Bone Regeneration in a Large Animal Model Featuring a Modular Off‐the‐Shelf Soft Callus Mimetic

Abstract

Implantation of engineered cartilage with soft callus features triggers remodeling to bone tissue via endochondral bone regeneration (EBR). Thus far, EBR has not progressed to the level of large animals on the axis of clinical translation. Herein, the feasibility of EBR is aimed for a critical‐sized defect in a large animal model. Chondrogenesis is first induced in goat‐derived multipotent mesenchymal stromal cells (MSCs) by fine‐tuning the cellular differentiation process. Through a unique devitalization process, two off‐the‐shelf constructs aimed to recapitulate the different stages of the transient cartilaginous soft callus template in EBR are generated. To evaluate bone regeneration, the materials are implanted in an adapted bilateral iliac crest defect model in goats, featuring a novel titanium star‐shaped spacer. After 3 months, the group at the more advanced differentiation stage shows remarkable regenerative capacity, with comparable amounts of bone regeneration as the autograft group. In contrast, while the biomaterial mimicking the earlier stages of chondrogenesis shows improved regeneration compared to the negative controls, this is subpar compared to the more advanced material. Concluding, EBR is attainable in large animals with a soft callus mimetic material that leads to fast conversion into centimeter‐scale bone, which prospects successful implementation in the human clinics.

This work highlights the feasibility of endochondral bone regeneration in a large animal model. To achieve this, allogeneic mesenchymal stem cells are stimulated to form two variants of a (nonliving) soft callus mimetic material. Upon implantation in a bone defect, one of the materials matches the current clinical gold standard treatment in bone regenerative properties.

1. Introduction

An established approach in bone tissue engineering (BTE) mimics fetal bone development and secondary fracture repair, which involve a cartilaginous intermediate stage and is known as endochondral bone formation.^{[ 1 ]} This regenerative approach involves the implantation of an engineered cartilage tissue that is gradually replaced by bone in the host. Using patient‐derived mesenchymal multipotent stromal cells (MSCs), these cartilaginous tissues can be formed via differentiation in high density pellet cultures or in combination with a carrier biomaterial. A multitude of studies have shown that these cell‐based chondrogenic implants induce endochondral bone regeneration (EBR) in small animal models.^{[ 2} , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ^{8 ]}

It is widely acknowledged that the use of autologous cells for clinical applications requires a tedious two‐stage process, which ultimately yields substitutes with short shelf‐lives and increased patient discomfort. Particularly in generating cartilaginous tissues, this entire process is highly unreliable due to the unpredictable and variable capacity of patient‐derived MSCs in forming cartilaginous tissues. It is therefore imperative for the field to engineer a one‐stage off‐the‐shelf biomaterial that mimics the cartilaginous template required for EBR. This would improve EBR reproducibility, ease logistics, and reduce the cost and time to treat patients. Such a biomaterial with impressive bone regenerative potential was previously established by our group. The substitute was developed from donor cells and underwent a mild devitalization process prior to implantation.^{[ 2 ]}

In the present study, we aimed to present the first‐ever proof of concept of EBR for a critical‐sized defect in a large animal model, featuring our off‐the‐shelf devitalized callus mimetic material. In comparison to small animal species, large animals (e.g., small ruminants) more closely resemble humans; physiologically, immunologically, and with regard to anatomical size.^{[ 9} , ¹⁰ , ¹¹ , ^{12 ]} In the present study, we selected goats owing to their similarity in bone tissue macro‐ and microstructure and composition.^{[ 12} , ¹³ , ¹⁴ , ¹⁵ , ^{16 ]} Further, bone remodeling rates and sequence of events during bone graft incorporation in goats are more comparable to humans than to small animals.^{[ 17} , ¹⁸ , ^{19 ]} For the model, we chose a bilateral iliac wing defect as it is a reproducible model which allows for convenient evaluation of a critical‐sized bone defect with a volume in the range of cubic centimeters.^{[ 20 ]} The iliac crest is structurally comparable to the metaphyseal region of long bones as it is composed of cortical and cancellous bone.^{[ 20} , ^{21 ]} While it is not directly load‐bearing since it is a circular defect in intact bone, this model is comparable to long bone defects stabilized with external or internal fixation devices.^{[ 20 ]}

The strategy of differentiating patient‐derived stem cells, in particular, MSCs is a promising approach for EBR. In general, MSCs isolated from different species have shown to require different culture conditions for optimal chondrogenesis.^{[ 22} , ^{23 ]} For example, human, bovine, and equine MSCs undergo chondrogenic differentiation in the presence of transforming growth factor beta (TGFß) alone, while rat and canine MSCs require additional supplementation of bone morphogenetic protein‐2 (BMP‐2).^{[ 6} , ²⁴ , ²⁵ , ²⁶ , ^{27 ]} As there is no current consensus on the optimal culture conditions for bone marrow‐derived goat MSCs, we first explored the effects of TGFβ +/− BMP‐2 on their in vitro chondrogenic differentiation. The composition of the storable callus implants was evaluated for chondrogenic and hypertrophic markers, potentially indicating their capacity for endochondral remodeling into bone. Using a modular approach, a cm‐scale cartilaginous construct was implanted for 3 months within a goat iliac crest and compared against the current clinical gold standard, autograft bone.

2. Experimental Section

2.1. Study Design and Overview

Figure  1 illustrates the overall experimental outline of both in vitro characterization and in vivo assessment. For in vitro characterization of the devitalized callus mimetic, goat MSCs were isolated, expanded, and encapsulated in a collagen type I hydrogel before exposure to two different culture conditions (+TGFβ, +/− BMP) to stimulate chondrogenic differentiation. After 28 days, the cartilaginous spheroids differentiated with TGFβ or TGFβ/BMP were devitalized and characterized for the presence of chondrogenic markers via histological staining and biochemical analyses.

Figure 1.

Overview of the in vitro and in vivo process and study designs. Top: For the in vitro characterization, goat MSCs were isolated, expanded, and embedded in collagen hydrogels. MSCs were differentiated in chondrogenic differentiation medium, supplemented with either TGβ1 alone or also in combination with BMP‐2. After 28 days of culture, spheroidal cultures from both conditions were devitalized and analyzed for markers indicating the presence of hyaline or hypertrophic cartilage tissue. Bottom: For in vivo implantation, the following groups were included: empty defect, fibrin carrier, bone autograft, and devitalized cartilaginous spheroids differentiated with TGFβ or with TGFβ/BMP. Samples were inserted into a titanium spacer that divided the critical‐sized circular defects (17 mm Ø; 8–10 mm height) into three equal parts. Fluorochromes (calcein green and oxytetracycline) were administered at different time points to assess the progress of bone formation. After 3 months, samples were explanted, and new bone formation was evaluated.

A bilateral iliac crest defect model was performed in six goats. With a star‐shaped titanium spacer, each defect was divided in three equal compartments. The following experimental groups were defined: empty defect (n = 4), empty fibrin carrier (n = 4), autologous bone graft (n = 5), devitalized cartilaginous spheroids cultured with TGFβ (n = 6), and devitalized cartilaginous spheroids cultured with TGFβ/BMP (n = 6) (Table S1, Supporting Information). Fluorochromes were administered after 1 and 2 months to assess bone formation dynamics. All goats were euthanized at 3 months. The extent of bone formation in each group was evaluated quantitatively with micro‐computed tomography (micro‐CT) and histomorphometry. All surgical procedures and animal care were performed in compliance with the Central Authority for Scientific Procedures on Animals (Dutch national CCD) and of the Local Animal Welfare Body (protocol number 8831‐1‐01) in accordance with the ARRIVE guidelines.^{[ 28 ]}

2.2. Isolation, Culture, and Expansion of gMSCs

Goat MSCs (gMSCs) were isolated from a biopsy of the iliac wing of adult female Dutch milk goats (Capra hircus) (N = 4). The bone was flushed with MSC expansion medium composed of α‐minimum essential medium supplemented with 10% heat‐inactivated fetal bovine serum (S14068S1810, Biowest, Nuaillé—France), 0.2 × 10⁻³ m L‐ascorbic acid 2‐phosphate (A8960, Sigma‐Aldrich, St. Louis, USA), 100 U mL⁻¹ penicillin with 100 mg mL⁻¹ sreptomycin (15140, Invitrogen) using a 19G needle. Bone debris was removed using a cell strainer (70 µm). Adherent cells were passaged at 80% confluency until passage 4. MSC multilineage potential was confirmed as reported previously.^{[ 29 ]}

2.3. Generation of Devitalized Cartilaginous Spheroids

Cartilaginous spheroids were prepared as previously described.^{[ 2 ]} Briefly, 20 × 10⁶ mL⁻¹ gMSCs at passage 4 were encapsulated in collagen type I gel droplets (4 mg mL⁻¹), according to the manufacturer's instructions (354249, Corning, New York, USA). Briefly, the pH of the collagen solution was neutralized with 0.023 mL of 1 m NaOH per mL of collagen solution. 50 µL of collagen hydrogel containing ≈1 million gMSCs were pipetted into 96 well round (U) bottom plate and allowed to gelate at 37 °C for 30 min before transferring into a 48 well plate filled with chondrogenic differentiation medium composed of high glucose Dulbecco's modified eagle medium (31966, Thermo Fisher Scientific), 1% ITS (insulin‐transferrin‐selenium)+ premix (354352, Corning), 100 × 10⁻⁹ m dexamethasone (D8893, Sigma‐Aldrich), 0.2 × 10⁻³ m L‐ascorbic acid 2‐phosphate, 100 U mL⁻¹ penicillin with 100 mg mL⁻¹ streptomycin, 10 ng mL⁻¹ transforming growth factor‐β1 (TGFβ1; Peprotech, New Jersey, USA) with or without the addition of 100 ng mL⁻¹ bone morphogenic protein‐2 (BMP‐2; InductOS, Wyeth/Pfizer, New York, USA). Samples took on a spheroidal shape and were cultured for 28 days. The first 4 days, medium was refreshed daily and afterward every 3rd day. After 28 days, spheroids were devitalized by a mild procedure including lyophilization as previously described.^{[ 2 ]}

2.4. Histological and Immunohistochemical Staining of the In Vitro Samples

For histological stainings, both vital and devitalized spheroids were washed with phosphate buffered saline (PBS) and fixed with 4% formaldehyde overnight. After fixation, samples were dehydrated in a series of increasing ethanol solutions (70–100%), followed by xylene and embedded in paraffin. Samples were sliced into 5 µm thick sections using a microtome (Microm HM340E; Thermo Fischer Scientific). Qualitative analysis for chondrogenic potential was carried out via Safranin‐O and Toluidine blue staining to visualize glycosaminoglycans (GAGs). For Safranin‐O, the sections were deparaffinized, washed with deionized water, and stained with Weigert's hematoxylin. After incubation in running tap water, the samples were rinsed with deionized water, and stained with 0.4% w/v fast green (Sigma‐Aldrich) to visualize collagenous matrix (green). After washing with acetic acid (1% v/v), the samples were incubated with 0.125% w/v Safranin‐O to stain proteoglycans (red). For Toluidine blue staining, deparaffinized and rehydrated samples were stained with 0.4% Toluidine blue solution (Sigma‐Aldrich) in 0.1 m sodium acetate (pH 4; Sigma‐Aldrich) to stain proteoglycans (purple) and counterstained with 0.2% w/v fast green (Sigma‐Aldrich). The stained sections were dehydrated with 96% ethanol followed by 100% ethanol and xylene before mounting in Eukitt Quick‐hardening mounting medium (Sigma‐Aldrich).

Immunohistochemical staining was performed to visualize collagen type II protein and vascular endothelial growth factor (VEGF) protein expression. Endogenous peroxidase activity was blocked by incubating samples with 0.5% v/v H₂O₂. For collagen type II protein, antigen retrieval was carried out by sequential incubation with 1 mg mL⁻¹ pronase and 1 mg mL⁻¹ hyaluronidase at 37 °C for 30 min, each. For VEGF, antigen retrieval was carried out by incubating sections in 10 × 10⁻³ m citrate buffer (pH 6) at 95 °C for 20 min. Blocking of aspecific protein was carried out by incubation with 5% w/v bovine serum albumin‐PBS at room temperature for 30 min. For collagen type II and VEGF, primary antibody (II‐II6B3) concentration was 0.6 µg mL⁻¹ and anti‐VEGFA antibody (VG‐1; ab1316) concentration was 5 µg mL⁻¹, respectively, both followed by BrightVision HRP‐anti‐mouse IgG (VWRKDPVM110HRP) as a secondary antibody. The staining (brown) was developed by 3,3′‐diaminobenzidine (DAB) oxidation. To visualize the nuclei, sections were counterstained with hematoxylin. Lastly, the slides were dehydrated and mounted with Eukitt Quick‐hardening mounting medium. A mouse isotype (X0931, Dako) was used as negative control at a concentration matched with the primary antibody. The sections were visualized using an Olympus BX51 microscope (Olympus DP73 camera, Olympus).

2.5. Biochemical Analysis

As a measure of chondrogenic differentiation, quantification of GAG content in the devitalized spheroids was carried out. At specific time points, devitalized spheroids (TGFβ, TGFβ/BMP) were digested overnight at 60 °C in papain digestion buffer (250 µg mL⁻¹ papain, 0.2 m NaH₂PO₄, 0.1 ethylenediaminetetraacetic acid, and 0.01 m DL‐cysteine hydrochloride; Sigma‐Aldrich). The total amount of GAGs was determined with a 1,9‐dimethyl‐methylene blue (DMMB pH 3.0; Sigma‐Aldrich) assay. As a standard, a series of dilution of chondroitin sulfate C (Sigma‐Aldrich) was used. Absorbance values were detected at 525 and 595 nm. DNA content was measured using the Quant‐iT PicoGreen assay carried out according to the manufacturer's instructions. The fluorescence was measured at 485 nm excitation and 535 nm emission wavelength. A DNA standard curve was used to quantify the DNA content in the samples. The total GAG content as quantified by the DMMB assay of the devitalized spheroids was normalized to the DNA measurement (DMMB/DNA). Alkaline phosphatase (ALP) activity of the devitalized spheroids was measured using the p‐nitrophenyl phosphate (pNPP) substrate system (N2765; Sigma). Several concentrations of ALP with a known activity (U mL⁻¹) were included as standard curve. The spheroids and the standard series were individually incubated with the pNPP substrate at 37 °C for 8 min. Absorbance was measured at 405 nm with 655 nm as a reference wavelength.

2.6. Isolation of Fibrin

Fibrin was isolated by ethanol precipitation following a published protocol.^{[ 30 ]} Allogeneic whole goat blood was collected from a single donor. Donor plasma was prepared at room temperature immediately after blood collection by centrifugation at 400 x g to remove all cellular components. The supernatant containing plasma was transferred into a fresh tube and stored at −20 °C. To precipitate the fibrin, stored plasma was thawed and mixed with 0.176 mL ethanol mL⁻¹ before incubation in an ice bath for 20 min. Remaining plasma was separated from fibrin pellets by centrifuging at 1000 x g at 4 °C for 10 min. The fibrin pellet was isolated and warmed at 37° C to allow solubilization.

2.7. Construct Preparation for In Vivo Implantation

Each critical size defect was divided into three parts (≈0.6 cm³) by a custom‐made titanium star‐shaped spacer as depicted in Figure 1. For each sample of the TGFβ and TGFβ/BMP groups, 35 spheroids were embedded in the allogeneic goat fibrin, which was clotted for 30 min at 37 °C with 10 IU mL⁻¹ thrombin (Tisseel, Baxter) and 20 × 10⁻³ m calcium chloride solution in a custom‐made polydimethylsiloxane mold. Fibrin controls (600 µL) were fabricated in an analogous fashion without spheroids.

2.8. Animal Experiment and Surgical Procedure

All surgical procedures and animal care were performed in compliance with the Central Authority for Scientific Procedures on Animals (Dutch national CCD) and of the Local Animal Welfare Body (protocol number 8831‐1‐01) in accordance with the ARRIVE guidelines.^{[ 28 ]} For this study, bilateral iliac crest defects were made in adult Dutch milk goats (Capra hircus, VOF De Römer; n = 6; females; age > 3 years; weight range 63–87.5 kg), which also underwent mandibular condyle replacement for an unrelated study. Animals were acclimatized for a minimum of 1 week before surgery. Because of the mandibular condyle replacement, animals were fed with pre‐moistened chow (standard chow, beets, and grass pulp) 1 week prior to surgery until maximum 1 week post‐surgery. The surgical procedure was performed under general anesthesia. Anesthesia was induced with detomidine (intramuscular (i.m.), 0.04 mg kg⁻¹), and propofol (intravenous (i.v.), 2 mg kg⁻¹) and maintained with propofol (i.v. 10 mg kg⁻¹ h⁻¹), cisatracurium (0.09 mg kg⁻¹ h⁻¹), and sufentanil (0.003 mg kg⁻¹ h⁻¹). Pre‐operative pain medication consisted of a buprenorphine patch (transdermal, 35 µg h⁻¹) and meloxicam (0.5 mg kg⁻¹) was given intraoperatively. Post‐operative pain was managed with buprenorphine patch (transdermal, 35 µg h⁻¹) for 3 days. Antibiotic prophylaxis consisted of amoxicillin‐clavulanate potassium (i.v. 10 mg kg⁻¹) was given just before the surgical procedure and penicillin (10 000 IE kg⁻¹), streptomycin (10 mg kg⁻¹) were given intramuscularly for 5 days including the day of surgery.

For the surgical procedure, the animals were placed in sternal recumbency. The skin over the pelvic area was shaved and disinfected and draped in routine fashion. A curved incision was made ventrodorsally over the cranial aspect of the iliac crest. The muscles were detached from the iliac wings. A 17 mm ⌀ diameter defect was created with a trephine drill under constant cooling with sterile saline. The depth of each defect varied between goats, based on individual anatomy, but on average had a depth of around 8 mm. The harvested autologous bone graft was morselized and kept in wet surgical gauze. The 3D‐printed star‐shaped spacer (Materialise, Belgium) was implanted press‐fit into the defect. The TGFβ, TGFβ/BMP, and fibrin groups were implanted as prepared. The autologous bone graft was clotted in a syringe with fresh blood before implantation. The implant locations of the experimental groups were predetermined by block randomization. The fascia and skin were sutured in multiple layers using resorbable sutures. Calcein green (10 mg kg⁻¹ in 2% w/v NaHCO₃) and oxytetracycline (16 mg kg⁻¹) were administered intravenously and intramuscularly, 1 and 2 months after surgery, respectively. The animals were euthanized with an overdose of pentobarbital (i.v., to effect) 3 months after surgery. Postmortem, the surrounding soft tissue was separated and removed from the iliac wings and the defect together with ≈1 cm of surrounding bone was removed using an oscillating saw. Explanted samples were rinsed with PBS before fixation in 4% w/v formaldehyde solution.

2.9. Micro‐CT Scanning and Analysis

After explantation, the samples were scanned with the micro‐CT (Quantum FX micro‐CT; Perkin Elmer, Waltham, MA, USA), with scan settings: Field of view (FOV): 30 mm, voltage: 90 kV, current: 160 µA, scan time: 3 min. The samples were scanned within a copper filter to reduce image scatter from the titanium spacer. Using ImageJ software, reconstructed 3D images were generated from the scans as shown in Figure 3C. After setting a global threshold of 40, the mineralized tissue volume of 50 micro‐CT slices central in the defects was measured with the ImageJ software, BoneJ plugin.

Figure 3.

In vivo evaluation of bone regeneration in goat iliac crest defects. A) A 3D representation of the titanium star‐shaped spacer designed using TinkerCAD software, dividing the circular bone defect into three equal compartments. B) The spacer is shown press‐fitted in the os ilium defect, while holding callus mimetic (top left), autograft bone (top right), and fibrin control (bottom). C) Top row: Representative 3D reconstructions of micro‐CT slices after 12 weeks showing mineralized tissues. Bottom row: Methylene blue/basic fuchsin‐stained MMA tissue sections, showing bone in bright pink and spheroid remnants and soft tissue as a shade of blue‐purple. D) Quantification of mineralized tissue volume from micro‐CT scans shown in a box and whisker plot. E) Percentage of total bone (bone and bone marrow) as determined from histomorphometry of MMA sections. F) Heatmap showing the quantification of total bone (bone and marrow area‐%) on histological sections of the central region of the defect (see also Table S2, Supporting Information) for all groups. G) Differences in total bone formation of the central region of the defect. *p < 0.05, **p < 0.01, ***p < 0.001; ****p < 0.0001; ns: nonsignificant.

2.10. Methylmethacrylate (MMA) Section Preparation

Upon fixation, the samples were dehydrated in a series of ethanol solutions (70, 90, 95, and 100% v/v for 24 h each step) and embedded in MMA. Excess bone tissue surrounding the samples was shaved down to approximately the edge of the defect using an automated sandpaper machine. Cross‐sectional sections (10 µm) of the explanted samples were made using a diamond‐coated hard tissue microtome (Leica, Microsystems SP 1600, Nussloch, Germany). Sections were stained with methylene blue and basic fuchsin to allow for histological and histomorphometrical analysis.

2.11. Quantitative and Qualitative Assessment of Bone Formation

A cross‐sectional overview of each sample was acquired using a thunder imaging system (Leica Microsystems, Germany). The region of interest (ROI) for the implants of all groups was outlined by the drilling edge on the periphery of implants and the two titanium edges of the spacer on the inside that were adjacent to the specific implant (see Figure S1A, Supporting Information). All regions within the ROI containing bone or bone marrow were manually selected using the quick selection tool on Adobe Photoshop 23.2.2. The number of pixels for each selected area was quantified using the “Record Measurement” function on the Measurement Log. The total amount of bone (bone and bone marrow) per group was expressed as percentage of the total cross‐sectional ROI for each group. Total bone was quantified at two different depths of the implant (anterior and middle). Analysis of sections was blinded and scored independently by two scientists and the results are presented as an average. Fluorochromes were visualized using confocal microscopy (Leica DMi8). To detect the fluorochromes calcein green (em 495 nm, ex 515 nm) and oxytetracycline (em 380, ex 510), sections were imaged using a 10x objective.

2.12. Statistics

Results were expressed as mean ± standard deviation (SD). For comparisons of spheroid diameter and biochemical analysis (GAG and DNA content), a two‐way analysis of variance (ANOVA) followed by post hoc Sidak's multiple comparisons test (GraphPad Prism 6, San Diego, CA, USA) was used. For ALP activity, and unpaired two‐tailed t‐test was carried out. Histomorphometrical data (mean ± SD) were analyzed by a two‐way ANOVA test followed by Tukey's multiple comparison test. Differences were considered to be statistically significant for p < 0.05.

3. Results

3.1. In Vitro Chondrogenic Differentiation

Spheroids exposed to TGFβ/BMP for a period of 28 days showed an intense Safranin‐O staining, homogenously distributed throughout the spheroid (Figure  2A), confirming chondrogenesis and indicating abundant GAG accumulation in the extracellular matrix (ECM). In contrast, the Safranin‐O staining was less homogenously distributed in spheroids cultured in TGFβ alone, where a distinct green collagen‐rich area was observed at the periphery of the spheroid. Similarly, toluidine blue staining showed metachromatic staining throughout TGFβ/BMP spheroids, while TGFβ spheroids displayed weaker metachromasy, particularly in the outer zone of spheroids. Consistent with histology, quantitative GAG analysis showed that the presence of BMP‐2 in the differentiation medium enhanced the amount of GAG/DNA present in the spheroids. GAG/DNA content per spheroid of the TGFβ/BMP group was significantly enhanced with an approximately twofold increase in comparison to TGFβ spheroids on day 28 (Figure 2B). Figure 2C shows that spheroids of the TGFβ/BMP group were significantly larger than TGFβ only group, despite similar DNA content (Figure S2, Supporting Information) suggesting that the larger size could be attributed to greater matrix deposition in the TGFβ/BMP group.

Figure 2.

In vitro chondrogenesis of goat MSCs differentiated with TGFβ or TGFβ/BMP. A) Sections of cultured spheroids from both culture conditions (TGFβ, TGFβ/BMP) were stained for GAGs (Safranin‐O and Toluidine blue staining), collagen type II, and VEGF at day 28. B) GAG content normalized to DNA content (n = 3, N = 3; each colored dot representing gMSCs of a different donor). C) Diameter of vital spheroids after the culture period for both culture conditions measured using a stereomicroscope. D) ALP activity per spheroid is shown for both culture conditions. Scale bars: 100 µm; in inserts: 500 µm. *p < 0.05; **p < 0.01.

Morphologically, the cells surrounded by cartilage‐specific ECM displayed morphology consistent with a chondrogenic phenotype, with rounded cells residing within a lacuna. Immunohistochemical analysis of cartilage‐specific type II collagen staining showed homogenous distribution throughout both groups. Similarly, both culture conditions were positive for expression of VEGF, an angiogenic marker typically present in early hypertrophy. In a separate experiment, ALP activity of the spheroids was quantified in the supernatant. ALP activity was more than twofold higher in the TGFβ/BMP group in comparison to the TGFβ‐only group (Figure 2D).

3.2. Post‐Surgery Observations

Figure  3A,B shows the titanium star‐shaped spacer reconstructed as a 3D representation and press‐fitted in the os ilium defect. Overall, all animals tolerated the surgery well and exhibited no signs of adverse clinical reactions during the postoperative period. One sample from the fibrin group was compromised during implantation and therefore excluded from all analysis. A total of five implants were excluded from this study, as four implants were controls for a separate nonrelated study and one implant from the control group (Fibrin) was compromised during the implantation.

3.3. In Vivo Bone Formation

Overall, 3D reconstructions of micro‐CT slices showed mineralized tissue formation in the defect in the peripheral region for all groups (Figure 3C). Quantitatively, a trend was observed for a higher median mineralization in autograft and TGFβ/BMP groups, albeit nonsignificant (Figure 3D). Consistent with the micro‐CT reconstructions, visual observation of the methylene blue‐basic fuchsin stained sections also showed bone formation from the periphery of the defect, indicative of osteo‐conduction (Figure 3C). The degree of healing was variable and none of the defects had fully healed toward the defect center. Instead, all groups presented a dense connective tissue layer of variable thickness adjacent to the titanium spacer. However, bone formation progressed furthest into the defect for the autograft and allogeneic devitalized cartilage group with TGFβ/BMP (Figure 3F). Noteworthy, TGFβ/BMP significantly outperformed the TGFβ group at the center of the defect in terms of bone formation (located as indicated in Figure S1, Supporting Information). In more detail, the autograft and both devitalized callus mimetic groups not only displayed new bone formation but also fatty and red marrow (stained blue‐purple) formation and the presence of osteoid, lined with osteoblasts (Figures  4 and  5 ). Microscopically, parts of the newly formed bone in all defects consisted of plexiform (or fibrolamellar) bone (Figure S3, Supporting Information). This bony structure is characteristic in large, fast growing animals and is virtually absent in humans.^{[ 31} , ^{32 ]} Closer examination of both devitalized callus mimetic groups showed the formation of woven bone surrounding regions of mineralized cartilage (as displayed by staining metachromasia; different color to surrounding bone) stipulated to actively be undergoing remodeling due to the simultaneous presence of multinucleated cells, e.g., osteoclasts/chondroclasts. Woven bone that formed toward the core of the circular defect was also lined by osteoid, indicating active bone formation at this location during the final time point (Figure 4). In contrast, bone that had formed toward the edge of the defect was more mature with lamellar bone features, covered with bone lining cells. Volkmann's canals and Haversian (central) canals were also visible in multiple regions of the defect (Figure 4). Quantitatively, the mean percentage of newly formed bone (BV/TV; BV = bone and bone marrow) based on histomorphometry was 61.32 ± 11.51% for the autograft and 60.34 ± 15.63% for the TGFβ/BMP group. In contrast, the TGFβ group, fibrin carrier, and empty defect showed 45.30 ± 4.83%, 36.43 ± 23.43%, and 28.35 ± 15.77% BV/TV, respectively. The distribution of bone formation of each individual goat can be found in Figure S4 in the Supporting Information. Quantification of BV/TV in only the inner core of the defect (see Figure S1B, Supporting Information) showed little to no bone in‐growth in the controls and TGFβ group, indicating that new bone formation induced by the implants was enhanced only in the autograft and the TGFβ/BMP groups.

Figure 4.

Histological examination of newly formed bone tissue in the two devitalized callus mimetic TGFβ and TGFβ/BMP groups after 12 weeks. Top row shows two overview images of the groups to localize following magnifications. Remodeling (second row): Multinucleated cells (presumably chondroclasts and/or osteoclasts) were identified that are resorbing the devitalized callus mimetic spheroids (mineralized areas indicated by *). New bone formation (third row): Surface osteoblasts aligned in a linear array to form osteoid on top of woven bone. bottom row: Lamellar bone of the osteons can be discerned. Volkmann's (transverse) canals and Haversian (longitudinal) canals can be observed in osteons of both TGFβ and TGFβ/BMP groups, respectively. MnC: multinucleated cells; BM: bone marrow; Ob: osteoblast; Os: Osteoid; WB: woven bone; LB: lamellar bone; VC: Volksmann's canal; HC: Haversian canal. Scale bars: (black) 500 µm; (white) 100 µm.

Figure 5.

Histological examination of newly formed bone tissue of the autologous group. A) Overview image of the histology for the autologous bone implants to indicate localization of the magnifications. Remodeling box: Multinucleated cells (presumably chondroclasts and/or osteoclasts) were identified that are resorbing implanted autograft bone (indicated by *). New bone formation box: Surface osteoblasts aligned in a linear array to form osteoid on top of woven bone. Lamellar bone box: Lamellar bone of the osteons can be discerned. MnC: multinucleated cells; BM: bone marrow; Ob: osteoblast; Os: Osteoid; WB: woven bone; LB: lamellar bone. B) Close‐up images of the autograft group and the interface (indicated with ▼) between new woven bone and postulated (dead) grafted bone that has not yet been remodeled into new bone. The fluorochrome incorporation in these same sections is also shown. Scale bars: (black) 100 µm; (yellow); 200 µm; (white) 500 µm.

3.4. Fluorochromes

The sequential administration of the calcium‐binding fluorescent dyes calcein green and oxytetracycline at 1 and 2 months post‐surgery, respectively, provides a dynamic insight in bone formation during the entire experiment. Upon explanation at 3 months, the detection of both administered fluorochrome labels indicated that the bone formation occurred at respective time points in all groups (Figure  6 ). In the empty, fibrin, and TGFβ groups, calcein was exclusively found at the periphery, in close contact with the surrounding bone, indicating bone ingrowth from the defect edges (osteo‐conduction). While this was present also in the TGFβ/BMP and autograft samples, in these groups calcein incorporation was also detected in the center of the defect, indicating a more homogenous bone formation throughout the entire area, highlighting the osteo‐inductive properties of the implants. Oxytetracycline presence in the TGFβ/BMP and autograft group also highlighted that bone formation was still ongoing after 2 months in the defect core, whereas this was not observed in the empty, fibrin, and TGFβ groups. Interestingly, fluorochrome labeling highlighted that in the autograft group new bone was formed next to the interface between newly formed woven bone and mature lamellar bone (Figure 5B). This, together with the cell morphology in some lamellar bone segments of the autograft group may indicate that those parts correspond to the dead bone that was implanted during the surgery and that it has not yet been remodeled.

Figure 6.

Progression of bone formation spatially and temporally, as marked by fluorochrome labels. A) Representative confocal images of the defect areas capturing the fluorochromes calcein (green, 1 month) and oxytetracycline (pseudo‐colored in red, 2 months). Areas in highlighted white rectangles are shown at higher magnification in the second row. B) Cross‐section of the implanted titanium star‐shaped spacer. The central region of the defect is discerned by a dashed line. The area inside the dashed line (in yellow) was used to determine fluorochrome presence as shown in panel (C). C) A heatmap of the detection of each fluorochrome (as percentage of the number of defects where they were observed) at the central region of the defect. Scale bar: 500 µm.

4. Discussion

Thus far, efforts reflecting the conventional tissue engineering paradigm of combining cells and bioactive molecules within a biomaterial scaffold to regenerate bone have failed to progress from bench to bedside. In order to realize the implementation of BTE as common practice, clinical experience, knowledge of biological mechanisms, and commercial practicality need to come together to nurture the idea. Our group has been committed to creating an off‐the‐shelf modular construct for EBR. The rationale for the development of a devitalized callus mimetic was to use the cues embedded within its ECM to drive ossification in vivo. The use of an engineered allogeneic bone substitute would present some major advantages in comparison to the current clinical gold standard. First, it mitigates the need for autologous bone harvesting and its associated donor site morbidities. Second, it overcomes the intrinsic limitations of autologous bone grafting, such as tissue availability for the majority and poor graft quality due to the presence of pre‐existing patient comorbidities in a subset of patients.^{[ 33} , ^{34 ]} And third, it would present a scalable and storable solution that can be beneficial for many patients.

Hitherto, the development of an engineered bone substitute with the capacity to induce bone regeneration at least comparable to gold standard autografting in large bone defects has remained elusive. Based on our selected species and model, we aimed to validate the feasibility of achieving EBR with a novel off‐the‐shelf callus mimetic in a large preclinical model for the first time. With an appropriate choice of species and model, data generated from preclinical large animal studies can provide further support for the progression toward the human clinical phase. In the present study, our novel off‐the‐shelf callus mimetic exhibits impressive regenerative properties similar to the ones of bone autograft in a clinically relevant orthotopic defect of a goat model. These regenerative properties come with an implant consisting of nonliving engineered tissue, which overcomes several limitations associated with the use of living engineered constructs for EBR applications. Considering translation into the clinics, devitalization prior to implantation not only ameliorates an immune response but also provides the option to store and transport constructs, which is a significant logistical advantage over living tissue implants. Furthermore, the use of allogeneic cells to produce the cartilaginous template allows for pre‐screening of MSCs for a high chondrogenic potential.^{[ 6 ]}

Chondrogenic differentiation of MSCs in a high cell density pellet culture leads to the deposition of cartilage‐specific ECM consisting of GAGs and collagen type II. The extent of in vitro chondrogenic differentiation can be manipulated based on the presence of soluble growth factors and is stipulated to influence bone regeneration.^{[ 35 ]} Further, it is known that MSCs isolated from different species and tissue of origin have different growth factor requirements. We first addressed this issue by determining the optimal culture conditions for chondrogenic differentiation of goat‐derived MSCs. In both culture conditions, TGFβ was included, as it is a known regulator of chondrogenic differentiation of human MSCs in vitro and in vivo.^{[ 36} , ^{37 ]} Notably, animal‐derived MSCs often require the addition of BMP‐2 for chondrogenesis^{[ 26} , ^{27 ]} and the combination of TGFβ1 and BMP‐2 is known to induce expression of molecular markers characteristic of pre‐ and hypertrophic chondrocytes in differentiated MSCs.^{[ 38} , ^{39 ]} Therefore, our second media composition also included BMP‐2. After 28 days, spheroids differentiated in the presence of BMP‐2 showed greater induction of chondrogenesis with a higher deposition of GAGs in the matrix. The TGFβ/BMP group also appeared to have acquired a more hypertrophic phenotype with enhanced ALP activity and VEGF expression in comparison to the TGFβ group. This is consistent with our findings when differentiating human‐derived MSCs in the presence of BMP‐2, where we found that the presence of BMP‐2 leads to enhanced VEGF content per spheroid when compared to TGFβ alone (Figure S5, Supporting Information). The increased presence of hypertrophic markers in the TGFβ/BMP group may be responsible for the improved bone regenerative outcome of this group in the goat model. Indeed, expression of hypertrophic chondrogenic markers in vitro is associated with improved in vivo endochondral bone regenerative potential.^{[ 5} , ^{8 ]}

To investigate whether this difference in GAG content and hypertrophic markers in vitro influenced EBR in vivo, both groups of spheroids were implanted into the orthotopic defect. For ethical considerations, as this study represented a first proof of concept, we aimed to reduce the number of animals to the lowest quantity possible. For this, we implemented the use of a titanium spacer in the defects, thereby increasing the number of groups per animal by threefold, while still creating a critical‐sized defect. After 3 months, micro‐CT analysis showed that the TGFβ/BMP and autograft group showed a higher trend in volume of mineralization. While micro‐CT permits easy reconstruction of images and analysis of mineralization volume, it cannot be used to distinguish between different tissue types (e.g., automineralization, mineralized fibrous tissue or cartilage, newly formed bone, and autograft bone).^{[ 40 ]} Therefore, histomorphometry was carried out as a more reliable measurement of bone formation. Histomorphometrical analysis revealed comparable amounts of bone in the TGFβ/BMP and autograft groups after 3 months. Our data indicate that, while the chondrogenic differentiation induced by TGFβ alone is sufficient to induce new bone formation in an empty defect, the presence of BMP2 in the differentiation media is required to achieve improved bone regeneration in goats.

In the empty defect, this primary bone structure was present toward the edge of the empty defect which most likely occurred from the ingrowth of bone originating from trabecular bone of the lateral wall and exposure of bone marrow, composed of a variety of osteogenic precursors and angiogenic cells.^{[ 41} , ^{42 ]} The higher bone area‐percentage observed in some samples the fibrin carrier group, in comparison to the empty defect group, may be attributed to the inherent properties of fibrin which has been shown to promote angiogenesis and osteogenesis in bone regeneration.^{[ 43 ]} Nevertheless, high variability in bone formation is observed, possibly because of the inherent variability associated with fibrin, the intrinsic characteristic of this natural polymer.

Bone formation in the TGFβ/BMP and autologous group was comparable at the core region of the defect which was reflected by the fluorochrome labeling. Closer examination of the TGFβ/BMP group shows that bone was formed as a surface‐related phenomenon where mineralization was first deposited on the surface of the devitalized spheroids before progressing inward to convert the inside of the callus mimetics. After 3 months, most of the devitalized spheroids were resorbed and replaced by woven bone. In the TGFβ/BMP group, the presence of multinucleated resorbing cells and osteoid deeper within the defect area indicates that maturation and remodeling of newly formed bone is still on‐going upon explantation. Similarly, the autograft group displayed on‐going remodeling toward the core of the defect. However, closer histological examination indicates that part of the bone in the defect of the autograft group was attributed to the primary bone graft placed at surgery (Figure 5B); indicating that not all bone within the autologous group had undergone remodeling and that part of the bone quantified was actually the implanted dead autograft bone. The presence of dead bone in the autologous group suggests that, although comparable total bone volume was observed by micro‐CT and histomorphometry in the TGFβ/BMP group and autologous graft, the bone in the TGFβ/BMP group consists exclusively of newly formed bone, whereas the autograft group includes dead grafting material that has not been remodeled. This would imply that the TGFβ/BMP group had potentially outperformed the autograft group in terms of total new bone formation.

Overall, our data show that spheroids exposed to TGFβ/BMP led to the formation of larger spheroids than TGFβ only, which benefited from enhanced GAG synthesis and ALP activity, leading to improved bone formation. While the microscopic histological analyses highlight the presence of a physiological bone morphology within the defect, whether the newly formed bone has mechanical properties similar to native bone has not yet been determined in the current study. Future studies in load‐bearing locations will further focus on the functionality of the newly formed bone within the defect in addition to its integration with the surrounding tissue. Nevertheless, the results of this study show for the first time, the feasibility of EBR in large animal species.

5. Conclusion

To the best of our knowledge, this is the first study to show the feasibility and efficacy of endochondral bone regeneration in a large animal model, bringing the field closer to application in humans. Using our devitalized callus mimetic spheroids, a bottom‐up approach can be adopted to tailor construct geometry to match the shape of complex or critical‐sized defects. Biofabrication technologies will propagate such developments. Furthermore, ongoing efforts addressing the upscaled manufacture of modular constructs under GMP conditions will contribute to the realization of its clinical implementation.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

de Silva L., Longoni A., Staubli F., Nurmohamed S., Duits A., Rosenberg A. J. W. P., Gawlitta D., Bone Regeneration in a Large Animal Model Featuring a Modular Off‐the‐Shelf Soft Callus Mimetic. Adv. Healthcare Mater. 2023, 12, 2301717. 10.1002/adhm.202301717

Data Availability Statement

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