A total of 72 male New Zealand rabbits (age, 24 weeks; weight, 2.5±0.2 kg) were purchased from the Experimental Animal Center of Zhejiang University of Traditional Chinese Medicine (Hangzhou, China). The animals were bred and maintained under a 12 h light/dark cycle with ad libiutm access to food and water. Room temperature and relative humidity were set at 25±3°C and 60±15%, respectively. All procedures and animal care were approved by the Institutional Animal Care (Hangzhou, China) and Use Ethical Review Committee of Wenzhou Medical University (Wenzhou, China).

MSCs from rabbits were isolated and cultured as previously described (17). Briefly, bone marrow was extracted from the rabbit femur and purified by density gradient centrifugation (1,600 × g; 4°C; 25 min) using Percoll solution (Sangon Biotech Co., Ltd., Shanghai, China). The cells were cultured in Dulbecco's Modified Eagle medium (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) with 10% fetal bovine serum (Gibco) and incubated at 37°C in humidified air with 5% CO₂ for 24 h. Non-adherent cells were discarded when the medium was refreshed, which was performed every 3–4 days. Following culturing for three passages in vitro, the cells were detached and incubated with the anti-rabbit antibodies, CD34 (1:300; cat. no. bs-2042R) and CD44 (1:300; cat. no. bs-0521R) at 4°C overnight. All antibodies were purchased from BIOSS (Beijing, China). Cells were then observed under an inverted fluorescence microscope following immunofluorescence staining at a magnification of ×200. Following 3–4 passages, the MSCs were harvested and used for cell implantation. Cultured MSCs were labeled with 5-bromo-2-deoxyuridine (BrdU; 10 mg/l; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) overnight at 37°C according to the manufacturer's protocol prior to transplantation. Following labeling, the cells were washed five times with PBS to remove all excess unbound BrdU. The MSCs were subsequently harvested (~1×10⁷ cells per implantation) and resuspended in IMDM media for further use.

A rabbit model of ANFH was established according to the Shwartzman response method with modifications: 20 µg/kg Escherichia coli endotoxin (LPS) was utilized instead of the 50 µg/kg dose utilized in the literature (18). Healthy New Zealand rabbits were separated randomly into the normal (n=3) and model groups (n=7; 4 of 7 rabbits succumbed). A total of 20 µg/kg LPS (Sigma-Aldrich) was injected through an ear vein to induce bone necrosis and this process was repeated after 24 h. Following the second administration of LPS, 20 mg/kg MP (Pharmacia and Upjohn; Pfizer Inc, NY, USA) was injected through the gluteus muscle three times with an interval of 24 h between each dose.

All rabbits were examined using a GE Signa EXCITE 1.5T MRI machine (GE Healthcare, Chicago, IL, USA). At 3 weeks an orthogonal head coil was placed on the rabbits while they were anesthetized, with its center located on the femoral head. The layer thickness was 3 mm and the field of view was 240 mm². The changes in femoral head signal and the surrounding soft tissue were observed.

The rabbits were anesthetized using intravenous administration with 3% sodium pentobarbital (30 mg/kg) in an ear vein and the animals were subsequently sacrificed via air embolism. The left femoral head was divided into two parts along the coronal plane of the central hole and fixed in 10% formaldehyde solution (Sangon Biotech Co., Ltd., Shanghai, China) at 4°C for 48 h, decalcified for 3 months with 0.27 mol/l EDTA-2Na (Sangon Biotech Co., Ltd.) at 25°C with pH 7.4. H&E and Sudan III (performed on 5 µm and 20 µm sections obtained from the same part of the femoral head, respectively, at room temperature for 15 min), and PATH staining (performed on 7 µm sections at room temperature for 24 h; all Sigma-Aldrich; Merck KGaA) was performed to observe histopathological changes using a light microscope (Olympus Corporation, Tokyo, Japan). Following H&E staining, the number of empty bone cells were counted at a magnification, ×400 to calculate the empty bone trap rate. Following Sudan III staining, the fat embolism number (number of blood vessels with fat embolization in the femoral head cartilage) was counted by eye using a microscope at magnification, ×200. Following PATH staining, the thrombosis number was observed and the thrombosis rate was calculated (thrombosis number/total blood vessels).

The rabbits in the model group were sacrificed at 1, 2 and 3 weeks following the final injection of MP. RNA was extracted from the right femoral head using a Qiagen RNeasy Mini kit, which included the RNA buffer (cat. no. 74104; Qiagen, Inc., Valencia, CA, USA) after the bone was milled (SPEX 6870 Freezer Mill; SPEX SamplePrep, Metuchen, NJ, USA) in liquid nitrogen. Samples were obtained from the femoral heads of the normal group and animals were sacrificed at the same time points as the model group. The sequences of rabbit MCP-1, SDF-1, BMP, VEGF and β-actin genes were obtained from GenBank (https://www.ncbi.nlm.nih.gov/nuccore). The primer sequences used are listed in Table I. The corresponding primers were synthesized by Sangon Biotech Co., Ltd. (Shanghai, China).

Following RNA extraction from the right femoral head a PrimeScript™ 1st Strand cDNA Synthesis kit (Takara Bio, Inc., Otsu, Japan) was used to reverse transcribe the RNA into cDNA according to the manufacturers' protocol. RT-PCR was performed according to the following thermocycling conditions: 94°C for 3 min, followed by 40 cycles of 94°C for 30 sec, 52°C for 30 sec and 72°C for 20 sec. The relative quantification of each gene mRNA was calculated relative to the β-actin mRNA.

The 24 model rabbits were randomly divided into the following four groups (n=6 in each): Danshen (cat. no. Z33020177-2009; lot no. 0908213; Chiatai Qingchun Bao Pharmaceutical Co. Ltd., Hangzhou China) injection group (2 ml per rabbit), the MSCs group (1×10⁷ cells per rabbit), the danshen combined with MSCs group (2 ml danshen + 1×10⁷ MSCs per rabbit) and the model group, which were administered physiological saline (2 ml per rabbit). The BrdU-labeled MSCs were extracted from rabbits through density gradient centrifugation and adherence screening at the third generation as described above. BrdU has been previously confirmed to have no inhibiting effect on MSC proliferation (19). The rabbits were administered their respective treatments in two sides of the femoral artery at 4 weeks following the final injection of MP. All rabbits were administered penicillin 3 days later to prevent infection following treatment. At 3 and 6 weeks following treatment, 3 rabbits from each group were sacrificed. Immunohistochemistry analysis was used to determine the protein expression of VEGF and BMP-2 and the number of MSCs labeled with BrdU to evaluate the ability of MSCs to migrate to the necrotic area. Femoral head samples were fixed in 10% formaldehyde for 48 h at room temperature, cut into 3–4 µm sections and embedded in paraffin (Sangon Biotech Co., Ltd.). For VEGF, BMP-2 and BrdU immunohistochemical staining, femoral head sections were deparaffinized in xylene and dehydrated with graded ethanol. After washing with distilled water, tissue peroxidase was blocked using 3.0% hydrogen peroxide in methanol for 10 min at room temperature. For antigen retrieval using citric acid buffer, slides were heated at 120°C for 20 min and then cooled for 15 min at room temperature. Following washing with PBS solution, slides were incubated with VEGF (1:100; cat no. SC-7269; Santa Cruz Biotechnology, Inc., Dallas, TX, USA), BMP-2 (1:100; cat no. SAB1411278) and BrdU (1:100; cat. no. B2531; both Sigma-Aldrich; Merck KGaA) antibodies at 37°C for 1 h. Samples were then washed with PBS in triplicate and incubated with anti-mouse horseradish peroxidase (HRP) conjugated immunoglobulin G (IgG; cat. no. 7076) or anti-rabbit HRP-conjugated IgG antibodies (cat. no. 7074; both CST Biological Reagents Co., Ltd., Shanghai, China) diluted (1:2,000) by Antibody Diluent Reagent Solution (cat. no. 1956331A; Thermo Fisher Scientific, Inc.) for 30 min at 37°C. Any immune reaction was detected using 3,3′-diaminobenzidine. Sections were then counterstained at room temperature for 10 min using Meyer's hematoxylin, dehydrated and mounted. The known positive samples were then utilized as positive controls. Following immunohistochemical staining using the Histostain-Plus kit (Mai Bio Co, Ltd., China), the cell number was counted under a light microscope at magnification, ×400.

All statistical analyses were performed using SPSS software, version 13 (SPSS, Inc., Chicago, IL, USA). Data are presented as the mean ± standard deviation and all tests were performed in triplicate. Statistical comparisons between two groups were made using Student's t-test. One-way analysis of variance followed by a post hoc Tukey's test was used to analyze differences among multiple groups. P<0.05 was considered to indicate a statistically significant difference.

There was no clear abnormality observed in the normal group femoral heads (Fig. 1A). In T2 weighted imaging, MRI revealed a normal joint space of the femoral head in the rabbit model and an infusion in the joint cavity. The right femoral head exhibited scattered spots in the medium-high signal region and inhomogeneous low signal mixing. MRI images of the model group revealed that the femoral head was not smooth (Fig. 1B).

H&E staining revealed that in the normal group, the trabecular bone was complete and arranged, the bone marrow was rich in hematopoietic cells and osteoblasts were observed on the surface of the trabecular bone (Fig. 1C). Cancellous bone near the femoral neck had undergone bone necrosis at week 4 in the model group; this was accompanied by neutrophil aggregation (Fig. 1D).

Sudan III staining demonstrated that there was no notable steatosis or lipid embolism in the liver and kidneys in the model group (Fig. 1E). Lipid droplets filled with cellular fluid were observed and the nucleus appeared misshapen under pressure. In the normal group, the lipid droplets in the bone cells were not clear, the fat cells in the bone marrow were smaller compared with the model group and the hematopoietic cells were observed in the voids of the bone marrow (data not shown).

The model group exhibited a large number of fibrin embolus in the femoral head cartilage, which was observed by PATH staining (Fig. 1F). The presence of fibrin embolus was less common in the control group (data not shown).

The empty bone rate, fat embolism rate and thromboembolism rate were calculated according to the aforementioned results and are presented in Fig. 2. The empty bone, fat embolism and thromboembolism rates were all significantly increased in the model group compared with the normal group (P<0.05).

At 3 weeks, few BrdU-labeled MSCs were detected in the normal group (Fig. 3A), whereas a greater number of BrdU-labeled MSCs were observed in the model group (Fig. 3B). The nuclei were stained brown. There was a significantly higher number of MSCs observed in the model group compared with the normal group (P<0.05; Fig. 3C).

RT-PCR revealed that the chemokine expression of MCP-1 and SDF-1 was significantly increased in the model group compared with the normal group at all time points measured (P<0.05; Fig. 4). These results suggest that in the necrotic femoral head MCP-1 and SDF-1 mRNA expression was increased and continued for an extended period of time.

At 3 weeks following the final administration of MSCs and danshen, the mRNA expression of BMP-2 and VEGF in the femoral head and surrounding area was significantly increased in the groups treated with MSCs, danshen, or a combination of the two compared with the model group (P<0.05; Fig. 5), Similar results were observed at 6 weeks. There was no significant difference observed between the results at 3 weeks and the results at 6 weeks. These results suggest that the expression of BMP-2 and VEGF mRNA is increased by treatment with danshen or MSCs, but this difference is not time-dependent. The group that received MSC in combination with danshen demonstrated a significant increase in the level of BMP-2 compared with the single treatment groups (P<0.05). The mRNA expression results were consistent with the immunohistochemistry results.

Immunohistochemistry staining was performed on tissue samples from each treatment group at 3 and 6 weeks following treatment (Fig. 6) The immunohistochemistry results revealed that the protein expression rates of BMP-2 and VEGF in the femoral head at 3 and 6 weeks following treatment were significantly higher in all groups treated with MSCs and/or danshen compared with the model group (P<0.05; Fig. 5B). However there was no significant difference between the BMP-2 and VEGF expression in all of the groups at week 3 compared with week 6. This suggests that the expression of BMP-2 and VEGF may be increased by the injection of danshen and MSCs, either individually or in combination, but this is not time-dependent. The positive expression rate of BMP-2 and VEGF in the MSCs and danshen group was significantly increased compared with that of the MSCs or danshen only groups (P<0.05). At week 6 the MSCs with danshen group had a significantly increased BMP-2 expression compared with week 3 (P<0.05), however this was not observed for VEGF expression. This suggests that the effect of danshen combined with MSCs was higher compared with a single administration of either danshen or MSCs and that BMP-2 expression increased over an extended period of time.