A biodegradable polyurethane adhesive augments tendon–bone interface healing in a rat Achilles repair model

Abstract

Aims

Surgical repair infrequently restores the native tendon–bone interface (TBI, also referred to as the enthesis), resulting in excessive scar formation, compromised biomechanics, and frequent re-tears. A simple, biodegradable surgical adjunct that supports tendon–bone healing could augment standard repair strategies. In this study, we evaluated a biomimetic, amino acid-based biodegradable polyurethane adhesive as an intraoperative adjunct to standard suture repair in a rat Achilles tendon–calcaneus model.

Methods

A total of 90 female Sprague-Dawley rats (8 weeks old) were randomly assigned to two groups: suture-only and suture + adhesive (n = 45 per group). In the enthesis injury model, after transection and drilling, the polymer adhesive was freshly prepared and applied directly to the tendon–bone interface prior to suture tightening. The contralateral limb served as an intact control. Animals were sacrificed at four-, eight-, and 12-week timepoints, and tissues were evaluated using histological, imaging, micro-CT, and biomechanical analyses.

Results

Compared with suture-only repair, the suture + adhesive treatment was associated with more organized fibrocartilage formation, improved collagen I/III organization, greater proteoglycan deposition, and a smaller, more compact repair interface. When normalized by cross-sectional area, the adhesive group exhibited higher ultimate stress (tissue-level mechanical strength), although ultimate stress remained below that of intact controls. Whole-construct mechanical properties were comparable between groups, suggesting that the adhesive served as a biological and structural adjunct that was associated with improved intrinsic tissue quality, rather than increased construct-level strength.

Conclusion

These findings suggest that a biodegradable polyurethane adhesive is associated with improved structural organization and intrinsic tissue-level mechanical properties of the repaired tendon–bone interface. Its straightforward intraoperative use and favourable biocompatibility profile support further investigation of its potential as an adjunctive strategy for tendon–bone repair.

Cite this article: Bone Joint Res 2026;15(9):1111–1122.

Article focus

• Does adjunctive application of a biodegradable polyurethane adhesive enhance structural healing and organization of the tendon–bone interface compared with standard suture repair alone?

• Does adhesive augmentation increase fibrocartilage formation and matrix composition (proteoglycan deposition) at the repair site?

• Does adhesive treatment improve intrinsic tissue-level mechanical quality of the repaired interface without acting as a primary load-bearing element?

Key messages

• Adjunctive use of a biodegradable adhesive is associated with improved fibrocartilaginous organization, greater proteoglycan deposition, and reduced hypertrophic remodelling of the tendon–bone interface.

• Adhesive-augmented repairs show higher area-normalized ultimate stress, indicating improved intrinsic tissue quality despite similar construct-level mechanical properties.

• A simple, intraoperatively applicable biomaterial was associated with improved structural healing of the tendon–bone interface and represents a potentially practical adjunct to standard repair.

Strengths and limitations

• This study used a well-established in vivo enthesis repair model with multimodal assessment (histology, imaging, micro-CT, and biomechanics), standardized region-of-interest definition, and blinded histological evaluation.

• Structural and functional outcomes were evaluated over multiple healing timepoints, including area-normalized mechanical analysis to assess intrinsic tissue quality.

• This rat model was single-sex (female), with limited healing duration relative to human repair, as well as a lack of mechanistic, immunological, or degradation-kinetics analyses, including no direct in vivo assessment of inflammatory response.

Introduction

The tendon–bone interface (TBI), or enthesis, is a highly specialized region that enables efficient force transmission from muscle to skeleton. It comprises a graded, four-zone structure (tendon, unmineralized fibrocartilage, mineralized fibrocartilage, and bone), which minimizes stress concentration and ensures mechanical continuity across the soft–hard tissue junction.^1-3 Surgical repair of tendon–bone injuries, such as the Achilles tendon enthesis, rotator cuff tendon insertions, or anterior cruciate ligament (ACL) insertions, infrequently restores this native architecture.⁴ Instead, healing typically produces a fibrovascular scar lacking zonal organization, leading to inferior biomechanics and high re-tear rates that can reach 40% to 60% in rotator cuff repairs.^5-7 This persistent failure to regenerate a functional enthesis remains a major clinical challenge and a key contributor to long-term repair failure.

To address this challenge, numerous regenerative strategies have been investigated, including advanced fixation devices, such as demineralized bone fibre implants, which not only serve as mechanical fixation devices but also possess osteoconductive properties and are already used clinically,⁸ as well as preclinical strategies such as growth-factor delivery systems, stem cell-based therapies, exosome-loaded biomaterials, and multiphasic scaffolds designed to reproduce the native gradient.^9-15 Although many approaches show promise in preclinical models, clinical translation remains limited by high cost, complex manufacturing, scalability issues, and uncertain long-term safety or efficacy. Moreover, many biological or advanced scaffold-based strategies require specialized handling, regulatory complexity, or prolonged preparation that limit their practicality in routine surgical settings. Even successful biological interventions often fail to re-establish the hierarchical structure or provide adequate mechanical integrity for sustained loading.

Biomimetic materials provide an attractive alternative because they can be applied intraoperatively to support the healing environment while avoiding the limitations of cellular or growth-factor therapies. Several injectable hydrogels and functionalized scaffolds have enhanced enthesis repair by offering physical support and bioactive cues.^10,12,14,16 Adhesive biofilm application and adhesive hydrogels have also been reported to improve tendon-to-bone healing.^17,18 However, many such systems still require complex delivery methods, integrate poorly with standard surgical workflows, or degrade into cytotoxic by-products that impair healing. Medical adhesives offer a simple intraoperative augmentation strategy, yet current options such as polymethyl methacrylate (PMMA), cyanoacrylates, and aromatic polyurethane glues are constrained by cytotoxicity, poor biodegradability, or harmful degradation products.^19-22 As a result, there remains a need for biodegradable, biocompatible surgical adhesives that can be readily integrated into standard tendon–bone repair procedures, and that support structural healing and functional restoration rather than merely providing fixation.

Biodegradable adhesive systems reported in the literature include fibrin-based sealants, synthetic polyurethane adhesives, and bioadhesive hydrogels. Fibrin sealants are widely used for soft-tissue fixation and haemostasis, whereas synthetic polyurethane adhesives and hydrogel-based systems have been explored for tissue sealing and regenerative applications. However, many of these approaches primarily function as bonding or sealing agents, and may not actively support structural organization of the healing interface or integrate seamlessly with standard surgical workflows.

To address this need, we evaluated a bioinspired amino acid-based polyurethane adhesive modelled after mussel adhesive proteins. Synthesized from L-lysine ethyl ester diisocyanate (LDI), 3,4-dihydroxy-L-phenylalanine (DOPA), cysteine, glycerol, and amino-poly(ethylene glycol) (BAPPG), the material degrades into small, biocompatible molecular species, minimizing the risk of long-term toxicity or foreign body reaction.²³ Its in situ polymerization forms a porous, conformable matrix that fills the tendon–bone defect during standard repair, providing local stabilization and a permissive environment for endogenous healing. Rather than serving solely as a bonding agent or replacing sutures, the adhesive is intended as an adjunct to conventional fixation that provides local support to the healing interface. Although the molecular components of this polyurethane system have been described previously, its use as a simple, intraoperative adjunct for tendon–bone repair, and its effects on enthesis structure and tissue-level mechanics in vivo, have not been systematically evaluated.

We hypothesized that combining this adhesive with standard suture repair would promote the formation of a more organized fibrocartilaginous tendon–bone interface, enhance matrix composition, and improve tissue-level mechanical properties compared with suture repair alone. A rat Achilles tendon–bone repair model was used to evaluate histological, imaging, and biomechanical outcomes. The aim of this study was to determine whether adjunctive use of a biodegradable adhesive is associated with improved structural and mechanical outcomes during tendon–bone healing in vivo, rather than to define its molecular mechanisms of action.

Methods

Materials

L-lysine ethyl ester diisocyanate (LDI; AAH6438406, Thermo Fisher Scientific, USA), L-3,4-dihydroxyphenylalanine (DOPA) methyl ester hydrochloride (D1507, Sigma Aldrich, USA), L-cysteine ethyl ester hydrochloride (CYS, C121908-25G), glycerol (GLY, 356352 M), and O,O′-bis(2-aminopropyl) polypropylene glycol-block-polyethylene glycol-block-polypropylene glycol (BAPPG; Cat. No. 14526-500 mL) were purchased from Millipore Sigma (USA). All other reagents were of analytical grade and used as received without further purification. The adhesive formulation and synthesis were based on a previously described biocompatible polyurethane platform.²³

Animals

Female Sprague–Dawley rats (eight weeks old; Charles River Laboratories, USA) weighing about 200 g were used. Female rats were selected in order to minimize variability related to body size, aggression, and postoperative activity, which can confound tendon–bone healing in group-housed males. Compared with males, female rats exhibit lower levels of inter-animal aggression, which facilitates group housing, reduces stress-related variability, and minimizes the risk of injury-related confounders. In addition, female rats typically show more stable body weight trajectories over time, which is advantageous for longitudinal studies and repeated dosing. All procedures were approved by the Institutional Animal Care and Use Committee and performed in accordance with ARRIVE guidelines, and we have included an ARRIVE checklist. Animals were housed under a 12-hour light/12-hour dark cycle at 21°C with ad libitum access to food and water.

Polymer adhesive synthesis

The adhesive was synthesized from LDI, GLY, BAPPG, DOPA, and CYS following a modified protocol (Figure 1).²³ Briefly, 0.92 g GLY (10 mmol; 30 mmol –OH groups) was added to a dry, nitrogen-flushed round-bottom flask. LDI (6.78 g, 30 mmol; 60 mmol –NCO groups) was slowly injected and stirred for three days at room temperature to yield a LDI–GLY pre-polymer. Reaction progress was monitored by Fourier transform infrared spectroscopy (FT-IR) (Supplementary Figure 1). Separately, DOPA methyl ester hydrochloride (0.5 g, 2.02 mmol), CYS ethyl ester hydrochloride (0.38 g, 2.05 mmol), and BAPPG (8 mL, MW ≈ 600, density 1.035 g mL⁻¹; 13.8 mmol; 27.6 mmol –NH₂ groups) were combined to prepare ‘solution A.’ Equal volumes of LDI–GLY pre-polymer and solution A were mixed to initiate polymerization, which occurred within approximately five minutes (Figure 1e) to form a porous foam (Figure 1c) that served as both an adhesive and a temporary scaffold for the Achilles tendon–calcaneus interface. No additional crosslinkers, growth factors, or bioactive additives were incorporated.

Fig. 1.

Adhesive synthesis and properties. Photographs show the a) isocyanate-terminated L-lysine ethyl ester diisocyanate (LDI)–glycerol (GLY) prepolymer and b) 3,4-dihydroxy-L-phenylalanine (DOPA)–cysteine (CYS)–O,O′-bis(2-aminopropyl) polypropylene glycol-block-polyethylene glycol-block-polypropylene glycol (BAPPG)-containing solution, both of which are viscous liquids before mixing. c) Upon combining equal volumes of the two components, rapid polymerization occurred within five to ten minutes at room temperature, producing a reactive wet foam that d) subsequently matures into a porous, sponge-like solid. e) A schematic illustrates the polymerization process, the chemical composition of the final polyurethane adhesive, and the intended interfacial interactions between the adhesive and tissue surfaces.

Hydrolytic degradation analysis

Hydrolytic degradation of the adhesive polymer was evaluated in vitro by incubating polymer samples (approximately 1 mL) in phosphate-buffered saline (PBS) at 37°C. Samples were placed in dialysis tubing (molecular weight cutoff ~ 7 kDa) and immersed in 10 mL of PBS within sealed tubes (n = 3). At predetermined time points, 200 μL aliquots of the incubation medium were collected and replaced with an equal volume of fresh PBS to maintain constant volume. Degradation was assessed by measuring cumulative release of degradation products over time (Supplementary Figure 4B).

Enthesis injury model

A total of 90 rats were randomly assigned to two groups: suture-only and suture + adhesive (n = 45 per group). A rat Achilles tendon–calcaneus enthesis injury model was established as described previously.¹⁵ Under isoflurane anaesthesia, a 12 mm to 15 mm lateral incision was made to expose the enthesis. 1) Suture-only group: the enthesis was transected, a 1 mm transosseous tunnel drilled through the calcaneus, and the tendon reattached with a transosseous suture (Supplementary Figure 2A). 2) Suture + adhesive group: after preparation of the transosseous tunnel, freshly mixed adhesive was immediately applied to the repair interface using a sterile applicator. The adhesive components were prepared using two 50 μL syringes containing the respective co-polymers. Immediately prior to application, approximately 10 μL of each component was dispensed onto a sterile glass surface and mixed thoroughly. The freshly mixed polymer was then collected using a modified wide-bore pipette tip (approximately 5 mm of a 200 μL pipette tip was trimmed to accommodate the viscous material) and applied directly to the tendon–bone interface. The material was placed between the posterior surface of the calcaneus and the transected tendon edge prior to suture tightening. Following fixation, any excess polymer extruded from the interface was carefully removed using sterile gauze, and the site was gently rinsed with PBS to ensure consistent localization of the adhesive.

This procedure was performed immediately after mixing to ensure consistent handling and polymerization behaviour. The adhesive was used as an adjunct to, not a replacement for, standard fixation. All surgical procedures and adhesive applications were performed by a single surgeon (VP) using a standardized technique. The contralateral limb served as an intact control (Supplementary Figure 2B).

Postoperative care and euthanasia

Rats received neomycin (1.17 g L^-1) in drinking water for seven days and carprofen (5 mg kg^-1 subcutaneously) daily for four days post-surgery. Wounds were monitored daily. Animals were euthanized at four, eight, or 12 weeks by CO₂ inhalation followed by bilateral thoracotomy, in accordance with American Veterinary Medical Association (AVMA) Guidelines (2020).²⁴

Sample preparation

For biomechanical testing, calf muscles were dissected, the plantaris tendon and associated ligaments transected, and the tibia removed just above the ankle joint. Following dissection, specimens were immersed in 10 mL of PBS and stored at −20°C until biomechanical testing. Prior to testing, samples were thawed and kept hydrated throughout preparation and testing. This storage method is commonly used to preserve tendon tissue integrity and enable standardized batch mechanical analysis.^25-27

Histological evaluation

For histological analysis, the Achilles tendon–calcaneus complexes were harvested en bloc at the designated endpoints. Specimens were fixed in 4% formalin for 24 hours. Decalcification was performed using Formical-4 solution with solution exchange every 48 hours for a total of two to three exchanges. Completion of decalcification was verified using a chemical assay (ammonium oxalate and ammonium acetate), with absence of turbidity indicating effective calcium removal.

Following decalcification, samples were dehydrated through a graded ethanol series, embedded in paraffin, and sectioned in the sagittal plane. Tissue blocks were trimmed to ensure clear visualization of the tendon–bone healing interface, and serial sections (5 μm thick) were obtained at defined intervals. For all specimens, histological analyses were performed using sections obtained from comparable anatomical levels of the tendon–bone interface according to a standardized trimming and sectioning protocol.

Histological processing was performed by the institution’s Biospecimen Core using standard validated protocols. For haematoxylin and eosin (H&E) staining, slides were stained with haematoxylin, rinsed, differentiated in 0.3% acid alcohol, counterstained with eosin, dehydrated, and mounted.

For Safranin O/Fast Green (SO/FG), slides were stained with Fast Green (five minutes), rinsed in 1% acetic acid, stained with 0.1% Safranin O (five minutes), dehydrated, cleared in xylene, and mounted.

Picrosirius red staining (PRS) was performed in our laboratory using a commercially available kit (ab150681, Abcam, UK). Briefly, paraffin-embedded sections were deparaffinized in 100% xylene, followed by a 50:50 xylene/alcohol solution, and then rehydrated through a graded ethanol series (90%, 70%, 50%, 25%, and 0%), with each step performed for two minutes. Slides were subsequently stained with picrosirius red solution for 60 minutes, dehydrated through a graded ethanol series (0% to 100%), cleared in 100% xylene (with destaining performed if necessary), air-dried, and mounted using Cytoseal 60 with a coverslip.

All histological evaluations, image selection, and qualitative assessments were performed by investigators blinded to treatment allocation (VP, JZ). Representative images were selected from sections obtained at comparable anatomical levels using the standardized sectioning procedure described above.

Quantitative proteoglycan analysis

Three slides per sample (n = 3 per group) were imaged under identical settings. Proteoglycan (PG) expression was quantified using a custom OpenCV-based Python script that segmented red/violet pixels corresponding to PG staining. Total stained area was compared across groups.¹² Thresholding parameters were fixed across all images and verified visually to ensure consistent segmentation. Image acquisition and quantitative analysis were performed using identical acquisition settings and thresholding criteria for all specimens.

Picro sirius red staining and polarized light microscopy

Paraffin-embedded tissue blocks were sectioned into 5 µm slices, and the sections with the same number from each rat were stained using a Picro Sirius Red kit (Cat. #ab150681, Abcam) as per the manufacturer’s instructions. Stained tissue sections were analyzed under a polarized light microscope (Nikon, Japan).

Cross-sectional area measurement

Micro-CT analysis was performed using a Scanco µCT 50 system (Scanco Medical, Switzerland). Rat ankle joint specimens were scanned at a voxel size of 10.3 µm, with a tube voltage of 55 kVp, current of 145 µA, 0.36° rotation step over a 180° angular range, and 1,000 ms exposure time per projection. Specimens were scanned in 70% ethanol to maintain tissue integrity. 3D reconstruction and image visualization were performed using the manufacturer’s software.

To ensure consistent specimen orientation, all reconstructed datasets were aligned using the longitudinal axis of the calcaneus together with standardized anatomical landmarks corresponding to the posterior Achilles tendon insertion.

Following reconstruction, regions of interest (ROIs) were manually defined at the tendon–bone interface using these consistent anatomical landmarks. In the repaired groups, newly formed attachment regions were defined as surface irregularities exceeding approximately two voxels (~ 20.6 μm) relative to the native cortical surface, based on system resolution. The intact controls were analyzed using the corresponding native enthesis as the anatomical reference. The newly formed enthesis was segmented using identical criteria for all specimens. The 3D reconstruction was used to identify anatomically corresponding measurement locations, and cross-sectional area measurements were subsequently obtained from anatomically matched reconstructed slices using ImageJ (National Institutes of Health, USA). ROI identification and morphometric measurements were performed in a standardized manner by investigators blinded to treatment allocation (LL, VP). The same predefined anatomical landmarks and segmentation criteria were applied uniformly to all specimens to minimize user-dependent variability. Measurements were obtained from five specimens per experimental group and six intact controls.

Biomechanical testing

Samples (n = 11 per group at eight weeks) were mounted in custom-designed clamps and loaded at 0.5 mm·s⁻¹ to failure. Ultimate load and displacement were recorded; stiffness was calculated from the linear region of the load–displacement curve (≥ 25 N over a 0.5 mm interval). Because the morphology and cross-sectional area of the repaired tendon–bone interface differed between experimental groups, ultimate stress was calculated by normalizing ultimate load to the micro-CT-derived cross-sectional area of the newly formed enthesis. This normalization was performed to assess the intrinsic tissue-level mechanical properties of the repaired interface independent of repair size.

Additional tests were performed on intact tendons with sutures to assess the mechanical contribution of suturing alone. Failure mode (enthesis rupture, midsubstance tear, or suture pull-out) was recorded for each specimen.

In vitro evaluation of adhesive properties

For mechanical testing, chicken femur bones (n = 3) were used. Bone samples were first cut transversely, perpendicular to the longitudinal axis. Subsequently, 1.5 cm to 2.0 cm of bone was trimmed from both sides of the cut, as described previously, ensuring that the final bone ends were ≤ 0.6 cm in width to fit securely into the testing clamps.

Freshly mixed adhesive was prepared using two 1 mL syringes containing the respective co-polymers. Approximately 400 μL of each component was dispensed onto a glass surface and mixed immediately prior to application. The resulting viscous polymer was collected using a modified wide-bore pipette tip (approximately 3 mm of a 1,000 μL pipette tip was trimmed) and applied directly to the prepared bone surfaces.

The adhesive was placed between the opposing cut surfaces of the chicken femur diaphysis, and the samples were aligned under gentle compression to ensure uniform contact. A consistent volume sufficient to fully cover the bonding interface without excess overflow was used for each specimen. The bonded interface was covered and allowed to cure at room temperature under identical conditions for 48 hours prior to biomechanical testing. After curing, each end of the bonded bone was secured in opposing clamps of the MTS testing system. Tensile testing was performed using a 1,000 N load cell at a displacement rate of 0.01 mm·s⁻¹ until adhesive failure occurred. Load–displacement curves and corresponding numerical data were recorded for subsequent analysis.

Statistical analysis

Quantitative data are expressed as mean (SD). Distribution and variance were evaluated by Q–Q plots and boxplots. Two-group comparisons (PG quantification) used Welch’s two-tailed t-test. Three-group comparisons (morphometric and biomechanical analyses) used one-way analysis of variance (ANOVA) with Tukey’s post hoc test. Significance was defined as p < 0.05 (or FDR (false discovery rate)-adjusted p < 0.05). Analyses were performed using GraphPad Prism v. 9 (GraphPad Software, USA) and Python (SciPy, statsmodels).

Results

Polymer adhesive synthesis

The polyurethane adhesive was synthesized in two steps. An isocyanate-terminated LDI–glycerol pre-polymer was first produced by reacting LDI and glycerol (3:1 molar ratio) for three days at room temperature. The FT-IR spectrum of LDI displayed a characteristic absorption band at 2,260 cm⁻¹ corresponding to –NCO groups (Supplementary Figure 1A), whereas glycerol showed no absorbance in the 2,000 to 2,500 cm⁻¹ region (Supplementary Figure 1B). After three days, a new band appeared at 1,726 cm⁻¹ (-NHCOO- formation) accompanied by decreased –NCO intensity at 2,252 cm⁻¹ (Supplementary Figure 1C). Approximately 50% of free isocyanate groups remained, confirming successful synthesis of the LDI–GLY pre-polymer (Supplementary Figure 1D).

In the second step, DOPA and CYS were dissolved in BAPPG to prepare ‘solution A’. Both the pre-polymer (Figure 1a) and solution A (Figure 1b) were light-yellow viscous liquids. Mixing equal volumes triggered rapid polymerization within approximately five to ten minutes at room temperature (Figure 1e), generating a foamed polymer (Figure 1c) that subsequently matured into a porous, sponge-like solid (Figure 1d). This synthesis reproducibly yielded a conformable, porous material suitable for intraoperative application.

Polymer adhesive mechanical properties

We quantified the mechanical properties of the adhesive and measured a breaking load of 0.33 N (SD 0.06), an elongation of 9.66 mm (SD 4.31), and an ultimate stress of 11.5 kPa (SD 1.9) (Supplementary Figure 4A). These values are relatively low and reflect the inherently soft, compliant nature of the material. Due to the limited mechanical strength and the small cross-sectional area of the adhesive samples, we did not extend the in vitro biomechanical testing to a tendon-to-bone model. Nevertheless, the adhesive was readily mixed and applied, conformed to the irregular repair interface, and polymerized within a practical working time while remaining sufficiently compliant to function as an adjunctive gap-filling scaffold, rather than a rigid primary fixation material.

Hydrolytic degradation of the adhesive polymer was assessed in PBS at 37°C by monitoring cumulative lysine release over a six-week period. The polymer remained relatively stable during the first week, with minimal hydrolysis. After this initial period, lysine release increased progressively, reaching approximately 50% by day 25 and 75% by day 42, indicating time-dependent polymer degradation (Supplementary Figure 4B). These findings suggest that while the polymer retains structural integrity during the early post-application phase, it undergoes substantial hydrolytic degradation over time, which may influence long-term tissue remodelling and adhesive bioresorption.

Histological assessment (H&E, SO/FG)

Histology revealed distinct healing patterns between groups. In the suture-only repair, the interface consisted of disorganized fibrocartilage throughout the study (Figures 2a, 2e and 2i). In contrast, the suture + adhesive group exhibited progressive structural organization. At four weeks, a nascent transition zone containing bone, mineralized cartilage, unmineralized cartilage, and tendon was evident (Figure 2b). By eight weeks, the zone thickened and became structurally integrated (Figure 2f), and by 12 weeks, a continuous, native-like enthesis was observed (Figure 2j). These features were consistently observed across specimens within each group.

Fig. 2.

Representative histological evaluation of tendon–bone healing at a) to d) four weeks, e) to h) eight weeks, and i) to l) 12 weeks. Haematoxylin and eosin (H&E) staining shows disorganized fibrocartilage and poor structural continuity at the tendon–bone interface in the suture-only group (a, e, i, red arrows), whereas the suture + adhesive group exhibits progressive structural organization and formation of a multilayered interface morphology by 12 weeks (b, f, j, blue arrows). Safranin O/Fast Green (SO/FG) staining reveals weak, patchy proteoglycan (PG) deposition in the suture-only group (c, g, k, red and green arrows) versus broader and more continuous PG staining in the suture + adhesive group (d, h, l, green and yellow arrows). m) Quantitative image analysis confirms significantly greater PG deposition in the suture + adhesive group than in the suture-only group at all timepoints. t, tendon; b, bone. N = 9 for the suture-only and suture + adhesive groups. Black bars: 200 μm.

Safranin O/Fast Green staining corroborated these findings. The suture-only group showed sparse, patchy proteoglycan (PG) deposition (Figures 2c, 2g and 2k), whereas the adhesive group displayed robust and spatially continuous PG accumulation increasing from four to eight weeks and persisting at 12 weeks (Figures 2d, 2h and 2l). A well-defined PG gradient spanning bone-to-tendon was consistently observed. The spatial distribution of PG staining suggested a more organized fibrocartilaginous transition zone in the adhesive-treated interfaces.

Adhesive treatment is associated with increased proteoglycan deposition

Quantitative image analysis confirmed significantly greater PG deposition in the suture + adhesive group than in the suture-only group at all timepoints (Figure 2m). PG-positive pixels in the suture-only group measured 143,791 (SD 66,543) at four weeks, 854,193 (SD 560,235) at eight weeks, and 519,537 (SD 348,501) at 12 weeks, whereas the adhesive group exhibited 797,660 (SD 337,234) (p < 0.001), 2,532,297 (SD 426,643) (p < 0.001), and 2,188,778 (SD 755,524) (p < 0.001, all Welch’s two-tailed t-test), respectively. Across all timepoints, adhesive-treated interfaces showed a broader and more intense PG-positive zone, indicating greater accumulation of fibrocartilaginous matrix at the repair site.

Collagen I and III organization in PRS

Application of the adhesive prior to enthesis suturing resulted in improved collagen organization, maturation, and spatial distribution of collagen types I and III. Bright-field images (Figures 3a and 3e) and polarized-light PSR images at eight weeks (Figures 3b to 3d, 3f, 3g and 3h) demonstrated more highly organized collagen fibres and a reduced healing-site volume in the adhesive + suture group compared with the suture-only group. High-magnification polarized images of the healing region in the suture-only group revealed poorly organized, weakly expressed collagen type I fibres and disorganized collagen type III fibres. In contrast, the adhesive + suture group exhibited strongly expressed, well-organized thin collagen type I fibres and uniformly distributed, organized collagen type III fibres.

Fig. 3.

Representative Picrosirius Red-stained sections of the repaired tendon–bone interface at eight weeks after surgery. Bright-field and polarized-light overview PSR images demonstrate more highly organized collagen fibres and reduced healing-site volume in the suture + adhesive group compared with the suture-only group (black and white arrows; a and b vs e and f, respectively). High-magnification polarized images of the healing site in the suture-only group reveal poorly organized, weakly expressed collagen type I fibres and disorganized collagen type III fibres (orange-red and yellow-green birefringence, respectively; g and h), whereas the suture + adhesive group shows strongly expressed, well-organized thin collagen type I fibres and uniformly distributed, organized collagen type III fibres (orange-red and yellow-green birefringence, respectively; c and d).

Adhesive treatment is associated with a more compact repair morphology

Micro-CT at eight weeks revealed marked morphological differences between groups (Figures 4a to 4f). 3D reconstructions and corresponding surface views demonstrated a larger, irregularly expanded attachment region in the suture-only group compared with the smaller and more compact morphology observed in the suture + adhesive group and intact controls.

Fig. 4.

Polymer adhesive treatment is associated with a more compact repair morphology at the tendon–bone interface. Micro-CT reconstructions reveal an enlarged repair interface in the suture-only group (a and d), whereas the suture + adhesive group (b and e) exhibits a smaller and more compact interface morphology that more closely approaches the intact junction (c and f). All reconstructed specimens are displayed in the same posterior viewing orientation. Additional views with reduced or no overlay are provided to improve visualization of the underlying morphology and the newly formed attachment region (d to f). The newly formed enthesis region was defined using consistent anatomical landmarks and segmented based on surface irregularities exceeding approximately two voxels (~ 20.6 μm) relative to the native cortical surface (as illustrated in d to f). Histological assessment using Safranin O/Fast Green (SO/FG) staining demonstrated a thickened and disorganized enthesis healing site in the suture-only group (g and j), compared with a thinner and more organized interface in the suture + adhesive group (h and k), while the normal enthesis was observed in intact controls (i and l). m) Quantitative analysis of micro-CT data demonstrated that the repair cross-sectional area was significantly reduced in the adhesive-treated group compared with the suture-only group, although it remained greater than that of uninjured controls. n) Consistent with these findings, histological measurements demonstrated significantly increased enthesis thickness in the suture-only group compared with the adhesive-treated group. o) There was no difference in body weight between the experimental groups. These data are consistent with reduced repair tissue expansion rather than complete restoration of native morphology. Non-absorbable 5-0 polypropylene sutures were used in both the suture-only and suture + adhesive groups. N = 5 for the suture-only and suture + adhesive groups; n = 8 for the intact control group.

SO/FG staining demonstrated a significantly greater enthesis thickness in the suture-only group than in both the adhesive-treated group and intact controls (Figures 4g to 4i). The newly formed enthesis appeared enlarged in the suture-only group, whereas the adhesive-treated group exhibited a thinner and more organized repair interface (Figures 4j to 4l).

Quantitatively, the suture-only group displayed a significantly greater repair area (3.18 mm² (SD 0.81)) than both the adhesive group (2.07 mm² (SD 0.42)) and intact controls (1.16 mm² (SD 0.12)) (Figure 4m). Similarly, enthesis thickness was significantly greater in the suture-only group (3.09 mm (SD 0.18)) than in the suture + adhesive group (1.95 mm (SD 0.15)) (Figure 4n).

No significant difference in body weight was observed between the suture-only (264.5 g (SD 10.68)) and suture + adhesive (270.66 g (SD 16.01)) groups (Figure 4o), indicating that the observed morphological differences were unlikely to be attributable to differences in body weight.

Collectively, these findings demonstrate that adhesive-treated repairs exhibited a smaller and more compact tendon–bone interface than suture-only repairs, while remaining larger than the native enthesis.

Adhesive improves tissue-level mechanical properties

At the whole-construct level, ultimate load, displacement, and stiffness did not differ significantly among the suture, suture + adhesive, and intact groups (85.36 N (SD 21.22), 88.24 N (SD 16.10), and 81.87 N (SD 11.43); p > 0.05) (Figures 5a to 5c).

Fig. 5.

Polymer adhesive treatment is associated with improved tissue-level mechanical properties of the repaired tendon–bone interface. a) to c) Quantitative analysis showed no significant differences in ultimate load, displacement at failure, or stiffness among the three groups: suture-only, suture + adhesive, and intact (p > 0.05, one-way analysis of variance with Tukey’s post hoc test). d) In contrast, area-normalized ultimate stress (ultimate load normalized to the micro-CT-derived cross-sectional area of the newly formed enthesis) was significantly higher in the adhesive-treated group than in the suture-only group, although it remained below that of intact controls. Because the adhesive-treated group exhibited a smaller repair cross-sectional area (Figure 4m) while maintaining comparable construct-level mechanical properties, normalization of ultimate load to repair area yielded a higher tissue-level ultimate stress. Non-absorbable 5-0 polypropylene sutures were used in both the suture-only and suture + adhesive groups. Additional testing demonstrated that the suture contributed approximately 11.8 N to the ultimate load (Supplementary Figure 3). N = 11 for the suture-only and suture + adhesive groups and n = 6 for the intact control group.

However, when normalized to cross-sectional area, ultimate stress was substantially higher with adhesive treatment (42.48 MPa (SD 7.75)) than with suture-only repair (26.80 MPa (SD 6.66); p < 0.001), although still below intact control (70.31 MPa (SD 9.81)) (Figure 5d). One-way ANOVA (p < 0.001) and Tukey tests confirmed significant pairwise differences (p < 0.001).

Because adhesive-treated repairs exhibited a significantly smaller repair cross-sectional area (Figure 4m) while maintaining construct-level mechanical properties comparable to those of the suture-only group, normalization of ultimate load to repair area demonstrated a higher tissue-level ultimate stress. These findings indicate that comparable mechanical loads were supported by a smaller repair interface in the adhesive-treated group.

Mechanical tests on intact tendons repaired by sutures yielded an ultimate load contribution of 11.82 N (SD 1.76) (Supplementary Figure 3), indicating that the observed differences in tissue-level properties were not attributable to suture mechanics alone.

Discussion

This study demonstrates that a biomimetic, amino acid-based polyurethane adhesive is associated with improved structural tendon–bone healing when used as an adjunct to standard surgical repair. When applied intraoperatively to fill and support the tendon–bone interface, the adhesive was associated with more organized interface morphology, reduced repair tissue expansion, and higher tissue-level mechanical properties compared with suture repair alone. These findings suggest that this biodegradable adhesive can function not only as a bonding material but also as a temporary structural support during healing, thereby promoting improved structural organization and tissue-level mechanical properties at the healing tendon–bone interface beyond its role in mechanical fixation alone.

Restoring a continuous, graded transition – from tendon to unmineralized fibrocartilage, mineralized fibrocartilage, and bone – is essential for effective load transfer and long-term durability.^1-3 Traditional suture repair typically produces fibrovascular scar tissue lacking this zonal hierarchy and remaining mechanically inferior.^5-7 In contrast, adhesive augmentation was associated with the formation of a more organized fibrocartilaginous interface exhibiting features of the native hierarchical structure, suggesting that local structural support during healing may facilitate more organized tissue architecture. Although the repaired interface did not fully recapitulate the native enthesis, the improved structural organization observed in the adhesive-treated group compares favorably with findings reported for more complex regenerative approaches, including advanced scaffold- and stem cell-based strategies.^{9,10,12,14-16} Notably, the present adhesive can be applied intraoperatively without substantially altering standard surgical procedures, potentially offering a practical approach for augmenting tendon–bone repair.

The adhesive’s beneficial effects may reflect a combination of structural and biochemical influences, although the present study was not designed to distinguish among these potential mechanisms. During application, the material foams in situ to form a porous matrix that stabilizes the tendon–bone interface while potentially permitting cellular infiltration and matrix deposition. In parallel, hydrolytic degradation of the polymer under physiological conditions yields small, biocompatible molecules such as lysine (Supplementary Figure 4B). Lysine is a naturally occurring amino acid with established roles in collagen synthesis, extracellular matrix remodelling, and wound healing. Therefore, it is plausible that local release of lysine during adhesive degradation could contribute to matrix formation and interface maturation. However, because in vivo lysine concentrations, local degradation kinetics, inflammatory responses, and cell-specific biological effects were not directly evaluated, the present study cannot distinguish between the relative contributions of structural support and biochemical activity to the observed healing outcomes. Accordingly, any proposed biochemical mechanism should be considered hypothesis-generating and will require direct investigation in future studies.

Histological analysis revealed increased proteoglycan deposition in the adhesive group, particularly at the intermediate stage of healing. This finding may be interpreted in two non-mutually exclusive ways: it may reflect either enhanced fibrocartilaginous matrix formation or a transient accumulation of immature cartilaginous tissue during repair. Proteoglycans are abundant in early fibrocartilage and later become spatially restricted during maturation. Accordingly, the increased proteoglycan deposition observed in the adhesive-treated group is consistent with active matrix remodelling rather than necessarily indicating a fully mature enthesis. Because proteoglycan levels remained relatively stable between eight and 12 weeks despite continued polymer degradation, the present findings do not support a direct relationship between degradation products and proteoglycan accumulation. Longer-term studies incorporating molecular and histological markers of fibrocartilage maturation will be required to determine whether these changes progress toward a more mature tendon–bone interface.

At the tissue level, the structural differences observed by micro-CT and histology were accompanied by higher area-normalized ultimate stress in the adhesive-treated group, despite similar whole-construct load-bearing properties. Because the adhesive-treated repairs exhibited a significantly smaller repair cross-sectional area while maintaining comparable ultimate load, normalization of load to repair area demonstrated a higher tissue-level ultimate stress. This finding indicates that comparable mechanical loads were supported by a smaller repair interface and is consistent with improved structural efficiency of the repaired tissue rather than increased repair tissue volume. Although the calculation of ultimate stress depends on accurate measurement of repair cross-sectional area, standardized micro-CT-based morphometric analysis and predefined anatomical landmarks were used to minimize measurement variability. When interpreted together with the histological and morphometric findings, the combination of reduced repair tissue expansion and higher tissue-level ultimate stress suggests that adhesive augmentation was associated with the formation of a smaller, more organized, and mechanically competent repair interface rather than simply increasing tissue mass.²⁸

Relative to other regenerative strategies, such as growth-factor delivery, stem cell implantation, exosome-loaded scaffolds, and engineered gradient materials, the findings of the present study are generally consistent with previous reports while offering several practical advantages.^9-14 Specifically, the adhesive can be applied intraoperatively without specialized equipment, integrates directly with existing surgical workflows, and does not require cell processing, exogenous biologics, or extended preparation time. Importantly, unlike many existing adhesive or scaffold systems that primarily function as sealing or bonding agents, this material is intended to provide structural support to the healing interface while creating conditions that may facilitate tissue repair. In this sense, its primary contribution is not conceptual novelty in materials chemistry, but rather the demonstration of a simple, scalable, and potentially clinically applicable augmentation strategy for tendon–bone repair.

Several design elements strengthen the clinical relevance of this study. Non-degradable sutures ensured repair stability and allowed isolation of the adhesive’s adjunctive effect, while postoperative unrestricted movement provided a healing environment that more closely approximates early postoperative mobilization in clinical settings. Nevertheless, future studies using degradable sutures and controlled loading are warranted to further define how the mechanical environment interacts with adhesive-augmented repair.

Some limitations of this study should be acknowledged. First, only female rats were used. Although female rodents are often selected to reduce variability related to body size and aggressive behaviour, sex-dependent differences in healing responses may exist; therefore, future studies incorporating both sexes are warranted. Second, the use of a rat model and a 12-week study duration limits extrapolation to human tendon-to-bone healing, which occurs over longer timeframes and under more complex physiological loading conditions. Third, this study did not directly assess in vivo degradation kinetics, local concentrations of degradation products, inflammatory or immune responses, or lineage-specific molecular markers, and therefore cannot determine the precise biological mechanisms responsible for the observed improvements in healing. Notably, qualitative histological evaluation did not reveal evidence of excessive inflammatory response in the adhesive-treated group, although dedicated immunological analyses will be required for definitive assessment. Fourth, biomechanical testing of the adhesive alone at the tendon–bone interface was not performed. However, in vitro testing using a larger chicken femur model demonstrated relatively low adhesive mechanical strength, suggesting that the adhesive itself is unlikely to function as a primary load-bearing component. Finally, comparator adhesives and material-only control groups were not included, precluding definitive attribution of the observed effects to specific chemical components versus non-specific physical gap-filling or stabilization. Collectively, these limitations highlight important directions for future work, including mechanistic studies, benchmarking against clinically used materials, and validation in large-animal models under physiological loading conditions.

In summary, this study provides in vivo evidence that adjunctive use of a biodegradable polyurethane adhesive is associated with improved structural organization and tissue-level mechanical properties at the tendon–bone interface. While the molecular mechanisms remain to be elucidated, the findings suggest that a simple, biocompatible surgical material can support improved tendon–bone healing when used as an adjunct to standard repair. These results support further investigation of biodegradable adhesive augmentation as a practical strategy for enhancing tendon–bone repair and provide a rationale for future translational studies in large-animal models and, ultimately, clinical evaluation.

Data Availability

The data that support the findings of this study are available to other researchers from the corresponding author upon reasonable request.