Integrative cartilage repair using acellular allografts for engineered structure and surface lubrication in vivo

Introduction

Articular cartilage is an essential tissue within the joint that functions as a load-bearing and low-friction surface to enable locomotion. Injury or damage to articular cartilage does not spontaneously heal and can lead to degeneration and osteoarthritis (OA). The loss of cartilage due to disease or overuse leads to an inability for the remaining tissue to maintain mechanical function properly when exposed to daily activities. In the United States, over 32 million people suffer from osteoarthritis of the knee and the majority of those affected experience a lower quality of life and loss of function due to the associated knee pain¹. If left untreated, injury sites in the cartilage often progress to advanced joint osteoarthritis². While there are medical treatments and surgical approaches to ameliorate the pain and loss of function, the most likely option to completely treat the condition and restore function at late-stage OA is a total knee replacement using metal and polymer components. The most robust knee replacements typically only last 15–20 years³, which is problematic since over 50% of patients requiring treatment for osteoarthritis of the knee are under the age of 65⁴.

A significant hurdle for patients suffering from osteoarthritis is the lack of effective therapies to address cartilage defects leading to osteoarthritis and minimize the need for more invasive replacement surgeries. Two of the most successful current clinical techniques for tissue repair are osteochondral (bone and cartilage) autologous transplantation and osteochondral allograft transplantation. Autologous transplantation involves taking cartilage tissue from a non-load-bearing region of a patient’s knee and moving it to the defect region. While long-term follow-up studies show success⁵, this procedure is limited by the defect size that can be repaired, as tissue is sourced from a different part of the knee and can lead to donor site morbidity and associated complications. Osteochondral allografts, or transplant of osteochondral tissue from healthy cartilage of a recently deceased donor, is the most successful therapy to date to delay osteoarthritis progression, and thus a full knee replacement, showing an 82% allograft survival rate after 10 years⁶. Unfortunately, the quantity of osteochondral allografts available is limited by reliance on healthy donor cartilage and the need for immediate harvest and implant into recipient for optimal success⁷.

Considering the increase in cartilage injury leading to degeneration, and patients with OA⁸, there exists a substantial need for cartilage defect therapies that (1) delay OA progression and a total joint replacement, (2) can be delivered in one surgery, (3) have a long shelf life, and (4) functionally repair tissue damage. Acellular osteochondral implants, where bone and cartilage tissue are taken together from a donor knee and processed to selectively remove cellular material, may preserve the complex architecture and molecular composition of an osteochondral allograft. If a full acellular osteochondral implant effectively promoted regenerative repair, surgeons would have a low-cost, structurally-mature treatment option to address cartilage defects leading to advanced joint damage and OA. Thus, decellularization and associated processing represent a straightforward and relatively simple method of acellular allograft preparation, while providing a potentially ideal natural tissue scaffold for regeneration. Though an acellular osteochondral implant maintains the structural and compositional properties of native cartilage, and may restore tissue architecture with targeted characteristics, the dense extracellular matrix of articular cartilage with low porosity makes it unclear whether recellularization is possible, and whether implants will integrate with surrounding tissue to provide long-term function and repair. This study was designed to directly test the long-term efficacy of allograft tissue, decellularized with minimal processing to retain structure and composition, and implanted into a critical-sized defect. We compared our acellular allograft implant to an untreated defect in a contralateral joint that was repaired by drilling into the subchondral bone to create a blood clot, similar to what may be commonly observed in a microfracture/microdrilling procedure (Fig. 1A).

Integrative cartilage repair using acellular allografts for engineered structure and surface lubrication in vivo — **Fig. 1: Tissue engineering strategy for integrative osteochondral repair using acellular tissues, sheep surgery, and experimental timeline.**

We have previously evaluated acellular allografts in short-term (3-month) studies and demonstrated success in immediately providing a structural repair in a defect, but the implants showed limited integration and migration of chondrocytes into the acellular implant⁹. However, these preliminary studies were of inadequate duration and location as the implantation and analysis were conducted in non- or low-load bearing regions of the trochlear groove at only 3 months post-surgery. From the previous study using an animal model to create acellular osteochondral materials, we know that the process of decellularization (Fig. 1A) slightly alters the biochemical makeup of the tissue by decreasing the relative percentage of glycosaminoglycans and increasing the relative percentage of collagen⁹, however the architecture of the tissue remains largely unchanged. We confirmed that our methods removed cells while maintaining the collagen structure around the previously formed chondrons in the tissue (Fig. 1A). Further in vivo investigation is needed to determine the success of a decellularized osteochondral plug in the load-bearing region of the femoral condyle, a location primarily impacted during the pathogenesis of OA. Understanding the long-term behavior and repair of a scaffold in a large animal model is paramount for predicting clinical success in humans. In this work, we aimed to measure and characterize the success of an allograft repair by investigating the structure, composition, function, and integration of an acellular osteochondral implant in a common large animal, sheep (ovine) model after 6 months in vivo. We prepared acellular allografts with minimal processing, removing cellular components, while maintaining important cartilage-specific structural layers, mechanics, and heterogenous composition throughout the depth of the tissue. We defined the functional, structural, and mechanical characteristics of the osteochondral repair, integration zone between implant and native tissue, and the surrounding tissue using multiple analysis techniques, including displacement-encoded MRI to assess regional heterogeneity in tissue-level biomechanics (Fig. 1B). We show that acellular allografts result in a functional and integrative osteochondral repair with potential for clinical translation, but also with limitations that suggest the long-term maintenance of the implant may suffer from a lack of cellularity and cellular migration into the repair tissue.

Results

Acellular allografts restored functional properties of native articular cartilage

To investigate the functional response of acellular allografts in load-bearing regions of the femoral condyle, we analyzed whole limbs using displacement-encoded (dualMRI^10,11) MRI to quantify tissue strain, in addition to MRI relaxivity. An acellular allograft from a donor sheep was implanted into a defect in the femoral condyle on the right limb of a recipient sheep. During the surgery, we made an identical defect in the condyle contralateral (left) limb of the animals, which we left empty as an untreated control (Fig. 1A). After 6 months when animals were able to move freely, functional MRI measurements indicated that the implants dissipated loading and maintained hydration, compared to untreated defect joints (Fig. 2). The average strain of the implant joints under loading was significantly lower, ~50% lower for maximum (1^st principal) strain and ~60% lower for minimum (2^nd principal) strain and max shear strain, compared to the untreated defect joint (Fig. 2C, D). Additionally, it is known that hydration plays a vital role in the ability of articular cartilage to cushion the joint from cyclic compressive forces and to recover after loading^12,13,14. MRI relaxivity measurements before and after loading demonstrated that joints with the implant retained higher T₂ (spin-spin relaxation) values than the untreated defect joints and the tissue that formed in the untreated defect showed the largest drop in T₂ values from before to after loading (Fig. 2E). However, it is important to note that the differences in T2 values are trends, not statistically significant (p < 0.05) differences.

**Fig. 2: MRI whole-limb functional analyses indicate mechanical and structural similarities between the implant and native cartilage.**

Implant architecture and structure recapitulated native tissue

To evaluate the structure and architecture of the implant and untreated defect we analyzed fixed tissue slices with histological staining and previously frozen tissue sections using Raman spectroscopy to define compositional elements of the repair. After 6 months in vivo, hematoxylin and eosin staining showed that implants maintained a distinct tidemark, with key division between cartilage and bone regions (Fig. 3A). When implanted tightly (i.e., press-fit, Supplementary Fig. 5) next to native tissue, implants also maintained surface continuity from implant to native tissue. The acellular implants structurally recapitulated the thickness and typical distinct zonal layers of articular cartilage. Each cartilage zone possessed key structural differences which are essential for function, including lubrication, load support, and anchoring to subchondral bone. In contrast, the repair tissue in the empty defect failed to maintain a smooth surface with native cartilage, and instead displayed an incongruous surface even in the most successful empty joint repair (Fig. 3A). While the implants demonstrated maintenance of surface continuity with the native tissue, reduced glycosaminoglycan (GAG) content was visualized in the cartilage layer of the implant (Fig. 3A, B). Additionally, the relative level of types I and II collagen staining was consistent between the implant and the native tissue, while the tissue in the empty defect showed increased type I collagen staining (Fig. 3D).

**Fig. 3: Structure and composition of implants closely matched native tissue.**

The implant and defect repair further manifested key differences in composition. We compared the amplitude of common peaks identified from Raman spectra^15,16,17,18 for collagen, chondroitin sulfate, phosphate, and non-collagenous proteins on the surface of three regions in each joint: native cartilage, integration zone, and tissue (i.e., implant or defect) repair (Fig. 3C). We found consistency across all native tissue, despite whether the joint had an implant or an untreated defect. Native tissue displayed a collagen to chondroitin sulfate (CS) ratio of ~2, a collagen to phosphate ratio of ~14, and a collagen to non-collagenous protein ratio of ~1.3. The implants contained a very similar composition to the native tissue: collagen:CS of 1.6, collagen:phosphate of 15, and collagen:non-collagenous protein of 2.1. Alternatively, the composition of the defect repair was different, especially in phosphate content which is an indicator of mineralization. We measured collagen:CS ratio of 1.1, collagen:phosphate of 2.9, and collagen:non-collagenous protein of 2.8, which results in differences from the native tissue by 45%, 79%, and 15%, respectively. While not statistically significant (p > 0.05) due to the high levels of variance in samples of each group, the trends indicate that implant tissue structure and composition matches native tissue. The integration zone between the implants and native tissue in the animals receiving the acellular implant treatment also mimicked the composition of tissue found in the animals with an untreated defect.

Acellular allografts restored key mechanical and tribological properties of cartilage

To determine whether the mechanical properties in the repair matched native tissue, we quantified Hertz contact and relaxation moduli using micro indentation (Fig. 4A), and measured surface tribological properties via atomic force microscopy (AFM) (Fig. 4C). The implant did not demonstrate significant differences from native tissue under indentation, demonstrating a Hertz contact modulus of 393 kPa and an equilibrium modulus of 120 kPa, compared to native tissue moduli of 321 kPa and 148 kPa, respectively (p > 0.05) (Fig. 4B). Defect repair sites alternatively showed a contact modulus of 414 kPa when indented, and while not significant (p < 0.05), the equilibrium modulus was nearly the same as the Hertz contact modulus at 316 kPa (Fig. 4B).

**Fig. 4: Acellular allografts restore native cartilage mechanical and surface properties.**

The surface of native articular cartilage plays a vital role in the function of the tissue under loading and movement. Using AFM, we showed that the implant restored surface roughness and friction to values nearing that of native tissue (Fig. 4D, E). We measured a surface roughness on the implant of 300 nm, similar to the 321 nm roughness of native tissue, which matches previously reported roughness values of the cartilage surface¹⁹. In the defect repair, we measured a significantly higher surface roughness of 500 nm (p < 0.001) (Fig. 4E). Importantly, we measured a friction coefficient on the implant surface of 0.43, not significantly different from the native tissue friction coefficient on both types of joints, i.e., 0.28 in the defect joint and 0.38 in the implant joints (p > 0.05). The AFM elastic modulus under indentation at the micrometer length scale (2 µm probe tip) indicated similarities between native tissue and the implant with a much softer measurement on the surface of the defect tissue (50–51 kPa on native tissue surface, 52 kPa on implant surface, 6.5 kPa on defect surface) (Fig. 4E). The difference in modulus measured by micro indentation (Fig. 4B) and AFM (Fig. 4E) is not unusual, as AFM uses a 2 µm probe tip and typically measures small changes in modulus throughout the surface while micro indentation uses a 100 µm probe to measure material properties of the tissue. Consequently, these tests were explicitly conducted at different scales, which leads to well-known differences in properties²⁰. Micro indentation measures represent properties closer to the ‘bulk’ scale, whereas AFM with a small probe tip measures at a cellular or sub-cellular scale. Friction and surface roughness properties were independently considered, as friction measured by AFM is not necessarily related to roughness of the cartilage surface and is instead influenced more by adhesion and plowing effects²¹. Furthermore, the friction coefficient is not necessarily a good indicator of wear resistance²².

Bone regions of acellular allografts functionally integrated with native tissue

To investigate the integration and quality of repair in the bone region of the treated and untreated defects, we used microCT imaging to quantify bone volume fraction, trabecular thickness, and trabecular spacing throughout the repair (implant or untreated defect) and integration regions (Fig. 5A, B). We observed integration of the bone regions with native tissue in both the implant and untreated defect joints. However, the bone in the implant repair displayed thicker trabeculae, smaller trabecular space, and increased bone volume fraction as compared to the defect joints (Fig. 5C). Furthermore, the microarchitecture of the bone in implant joints was not significantly different than the native tissue. On the contrary, in the untreated defect joints, both the bone volume fraction and trabecular thickness were significantly lower than in native tissue (Fig. 5C).

**Fig. 5: Acellular allografts demonstrate bony integration with native tissue.**

Implants showed variable recellularization but demonstrated interfacial cellularity and integration

To evaluate the integration of the repair and surrounding native tissue, we analyzed histologically stained tissue slices to visualize integration and quantify cell number in the native tissue, integration zone, and repair (implant or untreated defect). We noted high cellularity in subchondral bone regions and evidence of remodeling. In contrast, we observed inconsistent recellularization of chondral regions, with cell migration typically less than ~750 µm from the implant articular or lateral surfaces toward the central implant regions. Masson’s trichome staining of the repair regions (implant or defect) centered around the critical integration zone (Fig. 6A) illustrated that the integration region varied widely depending on the size of the surgical gap between the implant and native tissue. With a small gap, no new tissue filled the integration zone, but the continuity of the surface greatly improved. With a larger space between implant and native cartilage, highly cellular tissue nearly identical to the tissue in the defect repair filled the space. We quantified the cellularity in each region (i.e. native, integration zone, and repair) and confirmed that cellular density in the integration regions was nearly identical to cell density in the empty defect repair (Fig. 6B). While the tissue in the integration regions was highly cellular and seemingly fibrotic in composition, the fibrotic repair tissue succeeded in laterally integrating the implant with the native cartilage. Unfortunately, despite high cellularity in the integration zones, the implants themselves did not contain many cells (Fig. 6B), illustrating a limited ability of cells to migrate much more than 750 µm into the implants. From previous work done with a 3-month sheep study, we know that chondrocytes demonstrated low migration from the surface into dense extracellular matrix of the cartilage tissue, due to the limitation of their nucleus and cell size²³. Lack of cellularity in many of the implants is likely due to a myriad of factors, including high matrix density inhibiting cell migration, and that the cell type in integration regions is not cartilage specific.

**Fig. 6: Acellular allografts promote high cellularity in the integration zone with native tissue.**

Discussion

In this work, we evaluated an acellular allograft repair as a potential clinical therapy for articular cartilage defect repair to delay the progression of cartilage tissue damage, which utilizes the advantages of autografts and allografts while reducing source limitations. Decellularized tissue has been shown to provide superior regenerative potential in many clinical settings outside of orthopedics^24,25,26, and involves minimal manipulation to produce an acellular construct that matches native tissue architecture. We demonstrated that an acellular osteochondral implant provides cartilage-specific functional properties, and has the additional advantage that implants could be sourced from xenogenic (animal-derived) tissues as a means to overcome source limitations, or from non-viable human tissue to extend the gift of donation.

The acellular osteochondral implants evaluated here provided force dissipation, hydration, and improved articular cartilage function as compared to an untreated defect. After 6 months, the functional dualMRI data indicated that the osteochondral implants had physically integrated laterally into the surrounding native tissue as the implants were able to transfer strain throughout the joint, rather than concentrating the strain near cartilage damage, as seen in the dualMRI performed on untreated defects. Our functional biomarkers of repair success at six months post-surgery were promising and motivated us to look at the repair and integration of the implant in detail. MRI relaxivity data also demonstrated that the acellular implant and native tissue T₂ values were not significantly different, while the tissue in the untreated defect had marginally lower T₂ values (p = 0.079) (Fig. 2 and Supplementary Fig. 3). Since T₂ can be used as a surrogate of tissue water content²⁷, these data suggest that the implant preserved or restored critical hydration. Additionally, the architecture measurements taken with microCT of the bone regions of the implants (Fig. 5) further suggest that the tissue was experiencing loading, which in turn signaled to bone cells to remodel and strengthen trabeculae. The microCT data of bone structure supports our conclusion from MRI results that the implant experienced loading and dissipated the force. Histological assessment further supports the functional data, illustrating that the cartilage regions of the untreated defect repair were recessed from the surface (Fig. 3). As a result, the tissue in the untreated defect would have experienced less direct loading, leading to the thinner and more widely spaced bony trabeculae and lower bone volume fraction, as compared to implant and native tissue.

Surface lubrication and low friction are additional key components to a successfully engineered tissue construct for cartilage repair. Cartilage injury in OA leads to a loss of imperative surface lubrication that facilitates smooth sliding during joint movement²⁸. Surface measurements using AFM revealed that the acellular osteochondral implant restored surface roughness and friction values to those measured in the native tissue of the sheep (Fig. 4). Alternatively, the empty repair (untreated defect) formed a significantly rougher surface, suggesting the possibility of altered contact during sliding that may lead to increased wear of the cartilage counter face – a hallmark of further degradation in the joint. Interestingly, the integration region between the implant and native tissue demonstrates a lower modulus and frictional coefficient than either implant or native tissue. This can be attributed to numerous factors, but is likely because the integration zone is filled with less mature or poorly-formed tissue, or fibrocartilage, that does not present the same stiffness or surface structure found in native or implant cartilage.

Raman spectroscopy and collagen immunohistochemistry further elucidated the composition of the integration region. In contrast to the composition of acellular implants, untreated defects filled with non-cartilaginous tissue composed of higher levels of collagen I than either native or implant tissue. The Raman spectra analysis of tissue composition in each repair condition (implant or empty/untreated) compared to native tissue indicate that defects left untreated filled with a tissue that is high in collagen (45% increase from native tissue) and phosphate (a precursor to mineralization). The Raman spectra is supported by direct collagen I and collagen II immunohistochemistry of the tissue, demonstrating an increase of collagen I in both empty defects and integration regions where there was a gap between native tissue and the implant. The change in collagen to phosphate ratio from 14 to 15 in the implant and native tissue to 2.9 in defect repair (79% increase in phosphate content) is substantial (Fig. 3). High quantities of phosphate lead chondrocytes towards apoptosis, which is a first step in endochondral ossification and bone formation^29,30. The bulk mechanical profile of the defect repair supports the Raman spectral analysis of composition positing that the defect repair may be fibrotic and high in mineral content, contributing to limited tissue relaxation after loading (Fig. 4). It is useful to know that the analysis of mineral content using Raman was performed only on the cartilage surface, while the previously discussed bone analysis was performed on the bony region below the cartilage (~8 mm depth). Therefore, the BV/TV and mineral content cannot be correlated in this analysis. Unfortunately, the integration region between the implant and native tissue was composed of tissue with the same properties (composition, histology, mechanics, and surface tribology) as the tissue measured in the untreated empty defects.

Importantly, in all of the analyses where we compared repair or untreated tissue to native tissue, it is critical to think about the state of the native cartilage. It is well known that a cartilage surface defect can over time lead to biochemical changes within a joint, and in extreme cases result in OA impacting the entire joint. Additionally, the differential loading on the native tissue surrounding an injury could certainly contribute to changes in the native tissue. However, we did not measure any significant differences in the native tissue of the untreated joints from the native tissue in the treated joints. Furthermore, a previous study using the same age and species, and similar MRI parameters, measured T1ρ in an uninjured knee joint⁹. The previous study reported values of 90–100 ms⁹, and here we report very similar values in our ‘native’ region of ~90 ms. In the future, it would be useful to have an additional control group of animals that received no surgical interference, if animal numbers allowed. This parallel composition of the defect and integration zone is consistent with the idea that the gaps between implants and native tissue fill with the same blood and bone marrow components post-injury as that observed in the untreated defect.

Several important recent findings posit that integration with native tissue is critical to the long-term success of any engineered cartilage tissue. Many researchers have shown that it is essential for an implant to promote cellularity in order to successfully integrate with native tissue^2,31. In articular cartilage, promoting cellularity is a distinct challenge as the dense extracellular matrix limits cell movement²³, the tissue is non-vascular, and even native tissue displays a low cell density compared to many tissues. While our work demonstrates acellular allografts provide a structural and functional repair under loading similar to native tissue, integration with bone, and some physical (lateral) cartilage-cartilage integration, we also show that even after 6 months in vivo, implants failed to promote cellularity completely throughout implant cartilage despite high levels of cellularity in the integration zone, which is possibly a drawback to the repair method. At the same time, authors have reported that the loss of cells in cartilage does not contribute to further cartilage damage, and may actually protect cartilage structure in vivo³². We believe that targeted future work focusing on the role of cellular migration and implant maintenance in vivo could advance the clinical effectiveness of acellular osteochondral implants. Additionally, coating the implants or surgical site prior to implantation with cartilage-specific growth factors such as IGF-1 has been shown to stimulate cartilage-specific cell migration to the border regions^33,34. While we observed high cellularity in the integration region of the defect repair (Fig. 6), the cells were likely primarily of a fibroblast phenotype that failed to migrate into the acellular cartilage implant. Combining cartilage-specific growth factor treatment with an acellular allograft could be an interesting extension of the repair strategy explored here, with potential for improved integration and effectiveness of the repair. Furthermore, it is evident from the results presented here that the surgical technique and careful implantation also affects the success of the implant. In animals where the implant was tightly press-fit within native tissue, and the surface of the implant and native tissue were aligned, we observed improved surface continuity and functionality (Figs. 3 and 6 and Supplementary Fig. 5). Prioritizing specific and precise implantation techniques would result in a decrease in the volume of repair (e.g. fibrous) tissue in the integration region as well as superior surface continuity between the implant and native cartilage. Finally, recent work indicates that while the dense cartilage extracellular matrix limits cell migration into an acellular allograft, breaking down the tissue into particles facilitates cell migration while maintaining cartilage-specific structure and mechanics^23,35,36. Future work investigating an implant where the cartilage portion of the osteochondral implant is first pulverized could improve lateral integration, cellularity, and clinical effectiveness²³. However, it is not clear whether the use of pulverized and reconstituted cartilage would potentially lose the benefit of closely mimicking the tissue architecture and function that were key benefits of the acellular implant dicusse herein. Future studies may need to balance structure, function, and cellularity to improve integrative cartilage repair.

Long-term function of defect repair strategies in a relevant, non-human model is a critical aspect to any clinical intervention or engineered tissue construct prior to human clinical trials. Large animal models (e.g. in sheep, goat, dog, and horse) are useful in analyzing the effectiveness of a treatment and closing the gap between in vitro experiments and human studies^37,38. Sheep joints specifically demonstrate joint size, cartilage thickness, and limited intrinsic healing ability similar to human joints³⁹. As a result, many researchers utilize a sheep model³⁸ to show tibiofemoral defect treatment efficacy and limitations of microfracture with additional treatments^40,41,42, modified autologous plug transfer procedures^{43,44,45,46,47}, variations of allografts^9,48,49, and tissue engineered scaffolds^50,51. While all studies in load bearing regions of a sheep elucidate the potential clinical success of a technique or scaffold, many studies referenced above are only 1–3 months in duration, and thus do not indicate the long-term success. Additionally, sheep models are limited to answering specific questions. Unlike humans and dogs, it is difficult to manage post-operative non-load bearing or therapy sessions in sheep. Therefore, sheep are best used to evaluate repairs for which post-operative management is minimal, and animals can bear weight immediately³⁹.

In conclusion, the work presented here shows that after 6 months in a sheep knee, implants match the structural composition of surrounding native cartilage, provide bulk mechanical properties under compression, enable the critical capability of load transfer throughout the joint, and restore surface roughness and friction. Importantly, we show that acellular osteochondral implants are a viable and potential strategy to postpone osteoarthritis progression and patient pain in joints by restoring function, for months to years. However, the limitations of cellularity in the repair could limit cellular maintenance. Additionally, the multi-scale (i.e., whole joint to cellular level) analysis suite presented here provides a model for evaluating new cartilage repair strategies, including novel and non-invasive dualMRI biomarkers to incorporate strain measurements into osteoarthritis diagnosis and treatment evaluation. Using these analysis tools, we have demonstrated effectiveness of a surgical intervention that requires only one surgery, does not rely on a limited tissue source, can immediately restore articular cartilage structure and function, and maintains cartilage-specific function for at least 6 months in a relevant animal model. For the more than 50% percent of patients with osteoarthritis in their knee who are younger than 65³, acellular osteochondral implants could provide a potential solution to decrease pain and increase function until the age where a total joint replacement is recommended.

Methods

Decellularization of osteochondral allograft implants

Ovine osteochondral implants were harvested from the load bearing region of the distal femur of six (donor) stifle joints (aged < 3 years, female, acquired from local abattoir within 24 h of sacrifice) using the Osteochondral Allograft Transfer System (6 mm diameter; 10 mm thickness, OATS, Arthrex, Naples, FL). Implants were decellularized in 2% (w/v) Sodium Dodecyl Sulfate (SDS, Sigma Aldrich) at 37 °C under constant agitation for 8 h, washed in phosphate-buffered saline (PBS) and then incubated with 3.3 mg/ml deoxyribonuclease (DNase), 50 mg/ml ribonuclease (RNase), 1% (v/v) penicillin/streptomycin (P/S), 1% (v/v) fungizone at 37 °C under agitation for 24 h. To deactivate the DNase after treatment, implants were bathed in 0.02% (v/v) ethylenediaminetetraacetic acid (EDTA), with a final PBS wash. Acellular allografts were flash frozen in liquid nitrogen and lyophilized for 48 h prior to surgical implantation.

Surgical procedures for cartilage defect repair using decellularized allografts

Cartilage defect repair using acellular allografts was performed on six ewes (female) of Suffolk breed aged 2.25 ± 0.43 years, who all fell under a similar weight category (74.8 ± 8.4 kg). The surgical procedure was approved by the Purdue University Institutional Care and Use Committee (IACUC), and all experiments were carried out in accordance to IACUC procedures and policies. Using an end mill system, osteochondral defects were placed via drilled holes in the load bearing region of both medial condyles (6 mm diameter and 10 mm depth). An acellular allograft was press fit into the defect region on the right stifle joint, and the identical defect on the contralateral left joint was left empty as an untreated control. The joints were closed in layers, and the animal was allowed to move freely with no weight bearing limitations post-surgery. After six months, ewes were sacrificed, and stifle joints were frozen at −80 °C until subsequent analysis. Following MRI analysis (discussed below), repair tissue was harvested by exposing the stifle joint space and removing tissue with a scalpel. Implants and surrounding native tissue (10 mm diameter, 10 mm depth) were extracted from each joint and bisected, with one half stored at -80°C and the other half fixed for 48 h in 4% (v/v) paraformaldehyde (PFA).

Whole-joint analysis of structure and biomechanical function by MRI

Whole explanted joints were secured with polymethylmethacrylate into a custom-machined loading apparatus. The loading device fixed the tibia while allowing the femur to move in compression-distraction under mechanical stimulation (Fig. 2A, B), and each specimen was secured with a natural flexion angle of 135°. Joints were scanned in a 7.0 T MRI scanner (Bruker Medical GmbH, Ettlingen, Germany) for morphology, relaxometry, and strain mapping. For morphology imaging, Rapid Acquisition with Refocused Echoes (RARE) and Fast Low Angle Shot (FLASH) pulse sequences were applied in sagittal plane to cover the whole joint (Methods and sequence demonstrated in Supplementary Fig. 1). RARE imaging parameters were: TE/TR = 22.88/3000 ms; NA = 4; in plane Resolution = 0.25 × 0.25 mm²; image matrix size = 256 × 256 pixels²; slice thickness = 2.0 mm; RARE factor = 8; number of slices = 25. FLASH imaging parameters were identical, with the following modifications: TE/TR = 4.806/390.2 ms; NA = 4; flip angle = 30°. Thickness measurements of the full thickness of cartilage were obtained from morphology imaging using manual measurements (Fiji).

For relaxometry, the Multi Slice Multi Echo (MSME) sequence was used for T₂ (spin-spin relaxation) measurements, with the following imaging parameters: TE/TR = 7.781/3000 ms; number of slices = 25. The RARE sequence was used for T₁ρ relaxation measurements, with the following imaging parameters: TE/TR = 34.96/2578 ms; NA = 2; Repetition = 5; Echo spacing = 8.74 ms; RARE factor =8; spin lock strength=851 Hz; spin lock duration = 10, 45, 80, 115 ms. Relaxation measurements were obtained using monoexponential fitting routines of the time-course data.

For strain mapping, we utilized dualMRI^11,52,53, the stimulus was applied at the femur to compress and deform the joint space. The DENSE-FISP sequence was used to scan the joints with an xy displacement encoding gradient area of 0.507 π/mm. The scan was synchronized with cyclic loading at frequency of 0.33 Hz with 445 N compression and 100 N retraction. With the same slice resolutions and dimensions introduced before, FISP sequence had parameters of TE/TR = 2.505/5.01 ms; NA = 4; number of segmentation = 2. To study the effects of the external loading on the different repair types (i.e., implant or defect), analysis focused on the loaded cartilage images. The dualMRI loading setup was designed to align the implant or defect at the direct compression point, though they might not be located exactly at the cartilage-cartilage contact region during image. The displacement was smoothed using locally weighted linear regression smoothing (LOWESS) in MATLAB (MathWorks, Natick, MA). To avoid variation due to joint orientation, coordinate independent strains were calculated.

Structural analysis using histology

Matrix deposition, glycosaminoglycan content, and overall repair structural continuity and quality were evaluated using histology. One half of the extracted tissue from each joint was fixed for 48 h in 4% (v/v) paraformaldehyde, decalcified for 3 weeks in EDTA, dehydrated progressively with Ethanol (EtOH), embedded in paraffin, sliced into 5 μm sections using a microtome, and mounted on microscope slides. Sections were stained individually with Hematoxylin and Eosin (H&E) and Safranin-O/Fast Green (Histology Core, CU Denver Cancer Center). The International Cartilage Repair Society II (ICRS II) scoring paradigm was used to evaluate relative tissue quality in each repair (i.e. implant or defect)⁵⁴. Scoring was performed by three independent and blinded observers. Full ICRS II Scoring is included in Supplementary Fig. 4.

Surface property characterization by atomic force microscopy

The composition of tissue regions was quantitatively evaluated using non-destructive testing. Samples, which were frozen after harvest, were thawed and tested first using AFM, followed by microindendation, and finally Raman spectroscopy. Articular cartilage surface roughness, frictional coefficient, adhesion, compressive modulus, and topographical structures were measured using AFM (Keysight 5500 AFM, Keysight Technologies Inc., Santa Rosa, CA, USA). All measurements were taken on the articular surface of the implant in the three key regions of interest: native cartilage, integration zone, and tissue (i.e., implant or defect) repair. A cantilever with known geometry was used (2 µm borosilicate sphere, NovaScan), and the cantilever stiffness was pre-calibrated to 0.36 N/m by the thermal fluctuation method⁵⁵. Lateral calibration was determined using an improved wedge calibration method⁵⁶ and a TGF11 silicon calibration grating (Mikromasch). Each sample was affixed to a μ-Dish high grid-50 dish (Ibidi, USA) using a viscous cyanoacrylate and hydrated for all AFM testing. On the cut surface of the tissue >1 mm away from the potential surface measurement site, a small drop of india ink was applied with a needle to assist in identifying the native, integration zone, and repair sites once under high magnification. At each location, the macroscopic region of interest on the articular cartilage surface was approached and a brief scan (4 × 4) was performed to ensure the height change was in an acceptable range (~7 μm). After locating a measurable 40 × 40 μm area, sixteen independent locations of the scan area (4 × 4) were measured in force-volume mode, indented at 5 μm/s with a setpoint force of ~12 nN. To estimate the compressive modulus, force-displacement approach curves were fit to the Hertzian linear elastic model for a defined round tip geometry^57,58. Adhesion force was measured as the pull-off force upon tip separation from the surface during probe retraction. Both adhesion and compressive modulus were extracted from force-distance curves using PicoView 1.14 AFM analysis software. Immediately following and within the same scan area, a high-resolution contact scan (1028 × 1028 px, or 40 nm/px) was performed at 20 μm/s with a constant applied normal force of 20 nN to generate detailed topography, raw deflection, and lateral voltage trace/retrace signals of surface features. Areas where the controller overloaded the sample or did not interact with the surface were excluded. Topographical images were analyzed to calculate root-mean-square (RMS) surface roughness using Gwyddion 2.56 SPM analysis software⁵⁹. The averaged lateral voltage signal over each area was converted to friction force by multiplying by the lateral calibration constant and dividing by the normal force to yield the coefficient of friction.

Characterization of bulk mechanical properties via indentation

To assess the ex vivo bulk mechanical properties of the native cartilage, integration zone, and tissue (i.e., implant or defect) repair, micro indentation was performed at nine points spanning 1.6 mm (200 μm spacing between points) centered at the integration zone (Fig. 4). In preparation for micro indentation a flat, 1–2 mm section of tissue was sliced; the resulting test surface contained a cross-section of repair tissue, integration region, and native cartilage. Indentations (Hysitron TI 950 TriboIndenter, xZ-500 extended displacement stage) were performed on submerged samples using a 100 μm radius spherical probe. For each indent, the probe was first lifted off the sample surface, then indented to a depth of 20 μm at a rate of 240 μm/s, the indent was held for 45 s to permit force relaxation. A virtual contact point for each indent was determined following established methods⁶⁰. Hertzian contact modulus (assuming ν = 0.5) and equilibrium modulus (assuming ν = 0.0 after fluid drainage) were calculated at each indent site. To calculate the reported regional bulk mechanical properties, the moduli at each of the four points within each region were averaged.

Structural analysis using Raman spectroscopy

The composition of tissue regions was quantitatively evaluated using Raman spectroscopy with a 785 nm laser and spectra collected for wavenumbers 400–1600 cm⁻¹. Raman spectroscopy data was collected on an upright inVia microscope (Renishaw, Wotton-under-Edge, UK). The regions measured previously in AFM were first identified using brightfield microscopy at ×5 magnification, assisted by the India ink markings on the edge of each analysis location (described in section 5.5). Samples were submerged and a 63× immersion objective was focused with a 785 nm laser to illuminate a ~1 µm diameter spot on the surface. In each region, spectral maps consisting of sixteen measurements covering a 40 µm × 40 µm area were collected. At each point in the spectral map, cosmic rays were removed from the collected spectra, a linear baseline was subtracted and intensity normalized, and spectra were smoothed¹⁸. The sixteen spectral acquisitions that composed each spectral map were averaged, key peaks of interest were identified, and peak amplitude was calculated using custom R code following a previously established Raman spectra analysis approach⁶¹. Relative peak amplitudes for peaks of interest (amide III-1280 cm⁻¹, ({nu }_{2}) phosphate- 441 cm⁻¹, chondroitin sulfate- 1068 cm⁻¹, and phenylalanine- 1003 cm⁻¹)^15,16,17 were compared to calculate relative compositional differences in each region.

Quantitative imaging of osteochondral bone repair and remodeling

Trabecular bone volume fraction, trabecular thickness, and trabecular spacing were measured using micro-computed tomography (XRadia Versa 520, Zeiss, Dublin, CA, USA). The explants were scanned with a 0.4× objective, voxel size of 13.4 μm^3, energy settings of 50 kV, 3.0 W, 3 s exposure, 801 projections, and using the LE3 or LE4 filters. Automated centering and beam hardening corrections were applied using Scout-and-Scan Control System Reconstructor software (v 14.0.14829.38124). To analyze the bone structure, collected micro tomography images were imported into Dragonfly software (ORS v 4.1) and the Otsu algorithm⁶² was implemented to separate ranges of the histogram corresponding to bone and background noise. Distinct volumes to analyze were designated in three key regions: native cartilage, integration zone, and tissue (i.e., implant or defect) repair. In each region, trabecular bone volume fraction was calculated. A sphere fitting-method was used in each trabeculae or in the space between trabecula to calculate trabecular bone volume fraction, trabecular bone thickness, and trabecular bone spacing using Dragonfly Bone Analysis.

Cell quantification using histological images

Cellularity of all regions of interest and cell migration into the repair tissue was quantified in each of the three key regions using paraffin-embedded tissue slices (preparation detailed above) stained with Masson’s Trichrome (Newcomer Supply) following the manufacturer recommended staining protocol after tissue hydration with decreasing concentrations of EtOH. Each region (native, integration zone, and tissue repair) was isolated in a high-resolution image and standard ImageJ thresholding⁶³ was used to identify cells (blue/black spots) and quantify cell number/tissue area using Image J particle counting.

Detection of collagen I and II by immunohistochemistry

Collagen I and collagen II immunohistochemical sectioning and staining were performed by Premier Laboratory (Longmont, CO). Detection of Collagen I by immunohistochemistry (IHC) was carried out using a rabbit monoclonal antibody directed against Collagen I from rabbit (EPR7785 from Abcam). Detection of αCollagen II by IHC was carried out using a rabbit polyclonal antibody directed against αCollagen II from rabbit (203001 from MD Bioproducts).

Formalin fixed paraffin-embedded (FFPE) samples were sectioned at 5 µm and mounted onto charged slides. Slides were dried overnight, baked at 60°C, deparaffinized in xylene, rinsed in alcohol, rehydrated in water, and equilibrated in wash buffer (TRIS buffered saline with 0.05% (v/v) Tween 20; Dako, K8007). Slides were then loaded onto an AutostainerLink48 stainer (Dako) and the remaining steps were carried out at room temperature. Within the autostainer slides were incubated in 3.0% (v/v) hydrogen peroxide for 5 min, followed by proteolytic induced epitope retrieval (PIER) via Proteinase K for 5 min (Dako, S3020) for Collagen I staining. For Collagen II, within the autostainer slides were incubated in 3.0% (v/v) hydrogen peroxide for 5 min, followed by Chondroitinase ABC (from Proteus vulgaris from Sigma C3667-10UN). Next a serum free protein block was applied for 5 min (Dako, X0909), anti-Collagen I primary antibody for 30 min (1.09 µg/ml) or anti- αCollagen II primary antibody for 30 min (1.09 µg/ml), EnVision+ anti-Rabbit Labelled Polymer-HRP for 30 min (Dako, K4011), and DAB+ Chromogen Solution for 5 min (Dako, K4011). Wash buffer rinses were applied between appropriate reagents. Slides were then manually rinsed in tap water and counterstained for 6 min in an Automation hematoxylin (Dako, S3301). The slides were again rinsed in tap water, and wash buffer was used as the hematoxylin bluing reagent. The slides were then dehydrated in absolute alcohol solutions, cleared in xylene, and cover-slipped.

Statistical analysis

To test our hypotheses that treatment (osteochondral implant or empty defect) influenced articular cartilage architecture and repair (stiffness, surface roughness, adhesion, protein signature, bone composition) in vivo, one-way mixed model analyses of variance (ANOVAs) were performed on generalized linear models with treatment, animal, and implant location (left stifle joint or right stifle joint) as the predictors, and the resulting measurement variable as the response. All data sets utilized the tissue collected from contralateral joints in all six animals (n = 6 biological replicates). In all data sets, if treatment, animal, or implant location was found to be a significant predictor then the p-value significances between each group was calculated using Tukey’s Honest Significant Difference Test. The same statistical approach was used for data collected in each region of interest. To compare between regions with the same treatment, the ANOVA was performed on a generalized linear model with region, animal, and repair type as predictors. Statistical significance in all experiments was defined as p < 0.05. All statistical testing was performed using R software and the lmer package.