Enhanced healing of goat femur-defect using BMP7 gene-modified BMSCs and load-bearing tissue-engineered bone


  • Lian Zhu,

    1. Department of Plastic and Reconstructive Surgery, Shanghai Ninth People's Hospital, No. 639 Zhi Zhaoju Road, Shanghai 200011, P. R. China
    2. Shanghai Tissue Engineering Center, Shanghai Jiao Tong Medical University, Shanghai, P. R. China
    Search for more papers by this author
  • Dai Chuanchang,

    1. Department of Plastic and Reconstructive Surgery, Shanghai Ninth People's Hospital, No. 639 Zhi Zhaoju Road, Shanghai 200011, P. R. China
    Search for more papers by this author
  • Liu Wei,

    1. Shanghai Tissue Engineering Center, Shanghai Jiao Tong Medical University, Shanghai, P. R. China
    Search for more papers by this author
  • Cao Yilin,

    1. Department of Plastic and Reconstructive Surgery, Shanghai Ninth People's Hospital, No. 639 Zhi Zhaoju Road, Shanghai 200011, P. R. China
    2. Shanghai Tissue Engineering Center, Shanghai Jiao Tong Medical University, Shanghai, P. R. China
    Search for more papers by this author
  • Dong Jiasheng

    Corresponding author
    1. Department of Plastic and Reconstructive Surgery, Shanghai Ninth People's Hospital, No. 639 Zhi Zhaoju Road, Shanghai 200011, P. R. China
    • Department of Plastic and Reconstructive Surgery, Shanghai Ninth People's Hospital, No. 639 Zhi Zhaoju Road, Shanghai 200011, P. R. China, T: 0081-021-63138341-5120; F: 0081-021-5307-8025.
    Search for more papers by this author


Segmental defect regeneration is still a clinical challenge. In this study, we investigated the feasibility of bone marrow stromal cells (BMSCs) infected with adenoviral vector containing the bone morphogenetic protein 7 gene (AdBMP7) and load-bearing to enhance bone regeneration in a critically sized femoral defect in the goat model. The defects were implanted with AdBMP7-infected BMSCs/coral (BMP7 group) or noninfected BMSCs/coral (control group), respectively, stabilized with an internal fixation rod and interlocking nails. Bridging of the segmental defects was evaluated by radiographs monthly, and confirmed by biomechanical tests. Much callus was found in the BMP7 group, and nails were taken off after 3 months of implantation, indicating that regenerated bone in the defect can be remodeled by load-bearing, whereas after 6 months in control group. After load-bearing, it is about 5 months; the mechanical property of newly formed bone in the BMP7 group was restored, but 8 months in control group. Our data suggested that the BMP7 gene-modified BMSCs and load-bearing can promote bone regeneration in segmental defects. © 2009 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 28:412–418, 2010

Although bone is one of the few organs in the body that can spontaneously heal and restore function without scar, it has been recognized that repair of bone is not always satisfactory, and bone defects resulting from trauma or pathology are still a common and significant clinical problem. Tissue engineering has become a promising strategy for bone regeneration in recent years.

Research performed by Urist1 demonstrated that bone contained osteoinductive molecules, termed bone morphogenetic proteins (BMPs). The osteoinductive potential of recombinant BMPs has been demonstrated in animal models,1–4 as well as in clinical studies.5–7 The Food and Drug Administration (FDA) has approved the use of recombinant human BMP2 and BMP7 to enhance anterior spinal fusion and to treat tibial fractures, respectively.8 However, the results have been somewhat mixed, especially in human clinical trials. It usually requires large doses of the protein to be effective. In addition, the biological half-lives of these proteins are exceedingly short, whereas weeks are needed to stimulate a complete osteogenic response. Therefore, local gene therapy systems have been successfully used to deliver BMPs.9–11

Load-bearing is very important to bone formation, which has been described in classic literature, but many studies on tissue-engineered bone regeneration were conducted at non–load-bearing sites or in defects stabilized with stress-shielding devices.12, 13 In addition, those devices, such as bone plates, usually impede the remodeling of the newly formed bone at the defect site. Furthermore, stresses resulting from an inappropriate fixation also negatively affect long-term efficacy for the repair of long-bone defects.14

We have previously demonstrated that bone substitutes constructed with autologous osteogenically induced BMSCs loaded onto coral scaffold, along with the use of the internal fixation method, repair the femur defect, and it becomes a fully functional bone that can bear weight.15 However, it takes a long time for recovery of the femur defect, and we realized that it is crucial to achieve a functional reconstruction quickly for load-bearing of the femur using a tissue-engineering approach. Here, we report an enhanced healing of goat femur-defects by using BMP7 gene-modified BMSCs and load-bearing tissue-engineered bone.


Adenoviral Vector

The complementary DNA (cDNA) of human BMP7 (Gene bank no. BC008584) was inserted into an adenoviral vector. Viral stocks were amplified in 293 cells, and titered by optical density and standard plaque assay. This adenoviral vector, AdBMP7, was used to the autologous osteogenically induced BMSCs.

Generation of BMSCs

The BMSCs harvested from iliac aspirations were cultured, and all osteogenically induced in Dulbecco's modified Eagle's medium (DMEM; Gibco, Grand Island, NY) with 10% heated inactive fetal bovine serum (FBS), dexamethasone (10−8M), L-glutamine (0.3 g/L), β-phospherglycerol (10 mM), and ascorbic acid (50 µg/L) (Sigma, St. Louis, MO). Cells were incubated at 37°C in humidified atmosphere containing 95% air and 5% CO2. BMSCs were cultured for a minimum of 5 days without medium change to facilitate cell attachment to the culture dish. The nonadherent cells were then removed by medium replacement. Medium was replaced every 3 days. Once cells had become confluent (80%–90%), the cells were detached with 0.25% trypsin, 0.01% EDTA, and then sub-cultured and induced at a density of 5 × 105 per dish. Cells within three passages were used in the following experiments.

AdBMP7 Infection and Bone Formation In Vivo

When the adherent cells were confluent (80%), they were infected with AdBMP7 overnight at a multiplicity of infection (MOI) of 100. After 7 days, the infected and noninfected BMSCs, respectively, were harvested and adjusted to a density of 2.0 × 107 cells/ml with PBS. Using nine nude mice, 0.2-ml infected BMSCs' suspension was injected (i.m.) into the right thigh muscle, and 0.2-ml noninfected BMSCs into the left as control. Eighteen spots of injection were monitored for bone formation by radiographs after 3 weeks. The harvested samples were placed in neutral formalin, decalcified in the solution of EDTA, and treated with standard procedures for embedding, sectioning, and staining with hematoxylene.

Preparation of Cell–Coral Construct

Cell–coral constructs were prepared in detail, as previously described.15 Briefly, 5 × 107 cells were suspended in 2.5 ml of PBS and loaded into the cylinder of coral. The cell–coral construct was then placed in an incubator for 4 h. Subsequently, 30 ml of the above-induced medium were added, and the construct was completely submerged into the induced medium and incubated at 37°C in 5% CO2, followed by medium change every 2 days. Cell–cylinder hybrids were co-cultured in vitro for a period of 7 days prior to surgery, and were then implanted in the defect of the respective goats.

Surgical Procedure and Repairing Bone Defects

Twenty adult goats, aged from 12 to 14 months, 20.5 kg to 25 kg (an average of 23.4 kg), were licensed by the Animal Experiment Committee of Shanghai JiaoTong Medical University. The animals were anesthetized through intramuscular injection of ketamine (10 mg/kg). A segmental bone defect, 25 mm in length at the mid-portion of the right diaphysis of each goat, was created with an oscillating bone saw. Bone defects were stabilized with an internal fixation rod and interlocking nails. The operation procedure was detailed as previously described.15 Repairs were performed by implants with AdBMP7-infected BMSCs/coral group (10 goats, BMP7 group) and noninfected BMSCs/coral group (10 goats, control group).

Radiographic and Gray-Density Analyses

Radiographs of both sides of the femurs were obtained under general anesthesia on the first day, and 1–8 months after operation. One stepped aluminum wedge with graduated thickness was included in the field of each radiograph, acting as a standard for density correction of the radiographic technique. Radiographs were evaluated for evidence of new bone formation within the defect, and for development of bone union. The area of newly formed bone in the defect was digitized, and an average density of each radiograph taken at 1–4 months was obtained according to the standard gray density generated from the same stepped aluminum wedge by a computer program. The gray density representing newly formed bone (average density × area) within the defect was also quantitatively analyzed.

Load-Bearing of the Newly Formed Bone

Defect healing results were evaluated by postoperation radiographs and the gray density. As the callus was formed earlier in the BMP7 group than in control group, the upper and lower interlocking nails were removed to give newly formed bone in the defect a mechanical loading (weight-bearing) in the BMP7 group (n = 9) at 3 months, and in the control group (n = 9) at 6 months. One of the animals from each group was sacrificed at the time as non–weight-bearing engineered bone. The other animals were sacrificed after 2 months of weight-bearing. After the fixation rod was removed, the harvested femurs were used for gross observation, histology, and mechanical analysis.

Gross Observation and Histological and Biomechanical Analyses

The defect site of the right femur was assessed by gross observation, including osteo-juncture, length, color, and texture. Transected tissue from the middle portion of the original defect area in each group (n = 1 for BMP7 group at 3 months and 5 months, respectively; n = 1 for control group at 6 months and 8 months, respectively) was fixed in buffered 10% formalin in PBS for 72 h and decalcified in 8% HCl in PBS, 5% formic acid, and 7% aluminum chloride in PBS for another 48 h. The tissues were then embedded in paraffin; 5-µm thick sections were obtained and processed for hematoxylin and eosin staining. Additionally, the femurs harvested from both side legs of the same animal (n = 8 for BMP7 group at 5 months, and n = 8 for control group at 8 months, respectively) were immediately frozen stored at −80°C for mechanical testing. All the femurs were thawed for 24 h in a refrigerator before the test. Femurs were subjected to the mechanical test, and their normal left femurs served as a positive control. The diameters of the repaired defects and the corresponding diameters of the normal left femurs were measured. Each femur was placed in a three-point bending system, providing an unsupported length of 8 cm. Both ends of the tested femurs were not fixed. The condyles were directed upward, so that the anterior surface of the femur was placed in tension under the test. Load was applied in the posterior-to-anterior direction to the midpoint of the unsupported length, approximately in the middle of the repaired defect. The same approach was applied to the same point on the left femurs as a positive control. Testing was conducted in a hydraulic-materials testing machine (Shimadzu AG-20kNH, Kyoto, Japan) and stopped following fracture. Maximum breaking load were recorded. Deformation was measured using the strain gauge at a loading rate of 1.4 mm/min. Data from the strain gauge and the displacement of deformation were all recorded in 50 Newton increments. Then, based on the results, the bend intension and bend stiffness were determined.

Statistical Analyses

To observe gray-density changes, an ANOVA test was applied to analyze gray-density difference between the two groups at different time points, and p-value less than 0.05 was considered statistically significant. In addition, an unpaired t-test was applied to analyze the difference between engineered bones harvested from two groups in maximum breaking load strength, bend rigidity, bend intension, and bend displacement. A paired t-test was applied to analyze the mechanical property difference between engineered and normal femur of the same animals. p-Value less than 0.05 was considered statistically significant.


Bone Formation in Nude Mice

In all nine nude mice, radiographic evidence of bone formation was apparent in 3 weeks at the right thigh muscle (Fig. 1). The cartilage callus and mature osseous components were apparent, which meant the early formation of bone. There was no apparent inflammation around the newly formed tissue (Fig. 2). By contrast, no ectopic radiographic densities were detectable in the left thigh muscle in all the nude mice (Fig. 1).

Figure 1.

Three weeks after injection of BMSCs' suspension, a series of representative postoperative radiographs show new bone formation at the BMP7-treated right thigh muscle (arrow) and nothing at the control.

Figure 2.

Three weeks after injection of BMSCs' suspension, a series of histological evaluations show cartilage callus and mature osseous components at the BMP7-treated right thigh muscle (H&E).

Engineered Bone in the Defect of the Femur

Radiographic and Gray-Density Analyses

In the control group, a coral cylinder with a density less than that of the cortex could be found to bridge the two ends of the defect at day 1 postoperation. At 1 month, callus started to form at the interface between the bone cut ends and the implant. The interfaces were difficult to identify at 3-, 4-, and 5-months postoperation due to the filling in of new bone, and the density level of the repaired defects was uniformly distributed in the area. However, the density of the implant became higher than that of cancellous bone at 6 months, and approached that of the cortex at 8 months. Radiographic union, which was clearly observed, was found in all animals. (Fig. 3A).

Figure 3.

A series of postoperative radiographs show a goat femur-defect (arrows) treated with (A) control BMSCs alone, or (B) BMSCs-based BMP7 gene therapy. Treatment with BMP7 resulted in superior bone formation quickly across the defect.

In the BMP7 group, coral density is similar to that of the control group at day 1 postoperation. However, at 1-month postimplantation, callus was formed surrounding the implants in all animals, connecting the implants to the femur cutting ends, and the interfaces were difficult to identify. More surrounded callus was found at 2 months, and the shape of coral totally disappeared. At 3-months postimplantation, the density of the newly formed callus was higher than that of cancellous bone, just as that of the control group at 6 months, and newly formed cortex was found to bridge the defect at 5 months (Fig. 3B).

Gray-density quantitative analyses demonstrated that the BMP7 group maintained a relatively constant number of gray-density as new bone formation occurred quickly. An increase in gray-density number from day 1 to 3-months postoperation was observed around the defects and, at 4 months, the value reached down, possibly due to the bone remodeling after load-bearing. There was a significant difference in gray-density between the two groups at the period of 3 months (p < 0.05, Fig. 4).

Figure 4.

The plot for gray-density value shows that new bone mass increased bone formation in BMP7-treated animals (ANOVA; p < 0.05).

Gross, Histological Examination

After the fixations were removed, all repaired defects were grossly stable because new bone had been produced, newly formed cancellous bone with red color was observed grossly before weight-bearing, and white-colored remodeled cortex bone was observed after weight-bearing (Fig. 5). The diameters of healed defects in the control group were less than those in the BMP7 group (p < 0.05, Table 1). Normal trabecular and woven bone were uniformly formed in the BMP7 group at 3 months, and in the control group at 6 months, whereas both groups had no coral remaining (Fig. 6). Maturation of the regenerated bone unit occurred in the BMP7 group at 5 months, and in the control group at 8 months with evidence of irregular osteon formation, but the bone unit was higher and better in the BMP7 group than in the control group (Fig. 6).

Figure 5.

Gross examination shows a goat femur-defect (arrows) treated with (A) control BMSCs alone, or (B) BMSCs-based BMP7 gene therapy.

Figure 6.

A series of histological evaluations shows early, good bone formation across the defect in the BMP7-treated group (B), compared to the control group (A) (H&E, × 100).

Table 1. Biomechanical Test Resultsa
Data5 Months8 Months
Experimental FemurNormal FemurExperimental FemurNormal Femur
  • a

    Data present mean value ± SD (n = 8); t-test.

  • *

    Significant difference between femur group at 5 months and at 8 months (p < 0.05).

  • **

    Significant difference between experimental femur group at 8 months and normal femur group at 8 months (p < 0.05).

  • ***

    No difference between experimental femur group and normal femur group (p > 0.05).

Longest diameter (mm)19.03 ± 5.64*,***17.96 ± 7.2618.32 ± 5.30*,***18.29 ± 7.98
Shortest diameter (mm)18.13 ± 1.48*,***17.23 ± 1.3117.34 ± 1.92*,***17.16 ± 1.28
Maximum breaking load Pb(N)2294 ± 227*,***1281 ± 2761412 ± 201*,***1543 ± 112
Bend displacement f (mm) 1.89 ± 0.29*,*** 2.09 ± 0.192.13 ± 0.24*,** 2.03 ± 0.13
Rigidity EJ(N·M/ϕ) 7.53 ± 0.74*,*** 6.56 ± 0.48 6.22 ± 0.61*,*** 6.99 ± 0.87
Bend intension σr(Mpa)73.63 ± 6.42*,***59.67 ± 7.5354.19 ± 5.94*,**61.13 ± 7.24

Biomechanical Observation

The mean values of biomechanical parameters of regenerated bone in the BMP7 group was superior to those in the control group. Unpaired t-test revealed significant differences in maximum breaking load, bend rigidity, bend intension, and bend displacement between groups (p < 0.05, Table 1). Paired t-test of all the parameters comparing the contralateral normal femur with tissue-engineered segmental bone revealed similarities in the BMP7 group at 5 months (p > 0.05, Table 1), but differences with respect to bend intension and bend displacement in the control group at 8 months (p < 0.05, Table 1).


Scaffolds, cell, and growth factors are the three main factors for creating a tissue-engineered construct, which could be used to repair bone defects. The current study demonstrates that critically sized segmental goat defects of the femur can be quickly repaired by load-bearing and BMP7 gene-modified tissue- engineered bone.

A suitable scaffold for bone regeneration must be biocompatible to allow cell attachment, proliferation, ECM deposition, and could ultimately support in vivo bone regeneration. Coral is a commonly used bone graft substitute in clinic with successes in repairing bone defects. Comparing with other scaffolds, coral has more advantages for its microstructure of interconnecting pores similar in nature to cancellous bone, and its degradation rate matching the kinetics of new bone formation in vivo. Coral could be almost degraded after 4 months postimplantation. The rapid degradation of coral could facilitate the radiographic evaluation of healing defects, and diminish the potential risks that are associated with long-term foreign-body sequestration.15–18 We have successfully repaired bone defects of femora,15 crania,18 and mandible (data and photos not shown), with coral as scaffold.

Many researchers proved that scaffold alone or scaffold modified with growth factors is sufficient for bone regeneration and can repair small bone defects.2–4 However, native osteoblasts from surrounding tissue might not be fast or sufficient enough to efficiently generate bone tissue before the complete degradation of the scaffold. Studies using cells for bone regeneration have demonstrated that the seeded cells could not only provide an osteogenic cell source for new bone formation, but also secrete growth factors to recruit native cells to migrate into the defect site.19–21 Although many kinds of derived stem cells besides bone have been studied, such as adipose,20muscle,21 and skin,22 we concluded that bone marrow stem cell is the best candidate seeded cell for bone-engineering by our previous studies in animal models.15, 18, 23–26

Local gene therapy systems have the potential to provide more sustained protein production, deliver protein in a physiologic manner, and allow for the development of biological cellular delivery vehicles that can enhance bone repair. Adenoviral vectors are attractive because both dividing and nondividing cells can be transduced, and there is only short-term production of the gene of BMPs which meet the needs of bone formation. So, a number of laboratories have evaluated the potential of adenoviral BMP-2 to promote spinal fusion in a variety of preclinical models,27, 28 but little research has been done to evaluate the adeno-BMP7 to enhance healing of segmental weight-bearing bone defects. Adeno-BMP7-transduced BMSCs could express BMP7 up to 2–6 weeks,29, 30 inducing the implanted BMSCs and peripheral MSCs to differentiate into osteoblasts, trigger bone regeneration, and repair the defects. It is confirmed in this study that bone formation can be found in the muscle of nude mice when BMSCs are transfected by AdBMP7, and coral seeded with the AdBMP7-infected BMSCs induce impressive callus of bone 4 weeks after implantation and substantial union at the host–implant interface at 8 and 12 weeks, when the defected femur can afford the weight of the body.

In this article, we stabilized the tissue-engineered bone by intramedullary pin, which is clinically relevant. According to our clinic experiment of bone fracture treatment, callus bridging occurs at 3 months post-planted in the BMP7 group and indicates that the initial strength of the newly formed bone has assumed loading function in the defect of the femora, while it is at 6 months post-planted in control group. This is a load-sharing model since the loads are shared by the friction between the intramedullary pin and the contact areas in the medullary canal and by the engineered bone, indicating a stable biomechanical environment conductive to the healing and remodeling of the newly formed bone in the defect. After 2 months of load-bearing, the mechanical properties of femora in the BMP7 group show better than that in the control group.

In summary, this study investigated a tissue-engineering strategy for bone regeneration using gene-modified BMSCs and load-bearing. We found that critically sized segmental defects in the goat's femur have advanced radiological, histological, and mechanical healing using our tissue-engineered strategy of BMP7 gene transfer and load-bearing with intramedullary pins. Since the healing of the defect can be enhanced, it offers the potential for quick bone regeneration.


This study was supported by National Natural Science Foundation of China, National “973” and “863” Project Foundation.