Ex vivo gene therapy with stromal cells transduced with a retroviral vector containing the BMP4 gene completely heals critical size calvarial defect in rats

Article metrics


In order to develop a successful gene therapy system for the healing of bone defects, we developed a murine leukemia virus (MLV)-based retroviral system expressing the human bone morphogenetic protein (BMP) 4 transgene with high transduction efficiency. The bone formation potential of BMP4 transduced cells was tested by embedding 2.5 × 106 transduced stromal cells in a gelatin matrix that was then placed in a critical size defect in calvariae of syngenic rats. Gelatin matrix without cells or with untransduced stromal cells were the two control groups. The defect area was completely filled with new bone in experimental rats after 4 weeks, while limited bone formation occurred in either control group. Bone mineral density (BMD) of the defect in the gene therapy group was 67.8 ± 5.7 mg/cm2 (mean ± s.d., n = 4), which was 119 ± 10% of the control BMD of bone surrounding the defect (57.2 ± 1.5 mg/cm2). In contrast, BMD of rats implanted with untransduced stromal cells was five-fold lower (13.8 ± 7.4 mg/cm2, P < 0.001). Time course studies revealed that there was a linear increase in BMD between 2–4 weeks after inoculation of the critical size defect with 2.5 × 106 implanted BMP4 cells. In conclusion, the retroviral-based BMP4 gene therapy system that we have developed has the potential for regeneration of large skeletal defects.


Gene therapy is a promising modality that has been used for the delivery of diverse recombinant proteins in vivo. We are interested in developing gene therapy to increase the healing rate and regeneration of various types of osseous defects. Among the most potent growth factors that stimulate bone formation are the BMPs, multifunctional cytokines which are members of the transforming growth factor (TGF)-β superfamily.12 The osteoinductive potential of BMPs, especially BMP2, BMP4 and BMP7, makes them clinically valuable as alternatives to bone grafts.3456 In order to promote the healing of fractures or the filling-in of large bone defects, the BMPs should be delivered at adequate concentrations for a period of time sufficient to induce a strong bone formation response. Delivery of the BMP gene into an osseous defect has the potential to induce prolonged local synthesis of BMPs with a single application.

There are several reports that describe the use of vectors expressing the genes for BMP2, BMP4 and BMP7 in gene therapies to stimulate bone formation in a variety of systems (reviewed in 78). Most of the past studies on gene therapy with BMPs have used adenoviral vectors.91011121314 Accordingly, it has been shown that adenoviral transfer of BMP2 can stimulate bone formation at ectopic sites16 and promote healing of a segmental defect in long bones.914 In as much as the results of previous work using adenoviral vectors are promising and adenoviral vector systems have the advantage of high levels of gene expression acutely, adenoviral vector systems also have two important disadvantages: (1) these vectors, even with the use of ‘gutted’ adenoviral vectors, frequently elicit strong immune response, which is highly undesirable. More importantly, the immune response could not only produce adverse health effects in the host, but also could prevent repeated viral administrations. (2) Adenoviral vectors do not allow stable integration of the transgene into the host genome. The lack of stable integration of the transgene and the immune response would probably not allow for sustained expression of the therapeutic gene. In contrast, retroviral vector systems do not elicit substantial immune responses and allow stable integration of the transgene into the host genome and, therefore, could have important advantages over adenoviral vectors. In this regard, we have recently developed an MLV-based retroviral vector system that allows the secretion of biologically active BMP4 protein from transduced marrow stromal cells in amounts often exceeding 1 μg/106 cells/24 h.17 Consequently, we sought in this study to use our recently developed MLV-based BMP4 vector system to assess the usefulness of the retroviral vector-based system for gene therapy of bone repair and regeneration. In support of the contention that MLV-based vectors can be an effective means to deliver a bone growing gene for gene therapy of bone repair and regeneration, several recent reports have obtained promising results with MLV-based vectors in gene therapy of bone defects.181920

There are at least two main routes through which the delivery of a BMP gene to a local site can be accomplished to promote healing: (1) In vivo gene therapy; and (2) ex vivo gene therapy. Because of the low transduction efficiency with in vivo retroviral transfer, we have elected to develop ex vivo gene therapy. For this, we selected bone marrow stromal cells. Such cells are precursors to osteoblasts and are cycling cells and, thus, can be readily transduced with retroviral vectors. Additionally, such cells should be able to survive in their natural habitat within a bony lesion for long periods of time.

We chose a cranial critical bone defect model to assess the utility of our MLV-based ex vivo gene therapy approach to promote bone healing for the following reasons: (1) The healing of this cranial bone defect does not occur spontaneously over an extended observation period (eg for up to 12 weeks), and an addition of the appropriate amount of growth promoting substance can induce complete healing within 3–4 weeks.21 (2) It has recently been shown that implantation of gingival fibroblasts transduced with a BMP7 adenoviral vector in rats22 or autologous stromal cells transduced with a BMP7 MLV-based vector in rabbits20 was able to heal cranial bone defects within 4 weeks and 12 weeks, respectively. (3) The management of the surgical site is relatively easy. Unlike defects involving limbs or the spinal column, no stabilization is required, because the cranial site does not bear significant mechanical loads. Moreover, it is relatively easy to insert test substances such as ex vivo viral transduced cells into the defect and hold them tightly in place with skin sutures.23

In this study, we showed that implantation of a gelatin matrix impregnated with MLV-based BMP4 vector transduced stromal cells completely healed a large cranial critical size bone defect within 4 weeks in the rat. Therefore, this study provides strong evidence that retroviral-based gene therapy with a growth promoting gene has the potential for repairing large skeletal defects.


BMP4 expression in MLV-transduced stromal cells

In order to develop a successful ex vivo gene therapy for bone formation, cells are required that exhibit high-level gene expression and sufficient long-term expression to form large amounts of bone. We have been able to achieve these requirements with an MLV-based retroviral vector that has the gene for the mature human BMP4 protein coupled to a secretion signal from BMP2.17 In the present study, we verified the production of BMP by immunostaining and on Western blots using specific BMP4 antibodies in stromal cells transduced with BMP4 or control MLV virus. A high level of BMP4 transduced cells (>60%) stained positive for the transgene (Figure 1a), while cells transduced with a control LacZ vector show only background staining Figure 1b. In order to quantitate BMP4 expression by stromal cells transduced with BMP4 vector, cells were lysed in electrophoresis buffer for Western blotting with BMP4 antibody. Figure 1c shows that stromal cells transduced with BMP4 showed an immunoreactive band at the anticipated molecular weight of 24 kDa (lane 1), while control transduced stromal cells exhibited no band at this position (lane 2). Quantitative densitometry with known quantities of BMP4 standard on the same blots were performed with various cell preparations. From these experiments, it was estimated that cell lysates contained 0.3–0.5 μg of BMP4 per 1 × 106 cells. This high expression level was maintained at least for the first 4 weeks of continuous culture after transduction. After this initial period, cells continued to express the protein, but expression levels decreased to about half the initial level within 8 weeks after transduction (data not shown).

Figure 1

BMP4 immunostaining and Western blot analysis. Rat bone marrow stromal cells were transduced with MLV retrovirus coding for the BMP4 transgene as described in methods. Panel (a) shows immunostaining with BMP4 antibody of cells 7 days after transduction. Control cells without the BMP-transgene show only minimal background staining (panel b). Panel (c) shows a Western immunoblot after visualization with BMP4 antibody. Cell lysate corresponding to 1 × 105 cells containing the BMP4 gene were loaded in lane 1, while lane 2 contains lysate from cells transduced with MLV LacZ control vector. Lane 3 contains 10 ng of recombinant BMP4 standard. The position and size of molecular weight marker proteins (BioRad) is indicated on the right. A 24 kDa band corresponding to the BMP4 standard (marked by arrow) is seen in cell lysates from bone marrow stromal cells transduced with BMP4 vector (lane 1), but not in control cells (lane 2). A nonspecific protein band of 80 kDa was routinely observed in lysates from both transduced and untransduced cells.

Healing of calvarial defect and quantitation of bone formation after implantation of transduced stromal cells

In rat calvarial critical defects that were implanted with 2.5 × 106 stromal cells expressing BMP4, the entire defect was filled in with bone that appeared to be of higher bone mineral density (BMD) than the peripheral area surrounding the defect, as shown by X-ray imaging after 4 weeks (Figure 2b). In contrast, very little bone fill-in had occurred in controls that had received either gelatin matrix (Gelfoam) alone without cells Figure 2a, or Gelfoam containing control bone marrow stromal cells that had not been transduced with growth factor Figure 2c. To quantitate the amount of newly formed bone, BMD at the defect and surrounding peripheral area was measured by DEXA (Figure 3). BMD in the defect of rats implanted with stromal cells expressing BMP4 was 67.8 ± 5.7 mg/cm2 (s.d., n = 4). This was 119% of the BMD measured in the periphery surrounding the defect in controls (57.2 ± 1.5 mg/cm2, n = 4), suggesting that the defect has been completely filled in with new bone. In contrast, BMD in the defect area from controls without BMP was five- to six-fold lower. Control BMD in the defect area was not different in rats implanted with Gelfoam alone (11.6 ± 9.1 mg/cm2, n = 3) or Gelfoam containing untransduced stromal cells (13.8 ± 7.4 mg/cm2, n = 4). Similarly, when control cells were transduced with MLV-lacZ vector containing the same backbone as MLV-BMP4, they exhibited low bone formation like untransduced cells (see controls in Figures 4 and 5).

Figure 2

Fill-in of calvarial critical defect after implantation of BMP4 expressing stromal cells. Calvariae were harvested for analysis 4 weeks after implantation. Digital X-ray images were obtained with a Faxitron MX-20 specimen radiograpy system. Only a single representative image from each group is shown. The image to the left is from a control implanted with a gelatin sponge containing no cells (a). The calvaria in the center was implanted with 2.5 × 106 BMP4 expressing stromal cells (b), while the calvaria to the right received 2.5 × 106 untransduced control stromal cells (c).

Figure 3

Comparison of BMD in rat calvarial defect 4 weeks after implantation of BMP4-modified or control stromal cells. Density measurements were performed by Piximus X-ray densitometry in calvariae 4 weeks after surgery. A control group was implanted with a gelatin sponge containing no cells (open columns, n = 3). BMP4-expressing stromal cells (solid columns, n = 4) or untransduced control stromal cells (hatched columns, n = 4) were both implanted with gelatin sponges containing 2.5 × 106 cells. The normal bone at the periphery of the defect and circular defect areas were measured as described in the text. (a) P < 0.02 versus no cell control or control cells; (b) P < 0.001 versus no cell control or control cells.

Figure 4

Time course of defect healing after implantation of BMP-producing stromal cells. Rat calvarial defects were implanted with 2.5 × 106 cells and BMD was determined after 14, 21 and 28 days. The values are the average and s.d. (n = 4) from samples implanted with cells transduced with BMP4 (-♦-; b, P < 0.001 versus control). Control animals were implanted with cells transduced with a reporter vector (-▪-).

Figure 5

Effect of cell number on healing of calvarial defect after 2 weeks. Calvarial defects were implanted with gelatin sponges containing 2 × 105, 1 × 106, or 5 × 106 BMP4-expressing stromal cells, or 5 × 106 control stromal cells. Bone mineral density was determined 2 weeks after surgery in the area surrounding the defect (open columns) and the defect area (filled columns; a, P < 0.01 versus control defect area).

In rats implanted with stromal cells expressing BMP4, no appreciable differences were found in the center of the defect when compared with the outer edge in BMP4 implants (data not shown). This most likely indicates that bone formation occurs throughout the defect, and not just at the edge adjacent to the old bone, as occurs without implantation of growth factor. Interestingly, the BMD of the peripheral area surrounding the defect was 126% (72.2 ± 4.7 mg/cm2, n = 4) in implants of BMP4 cells, when compared with samples without cells (57.2 ± 1.5 mg/cm2). Controls with untransduced cells were not elevated (103% of the control sample without cells; 58.8 ± 1.6 mg/cm2). Compared against both controls, the increase obtained with the BMP4-expressing cells was significant (P < 0.02, ANOVA). This indicates that the BMP4 produced by the implanted cells resulted not only in filling of the defect with new bone, but also in increased bone formation in the area surrounding the defect.

Time and cell number effect on bone formation

Figure 4 shows that the BMD increased in a linear fashion with time in the defect area of rats implanted with BMP4 expressing cells. At all the time-points measured, BMD in the defect area of the experimental group was significantly higher compared with the control group 2–4 weeks after implantation (P < 0.001). In control rats, BMD increased slowly between weeks 2 and 3, from 4.3 ± 0.7 mg/cm2 to 10.6 ± 1.9 mg/cm2, followed by a more rapid increase between weeks 3 and 4, to 29.6 ± 7.0 mg/cm2.

Variation in the number of implanted BMP4-producing stromal cells indicated that the bone formation response was cell number-dependent at low levels (2 × 105 and 1 × 106 cells). However, measuring at the 2-week time-point, maximal BMD occurred with 1 × 106 cells in the implanted matrix, and increasing the cell number five times did not further enhance bone formation Figure 5. This may indicate that there is a limited pool of progenitor cells that can be induced to differentiate into bone forming cells by BMP4.

Histological studies

Histological evaluation of the newly formed bone within the defect showed that at 2 weeks there had been an initial deposition of a mixture of fibrous tissue and new bone. There was no evidence of cartilage formation. The presence of active osteoblasts was confirmed by the demonstration of strong alkaline phosphatase activity in the cell membranes of cells surrounding the bone after 2 weeks (Figure 6b), while no alkaline phosphatase staining was seen in typical sections that were implanted with control cells Figure 6a. Active bone formation at 4 weeks was indicated by the presence of plump osteoblasts (large basophilic cytoplasm and lighter staining Golgi) and by the presence of wide osteoid seams on bone surfaces (not shown). About 60% of the bone surfaces could be identified as forming surface by the presence of osteoid seams in defects implanted with BMP4-expressing cells.

Figure 6

Histology of defect area. The left panels show sections that have been stained for alkaline phosphatase 2 weeks after surgery. No alkaline phosphatase is seen in a typical section from a defect implanted with control cells (a), while strong alkaline phosphatase staining in implants with BMP4-expressing cells indicates active bone formation (b). The right panels show the extent of mineralization 4 weeks after implantation of control (c) and BMP4-expressing cells (d). Panel (c) shows the edge of the defect (arrow) after implantation of untransduced control stromal cells. Old lamellar bone (O) is stained dark blue with Goldner's stain. A short area of spongy new bone (N) was formed immediately adjacent to the edge of the defect. Additionally a thin layer of new bone was formed along the dura. The defect is covered with a fibrous membrane (F). Panel (d) is from an implant with BMP4-stromal cells and shows widespread cancellous new bone formation around the edge of the defect and around connective tissue (C) towards the defect center.

In the controls, bone formation measured at 4 weeks was largely confined to the area adjacent to the edge of the defect and at the junction of the gelatin sponge and the dura membrane Figure 6c. In the samples treated with BMP4-expressing cells, the bone formation occurred within and around the gelatin sponge in addition to the junction of the sponge and dura mater. The result was a thick layer of spongy bone that exceeded the thickness of the original calvaria Figure 6d. Examination of the bone deposits indicated that while the initial deposition of bone may have been unorganized, the bone being deposited at both 2 and 4 weeks was laminar in structure. By 4 weeks, the appearance of a few osteoclasts (identified by morphology and the presence of tartrate-resistant acid phosphatase activity) on the outer surfaces of the implants and the presence of scattered areas of marrow within the new bone indicated the initiation of remodeling within the implants (not shown).

Immunostaining of implants for BMP4 showed cells that expressed BMP4 after 2 weeks (Figure 7a), while only very few BMP4-positive cells were detected after 4 weeks (not shown). This suggests that expression of the BMP4 transgene occurs mostly in the initial implant phase, and does not extend significantly past the 2-week point. Control implants did not stain positively for BMP4 after 2 weeks Figure 7bor 4 weeks (not shown).

Figure 7

Immunostaining for BMP4. BMP4-producing cells were identified by immunostaining (brown color, arrows) of demineralized sections from the bones of rats implanted with cells on gelfoam gelatin matrix for 2 weeks. Sections were counterstained with toluidine blue to show the location of cells (dark blue). (a) Implant with BMP4-expressing stromal cells; (b) implant with LacZ stromal cells. The material of the gelfoam implant (labeled g) is stained light blue by toluidine blue. BMP4 producing cells are evident in the implants with BMP4 stromal cells (identified by arrows), but not in control implants.


A major objective of the present study was to evaluate the usefulness of our newly developed MLV-based retroviral vector to deliver sufficient amounts of biologically active BMP4 to heal a large bone defect. Two major problems that limited the usefulness of ex vivo retroviral vector systems to deliver a BMP gene in the past have been the low level of secretion of BMP protein in transduced cells and the relatively low titer. In this regard, our recently developed MLV-based vector system appeared to overcome these limitations.17 Accordingly, we reported that the use of the BMP2/4 hybrid expression construct, in which the propeptide domain of BMP4 was replaced by the corresponding propeptide domain of BMP2, greatly enhanced the secretion of mature and biologically active BMP4 by the transduced cells.17 The titer of our MLV-based BMP4 retroviral vector was in the range of 1 × 106 to 5 × 106 transduction units per ml.17 Although it is relatively low compared with adenoviral vectors, it is high compared with most available retroviral vector systems. Consequently, our recent advance in the development of a high titer MLV-based BMP4 retroviral vector that allows secretion of high levels of functionally active human BMP4 in transduced marrow stromal cells17 provides us with an important tool to test the feasibility of development of an MLV-based retroviral vector system for gene therapy of bone repair and regeneration. Indeed, during the course of our investigations, several recent reports presented results that were strongly supportive of the contention that MLV-based retroviral vector systems expressing a bone growth enhancing gene could be an effective means for gene therapy of bone defects.181920

In the present study, we sought to test whether ex vivo gene therapy using marrow stromal cells transduced with our MLV-BMP2/4 retroviral vector would promote bone formation in a calvarial critical size bone defect model in the rat. Several well-established critical size defect models have been used in the past for investigating the efficacy of biomaterials, growth factors and gene therapy vectors.356 In this study, we used an 8-mm cranial defect model in rats to evaluate the efficacy of ex vivo gene therapy. Several lines of evidence demonstrate that stromal cells transduced with MLV-based BMP2/4 retroviral vector can completely heal the defect within 4 weeks. First, evaluation of the extent of new bone formed by digital X-ray (Faxitron) revealed that the defect was completely filled with new bone in rats implanted with BMP4-expressing stromal cells, while new bone formation filled less than 20% of the area in rats implanted with gelatin sponge alone or gelatin sponge containing control stromal cells. Second, quantitation of the amount of new bone formed at the defect site by bone density measurements using DEXA revealed that bone density in the defect area of rats implanted with BMP4-expressing stromal cells was higher than the area surrounding the defect. In contrast, bone density of the defect area in rats implanted with control cells was less than 25% of the area surrounding the defect. Third, histomorphometric studies revealed evidence for initiation of bone formation at multiple sites in the defect area of rats implanted with BMP4-expressing stromal cells, while rats implanted with control cells had bone formation limited to the edges of the defect. Thus, we have compelling evidence to demonstrate the feasibility of MFG-based stromal cell-mediated gene therapy to heal large size cranial defects.

Although there are several studies which have shown evidence for skeletal regeneration in vivo using ex vivo modified stromal cells, myoblasts, or fibroblasts,101516222425 our study is unique in that it provides quantitative data on the extent of healing by valid bone density measurements. Furthermore, our data demonstrate that the healing of a critical calvarial defect is dependent on the time of exposure, as well as the number of BMP4-expressing stromal cells implanted. Time course studies revealed that there was a linear increase in BMD between 2 and 4 weeks after implantation of 2.5 × 106 BMP4 stromal cells in the calvarial defect. Subsequent experiments with different time-points indicated that there was a lag time of 1 week until bone formation started, and that little additional bone was formed between weeks 4 and 8 (not shown). Therefore, bone formation activity appears to be most active 1–4 weeks after implantation.

When varying numbers of MLV BMP4-expressing stromal cells were implanted, it was found that bone density was increased between 0.2 and 1 × 106 cells, but not between 1 and 5 × 106 cells at the 2-week time-point. We believe that the increased bone production is due to increased local production of BMP4. It is possible that the difference in bone density produced between 1 and 5 × 106 cells may become evident later than 2 weeks. Alternatively, there is a limited pool of progenitor cells available at the implanted site for BMP4 and that BMP4 produced by the highest dose of cell number did not have adequate progenitor cells for recruitment. Future studies are needed to address the limiting factors that contribute to BMP response at the implant site.

The amount of BMP4 produced by stromal cells after implantation depends on the ability of ex vivo modified stromal cells to survive at the implant site. In order to prevent migration of cells from the implant site, stromal cells were cultured overnight on a Gelfoam gelatin sponge. In addition to serving as a vehicle to transfer the cells and keep them localized in the defect area, the Gelfoam sponge may also serve as a three-dimensional scaffold that enhances cell recruitment, attachment and differentiation of local progenitor cells.2627 In our study, we found that neither Gelfoam alone nor Gelfoam containing control stromal cells was effective in completely healing the 8-mm calvarial defect in the rat model used in this study. In contrast, it was found in a previous study that stromal cells alone can fill smaller calvarial defects in a mouse model.28 It remains to be established whether the lack of effect of stromal cells to heal calvarial defects in the rat model compared with the mouse model is due to differences in the size of the defect, cell populations used, or species difference.

An important issue in ex vivo gene therapy studies relates to the time course of transgene expression in implanted cells in vivo. While our stromal cells transduced with MLV-based retroviral vector produced BMP4 protein for sustained periods of time in vitro, expression of BMP4 appeared to decrease much faster in implants. Immunohistochemical experiments detected a small number of BMP4-positive cells remaining after 2 weeks, while no cells staining for BMP4 could be detected after 4 weeks. We cannot presently distinguish whether this loss of transgene expression is due to gene silencing or loss of cells either by migrating out of the implant site, cell death due to lack of nutrition or from elimination by the immune system. Much additional work is needed to definitively determine the fate of transplanted cells. Nevertheless, cell fate and duration of gene expression are very important parameters for gene therapy, because escaping cells might form bone at distant extraskeletal sites. However, we have found no evidence of bone formation at extraneous sites when animals were examined 8 weeks after implantation.

An additional concern for this gene therapy approach is that continuous overexpression of BMP could be toxic, since it has been reported that overexpression of BMP7 from a strong viral promoter is toxic to cells in vitro.20 However, our implant cells expressing BMP4 do not exhibit signs of cytotoxicity in vitro.17 In addition, because only very few BMP4-expressing cells could be detected after 4 weeks in vivo, long-term overexpression of BMP4 does not occur in our system. Consequently, we do not foresee that potential cytotoxicity is a significant problem in our model.

Finally, histological examination of the healing defects showed that the presence of BMP4 producing stromal cells produced multiple sites of de novo bone formation in the Gelfoam-filled defect such that new bone formed throughout the entire implant area. Because there was no cartilage present, the new bone appears to develop in a direct manner similar to that reported for calvarial defects stimulated by the BMP present in demineralized bone particles.29 The bone being laid down at 4 weeks, while cancellous rather than cortical, appeared to have normal laminar architecture. New bone formation was still very active at 4 weeks as shown by the presence of large plump osteoblasts, wide osteoid seams and high alkaline phosphatase activity. Thus, consolidation of the cancellous bone was continuing and therefore could have resulted in regeneration of normal cortices. This densitometric and histologic study indicates that treatment with Gelfoam implants containing BMP4 producing stromal cells is an effective treatment for stimulating bone healing.

In conclusion, we have presented several lines of evidence to demonstrate that the repair of large size cranial defects can be accomplished by implantation of stromal cells expressing BMP4 transgene. These and other data support the concept that retroviral based BMP4 ex vivo stromal cell gene therapy has the potential for regeneration of human skull bone defects.

Materials and methods

Construction of BMP virus

A hybrid expression construct containing the signal sequence of hBMP2 and the mature sequence of BMP430 was prepared and incorporated into a retroviral backbone based on MLV. Details of the construct and biological activity of the secreted mature BMP4 from transformed and nontransformed cells are described elsewhere.17

Preparation and culture of rat bone marrow stromal cells and viral transduction

Stromal cells were prepared as previously described31 with minor modifications and cultured in Dulbecco's modification of Eagle's medium (DMEM) with 10% fetal bovine serum (FBS, Gibco/BRL Life Technologies, Rockville, MD, USA) at 37°C with 5% CO2. Briefly, Fisher 344 rats, 7–10 weeks old, were obtained from Harlan-Teklad, San Diego, CA, USA. Femora and tibiae were dissected and the marrow was collected by flushing the cavity with DMEM. The cells were washed twice with DMEM by low speed centrifugation, dispersed gently to generate a single cell suspension, and placed in 10-cm culture dishes with DMEM and 10% FBS. After 2 days, the nonadherent cells were removed by washing twice with phosphate buffered saline (PBS), and the adhering stromal cells were cultured until confluent. The cells were passaged in 5–7 day intervals and frozen stocks from the first passage were prepared for subsequent experiments.

For transduction, cells were plated at 50% confluence and left to attach overnight. The medium was removed and replaced with viral preparation at a concentration corresponding to 30–50 transforming units/cell in DMEM, 10% FBS, and 8 μg/ml of polybrene. After 8 h, the medium was removed, replaced with fresh viral preparation and left overnight. A third incubation with fresh virus was then performed for an additional 8 h. The transduced cells were expanded for one passage and frozen stock prepared. Cells that were used in animal experiments were in culture for 3–6 weeks after viral transduction. Control stromal cells used were either untransduced (for experiments described in Figures 2 and 6c) or stromal cells transduced with a MLV vector carrying the LacZ reporter gene (all other controls).

For implantation, gelatin sponges (Gelfoam; Pharmacia and Upjohn, Kalamazoo, MI, USA) were cut to the size of the defect and placed in six-well plates. Confluent cells were trypsinized, counted and suspended in complete medium after low speed centrifugation. The desired amount of cells was pipetted slowly on top of the Gelfoam in a volume of 200 μl and left to settle on the Gelfoam for 15 min. After an additional hour in the incubator, 5 ml of complete medium was added and the cells were allowed to attach to the matrix overnight.

Surgical procedures

Male Fisher 344 rats were obtained from Harlan–Teklad, and maintained at the Animal Research Facility of the Jerry L Pettis Memorial VA Medical Center for 7–14 days before the start of the experiment. All animal use procedures were approved by the Institutional Animal Care and Use Committee. Animals aged 7–8 weeks were anesthetized by intraperitoneal injection of ketamine and xylazine and the surgical site prepared by shaving and cleaning with disinfectant. A 3-cm incision was made over the calvariae and the skin held open by retractors. The periosteum was pushed to the side bilaterally and an 8-mm craniotomy defect was created with a trephine (Von Zabern Surgical Instruments, Riverside, CA, USA) attached to an electrical drill. Copious saline irrigation was applied during the procedure and the drilling was stopped when the defect area felt loose when probed. The calvarial disk was then removed by severing the remaining connections with a blunt surgical probe. Extreme care was taken to avoid damage to the dura mater, and occasional bleeding was stopped by temporary application of small pieces of Gelfoam. Gelfoam disks containing cells were then removed from the culture medium, washed 3× in PBS and inserted into the defect. The skin was closed using 4-0 silk sutures.

X-ray imaging (Faxitron) and densitometry (Piximus)

Animals were killed and calvariae dissected out for analysis. Bone formation was assessed in an X-ray specimen radiography system MX20 (Faxitron X-ray Corporation, Wheeling, IL, USA). X-ray images were obtained using Cronex 10T medical X-ray film at 26 keV for 6 s or by using automatic exposure by digital camera. Bone density was quantitated by dual-energy X-ray absorptiometry (DEXA) using a PIXImus soft-X-ray densitometer (Lunar, Madison, WI, USA) and analysis software version 1.45 provided by the manufacturer. A rectangular area encompassing the defect was defined as the total area of measurement. The 8-mm circular defect area was then determined and the peripheral area outside the defect was determined by subtraction. The peripheral area was approximately twice the size of the defect area.

Histology, immunostaining and Western blots

After X-ray imaging and densitometry, calvariae were fixed in 10% neutral buffered formalin overnight, cut in half and embedded either undecalcified in glycol methacrylate or in paraffin after decalcification. For glycol methacrylate embedding, the samples were agitated at 4°C in successive solutions containing 70%, 85%, and 95% glycol methacrylate (JB4 kit, Polysciences, Warrington, PA, USA). Samples were infiltrated with glycol methacrylate containing 0.09% benzoyl peroxide for 2 days in the cold with shaking, and then embedded in glycol methacrylate containing 2% solution B (JB4 kit) and 0.09% benzoyl peroxide. For paraffin embedding, samples were demineralized in 0.1% Na-citrate in 22.5% formic acid, dehydrated in graded alcohol series, and embedded in paraffin. Sections cut from glycol methacrylate blocks were stained with Goldner's Trichrome stain. Sections cut from paraffin blocks were stained for alkaline phosphatase activity and with toluidine blue.

For immunodetection of BMP4, paraffin sections were deparaffinized, rehydrated to H2O and incubated in 3% H2O2 in 70% methanol/PBS for 15 min. The samples were blocked with 5% normal horse serum for 30 min at room temperature. After rinsing twice with tap water for 5 min each, the following immunostaining procedures were carried out at 37 to 40°C with an automatic immunostainer (Ventana, AZ, USA). The sections were first incubated for 32 min with diluted monoclonal anti-BMP4 antibodies (Chemicon International, Temecula, CA, USA), rinsed with TBS and then incubated for 8 min in a biotinylated, rat-adsorbed, horse anti-mouse IgG antibody (Vector Laboratories, Burlingame, CA, USA). BMP4 was visualized by incubation of these specimens with streptavidin-horseradish peroxidase (HRP), diaminobenzidine and H2O2. The sections were then counterstained with toluidine blue (0.33%, pH 2.76). The specificity of immunostaining was determined by incubating sections in the absence of the primary antibody.

For immunostaining of cell cultures, the cells were fixed with formalin and pretreated with 3% H2O2. After an overnight incubation at 4°C with 1:250 diluted BMP4 antibody (Chemicon) in PBS and 0.5% bovine serum albumin, the cultures were rinsed and then incubated with 1:100 diluted biotinylated anti-mouse IgG (Vector Laboratories). Primary antigen-antibody complexes were detected with streptavidin-HRP as described in the instructions for the Vectastain kit (Vector). For Western blots, samples were separated in 10% polyacrylamide gels and transferred to 0.2 μ polyvinylidine difluoride membranes (BioRad, Hercules, CA, USA) by standard methods. BMP4 was detected with 1:1000 diluted monoclonal BMP4 antibody (R&D Systems, Minneapolis, MN, USA), followed by 1:10 000 diluted goat-anti mouse IgG-HRP (Pierce, Rockford, IL, USA). Secondary antibody was visualized by enhanced chemiluminescence (SuperSignal West Pico, Pierce).


Four animals were used for most treatment groups and values are expressed as mean ± s.d. Multiple comparisons between treatment groups were made with the Neuman-Keuls post-hoc test with a two-way analysis of variance (ANOVA).


  1. 1

    Wozney JM et al. Novel regulators of bone formation: molecular clones and activities Science 1988 242: 1528–1534

  2. 2

    Reddi AH . Role of morphogenetic proteins in skeletal tissue engineering and regeneration Nat Biotechnol 1998 16: 247–252

  3. 3

    Gerhart TN et al. Healing segmental femoral defects in sheep using recombinant human bone morphogenetic protein Clin Orthop 1993 293: 317–326

  4. 4

    Cook SD et al. Effect of recombinant human osteogenic protein-1 on healing of segmental defects in non-human primates J Bone Joint Surg Am 1995 77: 734–750

  5. 5

    Zegzula HD et al. Bone formation with use of rhBMP-2 (recombinant human bone morphogenetic protein-2) J Bone Joint Surg Am 1997 79: 1778–1790

  6. 6

    Sheehan JP et al. Molecular methods of enhancing lumbar spine fusion Neurosurgery 1996 39: 548–554

  7. 7

    Winn SR et al. Gene therapy approaches for modulating bone regeneration Adv Drug Deliv Rev 2000 42: 121–138

  8. 8

    Kirker-Head CA . Potential applications and delivery strategies for bone morphogenetic proteins Adv Drug Deliv Rev 2000 43: 65–92

  9. 9

    Baltzer AW et al. Genetic enhancement of fracture repair: healing of an experimental segmental defect by adenoviral transfer of the BMP-2 gene Gene Therapy 2000 7: 734–739

  10. 10

    Lieberman JR et al. The effect of regional gene therapy with bone morphogenetic protein-2-producing bone-marrow cells on the repair of segmental femoral defects in rats J Bone Joint Surg Am 1999 81: 905–917

  11. 11

    Riew KD et al. Induction of bone formation using a recombinant adenoviral vector carrying the human BMP-2 gene in a rabbit spinal fusion model Calcif Tissue Int 1998 63: 357–360

  12. 12

    Lou J et al. Effect of bone morphogenetic protein-12 gene transfer on mesenchymal progenitor cells Clin Orthop 1999 369: 333–339

  13. 13

    Alden TD et al. In vivo endochondral bone formation using a bone morphogenetic protein 2 adenoviral vector Hum Gene Ther 1999 10: 2245–2253

  14. 14

    Musgrave DS et al. Adenovirus-mediated direct gene therapy with bone morphogenetic protein-2 produces bone Bone 1999 24: 541–547

  15. 15

    Moutsatsos IK et al. Exogenously regulated stem cell-mediated gene therapy for bone regeneration Mol Ther 2001 3: 449–461

  16. 16

    Gonda K et al. Heterotopic ossification of degenerating rat skeletal muscle induced by adenovirus-mediated transfer of bone morphogenetic protein-2 gene J Bone Miner Res 2000 15: 1056–1065

  17. 17

    Peng H et al. Development of an MFG-based retroviral vector system for secretion of high levels of functionally active human BMP4 Mol Ther 2001 4: 95–104

  18. 18

    Stover ML et al. Bone-directed expression of Col1a1 promoter-driven self-inactivating retroviral vector in bone marrow cells and transgenic mice Mol Ther 2001 3: 543–550

  19. 19

    Oreffo RO, Virdi AS, Triffitt JT . Retroviral marking of human bone marrow fibroblasts: in vitro expansion and localization in calvarial sites after subcutaneous transplantation in vivo J Cell Physiol 2001 186: 201–209

  20. 20

    Breitbart AS et al. Gene-enhanced tissue engineering: applications for bone healing using cultured periosteal cells transduced retrovirally with the BMP-7 gene Ann Plast Surg 1999 42: 488–495

  21. 21

    Schmitz JP et al. Characterization of rat calvarial nonunion defects Acta Anat (Basel) 1990 138: 185–192

  22. 22

    Krebsbach PH et al. Gene therapy-directed osteogenesis: BMP-7-transduced human fibroblasts form bone in vivo Hum Gene Ther 2000 11: 1201–1210

  23. 23

    Kenley R et al. Osseous regeneration in the rat calvarium using novel delivery systems for recombinant human bone morphogenetic protein-2 (rhBMP-2) J Biomed Mater Res 1994 28: 1139–1147

  24. 24

    Lieberman JR et al. Regional gene therapy with a BMP-2-producing murine stromal cell line induces heterotopic and orthotopic bone formation in rodents J Orthop Res 1998 16: 330–339

  25. 25

    Engstrand T et al. Transient production of bone morphogenetic protein 2 by allogeneic transplanted transduced cells induces bone formation Hum Gene Ther 2000 11: 205–211

  26. 26

    Hollinger JO, Winn SR . Tissue engineering of bone in the craniofacial complex Ann NY Acad Sci 1999 875: 379–385

  27. 27

    Sweeney TM et al. Repair of critical size rat calvarial defects using extracellular matrix protein gels J Neurosurg 1995 83: 710–715

  28. 28

    Krebsbach PH et al. Repair of craniotomy defects using bone marrow stromal cells Transplantation 1998 66: 1272–1278

  29. 29

    Wang J, Glimcher MJ . Characterization of matrix-induced osteogenesis in rat calvarial bone defects: II. Origins of bone-forming cells Calcif Tissue Int 1999 65: 486–493

  30. 30

    Hammonds RG Jr et al. Bone-inducing activity of mature BMP-2b produced from a hybrid BMP-2a/2b precursor Mol Endocrinol 1991 5: 149–155

  31. 31

    Zhang RW et al. Expression of selected osteogenic markers in the fibroblast-like cells of rat marrow stroma Calcif Tissue Int 1995 56: 283–291

Download references


This material is based on work supported by a special appropriation to the Jerry L Pettis Memorial VA Medical Center, Musculoskeletal Disease Center. All work was performed with facilities provided by the Department of Veterans Affairs. We thank Dr D Strong for encouragement and helpful discussions and Dr C Nyght for establishing the Western blot procedures. We acknowledge the excellent technical assistance of Yue-hua Liu, Mitra Bhattacharyya, Cheryl Busse and Jann Smallwood.

Author information

Correspondence to DJ Baylink.

Rights and permissions

Reprints and Permissions

About this article


  • BMP
  • retroviral vector
  • stromal cells
  • bone regeneration

Further reading