Isolation of rabbit BMSCs and PB–EPCs.

Rabbit BMSCs: Anterior superior iliac spine was sterilized and draped conventionally and then exposed under aseptic condition. Puncture needle was slowly injected 1 cm below the anterior superior iliac spine to extract bone marrow aspirate. The extracted 5 mL of bone marrow aspirate was slowly transferred into a centrifugal tube along its wall and laid onto Percoll solution at a ratio of 1:1. Buffy coat was collected after centrifugation. BMSC complete medium with serum and penicillin–streptomycin solution was added to the buffy coat and mixed well. After dispersing the cell mass, the mixture was transferred into a plastic cell culture flask and then cultured in a 5% CO₂ incubator at 37°C. The medium was replaced after 48 h, and all suspended cells were removed. The medium was then replaced every 24 h. Changes in cell morphology were observed under an inverted phase-contrast microscope.

Rabbit PB–EPCs: Approximately 5 mL of venous blood sampled from the edge of the rabbit’s ear was mixed with Percoll solution for centrifugation. Cells were separated through centrifugation, cultured in EGM-2 BulletKit endothelial cell growth medium, and then inoculated on a fibronectin-coated plate. The plate was placed in 5% CO₂ incubator at 37°C, and the medium was replaced every 24 h. Changes in cell morphology were observed under an inverted phase-contrast microscope.

Flow cytometry.

When the density of BMSCs reached 80%, 2 mL of 0.25% pancreatic enzyme solution containing EDTA was added to the culture. Culture medium was added to stop the digestion after the cells detached from the plate. Cell mass in the medium was dispersed by blowing to obtain a cell suspension. PBS was added to the cell suspension to adjust the cell concentration to 106/mL; 20 μL of the corresponding monoclonal antibodies specific for rabbit CD29, CD34, CD45 were added to a flow cytometry tube and mixed with 100 μL of cell suspension. Cells incubated without antibody served as control. The mixture was incubated without light for 30 min at 4°C. Afterward, unbound antibodies were washed off.

Characterization of EPCs.

The cultured cells were confirmed to be EPCs by detecting the presence of both low-density lipoprotein acetylated DiI complex (acLDL) and FITC-labeled Ulex europaeus agglutinin I (UEA- I), as well as forming 2D networks on Matrigel, which were commonly referred to as EPC characteristics, as previously demonstrated [22, 23]. Briefly, cells were incubated with 2.5 μg/mL acLDL at 37°C for 3 h. The cells were then fixed with 2% paraformaldehyde for 10 min. After washing, the cells were counterstained with 10 μg/mL FITC-labeled UEA-1 for 1 h at 37°C. Stained cells were viewed using a fluorescent microscope. Matrigel complex was diluted at 1:1 ratio with EGM-2 on ice after plating the mixture for 3 h. The cells were seeded on a 24-well plate pre-coated with Matrigel matrix. Tubes were observed under an inverted light microscope.

Preparation of BMSCs expressing eGFP.

BMSCs cultured to the third generation with a density of 1 × 10⁵/mL were inoculated onto six-well culture plates. When the extent of cell confluence reached approximately 80%, 3 mL of culture medium and 1 mL of recombinant eGFP-carrying lentivirus (1.39 × 10⁸ copies/mL) were added to the culture. The culture flask was shaken slowly and then placed in an incubator at 37°C. For screening, BMSCs were transferred to DMEM/F-12 culture medium containing 0.5 μg/mL puromycin. The inserted exogenous plasmid was resistant to puromycin. Hence, we used puromycin to screen the stable expression of eGFP in BMSCs. After 3 weeks, cell growth and expression of GFP were observed. Transduced BMSCs, which were passaged thrice, were injected intravenously into the rabbit bone defect model.

SDF-1 and MCP-1 protein expression of BMSCs expressing eGFP in ex vivo cultures.

The transduced BMSCs were placed in a six-well plate with a density of 10⁵/mL. Cells were analyzed on days 3, 7, and 14. Culture medium was changed 48 h before the assay, and supernatant was collected from the cultured cells. Protein content was determined using SDF-1 and MCP-1 ELISA kits in accordance with the supplier’s instructions, with a minimum detectable concentration of 18 pg/mL. Absorbance was determined at 450 nm. SDF-1 and MCP-1 contents were calculated using the absorbance values of the reference standard and samples. Each experiment was repeated thrice.

Formation of biological bone from a combination of partially deproteinized bone (PDPB) and seed cells.

A pig vertebra obtained from the market was crushed to make bone fragments. These bone fragments were repeatedly washed with distilled water and immersed in ethanol for 24 h. The bone fragments were then immersed in distilled water for 30 min and in acetone for 24 h, with the solution pH adjusted between 7.0 and 7.2. Bones were dried in a drying oven to obtain PDPB after immersion. PDPBs were ground into bone fragments with sizes of 1.2 cm × 0.4 cm × 0.3 cm, which were washed with normal saline by using ultrasonic cleaner, and then dried and sterilized by radiation for later use. PDPBs were randomly divided into four groups: group 1, PDPBs cultured with BMSCs and PB–EPCs (10⁶/mL total at 1:1 ratio, each type cell 5╳10⁵/mL); group 2, PDPBs cultured with BMSCs alone (10⁶/mL); group 3, PDPBs cultured with PB–EPCs alone (10⁶/mL); and group 4, PDPBs cultured alone. Complexes were cultured for 7 days with 4 mL of complete medium. The culture medium of all groups were replaced daily. The 7-day-old complexes were fixed with 30 g/L pentodialdehyde, 1% osmium acid post-fixation, ethanol desiccation, and isopropyl acetate replacement for observation under a scanning electron microscope.

Establishment of rabbit radial defect model and grouping.

Healthy New Zealand white rabbits aged 4 weeks and weighing 800±10 g received intraperitoneal anesthesia via injection of 10% chloral hydrate at a dosage of 2 mL/kg. After routine sterilization and draping, 12 mm long defects were made in the middle radius of bilateral forearms under aseptic condition through skin incisions. The periosteum was then removed and a 12 mm critically sized defect was osteotomized using a bone-grinding instrument. A total of 48 rabbits were randomly divided into four groups (n = 12 each) to receive the following implants: Co-culture group, in which animals were implanted with PDPBs seeded with both BMSC and PB-EPC (i.e., injected BMSC would augment the number of seeded BMSC); BMSC group, in which animals were implanted with PDPBs seeded with BMSC alone (i.e., implanted BMSC would recruit injected BMSC); EPC group, in which animals were implanted with PDPBs seeded with EPC alone (i.e., implanted PB-EPC would recruit injected BMSC); and Unseeded group, in which animals implanted with PDPBs that were unseeded (i.e., baseline control to account for injected BMSC recruitment to the wound site). PDPBs of the four groups were precisely implanted in the defects using a stainless plate fixed with screws on the proximal and distal ends of the radius. The wounds were then sutured layer by layer. Rabbits were reared in separate cages under the same condition and injected with gentamicin sulfate at 2 × 10⁴ units per day to prevent infection. Rabbits that developed fast heartbeat and breathing received 10% chloral hydrate at a dose of 0.5 mL/kg. Rabbits were monitored daily for any complication or abnormal behavior following surgery. After 48 h, rabbits from each group were injected through the caudal vein with BMSC suspension with stable expression of eGFP at a concentration of 5 × 10⁶/mL. No cases of infection, fracture, or death occurred during the course of the experiment.

Quantitative real-time PCR assay.

PDPB-embedded rabbit bone tissues were harvested after the animals were sacrificed through an overdose of chloral hydrate at 8 weeks after surgery. Samples were ground and treated with Trizol to extract the total RNA of cells, as recommended by the manufacturer. Afterward, cDNA was immediately synthesized. Second-strand synthesis and amplification were performed by mixing 1 μL of cDNA, 2 μL of PrimerScript RT Master Mix, 1 μL of PCR forward primer, 1 μL of PCR reverse primer, and pure water to obtain a total volume of 10 μL. The following PCR conditions were used: 95°C for 5 min followed by 40 cycles of 95°C for 15 s, 57°C for 15 s, and 72°C for 30 s. The Ct value was recorded, and GAPDH was used as an internal reference. All reactions were performed in triplicate. Relative gene expression was calculated using the 2^−ΔΔCt method.

ELISA.

The SDF-1 and MCP-1 levels of each cell group were measured using ELISA. A specific ELISA kit was employed with a minimum detectable concentration of 18 pg/mL. Briefly, serum samples were collected at 8 weeks postoperatively. Standard wells, sample wells, and blank wells were diluted as instructed by the corresponding suppliers. Absorption at 450 nm was determined using a microplate reader. The SDF-1 and MCP-1 levels in each serum sample group were calculated using the absorbance values of the reference standard and samples. Each experiment was repeated thrice.

Western blot.

Subsequently, we investigated whether SDF-1 receptor CXCR4 and MCP-1 receptor CCR2 were involved in the homing ability of BMSCs. The material-embedded rabbit bone tissues were harvested after the animals were sacrificed through an overdose of chloral hydrate at 8 weeks after surgery. Protein lysates were obtained in a RIPA lysis buffer with protease inhibitors and protein concentrations determined using Bradford assay (BioRad). Approximately 50 μg of protein was separated on precast 10% SDS-PAGE denaturing gels, which were then transferred to PVDF membranes. Western blot analysis was performed with anti-rabbit CXCR4 and anti-rabbit CCR2 primary antibodies, as well as with relevant secondary antibodies by using standard techniques [24]. All values were averaged at least thrice and normalized to the constitutive actin expression.

Immunohistochemistry.

The material-embedded rabbit bone tissues were harvested at 2, 4, and 8 weeks after surgery and then fixed in 4% paraformaldehyde. A hard tissue slicer was used to cut the tissues into 4 μm-thick sections. Briefly, deparaffinized sections were washed with PBS. Subsequently, eGFP primary antibodies (1:100 dilution) were incubated overnight at 4°C. Polymer helper- and poly peroxidase-tagged anti-rabbit IgG were then dropped to the sections, which were later incubated at 37°C and developed with DAB solution. The sections were routinely sealed and examined under a microscope. Three sections were randomly selected for each sample, and five visual fields were randomly selected for each section. The positively stained area of eGFP was calculated using Image-ProPlus 6.0 image analysis system.

Histological analyses.

Sections were collected via immunohistochemistry at 2, 4, and 8 weeks after surgery. Sections of each group received Masson’s trichrome staining to study the collagen deposition and evaluate the new bone formation of the samples. The distribution of collagen deposition was detected as blue color by using Masson’s trichrome staining kit in accordance with the manufacturer’s instructions. The region of collagen deposition in the samples was calculated using Image J 4.5 version software. Three selected sections from randomly chosen high-power fields (HPFs) were quantified and analyzed. The number of pixels obtained from the three HPFs was summed.