set to 70 kV, the power is 7 W, 4 frames are superimposed, the angle gain is 0.72 degrees, the exposure time is 100 ms, and the scanning is completed by rotating once.

3D reconstruction and quantitative analysis were conducted between the scanned femoral defect sites. Histological and morphological analyses were conducted on the cartilage position, with calculation indicators including BMD (bone mineral density), BV/TV (bone volume/total volume), Tb. Th (trabecular thickness), etc.

2.10.3 Pathological experiment. Tissue samples were fixed in 4% paraformaldehyde, decalcified for 1–2 months, dehydrated through a graded ethanol series, and embedded in paraffin. Sections of 4–5 µm thickness were cut, mounted on glass slides, and baked prior to staining. For hematoxylin and eosin (H&E) staining, sections were deparaffinized, rehydrated, and stained with hematoxylin for nuclear visualization, differentiated in acid alcohol, blued in ammonia water, and counterstained with eosin for cytoplasm, followed by dehydration, clearing, and mounting with neutral resin, allowing evaluation of general tissue morphology and cellular distribution. For Masson's trichrome staining, deparaffinized and rehydrated sections were stained sequentially with hematoxylin for nuclei, acid fuchsin for cytoplasm, and aniline blue for collagen fibers, enabling assessment of collagen deposition and extracellular matrix maturation. For Alcian blue and periodic acid-Schiff (PAS) staining, deparaffinized and rehydrated sections were treated with Alcian blue (pH 2.5) to visualize acidic glycosaminoglycans and periodic acid-Schiff reagent for neutral proteoglycans, followed by counterstaining to differentiate the cartilage matrix and evaluate the maturity of the regenerated tissue. Similarly, for toluidine blue staining, sections were incubated with toluidine blue solution to highlight the distribution of chondrocytes and the metachromatic cartilaginous matrix.

2.10.4 Histological and immunohistochemical analysis. For immunohistochemical (IHC) and immunofluorescence (IF) labeling, sections underwent heat-induced antigen retrieval using EDTA (pH 8.0/9.0) or citrate (pH 6.0) buffers. After blocking endogenous peroxidase with 3% H₂O₂ and non-specific binding with 5% BSA or normal rabbit serum, the sections were incubated with primary antibodies against type I collagen (Col I), type III collagen (Col III), aggrecan (Acan), and osteocalcin (OCN). For IHC (Col I), signals were developed using DAB chromogen for matrix localization and counterstained with hematoxylin. For IF (Col III, Acan, and OCN), sections were incubated with Alexa Fluor-conjugated secondary antibodies and DAPI for nuclear visualization, enabling the evaluation of specific protein expression and the spatial coupling of osteogenic and chondrogenic processes. All stained sections were mounted with neutral resin and imaged to facilitate quantitative analysis of positive areas and fluorescence intensities via ImageJ.

3 Results and discussion

3.1 Characterization of HAP and KGN@Lip nanoparticles

To address the structural and functional differences between the osteogenic and chondrogenic regions in osteochondral repair, a bilayered hydrogel scaffold was designed with distinct compositions and architectures in each layer. The lower layer was constructed using GelMA incorporated with HAP nanoparticles to promote osteogenesis, while the upper layer consisted of GelMA loaded with KGN@Lip to induce chondrogenic differentiation. This bilayer configuration is intended to mimic the native osteochondral interface by delivering distinct biochemical signals and structural features in each layer.

The crystalline structure and phase purity of the synthesized HAP nanoparticles were verified by XRD. As shown in Fig. 1A, the diffraction peaks at 2θ values of 25.9°, 31.7°, 39.8°, 46.8°, 49.7°, and 53.2° corresponded well to the characteristic reflections of HAP, confirming its good crystallinity and defined crystal phase. TEM revealed that the HAP nanoparticles exhibited a uniform rod-like morphology (Fig. 1B), which facilitates their uniform dispersion within the GelMA matrix and enhances osteoinductive performance.

Fig. 1 Characterization of KGN@Lip nanoparticles. (A) XRD pattern of HAP. (B) Particle size distribution histogram of blank Lip. (C) Particle size distribution histogram of Lip@KGN. (D) TEM image of HAP. (E) TEM image of blank Lip. (F) TEM image of Lip@KGN.

For the chondrogenic layer, liposomes were employed as carriers for KGN not only to enable sustained release but also to improve its aqueous dispersibility, as KGN is a poorly water-soluble small molecule. By measuring the absorbance of the supernatant on the Lip@KGN after loading KGN, it was calculated that the drug loading of KGN was 14.70 ± 0.27%, and the encapsulation rate was 87.83 ± 0.16% (0.147 mg of drug KGN per milligram Lip@KGN). DLS analysis (Fig. 1C and D) showed that Lip had an average hydrodynamic diameter of 388.99 ± 3.90 nm, which slightly increased to 503.93 ± 4.45 nm after Lip@KGN. This moderate size increase indicated successful drug loading without compromising colloidal stability. TEM images (Fig. 1E and F) confirmed their spherical morphology and intact bilayer structure.

3.2 Structural evaluation of hydrogels

The successful chemical modification of gelatin and the formation of GelMA were confirmed using FTIR spectroscopy and ¹H NMR analysis. FTIR spectra (Fig. 2A) showed characteristic amide absorption bands in both Gel and GelMA, including the amide I band at 1627.1 cm⁻¹ corresponding to C=O stretching, the amide II band at 1523.0 cm⁻¹ resulting from coupled N–H bending and C–H stretching vibrations, and the amide III band at 1232.6 cm⁻¹ associated with N–H bending and C–N stretching. Remarkably, the intensities of all three amide peaks in GelMA were significantly higher than those observed in pristine Gel, indicating the introduction of methacrylate groups into the gelatin backbone and the formation of new amide linkages. Complementary ¹H NMR analysis (Fig. 2B) revealed two new proton resonance peaks at 5.3 and 5.6 ppm, corresponding to the vinyl protons (=CH₂) of the methacrylate moieties. Together, these findings confirmed the successful methacrylation of gelatin, ensuring that GelMA could undergo photocrosslinking to form stable hydrogel networks suitable for tissue engineering applications.

Fig. 2 Results of structural and property characterization of hydrogels. (A) FTIR spectra of gelatin and GelMA. (B) ¹H NMR spectra of gelatin and GelMA. (C) Stress–strain curves of the hydrogels. Macroscopic view and side view of the bilayer hydrogel scaffold (D) and side view of the radially oriented channels in the bilayer scaffold (E). (F) The printed model of the hydrogel. SEM-EDS mapping of the upper hydrogel layer (G) and the lower hydrogel layer (H).

To evaluate whether the 3D printed scaffolds faithfully reproduced the designed bilayer architecture and to assess their suitability for supporting cell infiltration, both macroscopic and microscopic structural analyses were performed (Fig. 2D–F). The digital computer-aided design (CAD) models (Fig. 2F) illustrate the precisely engineered bilayer configuration, featuring a top layer with radially oriented channels and a solid, non-porous base layer. Bright-field images revealed that the printed scaffolds accurately replicated these digital designs (Fig. 2D). Specifically, the upper layer formed a dual radial channel network with interconnected pathways, while the lower layer consisted of a solid cylindrical base without channels. The continuous radial channels in the upper layer provide an interconnected structure that may be favorable for cell infiltration from surrounding bone tissue into the defect site, although this effect requires further investigation. Incorporation of HAP into the bioink on the lower layer resulted in a slightly whitish appearance, confirming successful mineral integration. Inverted microscopy and side view observations further demonstrated the integrity of the bilayered structure and the continuity of interconnected pores in both layers, consistent with the preset digital parameters (Fig. 2E).

In order to assess both the micro-porous architecture and the spatial distribution of incorporated HAP and KGN@Lip, the scaffold was analyzed by SEM and SEM–EDS elemental mapping. SEM images revealed uniform and continuous micro-porous structures, providing a favorable microenvironment for cell attachment and proliferation. The lower layer containing HAP exhibited denser structures and numerous particulate features compared to the upper layer, consistent with the incorporation of mineral particles. SEM–EDS mapping confirmed the presence of calcium (Ca) and phosphorus (P) in the lower layer, while the upper layer loaded with KGN@Lip exhibited higher signals of carbon (C), nitrogen (N), and oxygen (O) (Fig. 2G and H).

3.3 Mechanical and physicochemical characterization of dual-component hydrogel scaffolds

Building upon the bilayer design introduced above, the mechanical and structural properties of the GelMA-based hydrogel were further analyzed to verify whether the distinct nanocompositions endow the scaffold with a gradient suitable for osteochondral repair. Owing to the distinct functional roles of cartilage and subchondral bone, the two layers were engineered to display graded mechanical strength, viscoelasticity, and degradation profiles that reflect the distinct structural features of the osteochondral interface. Compression testing (Fig. 2C) revealed distinct mechanical characteristics between the two layers. The GelMA/HAP hydrogel exhibited a significantly higher compressive strength (35.13 kPa at 40% strain) than the GelMA/Lip hydrogel (25.35 kPa), demonstrating the reinforcing effect of inorganic HAP. The rigid HAP nanoparticles acted as nanoscale fillers and crosslinking anchors, enhancing polymer chain aggregation and forming a compact 3D network that resisted deformation. In contrast, the GelMA/Lip hydrogel, with its flexible liposomal inclusions, presented a softer structure favorable for cellular encapsulation and cartilage tissue remodeling. Such stiffness variation reflects a relative difference between the two layers, which is consistent with the structural organization of the osteochondral interface. Although the absolute mechanical properties of the hydrogels remain lower than those of native tissues, the observed difference in stiffness between the layers aligns with the distinct mechanical characteristics of subchondral bone and cartilage.

Rheological characterization further supported this gradient design (Fig. 3A and B). In both frequency and time sweep tests, G′ consistently exceeded G″, confirming stable gel formation. The GelMA/HAP hydrogel exhibited higher storage and loss moduli compared to GelMA/Lip, reflecting enhanced elasticity and viscoelastic resistance. The presence of HAP nanoparticles restricts molecular motion and improves the mechanical robustness of the lower layer, while the upper layer containing Lip remains with more flexible properties that ensure both printability and functional integration between the two regions.

Fig. 3 Results of property characterization of hydrogels. (A) Storage modulus (G′) and loss modulus (G″) as a function of frequency. (B) Storage modulus (G′) and loss modulus (G″) as a function of time. (C) Swelling profile of the hydrogel in PBS. (D) Degradation curve of the hydrogel in PBS. (E) Degradation curve of the hydrogel in the collagenase II solution. (F) Cumulative release profile of KGN from the hydrogel scaffold.

Swelling and degradation analyses (Fig. 3C–E) further demonstrated the layer-dependent physicochemical properties of the hydrogels. In terms of swelling behavior (Fig. 3C), the GelMA/Lip hydrogel exhibited a higher swelling ratio, indicating greater water uptake capacity which may be attributed to the aqueous nature of liposomes and their potential influence on the GelMA network structure. As soft nanocomponents, liposomes might introduce microscopic steric hindrance during the photocrosslinking process and consequently lead to a relatively lower packing density of the polymer chains, which allows for increased hydration within the network. In contrast, the GelMA/HAP hydrogel showed a relatively lower swelling ratio, suggesting a denser network structure and enhanced structural stability. This phenomenon could potentially be explained by the interfacial interactions between the rigid HAP nanoparticles and the GelMA polypeptide chains, where these nanoparticles might serve as physical reinforcement points to strengthen the crosslinking network and effectively constrain the expansion of the hydrogel in an aqueous environment. For degradation behavior (Fig. 3D and E), consistent trends were observed under both PBS and enzymatic conditions. In PBS, the GelMA/HAP hydrogel degraded more slowly (38.98 ± 1.59% mass loss after 21 days), whereas the GelMA/Lip hydrogel showed a faster degradation rate (48.56 ± 1.95%). A similar difference was maintained in 2.5 U mL⁻¹ collagenase II solution (Fig. 3E), where both hydrogels exhibited accelerated degradation compared to PBS, but the GelMA/Lip group consistently degraded faster than the GelMA/HAP group at all time points. This result indicates that the GelMA/Lip hydrogel is more susceptible to enzymatic cleavage potentially due to its higher hydration and more porous network which likely allows for easier infiltration of enzymes. In contrast, the incorporation of HAP effectively retards the degradation rate of the GelMA network by potentially providing a physical shielding effect that might reduce the accessibility of enzymatic cleavage sites on the GelMA backbone.

3.4 KGN release behavior from the hydrogel scaffolds

To further evaluate the functionality of the bilayer hydrogel system, the release behavior of the bioactive molecule KGN was investigated (Fig. 3F). The cumulative release of KGN from the GelMA/KGN hydrogel reached approximately 59.91% within 21 d, showing a relatively rapid release profile. In contrast, the GelMA/KGN@Lip hydrogel exhibited a markedly slower release behavior, with only 44.52% released at 21 d and approximately 46.09% at 28 d. This plateau after 21 days suggests limited further release under the current conditions, which may be attributed to restricted diffusion of liposome-encapsulated KGN within the crosslinked GelMA network, as well as possible drug instability over prolonged incubation. This significant difference demonstrates that the liposomal encapsulation effectively delayed KGN diffusion through a dual barrier system, where the drug must sequentially traverse the liposomal phospholipid bilayer and the GelMA network before release into the medium. Although the higher swelling and faster degradation of the chondrogenic layer generally favor mass transport, the liposomal phospholipid bilayer appears to function as the primary rate-limiting factor that manages the release kinetics within this specific layer. Such a release profile may be beneficial for providing prolonged biochemical stimulation during the early stage of cartilage regeneration, during which continuous exposure to chondrogenic cues (such as KGN) is necessary to maintain matrix synthesis and phenotype stabilization. The sustained release from the KGN@Lip hydrogel enabling relatively prolonged chondrogenic stimulation potentially reflects the evolving nature of the scaffold, where the progressive erosion of the GelMA matrix might create new diffusion pathways to maintain a steady output of KGN. This may contribute to the orderly regulation of biochemical cues within the scaffold system, where the upper layer supports extracellular matrix production through sustained signaling while the lower HAP enhancement layer provides mechanical stability for subchondral bone regeneration.

3.5 Evaluation of the in vitro biocompatibility of BMSC by hydrogel

To investigate the biocompatibility and proliferative response of BMSCs to different hydrogel extracts, CCK-8 assays, live/dead staining, and cytoskeletal fluorescence staining were performed (Fig. 4). Cell viability in all groups remained above 100% at 24, 72, and 120 h, indicating that none of the extracts induced detectable cytotoxicity (Fig. 4A). Cells exposed to GelMA/HAP and GelMA/KGN@Lip extracts showed slightly higher viability than those in the GelMA and control groups at all time points, suggesting that incorporation of HAP or KGN@Lip into the GelMA matrix did not compromise cell survival and may have enhanced BMSC proliferation. This effect is likely related to Ca²⁺ ions released from HAP, which can regulate cellular metabolic activity,³³ and to the pro-proliferative influence of KGN.³⁴ Live/dead staining confirmed these observations (Fig. 4B). At 24 h, most cells were viable (green), with only a few dead cells (red) observed. By 72 and 120 h, cell density increased substantially while the proportion of viable cells remained high, reflecting stable biocompatibility of the extracts during prolonged culture. In addition, BMSCs in the GelMA/HAP and GelMA/KGN@Lip groups exhibited more uniform distribution and denser coverage, indicating a microenvironment favorable for adhesion and proliferation. Cytoskeletal staining revealed progressive elongation and alignment of F-actin fibers over time, accompanied by a transition in cell morphology from a short spindle shape to a long spindle shape and a spreading shape (Fig. 4C). No abnormal cellular morphology or nuclear alterations were detected, suggesting that soluble components of the hydrogels did not negatively affect cell structure or adhesion.

Fig. 4 Evaluation of the in vitro biocompatibility of BMSC by different hydrogel extracts. (A) The cell viability of BMSC at 24, 72 and 120 h was detected by the CCK-8 method; (B) live and dead staining images at different time points (green: live cells, red: dead cells); (C) fluorescence staining images of the cytoskeleton at different time points (red: F-actin, blue: nucleus). Data were expressed as mean ± standard deviation (n = 5), and statistical analysis was performed using one-way analysis of variance and Tukey post hoc test (***p < 0.001).

3.6 Evaluation of the osteogenic performance of hydrogels

The effects of different hydrogel extracts on BMSC osteogenic differentiation were evaluated by ALP staining and activity assays on day 7 under osteogenic induction, followed by Alizarin red S staining and quantitative analysis on day 21, and qRT-PCR analysis of osteogenic related genes (Fig. 5). On day 7, the GelMA/HAP extract group displayed the strongest ALP staining and the highest ALP activity (868.80 ± 15.53 U per mg protein), which was significantly higher than that of the control, GelMA, and GelMA/KGN@Lip groups (Fig. 5A and B). The upregulation of ALP, a key marker of early osteogenic differentiation, indicates that BMSCs had entered an active osteogenic initiation phase. This early osteogenic enhancement is primarily attributed to the release of Ca²⁺ and PO₄³⁻ ions from HAP. Soluble Ca²⁺ can regulate ionic homeostasis and activate the calcium-sensing receptor (CaSR),³⁵ thereby stimulating downstream PI3K/AKT and MAPK signaling pathways to upregulate osteogenic factors,³⁶ promote ALP synthesis, and facilitate early-stage differentiation.³⁷ By day 21, the GelMA/HAP extract group exhibited the most extensive formation of mineralized nodules, as shown by dense Alizarin red S staining. Quantitative analysis revealed a significantly higher absorbance (1.9934 ± 0.043) compared with the other groups (Fig. 5C and D). These results suggest that the inorganic ions released by HAP can continuously participate in the later mineralization process in the culture environment, promoting calcium deposition and extracellular matrix maturation. Gene expression analysis was consistent with these findings (Fig. 5E–G). RUNX2, Col-I, and OCN expression levels in the GelMA/HAP group were upregulated by approximately 1.686 ± 0.06, 1.720 ± 0.02, and 1.918 ± 0.06 times higher, respectively, showing statistically significant differences relative to the other groups (p < 0.001). Increased RUNX2 expression indicates activation of early osteogenic transcriptional programs, elevated Col-I reflects enhanced collagen matrix synthesis, and higher OCN expression corresponds to progression toward mature mineralization. In contrast, the GelMA/KGN@Lip extract had comparatively weaker effects on osteogenic markers, indicating limited osteoinductive potential under the current induction conditions.

Fig. 5 Effects of different hydrogels on osteogenic differentiation of BMSC. (A) Staining results of alkaline phosphatase (ALP) 7 days after osteogenic induction; (B) corresponding quantitative analysis of ALP activity; (C) Alizarin red staining results 21 days after osteogenic induction; (D) quantitative analysis of calcified nodule formation; (E)–(G) Relative mRNA expression levels of osteogenic related genes COL-I, OCN and RUNX2 after 21 days of induction. Data were expressed as mean ± standard deviation (n = 3), and statistical analysis was performed using one-way analysis of variance and Tukey post hoc test (**p < 0.01, ***p < 0.001).

3.7 Evaluation of chondrogenic properties of hydrogels

To evaluate the induction ability of different hydrogels on the chondrogenic differentiation of BMSC, Alcian blue staining, glycosaminoglycan (GAG) quantification, and qRT-PCR analysis of chondrogenic marker genes were performed. As shown in Fig. 6A and B, Alcian blue staining revealed weak matrix deposition in the control group, whereas all hydrogel-treated groups showed enhanced blue staining to varying degrees, indicating increased cartilage-like extracellular matrix formation. Notably, the GelMA/KGN@Lip group exhibited the most intense and widely distributed Alcian blue-positive staining, suggesting that KGN-loaded liposomes effectively promoted cartilaginous matrix deposition. Consistently, quantitative analysis of GAG content further confirmed this trend, with the GelMA/KGN@Lip group showing the highest GAG concentration among all groups (Fig. 6C), which was significantly higher than that of the control, GelMA, and GelMA/HAP groups (p < 0.001). These results demonstrate that the GelMA/KGN@Lip hydrogel markedly enhanced the synthesis and accumulation of cartilage-specific extracellular matrix.

Fig. 6 Effects of different hydrogels on chondrogenic differentiation of BMSC. (A) Low-magnification image stained with alcian blue; (B) Alcian blue-stained high-power image; (C) quantitative statistical results of glycosaminoglycans; (D) the relative mRNA expression levels of chondrogenic related genes SOX9, Col2a1 and Acan after induction. Data were expressed as mean ± standard deviation (n = 3), and statistical analysis was performed using one-way analysis of variance and Tukey post hoc test (**p < 0.01, ***p < 0.001).

The expression levels of key chondro-related genes SOX9, Col2a1 and Acan were detected by qRT-PCR (Fig. 6D). Compared with the control group, the chondro-related genes in all hydrogel treatment groups were upregulated, indicating that the three-dimensional hydrogel environment helps promote the differentiation of BMSC into cartilage phenotypes. Among them, the expression levels of SOX9 (1.77 ± 0.01 times), Col2a1 (1.76 ± 0.04 times), and Acan (1.80 ± 0.01 times) in the GelMA/KGN@Lip group were the highest, and were significantly increased compared with both the GelMA group and the GelMA/HAP group (p < 0.001). SOX9, a key transcription factor regulating cartilage differentiation fate, was significantly upregulated in the GelMA/KGN@Lip group, suggesting that this system effectively activated the chondrogenic transcription program of BMSC. Meanwhile, the synchronous upregulation of the downstream structural protein gene Col2a1 (encoding type II collagen) and the main component Acan of the extracellular matrix of chondrocytes further verified the formation of a mature chondroid phenotype. The comprehensive results show that the GelMA hydrogel loaded with KGN liposomes can significantly enhance the chondrogenic differentiation potential of BMSC, and its induction effect is significantly better than that of the control hydrogels without loading or containing HAP.

3.8 In vivo evaluation of osteochondral repair at 6 weeks

After establishing a cylindrical osteochondral defect with a diameter of 5 mm and a depth of 4 mm in the femoral trochlea, three types of scaffolds were implanted, including the GelMA/KGN@Lip monolayer, the GelMA/HAP monolayer, and the bilayer GelMA/KGN@Lip//GelMA/HAP scaffold, which is hereafter referred to as KGN@Lip//HAP-GelMA. Six weeks after surgery, the femora were harvested for macroscopic evaluation. As shown in Fig. 7A, the control group exhibited a clearly visible defect with marked depression and negligible tissue ingrowth. Although partial surface coverage was observed in the GelMA/HAP group, the defect center still showed collapse. In contrast, both the GelMA/KGN@Lip and KGN@Lip//HAP-GelMA groups displayed substantially improved gross morphology, with the regenerated tissue showing improved integration with the surrounding cartilage and a relatively smooth surface.

Fig. 7 Evaluation of the therapeutic effect on femoral defects 6 weeks after hydrogel stent implantation. (A) Macroscopic appearance images of the femurs after sampling in each group, with the initial defect areas marked by red circles. It can be seen that there are differences in the coverage of new bone-cartilage and the degree of defect closure among different treatment groups. (B) Three-dimensional reconstruction images of femur micro-CT and two-dimensional tomographic images (cross-sectional and sagittal planes), showing the volume filling of new bone at the defect site, the continuity of cortical bone and the recovery of trabecular bone structure. (C) Based on the bone morphometric parameters obtained from micro-CT analysis, including bone volume fraction (BV/TV), bone mineral density (BMD), trabecular bone number (Tb.N), and trabecular bone thickness (Tb.Th), they were used to quantify the bone repair effects of each group. (D) H&E staining map of the defect area.

To further validate the macroscopic appearance, micro-CT analysis was performed, which revealed consistent trends across groups (Fig. 7B). The control group showed almost no subchondral bone regeneration, reflected by a large discontinuity at the defect site. The GelMA/KGN@Lip and GelMA/HAP groups exhibited similar levels of bone repair, whereas the KGN@Lip//HAP-GelMA group showed nearly complete re-establishment of the bone interface, with only slight surface depression. These imaging results indicated that the bilayer scaffold provided the most pronounced structural restoration at the osteochondral site.

Building on these qualitative observations, quantitative microstructural parameters were analyzed to assess the extent of bone regeneration (Fig. 7C). Consistent with the imaging results, all scaffold groups showed progressive improvements compared with the control. BV/TV increased from 8–10% in the control group to 9–13% in the GelMA/KGN@Lip group, then to 16–17% in the GelMA/HAP group, reaching the highest value of 24–25% in the bilayer scaffold. BMD exhibited a similar stepwise increase, rising from about 0.18–0.19 g cm⁻³ in the control to about 0.28–0.29 g cm⁻³ in the KGN@Lip//HAP-GelMA group. Moreover, trabecular parameters (Tb.N and Tb.Th) reflected significant regeneration, particularly in the bilayer scaffold. Despite Tb.Th remaining slightly lower than native tissue, its moderate increase suggests that the newly formed bone was still undergoing early-stage remodeling, consistent with the micro-CT morphologies.

To correlate these medical images with tissue-level repair, histological analysis was conducted using H&E staining (Fig. 7D). Low-magnification images confirmed the repair tendencies observed in micro-CT, with minimal bone formation in the control and substantial structural restoration in the scaffold groups. High-magnification views further highlighted group-specific differences: the control group exhibited primarily fibrous tissue filling with limited new bone at the mid-defect region. In contrast, both GelMA/KGN@Lip and GelMA/HAP monolayer scaffolds supported the formation of continuous subchondral bone, albeit with fibrocartilage at the upper surface. Notably, the KGN@Lip//HAP-GelMA group demonstrated a markedly more organized osteochondral architecture, including regions with distinct morphological features, where chondrocytes appeared more regularly arranged compared with the monolayer groups. These characteristics collectively suggest that the bilayer scaffold promoted a more organized osteochondral repair compared with the other groups.

3.9 Extended in vivo evaluation of cartilage regeneration at 10 weeks

To evaluate the long-term restorative outcomes, we extended the observation period to 10 weeks, during which the regenerative process advanced from initial tissue filling to structural maturation (Fig. 8). Compared with the 6 week time point, all groups exhibited more advanced tissue regeneration and improved structural restoration. Notably, the bilayer KGN@Lip//HAP-GelMA scaffold maintained the most pronounced repair performance, characterized by superior matrix organization and enhanced interfacial integration.

Fig. 8 Evaluation of cartilage regeneration effect 10 weeks after hydrogel scaffold implantation. (A) H&E staining result of the injured area; (B) Alcian blue staining of the regenerated cartilage; (C) periodic acid-Schiff staining of the regenerated cartilage; (D) immunofluorescence staining of aggrecan in the regenerated cartilage; (E) fluorescence quantitative intensity of aggrecan.

H&E staining (Fig. 8A) revealed that the defect in the control group remained significantly depressed with distinct boundaries and was filled primarily with fibrous tissue and dense cell nuclei. Similarly, the GelMA/HAP group exhibited substantial fibrous tissue and dense nuclei without a continuous cartilage layer, although the total volume of new tissue was higher than that of the control. In contrast, the GelMA/KGN@Lip group formed a relatively continuous tissue coverage with improved surface smoothness and a more organized tissue structure. Notably, the KGN@Lip//HAP-GelMA group showed nearly complete defect filling with the smoothest surface, the clearest repair boundaries, and optimal tissue integrity. The deposition of the cartilaginous matrix was further characterized by specialized staining. Alcian blue staining (Fig. 8B) showed weak signals in the control group and limited, uneven staining in the GelMA/HAP group. Conversely, the GelMA/KGN@Lip group displayed significantly enhanced and continuous blue staining. The KGN@Lip//HAP-GelMA group exhibited the most intense Alcian blue staining with the richest matrix accumulation. Consistent with these findings, Safranin O-Fast Green staining (Fig. 8C) showed shallow staining and prominent fibrous components in the control and GelMA/HAP groups. In the GelMA/KGN@Lip and especially the KGN@Lip//HAP-GelMA groups, PAS staining was significantly intensified, with the latter showing a regenerated tissue thickness and maturity closest to native cartilage. These results were further validated by aggrecan immunofluorescence staining (Fig. 8D and E), where the KGN@Lip//HAP-GelMA group demonstrated the highest fluorescence intensity among all groups, followed by the GelMA/KGN@Lip group.

3.10 Extended in vivo evaluation of subchondral bone remodeling at 10 weeks

Fig. 9 shows that there are significant differences in the repair effects of the various hydrogel scaffolds in the regeneration of subchondral bone. Masson's trichrome staining (Fig. 9A) showed that the control group defect remained depressed, with disordered collagen deposition and limited bone regeneration. The GelMA/KGN@Lip group showed some tissue formation, but the subchondral structure remained relatively loose. In the GelMA/HAP group, a denser distribution of collagen fibers was observed, indicating more active collagen deposition. Notably, the KGN@Lip//HAP-GelMA group exhibited the most complete defect filling, with a dense, regular subchondral bone structure that appeared well-integrated with the host bone. The formation and maturation of the bone matrix were further evaluated through collagen typing and osteogenic markers. Type I collagen (Col I) immunohistochemistry (Fig. 9B) and Type III collagen (Col III) immunofluorescence (Fig. 9C) revealed weak positive signals in the control and GelMA/KGN@Lip groups, suggesting limited matrix deposition. In contrast, the GelMA/HAP group showed significantly enhanced expression of both Col I and Col III. The KGN@Lip//HAP-GelMA group displayed the strongest and most continuous fluorescence signals for these collagen types, which was further supported by quantitative analysis (Fig. 9E). Furthermore, osteocalcin (OCN) immunofluorescence (Fig. 9D), a marker of bone maturity, showed relatively weak signals in the control and GelMA/KGN@Lip groups. OCN expression was notably higher in the GelMA/HAP group, while the KGN@Lip//HAP-GelMA group exhibited the most widespread and intense OCN signal distribution.

Fig. 9 Evaluation of the regeneration effect of subchondral bone after hydrogel scaffold implantation for 10 weeks. (A) Masson staining results of the injured area; (B) immunohistochemical staining of type I collagen in subchondral bone; (C) immunofluorescence staining of type III collagen in subchondral bone; (D) immunofluorescence staining of osteocalcin in subchondral bone; (E) quantitative statistics of the expression of type I collagen, type III collagen and osteocalcin.

The observed differences in osteochondral repair among the four groups can be related to the distinct structural and compositional characteristics of the scaffolds. The GelMA/KGN@Lip scaffold, despite lacking inorganic osteoinductive components, exhibited enhanced cartilage-related repair, which may be associated with its radially oriented and interconnected porous architecture that facilitates cell infiltration, together with the sustained release of KGN, consistent with its reported role in promoting progenitor cell activity and chondrogenic differentiation. In contrast, although the GelMA/HAP scaffold showed more pronounced subchondral bone regeneration, its overall macroscopic repair outcome appeared less favorable, which may be attributed to its limited contribution to cartilage formation and surface continuity, both of which are critical for early defect coverage and structural integration. Notably, the bilayer KGN@Lip//HAP-GelMA scaffold integrates these complementary features within a spatially organized configuration, which may contribute to more uniform tissue filling and improved structural continuity, thereby supporting a more coordinated regeneration of cartilage and subchondral bone.

4 Conclusions

In this study, a 3D-printed bilayer hydrogel scaffold with a radially oriented architecture was developed to mimic the hierarchical structure of osteochondral tissue. Compared to traditional casting, 3D printing enables the precise one-step fabrication of complex radially oriented microchannels without the need for intricate demolding processes, while offering the flexibility for patient-specific customization. By integrating a GelMA/HAP layer with a GelMA/KGN@Lip layer, the construct provides spatially distinct biochemical microenvironments. In vivo results showed that the lower layer facilitated substantial bone formation and matrix mineralization, likely associated with the mechanical support and ion release (Ca²⁺ and PO₄³⁻) from HAP. Conversely, the upper layer supported cartilage-like tissue formation and matrix deposition, where liposomal encapsulation provided sustained KGN release to maintain chondrogenic signaling. Notably, the radially oriented microchannels appear to enhance nutrient diffusion and guide endogenous cell migration, contributing to more uniform tissue filling and organized interfacial integration. The combination of mechanically graded design, localized biochemical cues, and anisotropic structural guidance enables coordinated repair of both bone and cartilage compartments. While the proposed scaffold represents a promising platform for osteochondral defect repair, further studies are necessary to enhance and comprehensively evaluate its mechanical performance, including compressive modulus and interfacial strength, relative to native tissue.

Author contributions

Zhenbiao Wangzi: conceptualization, methodology, investigation, data curation, formal analysis, writing – original draft. Chengcheng Yang: investigation, data curation, formal analysis. Peng Yu: investigation, validation. Xiubiao Huang: investigation, resources. Nan Hu: investigation. Tuo Jiao: methodology, supervision. Hao Qi: conceptualization, supervision, writing – review & editing.

Conflicts of interest

The authors declare no competing financial interests.

Data availability

Data are available on request to the authors.

Acknowledgements

This work was supported by the Hainan “South China Sea Nova” Project (Grant No. NHXX-WJW-2023010).

References

1. J. Du, Z. Zhu and J. Liu, et al., 3D-printed gradient scaffolds for osteochondral defects: Current status and perspectives, Int. J. Bioprint., 2023, 9(4), 724.
2. D. Yue, L. Du and B. Zhang, et al., Time-dependently Appeared Microenvironmental Changes and Mechanism after Cartilage or Joint Damage and the Influences on Cartilage Regeneration, Organogenesis, 2021, 17(3–4), 85–99.
3. A. Cardoneanu, L. A. Macovei and A. M. Burlui, et al., Temporomandibular Joint Osteoarthritis: Pathogenic Mechanisms Involving the Cartilage and Subchondral Bone, and Potential Therapeutic Strategies for Joint Regeneration, Int. J. Mol. Sci., 2022, 24(1), 171.
4. N. Yildirim, A. Amanzhanova and G. Kulzhanova, et al., Osteochondral Interface: Regenerative Engineering and Challenges, ACS Biomater. Sci. Eng., 2023, 9(3), 1205–1223.
5. Q. Han, Y. Tong and L. Zhang, et al., Comparative efficacy of osteochondral autologous transplantation and microfracture in the knee: an updated meta-analysis of randomized controlled trials, Arch. Orthop. Trauma Surg., 2023, 143(1), 317–328.
6. F. Migliorini, N. Maffulli and A. Baroncini, et al., Allograft Versus Autograft Osteochondral Transplant for Chondral Defects of the Talus: Systematic Review and Meta-analysis, Am. J. Sports Med., 2022, 50(12), 3447–3455.
7. H. R. Matheus, Ş. D. Özdemir and F. P. S. Guastaldi, Stem cell-based therapies for temporomandibular joint osteoarthritis and regeneration of cartilage/osteochondral defects: a systematic review of preclinical experiments, Osteoarthritis Cartilage, 2022, 30(9), 1174–1185.
8. M. Tamaddon, H. Gilja and L. Wang, et al., Osteochondral scaffolds for early treatment of cartilage defects in osteoarthritic joints: from bench to clinic, Biomater. Transl., 2020, 1(1), 3–17.
9. X. Lin, Y. Zhang and J. Li, et al., Biomimetic multizonal scaffolds for the reconstruction of zonal articular cartilage in chondral and osteochondral defects, Bioact. Mater., 2025, 43, 510–549.
10. G. Tan, J. Xu and Q. Yu, et al., Photo-Crosslinkable Hydrogels for 3D Bioprinting in the Repair of Osteochondral Defects: A Review of Present Applications and Future Perspectives, Micromachines, 2022, 13(7), 1038.
11. W. Zhang, K. Zha and W. Hu, et al., Multifunctional hydrogels: advanced therapeutic tools for osteochondral regeneration, Biomater. Res., 2023, 27(1), 76.
12. H. Cao, L. Duan and Y. Zhang, et al., Current hydrogel advances in physicochemical and biological response-driven biomedical application diversity, Signal Transduction Targeted Ther., 2021, 6(1), 426.
13. X. Wang, W. He and H. Huang, et al., Recent Advances in Hydrogel Technology in Delivering Mesenchymal Stem Cell for Osteoarthritis Therapy, Biomolecules, 2024, 14(7), 858.
14. Z. Liu, W. Xin and J. Ji, et al., 3D-Printed Hydrogels in Orthopedics: Developments, Limitations, and Perspectives, Front. Bioeng. Biotechnol., 2022, 10, 845342.
15. J. Lu, Y. Gao and C. Cao, et al., 3D bioprinted scaffolds for osteochondral regeneration: advancements and applications, Mater. Today Bio, 2025, 32, 101834.
16. D. Ege and V. Hasirci, Is 3D Printing Promising for Osteochondral Tissue Regeneration?, ACS Appl. Bio Mater., 2023, 6(4), 1431–1444.
17. B. Zhou, X. Jiang and X. Zhou, et al., GelMA-based bioactive hydrogel scaffolds with multiple bone defect repair functions: therapeutic strategies and recent advances, Biomater. Res., 2023, 27(1), 86.
18. G. Im and R. Lin, Bioengineering for vascularization: Trends and directions of photocrosslinkable gelatin methacrylate hydrogels, Front. Bioeng. Biotechnol., 2022, 10, 1053491.
19. A. Kavitha Sri, C. Arthi and N. R. Neya, et al., Nano-hydroxyapatite/collagen composite as scaffold material for bone regeneration, Biomed. Mater., 2023, 18(3), 032002.
20. J. Khotib, M. A. Gani and A. S. Budiatin, et al., Signaling Pathway and Transcriptional Regulation in Osteoblasts during Bone Healing: Direct Involvement of Hydroxyapatite as a Biomaterial, Pharmaceuticals, 2021, 14(7), 615.
21. A. H. Hoveidaei, M. Sadat-Shojai and S. Mosalamiaghili, et al., Nano-hydroxyapatite structures for bone regenerative medicine: Cell-material interaction, Bone, 2024, 179, 116956.
22. T. Zhang, J. Li and Y. Wang, et al., Hydroxyapatite/Polyurethane Scaffolds for Bone Tissue Engineering, Tissue Eng., Part B, 2024, 30(1), 60–73.
23. Y. Xiong, B. Mi and Z. Lin, et al., The role of the immune microenvironment in bone, cartilage, and soft tissue regeneration: from mechanism to therapeutic opportunity, Mil. Med. Res., 2022, 9(1), 65.
24. K. Johnson, S. Zhu and M. S. Tremblay, et al., A stem cell-based approach to cartilage repair, Science, 2012, 336(6082), 717–721.
25. M. Zhang, W. Hu and C. Cai, et al., Advanced application of stimuli-responsive drug delivery system for inflammatory arthritis treatment, Mater. Today Bio, 2022, 14, 100223.
26. X. Liu, P. Liu and H. Li, et al., Application of kartogenin for the treatment of cartilage defects: current practice and future directions, RSC Adv., 2024, 14(45), 33206–33222.
27. S. Bhuyan, S. Swain and R. D. K. Misra, et al., Nanomedicine-Driven Approaches for Kartogenin Delivery: Advancing Chondrogenic Differentiation and Cartilage Regeneration in Tissue Engineering, Int. J. Nanomed., 2025, 20, 7443–7468.
28. M. Wytrwal, K. Szmajnta and M. Kucharski, et al., Kartogenin-loaded liposomes coated with alkylated chondroitin sulfate for cartilage repair, Int. J. Pharm., 2023, 646, 123436.
29. X. Niu, N. Li and Z. Du, et al., Integrated gradient tissue-engineered osteochondral scaffolds: Challenges, current efforts and future perspectives, Bioact. Mater., 2023, 20, 574–597.
30. S. Chen, H. Wang and V. L. Mainardi, et al., Biomaterials with structural hierarchy and controlled 3D nanotopography guide endogenous bone regeneration, Sci. Adv., 2021, 7(31), eabg3089.
31. S. Zheng, D. Li and Q. Liu, et al., Surface-Modified Nano-Hydroxyapatite Uniformly Dispersed on High-Porous GelMA Scaffold Surfaces for Enhanced Osteochondral Regeneration, Int. J. Nanomed., 2023, 18, 5907–5923.
32. Y. Liu, W. Xu and D. Zhao, et al., Bilayer Hydrogel Scaffold with Directional Micro-Channels and Mechanical Gradient to Promote Osteochondral Repair, Adv. Healthcare Mater., 2025, e3547.
33. M. N. Lee, H. Hwang and S. Oh, et al., Elevated extracellular calcium ions promote proliferation and migration of mesenchymal stem cells via increasing osteopontin expression, Exp. Mol. Med., 2018, 50(11), 1–16.
34. Y. Zhao, X. Zhao and R. Zhang, et al., Cartilage Extracellular Matrix Scaffold With Kartogenin-Encapsulated PLGA Microspheres for Cartilage Regeneration, Front. Bioeng. Biotechnol., 2020, 8, 600103.
35. R. Yanai, F. Tetsuo and S. Ito, et al., Extracellular calcium stimulates osteogenic differentiation of human adipose-derived stem cells by enhancing bone morphogenetic protein-2 expression, Cell Calcium, 2019, 83, 102058.
36. W. Zhao, Z. Zhang and L. Zheng, et al., Calcium-sensing receptor activating ERK1/2 and PI3K-Akt pathways to induce the proliferation of osteosarcoma cells, Clin. Exp. Pharmacol. Physiol., 2020, 47(3), 517–519.
37. J. Xi, H. Zang and L. Guo, et al., The PI3K/AKT cell signaling pathway is involved in regulation of osteoporosis, J. Recept. Signal Transduction Res., 2015, 35(6), 640–645.

---