ASA scaffolds were fabricated according to the freeze-drying and chemical cross-linking procedures [1011]. Briefly, 10 mL of canine blood was withdrawn by venipuncture and kept at 37℃ for 30 min for fibrin clot retraction. The blood sample was then centrifuged at 1,260 × g for 15 min to obtain the serum, after which 0.5 mL of 25% glutaraldehyde solution (Sigma-Aldrich, USA) was added to 5 mL of the serum obtained for cross-linking. This serum was transferred to a 1 mL disposable syringe and kept until solidification. The solution was subsequently moved to a freezer and maintained at -70℃ overnight. After cutting the syringe, the cross-linked frozen solution was lyophilized for 48 h in a freeze dryer (DFU-8603; Operon, Korea). Next, the cylindrical sponge-shaped, freeze-dried serum was obtained and rehydrated in graded ethanol series (100%, 90%, and 80%) by dipping in each concentration for 1 h. The obtained scaffold was then cut into 20 mm long columns and sterilized in 70% ethanol for 8 h. Next, the scaffold was neutralized in serum-free Dulbecco's Modified Eagle's Medium (DMEM; Gibco, USA), after which excess fluid was removed and the column-shaped scaffolds were placed in a 6-well plate (Thermo Scientific, USA) at one scaffold per well.

In this study, there were four experimental groups depending on the presence or absence of β-TCP and with or without the seeding of Ad-MSCs (panel A in Fig. 1). In groups with β-TCP, 5 mL of serum was added and simultaneously mixed with 0.5 mL of 25% glutaraldehyde solution and 300 mg β-TCP. This solution was then transferred to a 1 mL disposable syringe and rolled until solidification, after which it was processed as described above.

β-TCP granules with particle sizes of 200 to 300 µm were kindly provided by the Biomaterial Center, National Institute for Material Science, Tsukuba, Japan [21].

Canine Ad-MSCs were obtained by culturing to facilitate the proliferation of mononucleated cells from adipose tissues as previously described [1920]. Briefly, adipose tissues were aseptically collected from the gluteal region of a 2 year old beagle under general anesthesia. The adipose tissues were extensively washed with phosphate buffered saline (PBS), then minced with scissors, after which they were digested with 1 mg/mL collagenase type I (Sigma-Aldrich) for 2 h at 37℃. Next, the tissue samples were washed with PBS and centrifuged at 200 × g for 10 min. The resulting pellet containing the stromal vascular fraction was resuspended, filtered through a 100 µm nylon mesh, and then incubated overnight in DMEM with 10% fetal bovine serum (FBS; Gibco) at 37℃ in a 5% CO₂ humidified atmosphere. After 24 h, unattached cells and residual non-adherent red blood cells were removed by washing with PBS. The medium was replaced every 48 h until the cells became confluent. At 90% confluence, the cells were subcultured, and cells at passage 3 were used for seeding into the scaffolds.

A suspension of 1 × 10⁶ cells in 50 µL PBS was mixed with 250 µL purified bovine collagen solution (Nutragen; Advanced BioMatrix, USA) to improve the cell-adherent properties of the scaffolds and increase the rate of MSC-mediated bone formation [18]. The cells were seeded into the scaffolds placed in the wells of a 6-well plate and incubated in growth medium (DMEM supplemented with 10% FBS) at 37℃ and 5% CO₂ for one day before in vivo implantation.

All animal procedures were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee of Seoul National University (SNU-121217-1). Ten skeletally mature beagles (average weight of 8.95 ± 0.96 kg and average age of 1.3 ± 0.5 years) were used for the orthotopic implantation. The segmental resection of a portion of the ulna was performed bilaterally as previously described [22]. Briefly, dogs were sedated with an intravenous injection of acepromazine maleate (Sedaject; Samwoo Medical, Korea) at a dose of 0.01 mg/kg body weight and then premedicated with a subcutaneous injection of atropine sulfate (Jeil Pharmaceutical, Korea) at a dose of 0.1 mg/kg of body weight. Anesthesia was then induced by intravenous administration of a mixture of zolazepam/tiletamine (25 mg/mL, Zoletil 50; Virbac, France) at a dose of 5 mg/kg body weight and maintained with isoflurane (Aerane; Baxter, USA) in oxygen. Intravenously administered tramadol (Toranzin; Samsung Pharm, Korea) and cefazolin sodium (cefazolin injection; Chong Kun Dang, Korea) at a dose of 4 mg/kg and 22 mg/kg body weight were used as an analgesic and antibiotic, respectively. An anesthesia monitor (Datex-Ohmeda S/5; GE Healthcare, Finland) was used to monitor physiological factors including rectal temperature, oxygen saturation, end tidal CO₂, electrocardiogram, minimum alveolar concentration value, and pulse rate. The diaphysis of the ulna was exposed under sterile conditions using the caudal approach, after which the overlying periosteum was resected from the defect area. An eight hole, 2.7 dynamic compression plate (Synthes, Switzerland) was then contoured and applied to the medial portion of the ulna. The plate was subsequently removed, after which a 15 mm long osteoperiosteal segmental cortical defect was created at the mid-portion of the diaphysis with an oscillating saw (Stryker, USA) (panel B in Fig. 1). The defect was then filled with ASA scaffold (albumin [Alb], n = 4), ASA scaffold seeded with Ad-MSCs (Alb + Ad-MSCs, n = 4), albumin mixed with β-TCP scaffold (Alb + β-TCP, n = 4), albumin mixed with β-TCP scaffold plus Ad-MSCs (Alb + β-TCP + Ad-MSCs, n = 4), or nothing (control, n = 4) (panel C in Fig. 1). Next, the plate was reapplied, and the prepared scaffolds were implanted into the defect. Each dog received an implant at random on each ulna. After closing the soft tissue, a Robert Jones bandage was applied for 2 weeks. Two weeks after the surgery, all animals were active and fully capable of walking and bearing weight. The implants were harvested 16 weeks after implantation and used for histological examination (panel D in Fig. 1).

Lateral and craniocaudal radiographs of the antebrachium were made before and immediately after surgery, as well as 4, 8, and 16 weeks after implantation. All radiographs assessed the total area of new bone formation at the proximal and distal host cortex-implant interfaces. Radiographic imaging software (Infinitt Pacs; INFINITT Healthcare, Korea) was used to measure the area occupied by the newly formed bone.

A 5 cm segment of the bone including the defect site was harvested and fixed in 4% paraformaldehyde. The samples were then decalcified with hydrochloric acid, dehydrated in a series of ethanol solutions, and longitudinally embedded in paraffin. Next, samples were cut in the sagittal plane. The central longitudinal sections from each ulna were subsequently placed on the ground and cut to a thickness of 100 µm, after which they were stained with H&E and Massons's trichrome stain to evaluate new bone formation.

The stained sections from each group were observed under a light microscope and scanned using an attached digital camera and a NIS-Elements system (Nikon, Japan). For histomorphometric analysis, all groups were analyzed according to the following protocol. The entire implant area was viewed in six microscopic fields, and each field was captured using a digital camera and then subjected to histomorphometric analyses. The interface between the new bone and host bone in the histological picture was marked by measuring the distance from the screw hole in the radiographs. The newly formed bone area was estimated and converted into a percentage of the total implanted area using image processing and analysis software (ImageJ 1.49; National Institute of Health, USA).

Data were analyzed using the SPSS statistical analysis software (ver. 18.0; SPSS, USA). The Kruskal-Wallis test was used to assess differences among groups. A post-hoc test was performed along with a Mann-Whitney U test. A p value < 0.05 was considered to indicate statistical significance.

In all groups, reactive bone formation at the transversely cut edges of the host ulna was observed at 4 weeks. By 16 weeks, slightly increased bone formation was visible within the gap, especially along the scaffold in all groups except the control. In the control group, similar changes such as reactive bone formation were limited to areas near the interface with the host bone (panel A in Fig. 2). The implanted sites in groups containing β-TCP could be easily visualized because of the radiopacity of the β-TCP particles (panels D and E in Fig. 2).

Radiomorphometric analysis showed that all of the treated groups had significantly larger areas of increased opacity at the proximal and distal host cortex-implant interfaces compared to the control group (p < 0.05). In addition, the groups with β-TCP showed a smaller area of increased opacity at the proximal and distal host cortex-implant interfaces compared to the Alb + Ad-MSCs groups (p < 0.05) (Table 1).

In all experimental groups, new bone formation was observed in longitudinal sections throughout the segmental bone defect at 16 weeks after implantation. Histologically, no adverse host response was detected and newly formed bone was primarily observed at the cortex-implant interfaces, especially the proximal portion. Some of the areas in the middle section of the bone defect were filled with fibrous connective tissue (Fig. 3). In the β-TCP-applied group, the β-TCP remained in the middle section of the defect (Figs. 3 and 4). The Alb + Ad-MSCs group showed chondrocytic changes and new woven trabecular bone formation (Fig. 4).

The bone formation capacity was assessed from the histological images of all the groups by measuring the area of the newly formed bone (Fig. 3). The percentage of new bone formation out of the total area was significantly higher in all treated groups than the control group (p < 0.05) (Table 2). The histomorphometric analysis was consistent with the visual scoring, confirming that the Alb + Ad-MSCs group had a significantly larger area of bone formation than the Alb + β-TCP and Alb + β-TCP + Ad-MSCs groups (p < 0.05). Moreover, the Alb + Ad-MSCs group had significantly greater bone formation at the implant-to-host bone interfaces than the Alb group (p < 0.05) (Table 2).