Fresh porcine intervertebral disc annulus fibrosus tissues were purchased from a local market. Decellularization was performed using a previously reported chemical-enzyme combined method (Liu B. et al., 2024). Fresh annulus fibrosus tissues were placed in 50 mM Tris buffer and stirred overnight. Tissues were treated with 1% Triton X-100 in Tris buffer at 4°C for 24 h, and digested in Hank’s balanced salt solution containing 40 U/mL DNase (Macklin Biochemical, Shanghai, China), 20 mg/L RNase (Solarbio Life Sciences, Beijing, China), and 0.01% trypsin (Sigma, NY, United States) at 37°C for 4 h. Samples were stirred overnight at 4°C in 1% Triton X-100/Tris buffer. After decellularization, the tissue samples were lyophilized and then stored at −20°C for later use. To prepare decellularized matrix solution, the lyophilized decellularized tissues were ground into powder in liquid nitrogen, and dissolved at 4°C in a 1 mg/mL pepsin solution prepared in 0.01 N HCl. The pH was then neutralized using 1 M NaOH solution. The prepared decellularized matrix solution was stored at 4°C for further use. For histological staining, both fresh and decellularized annulus fibrosus tissues were fixed in buffered formalin at 4°C for 2°days, then subjected to standard dehydration and paraffin embedding. The paraffin-embedded samples were sectioned at a thickness of 5 μm, and stained with hematoxylin and eosin (H&E) and Masson’s Trichrome.

The dry weight of the freeze-dried decellularized annulus fibrosus matrix was measured, and the tissue samples were then homogenized using a grinder. A 300 μg/mL papain solution (Meryer Biochemical Technology, Shanghai, China) was prepared using 50 mM phosphate buffer containing 5 mM L-cysteine (Meryer Biochemical Technology, Shanghai, China) and 5 mM ethylenediaminetetraacetic acid (EDTA) (Adamas Reagent, Shanghai, China), and the samples were digested at 65°C for 16 h.

Collagen content was determined using the hydroxyproline assay based on the above digestion solution, following the instructions of the hydroxyproline assay kit (Jiancheng Bioengineering Institute, Nanjing, China). Since hydroxyproline accounts for 13.4% of collagen, the collagen content was estimated accordingly. All collagen content results were normalized to tissue dry weight and expressed in μg/mg. Glycosaminoglycan (GAG) content was measured using the dimethylmethylene blue (1,9-dimethylmethylene blue chloride, DMMB) assay (Duan et al., 2015) based on the digestion solution. All GAG content results were normalized to tissue dry weight and expressed in μg/mg. DNA content was quantified using the PicroGreen double-stranded DNA assay (Yeasen Biotechnology, Shanghai, China) based on the above digestion solution. All DNA content results were normalized to tissue dry weight and expressed in ng/mg.

Dissolve zinc nitrate hexahydrate (Sigma, MO, United States) at 0.07 g/mL in 7.5 mL of deionized water, and 2-methylimidazole (Macklin, Shanghai, China) at 0.1 g/mL in 7.5 mL of methanol. Mix the solutions and stir at room temperature for 2 h. Separate the product by centrifugation, wash the precipitate with ethanol, and freeze-dry it. Store the powder at −20°C. Analyze it via scanning electron microscope (SEM, SU8010, Hitachi, Tokyo, Japan), energy dispersive X-ray spectroscopy (EDX), X-ray diffraction (XRD, D2 PHASE, Bruker, Karlsruhe, Germany), and fourier transform infrared spectrometer (FTIR, Tensor 27, Bruker, Karlsruhe, Germany). The ZIF-8 nanoparticle diameters in SEM images were measured using ImageJ.

Following the previously described method (Rockwood et al., 2011), a silk fibroin solution was prepared by degumming and dissolving silk fibers. Tannic acid powder (Aladdin, Shanghai, China) was dissolved in double-distilled water to prepare a 0.3 g/mL tannic acid solution. Mixed solutions were prepared using different concentrations of silk fibroin solution, decellularized matrix solution, and tannic acid solution, and varying amounts of ZIF-8. These compositions were added to formulate five groups of adhesives.

Pig skin was purchased from a local market. Adhesive samples from different groups were evenly applied between two pieces of pig skin tissue (dimensions: 30 mm × 5 mm) pre-wetted with phosphate buffered saline (PBS), with an overlapping area of 5 mm × 10 mm. After applying finger pressure for 15 s, a universal testing machine (1710-111, Instron, MA, United States) was used to perform lap shear testing. The test was conducted at a constant stretching speed of 10 mm/min until complete separation of the two tissue samples, and the maximum lap shear strength was recorded.

Based on the results of tensile testing, three groups of hydrogels were selected for subsequent experiments: (1) silk fibroin/tannic acid (ST group), (2) ST combined with decellularized porcine annulus fibrosus extracellular matrix (ST/dECM group), and (3) ST/dECM supplemented with ZIF-8 (ST/dECM/ZIF-8 group). The freeze-dried adhesives from these three groups were applied for SEM and FTIR analyses.

For each group, 200 μL of the hydrogel sample was prepared. The mass of each hydrogel adhesive was weighed and recorded as m (original). The samples were then immersed in 50 mL of PBS solution at room temperature. At time points of 0.5, 1, 2, 4, 20, and 24 h, surface liquid was blotted off, and the samples were weighed to record their swollen mass as m (swollen). The swelling ratio was calculated as: swelling ratio  % = m swollen –   m original m original × 100 %

Hydrogels made from ST, ST/dECM, and ST/dECM/ZIF-8 with a dry weight of 17–20 mg were weighed and recorded as W₀. These samples were immersed in PBS solution at 37°C for 1, 3, 5, 7, and 14 days. At each time point, surface liquid was wiped off, and the dry weight of the hydrogel was measured and recorded as W_x. The samples were then returned to the PBS solution for continued incubation. The degradation rate was calculated using the following formula: d egradation  r ate  % = W x   –   W 0 W 0 × 100 %

Each group of prepared hydrogel samples was freeze-dried and ground into fine powder. The resulting powder was exposed to ultraviolet (UV) light for 24 h for sterilization. Then, 5 mg of the sample powder was weighed and added to 25 mL of complete culture medium, containing 1% penicillin/streptomycin (Hyclone, UT, United States), 10% fetal bovine serum (FBS, Cyagen Biosciences Inc., DE, United States), α-MEM culture medium (Hyclone, UT, United States). The mixture was incubated at 37°C for 24 h to obtain the extract, which was collected for subsequent experiments. The viability of passage 2 or 3 rat bone marrow mesenchymal stem cells (BMSCs, Cyagen Biosciences, DE, United States) was assessed after culturing in the extract-containing medium for 1, 2, and 3 days using the live/dead cell viability assay (Invitrogen, CA, United States) (Shi et al., 2021). Cell proliferation of passage 2 or 3 rat BMSCs cultured in the extract-containing medium for 1 and 3 days was evaluated using the MTT assay (Sigma, MO, United States) (Kong et al., 2021).

For chondrogenic differentiation, passage 2 or 3 rat BMSCs were co-cultured with adhesive hydrogels from different groups in chondrogenic differentiation medium and induced for 14 days. The chondrogenic differentiation medium consisted of: DMEM/F12 (Hyclone, UT, United States), 10% fetal bovine serum, 1% penicillin/streptomycin, 100 nM dexamethasone (Sigma, MO, United States), 0.2 mM ascorbic acid (Sigma, MO, United States), 1 mM sodium pyruvate (Sigma, MO, United States), insulin-transferrin-selenium (BD Biosciences, NJ, United States), and 10 ng/mL TGF-β3 (PetroTech, OK, United States) (Bai et al., 2022). For osteogenic differentiation, passage 2 or 3 rat BMSCs were cultured in osteogenic differentiation medium containing extracts of adhesive hydrogels from different groups and induced for 14 days (powder-to-medium mass/volume ratio = 5 mg: 25 mL). The osteogenic differentiation medium consisted of: DMEM/F12, 10% fetal bovine serum, 1% penicillin/streptomycin, 100 nM dexamethasone, 10 mM β-glycerophosphate (Sigma, MO, United States), and 50 μM ascorbic acid (Ogawa et al., 2004). Total RNA was extracted from rat BMSCs cultured in differentiation medium for 14 days using an RNA extraction kit (Servicebio, Hubei, China). The concentration and purity of the RNA were measured using a NanoDrop spectrophotometer (Thermofisher, DE, United States). Qualified RNA samples were immediately subjected to reverse transcription. cDNA was synthesized from the total RNA using a cDNA synthesis kit (Servicebio, Hubei, China), and quantitative real-time PCR (qPCR) was performed using SYBR Green (Servicebio, Hubei, China) on a real-time PCR system. GAPDH was used as the internal reference gene to normalize the expression levels of target genes. The relative expression of genes related to chondrogenic and osteogenic differentiation was analyzed using the comparative Ct (2^−ΔΔCT) method. All primers are shown in Supplementary Table S1.

Passage 2 or 3 rat BMSCs were cultured with different adhesives in chondrogenic differentiation medium and on the slides. The slides were washed with PBS. A 3% bovine serum albumin (BSA) solution was added to the reaction area to fully cover the sample, and the sample was incubated at room temperature. After discarding the blocking solution, primary antibody working solution against aggrecan (13,880-1 AP, ProteinTech, Hubei, China) was added. The samples were transferred to a culture chamber and incubated at 4°C for 16–18 h. Following incubation, the samples were incubated with a CY3-labeled goat anti-rabbit IgG secondary antibody (111-165-003, Jackson, MS, United States) at room temperature. After additional PBS washes, DAPI staining solution was added for nuclear staining. Coverslips were mounted using anti-fade mounting medium. The prepared samples were immediately imaged using a laser confocal microscope (Leica, Wetzlar, Germany). Semi-quantitative analysis of immunofluorescence intensity was performed using ImageJ software.

Passage 2 or 3 rat BMSCs were cultured with different adhesive extracts in osteogenic differentiation medium. After removing the culture medium, the samples were washed with PBS and then fixed with 4% paraformaldehyde (PFA) at room temperature. After fixation, the samples were washed again with PBS. The ALP staining kit (Beyotime, Shanghai, China) was applied. A chromogenic working solution was prepared by mixing BCIP stock solution, NBT stock solution, and ALP chromogenic buffer. The samples were covered with the staining solution and incubated at room temperature. Microscopic imaging was performed using a light microscope. The ALP activity assay kit (Beyotime, Shanghai, China) was applied. The chromogenic substrate was prepared, and p-nitrophenol standards were serially diluted to a working concentration. After removing the culture medium, RIPA lysis buffer (Beyotime, Shanghai, China) was added to each well, and the supernatant was collected for analysis. Gradient volumes of the standard solution and different testing samples were prepared and measured following the kit. Enzyme activity was calculated based on the standard curve, expressed in units of enzyme activity within the diethanolamine (DEA) buffer system. Total protein concentration of the samples was also determined, and the final result was presented as enzyme activity units per protein concentration.

The animal experiments were approved by the institutional animal care and use committee of Shanghai Sixth People’s Hospital affiliated to Shanghai Jiao Tong University School of Medicine (Approval No.: DWLL2025-0818). A total of 40 male Sprague-Dawley (SD) rats (weighing 250–300 g) underwent rotator cuff injury and repair surgery, as detailed below: Anesthesia was induced using 2%–3% isoflurane (Reward life science, Guangdong, China) delivered via an animal gas anesthesia system (EZSystem, CA, United States). After skin incision, the distal clavicle was located, and the coracoclavicular ligament was severed to expose the supraspinatus tendon insertion site (Supplementary Figure S2A,D). The tendon was detached from the humeral insertion using a scalpel, and the tendon-to-bone insertion site was carefully debrided. A bone tunnel was created in the humerus using an electric drill, and a 4-0 suture was passed through the tunnel. According to experimental group assignment, 200 μL of the corresponding adhesive hydrogel was implanted between the tendon and bone (Supplementary Figure S2B,E). The sutures were then passed through the torn end of the supraspinatus tendon and tied beneath the bone tunnel to secure the repair (Supplementary Figure S2C,F). The skin was closed using 3-0 sutures, and penicillin was administered postoperatively to prevent infection. The rats were divided into four groups. Defect group: rotator cuff injury and repair surgery without hydrogel implantation; ST group: repair surgery with implantation of ST group hydrogel; ST/dECM group: repair surgery with implantation of ST/dECM hydrogel; ST/dECM/ZIF-8 group: repair surgery with implantation of ST/dECM/ZIF-8 hydrogel. At 4 weeks post-surgery, 16 rats (4 per group) were euthanized for tissue collection and histological staining. At 8 weeks post-surgery, the remaining 24 rats (6 per group) were euthanized for histological staining, micro-CT analysis, and biomechanical testing of the humerus-supraspinatus complex (Supplementary Figure S2G).

The humerus-supraspinatus tendon units were dissected from the rats (Supplementary Figure S4). The samples were fixed in a universal tissue fixative, followed by decalcification, graded ethanol dehydration, and paraffin embedding. Longitudinal sections (5 μm thick) were cut along the long axis of the supraspinatus tendon and subjected to H&E and Masson’s trichrome staining. Tendon-to-bone healing at the reconstructed supraspinatus tendon insertion was evaluated using the tendon maturation scoring system proposed by Ide et al. (Ide et al., 2009). Histological images were reviewed in a blinded manner by two histologists. Additional sections of the humerus-supraspinatus tendon unit were also prepared for safranin O and picrosirius red staining analysis. For picrosirius red staining, images were obtained with a polarized microscope (DS-Ri2, Nikon, Tokyo, Japan). Five randomly selected areas underwent quantitative assessment using gray scale values. These values correspond to collagen birefringence against a dark background, with higher values correlating with greater collagen fiber maturity (Jiang et al., 2022). Semiquantitative analyses were performed using ImageJ software.

Tissue sections were sequentially immersed in xylene, followed by dehydration through a graded anhydrous ethanol series, and then rinsed with deionized water. Antigen retrieval was performed using EDTA buffer in a microwave oven. After cooling, 3% hydrogen peroxide solution was applied and incubated in the dark at room temperature. After washing with PBS, 3% BSA blocking solution was applied and incubated at room temperature. The tissue sections were then incubated overnight at 4°C in a humid chamber with diluted primary antibody iNOS (HA722031, Huabio, Zhejiang, China) and CD206 (ab300621, Abcam, MA, United States) solution. After PBS washing, species-matched secondary antibodies conjugated with horseradish peroxidase (ab205718, Abcam, MA, United States) were applied and incubated at room temperature. DAPI staining solution was added and incubated in the dark, followed by PBS washes. Finally, the sections were sealed with an anti-fade mounting medium. Fluorescence images were acquired using a laser confocal microscope.

The humerus-supraspinatus units were harvested from the rats and fixed in a universal tissue fixative to preserve tissue integrity. The samples were then scanned using a micro-CT scanner (nanoVoxel-2000, Sanying Precision Engineering, Tianjin, China). Scanning parameters were set as follows: X-ray source voltage at 85 kV, current at 100 μA, 360° rotation with 720 frames at 0.5° per frame, and a scan resolution of VoxelDis = 9.99 μm. Images were acquired using NanoVoxel Scan software version 2.1.702.0 (Sanying Precision Engineering, Tianjin, China), and three-dimensional reconstruction was performed using VoxelStudio Recon software version 2.5.1.25 (Sanying Precision Engineering, Tianjin, China). Image analysis was conducted using CT Vox software version 3.2 (Bruker, Karlsruhe, Germany). The region of interest (ROI) was the rotator cuff bone tunnel, and the analyzed parameters included bone volume fraction (BV/TV) and trabecular number (Tb.N).

The humerus of the experimental animals was fixed in a 1.5 mL centrifuge tube using hot-melt adhesive, ensuring full exposure of the articular surface of the humeral head. A scalpel was used to carefully dissect along the direction of the supraspinatus muscle fibers, preserving the integrity of the tendon-to-bone interface structure. The tendon was secured using sandpaper to ensure stability during testing. The attachment area of the tendon was measured using a vernier caliper. Biomechanical testing was conducted using a mechanical testing machine (HSS-DX1000, Heng Rui Jin Testing Machine Co., Ltd., Shandong, China). Each sample was initially preloaded with a tensile force of 0.1 N, followed by loading at a constant rate of 10 mm/min until failure. The load-displacement curve was recorded, and stiffness was calculated accordingly. The maximum load was directly obtained from the testing machine, and the ultimate stress was calculated by dividing the ultimate load at failure by the cross-sectional area of the tendon.

The data for DNA, collagen, and GAG content in the decellularized matrix were analyzed using an unpaired t-test. One-way analysis of variance (one-way ANOVA) followed by Tukey’s post hoc test was used for analyzing lap shear test mechanical properties, swelling ratio and degradation rate at different time points, qPCR results, MTT assay, semi-quantitative analysis of immunofluorescence intensity, ALP activity assay, Safranin O staining heterochromia, micro-CT results, and biomechanical testing data from ex vivo rat tendon-bone unit experiments. Histological scores were compared using the Kruskal-Wallis test. All statistical analyses were performed using GraphPad Prism version 9.0.0. Significance levels were indicated: no significance (ns), *p < 0.05, **p < 0.01, ***p < 0.001.

H&E and Masson’s trichrome staining were used to compare the porcine intervertebral disc annulus fibrosus tissue before and after decellularization (Figures 2A,B). The results indicated that the original nuclear structures had largely disappeared following decellularization, indicating successful removal of cellular components. Moreover, the tissue morphology of the annulus fibrosus was well preserved without extensive structural damage. To further quantitatively assess changes in extracellular matrix components after decellularization, biochemical assays were performed to measure the contents of DNA, collagen, and GAGs before and after treatment. The DNA content significantly decreased after decellularization (Figure 2C), meeting the standard for decellularized materials (less than 50 ng/mg). The collagen content in the decellularized annulus fibrosus remained comparable to that of the untreated group (Figure 2D). Although the GAG content in the decellularized group was reduced compared to the untreated group, it remained at a relatively high level overall (Figure 2E).

The synthesized ZIF-8 was characterized using SEM and EDX for elemental mapping. The diameter of ZIF-8 particles was approximately 80 nm. The EDX spectrum revealed characteristic peaks for Zn and N elements, corresponding to the metallic and organic components of the ZIF-8 ligand, respectively (Supplementary Figure S1A). Supplementary Figure S1B displays the XRD pattern of ZIF-8. The diffraction peak at 7.6° corresponds to the [011] plane, representing the most intense peak of ZIF-8 and indicating diffraction from a densely packed plane in a cubic crystal system. The peak at 10.6° corresponds to the [002] plane, a characteristic reflection of the cubic structure, indicating periodic arrangement along the c-axis. The 13.0° peak corresponds to the [112] plane, a higher-order diffraction peak that further confirms the cubic symmetry. The peak at 18.2° corresponds to the [222] plane, reflecting smaller interplanar spacing (Zhou et al., 2017). Additionally, the FTIR spectrum shown in Supplementary Figure S1C reveals an absorption peak at 1,585 cm^-1 attributed to C=N stretching vibration of the imidazole ring, a peak at 1,144 cm^-1 corresponding to C–N stretching vibration, and a peak at 420 cm^-1 associated with the Zn–N coordination bond vibration (Maghsoudi et al., 2024).

According to previous reports, in silk fibroin-tannic acid adhesives, the concentration of silk fibroin solution typically ranges from 5 wt% to 10 wt%, and tannic acid solution ranges from 0.1 g/mL to 0.6 g/mL (Luo et al., 2020; Guo et al., 2024; Kim et al., 2023). Since the silk fibroin solution would be further diluted in subsequent experiments, a higher concentration of 10 wt% silk fibroin solution was selected for this study. Additionally, a 0.3 g/mL tannic acid solution was chosen because its pH is close to the theoretical isoelectric point of silk fibroin, which facilitates rapid gelation through a dual mechanism: suppressing electrostatic repulsion between molecules and inducing hierarchical and orderly assembly of adjacent silk fibroin chains (Zhang et al., 2011). Based on previous literature, cartilage-derived dECM, used to induce chondrogenic differentiation of mesenchymal stem cells, is typically applied at concentrations between 1 wt% and 5 wt%. In this study, concentrations of 1.25 wt% and 2.5 wt% were selected for evaluation (Zeng et al., 2022; Li et al., 2024). ZIF-8 was chosen at 0.5 wt% and 1 wt% based on prior findings that indicate its biosafety and osteoinductive properties within certain concentration ranges (Tang et al., 2024).

To assess the wet-state tissue adhesive strength, five adhesive formulations were tested using a universal testing machine (Figure 2F). The five groups were formulated as follows: group A: 10 wt% silk fibroin solution and 0.3 g/mL tannic acid solution mixed at a volume ratio of 1:1; group B: 10 wt% silk fibroin solution, 1.25 wt% dECM solution, and 0.3 g/mL tannic acid solution mixed at a volume ratio of 2:1:2; group C: 10 wt% silk fibroin solution, 2.5 wt% dECM solution, and 0.3 g/mL tannic acid solution mixed at a volume ratio of 2:1:2; group D: 10 wt% silk fibroin and 2.5 wt% dECM mixture containing 0.5 wt% ZIF-8, mixed with 0.3 g/mL tannic acid solution at a volume ratio of 2:1:2; group E: 10 wt% silk fibroin and 2.5 wt% dECM mixture containing 1 wt% ZIF-8, mixed with 0.3 g/mL tannic acid solution at a volume ratio of 2:1:2. Figure 2G shows the representative tensile strength curves for groups A-E. The addition of dECM at different concentrations (groups B and C) to the base formulation (group A) did not significantly affect the adhesive mechanical strength (Figure 2H). Since a higher concentration of dECM is more favorable for chondrogenic differentiation of mesenchymal stem cells, group C was selected for further experiments. Based on the composition of group C, different amounts of ZIF-8 were added (groups D and E). The results showed that 0.5 wt% ZIF-8 (group D) did not negatively affect mechanical performance, whereas 1 wt% ZIF-8 (group E) led to a reduction in mechanical strength. Therefore, group D was chosen for subsequent studies. Based on the tissue adhesion testing results, three formulations were selected for further material, cellular, and animal experiments. They were ST group: 10 wt% silk fibroin solution and 0.3 g/mL tannic acid solution mixed at a 1:1 volume ratio; ST/dECM group: 10 wt% silk fibroin solution, 2.5 wt% dECM solution, and 0.3 g/mL tannic acid solution mixed at a 2:1:2 volume ratio; ST/dECM/ZIF-8 group: 10 wt% silk fibroin solution containing 0.5 wt% ZIF-8, 2.5 wt% dECM solution, and 0.3 g/mL tannic acid solution mixed at a 2:1:2 volume ratio.

As shown in Figure 2I, adhesives prepared using only tannic acid and dECM exhibited poor gelation and insufficient adhesive strength and were therefore excluded from further investigation. However, adding silk fibroin to the dECM prior to mixing with tannic acid significantly improved gelation performance (Figure 2J). Figures 2K–M show the adhesion performance of ST, ST/dECM, and ST/dECM/ZIF-8 adhesives in bonding porcine skin directly to a glass slide. In the ST/dECM/ZIF-8 group, green food dye was added to the adhesive and applied to the porcine skin (Figure 2N). Following mechanical manipulation such as stretching (Figure 2O) and twisting (Figure 2P) of the porcine skin, the ST/dECM/ZIF-8 adhesive demonstrated excellent adhesion. Furthermore, the ST/dECM/ZIF-8 adhesive remained well-attached to the porcine skin even under running water, indicating its strong tissue adhesion in dynamic wet environments (Figure 2Q).

The SEM results showed that the mixture of dECM and tannic acid alone failed to form a well-structured gel, whereas the other three formulations are able to form relatively dense hydrogel structures (Figure 3A). According to the FTIR spectra of the samples (Figure 3B), the shared broad peak at 3,276–3,291 cm^-1 corresponded to O–H and N–H stretching vibrations, indicating the presence of abundant hydrogen bonds in all samples. The common peak at 1,445 cm^-1 was attributed to CH₂/CH₃ bending or aromatic C=C stretching vibrations, typically found in proteins and phenolic compounds. Characteristic peaks of silk fibroin included 1,656 cm^-1 (amide I band), associated with C=O stretching in β-sheet structures; 1,512 cm^-1 (amide II band), attributed to N–H bending and C–N stretching; 1,313 cm^-1 (amide III band), associated with C–N and N–H vibrations reflecting secondary structural changes; and 1,082 cm^-1 and 1,016 cm^-1, corresponding to C–O vibrations. For tannic acid, the characteristic peak at 1700 cm^-1 was due to C=O stretching, while peaks at 1,607 cm^-1 and 1,534 cm^-1 corresponded to C=C stretching in aromatic rings, and 1,175 cm^-1 was due to C–O stretching in phenolic hydroxyl groups. In the ST, ST/dECM, and ST/dECM/ZIF-8 composite hydrogels, the characteristic tannic acid peak at 1700 cm^-1 disappeared or was weakened, likely due to hydrogen bonding or ionic interactions between ester or carboxyl groups in tannic acid and amino groups in silk fibroin, resulting in shifts or reductions in C=O peak intensity. No significant changes in the FTIR spectra were observed before and after the addition of dECM, possibly because the dECM content was low compared to silk fibroin and its major components were similar to those of silk fibroin, leading to overlapping peaks. In the ST/dECM/ZIF-8 composite hydrogel, new peaks appeared: 1,149 cm^-1 corresponded to C–N stretching in the imidazole ring of ZIF-8, 998 cm^-1 might be attributed to Zn–N coordination or imidazole ring deformation in ZIF-8, and 696 cm^-1 corresponded to out-of-plane bending of the imidazole ring, indicating successful incorporation of ZIF-8. These results suggested that hydrogen bonding, coordination interactions, and hydrophobic interactions played a dominant role in forming the multicomponent composite system, and the inclusion of ZIF-8 further enhanced the crosslinked network via metal coordination.

After 24 h, the swelling ratios of the three adhesives were 33.24% ± 2.17%, 28.83% ± 3.12%, and 30.74% ± 3.71%, respectively (Figure 3C). After 14 days, their degradation rates were 47.17% ± 2.78%, 44.12% ± 2.94%, and 43.15% ± 2.74%, respectively (Figure 3D). The three bioadhesives exhibited similar swelling and degradation behavior, indicating that the addition of dECM and ZIF-8 had no significant impact on the swelling or degradation properties of the adhesives.

Co-culture of rat BMSCs with the extracts of ST, ST/dECM, and ST/dECM/ZIF-8 adhesives demonstrated good cytocompatibility on days 1, 2, and 3 of culture (Figure 4A). The MTT semi-quantitative assay results indicated that rat BMSCs proliferated in all three adhesive extracts, with no significant differences in proliferation trends among the groups (Figure 4B).

After 14 days of co-culture of rat BMSCs with adhesives containing different formulations in chondrogenic induction medium, the expression levels of chondrogenic-related genes ACAN, and COL2 were significantly upregulated in both the ST/dECM and ST/dECM/ZIF-8 groups compared to the ST group (Figures 4C,D). Furthermore, immunofluorescence staining revealed that the protein expression levels of ACAN in the ST/dECM and ST/dECM/ZIF-8 groups were also significantly higher than in the ST group (Figures 5A,B). These results indicated that the addition of decellularized extracellular matrix to the adhesives promoted chondrogenic differentiation of rat BMSCs in vitro.

After 14 days of co-culture of rat BMSCs in osteogenic induction medium containing adhesive extracts with different formulations, the expression levels of osteogenesis-related genes RUNX2, and OCN were significantly higher in the ST/dECM/ZIF-8 group compared to the ST and ST/dECM groups (Figures 4E,F). After 14 days of co-culture in osteogenic induction medium, ALP staining results demonstrated a significant increase in ALP staining intensity in the ST/dECM/ZIF-8 group compared to the ST and ST/dECM groups, indicating a stronger osteogenic effect (Figure 5C). Additionally, semi-quantitative analysis of ALP activity assays further confirmed that the ST/dECM/ZIF-8 group had a more pronounced promoting effect on the osteogenic differentiation of rat BMSCs (Figure 5D).

Eight weeks after implantation of adhesives with different formulations, H&E staining revealed no significant damage to the heart, liver, spleen, lungs, or kidneys in rats (Supplementary Figure S3). These results indicated that the adhesives possessed good tissue compatibility with major organs.

To investigate the inflammatory response at the tendon-to-bone interface following rotator cuff repair, we performed iNOS and CD206 immunofluorescence staining on tissue sections collected 8 weeks post-surgery. Cells with high iNOS expression are indicative of M1-type macrophages, while those with high CD206 expression represent M2-type macrophages. INOS staining at 8 weeks post-operation revealed a high number of M1 macrophages at the enthesis in the Defect group, whereas the number of M1 macrophages was significantly reduced in the ST, ST/dECM, and ST/dECM/ZIF-8 groups (Figure 6A). CD206 staining demonstrated that the ST, ST/dECM, and ST/dECM/ZIF-8 groups had a significantly higher number of M2 macrophages at the tendon-to-bone interface compared to the Defect group (Figure 6B). These findings suggested that tannic acid in the adhesive might exert a strong immunomodulatory effect on the postoperative inflammatory response by promoting the activation of M2 macrophages and suppressing the excessive accumulation of M1 macrophages.

Histological evaluation of the tendon-to-bone interface at 4 weeks post-surgery revealed that although the ST, ST/dECM, and ST/dECM/ZIF-8 groups exhibited higher histological scores than the Defect group (ST vs. Defect: P = 0.169; ST/dECM vs. Defect: P = 0.083; ST/dECM/ZIF-8 vs. Defect: P = 0.054), no statistically significant differences were observed between groups (Figures 7A,B). At 8 weeks postoperatively, histological assessment of the tendon-to-bone interface at this time point showed significantly higher scores in the ST/dECM and ST/dECM/ZIF-8 groups compared to the Defect group. Although the scores were also higher than those in the ST group, the differences were not statistically significant (Figures 7C,D).

Polarized light microscopy with picrosirius red staining was used to distinguish collagen types: type I collagen appeared orange or red, while type III collagen appeared green. At 8 weeks postoperatively, the fiber orientation became aligned, particularly in the ST/dECM and ST/dECM/ZIF-8 groups, which showed notably improved structural organization (Figure 8A). Semi-quantitative analysis at 8 weeks post-surgery showed that the ST/dECM and ST/dECM/ZIF-8 groups had significantly higher values than the Defect group, and the ST group also showed some improvement (Figure 8B). These results confirmed that adhesive implantation promoted type I collagen deposition and organized fiber remodeling, thereby enhancing the quality of tendon repair.

For safranin O staining, cartilage appears red. At 8 weeks post-surgery, the ST/dECM and ST/dECM/ZIF-8 groups displayed larger red-stained areas at the tendon-to-bone interface compared to the Defect and ST groups, indicating that adhesives containing dECM promoted greater cartilage formation at the repair site (Figures 8C,D). These findings suggest that the addition of dECM enhanced cartilage regeneration at the tendon-to-bone interface.

The regeneration of the humeral bone tunnel in rats was evaluated using micro-CT at 8 weeks post-surgery (Figure 9A). The bone volume fraction (BV/TV) in the ST/dECM/ZIF-8 group was 30.67% ± 4.04%, which was significantly higher than that in the Defect group (19.67% ± 5.03%) and the ST group (18.00% ± 2.00%). Although the BV/TV was not significantly different from the ST/dECM group (21.00% ± 4.36%), it still showed a noticeable increase (Figure 9B). The trabecular number (Tb.N) in the ST/dECM/ZIF-8 group was 1.69 ± 0.06 1/mm, significantly higher than that in the Defect group (1.41 ± 0.09 1/mm), ST group (1.44 ± 0.08 1/mm), and ST/dECM group (1.48 ± 0.08 1/mm) (Figure 9C). These results indicated that the ST/dECM/ZIF-8 group significantly promoted bone tunnel healing at 8 weeks postoperatively compared to the Defect and ST groups.

No obvious gross morphological differences were observed among the humerus-supraspinatus complexes in each group at 8 weeks post-surgery (Supplementary Figure S4). Biomechanical testing of the humerus-supraspinatus units was performed using the platform (Figure 9D), following previously reported tensile loading directions (He et al., 2021; Chae et al., 2021). There were no significant differences in the cross-sectional area at the tendon-bone insertion site among the groups (Figure 9E). The maximum load to failure was significantly higher in the ST (14.25 ± 2.50 N), ST/dECM (15.38 ± 4.10 N), and ST/dECM/ZIF-8 (14.83 ± 3.43 N) groups compared to the Defect group (7.91 ± 1.04 N), although no significant differences were observed among the three adhesive groups (Figure 9F). Similarly, the maximum stress in the ST (2.08 ± 0.41 N/mm²), ST/dECM (2.19 ± 0.41 N/mm²), and ST/dECM/ZIF-8 (2.07 ± 0.47 N/mm²) groups was significantly higher than in the Defect group, but again, no significant differences were detected among the three adhesive-treated groups (Figure 9G). The maximum stiffness in the ST (10.33 ± 2.95 N/mm) and ST/dECM/ZIF-8 (10.74 ± 2.55 N/mm) groups was significantly higher than in the Defect group (3.96 ± 1.35 N/mm). Although the ST/dECM group (9.36 ± 4.64 N/mm) showed increased stiffness as well, no significant differences were found among the three adhesive-treated groups (Figure 9H). These results indicated that all three adhesive formulations significantly improved the biomechanical strength of tendon-to-bone healing in the rotator cuff repair model. The enhancement in mechanical performance might be attributed to the increased adhesive strength provided by the combination of silk fibroin and tannic acid used in the adhesives.