Converse Relationship Between In Vitro Osteogenic Differentiation and In Vivo Bone Healing Elicited by Different Populations of Muscle-Derived Cells Genetically Engineered to Express BMP4

Authors

  • Hairong Peng,

    1. Growth and Development Laboratory, Children's Hospital of Pittsburgh, Pittsburgh, Pennsylvania, USA
    2. Department of Orthopaedic Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author
  • Arvydas Usas,

    1. Growth and Development Laboratory, Children's Hospital of Pittsburgh, Pittsburgh, Pennsylvania, USA
    2. Department of Orthopaedic Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author
  • Brian Gearhart,

    1. Growth and Development Laboratory, Children's Hospital of Pittsburgh, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author
  • Anne Olshanski,

    1. Growth and Development Laboratory, Children's Hospital of Pittsburgh, Pittsburgh, Pennsylvania, USA
    2. Department of Orthopaedic Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author
  • Hsain-Chung Shen,

    1. Department of Orthopaedic Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
    2. Department of Orthopaedic Surgery, Tri-Service General Hospital, Taipei, Taiwan
    Search for more papers by this author
  • Johnny Huard

    Corresponding author
    1. Growth and Development Laboratory, Children's Hospital of Pittsburgh, Pittsburgh, Pennsylvania, USA
    2. Department of Orthopaedic Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
    3. Department of Molecular Genetics and Biochemistry, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
    • Departments of Orthopaedic Surgery, Molecular Genetics, and Biochemistry, and Bio-engineering, University of Pittsburgh, 4100 Rangos Research Center, Children's Hospital of Pittsburgh, 3705 Fifth Avenue, Pittsburgh, PA 15213, USA
    Search for more papers by this author

  • Dr Huard is the co-founder of Cook MyoSite. All other authors have no conflict of interest.

Abstract

In this study, we compared the use of primary muscle-derived osteoprogenitor cells (PP6 cells) for the delivery of BMP4 to improve bone healing to that of muscle-derived non-osteoprogenitor cells (PP1 cells). Surprisingly, the use of PP1 cells resulted in an improved outcome because of the lack of adverse responses to BMP4 involving cell differentiation, proliferation, and apoptosis.

Introduction: Although researchers frequently opt to use osteogenic cells for osteogenic bone morphogenetic protein (BMP)-based ex vivo gene therapy to improve bone healing, it remains unclear whether the osteogenic potential of a cellular vehicle affects the outcome of bone healing applications. Here we compared the use of muscle-derived non-osteoprogenitor cells (PP1 cells) to that of primary muscle-derived osteoprogenitor cells (PP6 cells) for the delivery of BMP4 to improve the healing of bone defects.

Materials and Methods: Two distinct populations of primary rat muscle-derived cells—PP1 and PP6—were selected, transduced with retroviral vectors to express BMP4 or a marker gene (LacZ), and implanted into critical-sized calvarial defects created in syngeneic rats. The bone healing was monitored radiographically and histologically at 7 and 14 weeks after implantation. Cellular responses to BMP4 were evaluated by alkaline phosphatase histochemical staining and RT-PCR of another osteogenic marker to indicate osteogenic differentiation, a cell proliferation assay and BrdU (bromodeoxyuridine) labeling to assess cell proliferation, and the TUNEL assay to determine apoptosis.

Results and Conclusions: In all animals (nine rats per group), transduced PP1 cells expressing BMP4 demonstrated significantly advanced healing compared with PP6 cells expressing BMP4 and control cells expressing LacZ. We found that constitutive BMP4 expression negatively impacted the in vitro proliferation and in vivo survival rates of PP6 cells, but not PP1 cells. BMP4 exposure also directly inhibited the proliferation and induced the apoptosis of PP6 cells, but not PP1 cells. The impairment in PP6 cell proliferation was directly associated with the osteogenic differentiation of these cells. These results indicate that PP1 cells are better suited than osteoprogenitor cells for use as cellular vehicles to deliver osteogenic BMP4 to improve bone healing and that cellular behavior in response to a particular gene can be used to predict the cells' performance as delivery vehicles in ex vivo gene therapy.

INTRODUCTION

IMPAIRED BONE FORMATION and incomplete bone healing continue to pose challenging problems in current clinical practice.(1) Many different growth factors have been used to improve bone formation and/or regeneration, with bone morphogenetic proteins (BMPs) among the most promising ones investigated to date.(2-5) Recently, ex vivo gene therapy to improve bone healing has garnered great interest because of its markedly enhanced potential compared with pure cell- or BMP protein-based therapy.(6-13)

In addition to the use of proper osteogenic growth factors, the selection of an effective cellular vehicle is critical to the success of any ex vivo gene therapy strategy. Various types of cells have been studied for their ability to deliver osteogenic BMPs to improve bone formation or bone healing. Particular attention has been focused on bone marrow stromal cells or mesenchymal stem cells,(7, 11) mesenchymal stem cell lines,(14, 15) marrow cells,(6) skin fibroblasts,(16) and muscle-derived cells.(9, 10, 12, 13, 17, 18)

Despite the importance of the cellular vehicle in ex vivo gene therapy, few studies have investigated the effects of osteogenic gene expression on the cellular vehicle itself or any subsequent effect on bone healing potential. Researchers have postulated that the use of osteogenic cells as cellular vehicles to deliver BMPs in gene therapy to improve bone healing might prove advantageous because of the dual roles played by these cells—the secretion of bone growth factor (i.e., BMP) and the production of bone matrix through cell differentiation into osteoblasts. However, we are aware of no study designed to directly compare the relative advantages of using primary osteogenic cells versus non-osteogenic cells in terms of their ability to heal bone defects through ex vivo gene therapy. We explored these important questions using muscle-derived cell-based bone gene therapy as a model system. Primary muscle-derived cells were chosen because of our ability to select two distinct cell populations: one that displayed osteogenic potential after stimulation with BMP2 or BMP4 and one that did not. After transduction with a retroviral vector expressing BMP4, we compared the osteogenic cells with the non-osteogenic cells in terms of their ability to heal critical-sized calvarial defects in rats. Finally, we investigated possible mechanisms that might explain any differences in the bone healing ability of these cells. Our findings support the motivating hypothesis that the response of a particular cellular vehicle to the autocrine effects of BMP4 can strongly influence the efficacy of the bone gene therapy protocol because of several mechanisms related to cell differentiation, proliferation, and survival.

MATERIALS AND METHODS

Preparation and characterization of rat muscle-derived cells

Rat muscle-derived cells were isolated using a preplate technique as described previously.(19, 20) In brief, muscle obtained through biopsy from adult Fischer 344 rats was minced and sequentially digested with 0.2% collagenase XI for 60 minutes, 0.3% dispase for 45 minutes, and 0.1% trypsin for 30 minutes The cell clumps were disrupted by passage through a 27G needle. At the end of preparation, these cells were suspended in DMEM supplemented with 10% FBS, 10% horse serum, 0.5% chick embryo extract, glutamine (2 mM), streptomycin (100 μg/ml), and penicillin (100 U/ml); the cells then were plated in type I collagen-coated flasks and incubated at 37°C with 5% CO2. After about 1 h, the medium containing nonadherent cells was transferred into a fresh collagen-coated flask. The cells that adhered onto the first flask were called PP1 and cultured in the growth medium described above. Serial replating of nonadherent cells was performed when 30-40% of the cells had adhered to a particular flask. The rat muscle-derived cells were separated into a total of six fractions or populations, sequentially named from PP1 to PP6. Vimentin immunostaining was performed as described previously,(11) using anti-vimentin monoclonal antibodies (Clone VIM 13.2; Sigma Diagnostics). Myogenic differentiation into myotubes was induced by culturing the cells in medium containing 0.5% horse serum for 1 week.

BMP4 and BMP2 stimulation of rat muscle-derived cells

Among the six populations of rat muscle-derived cells, the PP1, PP3, and PP6 cells were selected for further testing. After being treated with 100 ng/ml rhBMP4 (R & D Systems) or rhBMP2 (R & D Systems), the cells were harvested at designated time points, and their expression of alkaline phosphatase (ALP)—an early osteogenic marker—was evaluated by both cytochemical staining and analysis of ALP enzymatic activity. ALP staining was performed using the AP Kit 86-C (Sigma Diagnostics), and ALP enzymatic activity was assessed with the Sigma Diagnostics ALP reagent, ALP20, using the protocols recommended by the manufacturer.

RT-PCR analysis of other osteogenic markers

Cellular RNA was isolated using the RNeasy kit (Qiagen). For each reverse transcription, 5 μg of RNA was used, and the reactions were performed according to the procedure recommended by the manufacturer (Superscript First-Strand Synthesis System for RT-PCR; Invitrogen Life Technologies). One-fifth of the RT product was used for each PCR, using primers β-actin 1 (5′-GGTGGCCGCCCTAGGCACCAG-3′) and β-actin 2 (5′-CTCTTTAATGTCACGCACGATTTC-3′) to detect β-actin mRNA; ALP 1 (5′-GGCCAAGGATGCTGGGAAGTC-3′) and ALP2 (5′-GTCAGGATCCGGAGGGCCACC-3′) to detect ALP mRNA; and PTH-R1 (5′-CCAGTGCTCAGCTCCGCATATGCG-3′); and PTH-R2 (5′-CATGCCTAGGCGGTCAAATACCTC-3′) to detect parathyroid hormone receptor (PTH-R) transcripts. The PCR products were separated in 1.5% agarose containing ethidium bromide, and were photographed under UV light.

Retroviral transduction

Retroviral vectors expressing BMP4 or LacZ (CLBMP4 or CLlacZ) were generated as described earlier.(12) Rat muscle-derived cells were transduced separately with the retroviral vectors CLBMP4 and CLlacZ at a multiplicity of infection equal 5 (MOI = 5) in the presence of 8 μg/ml polybrene. One week after transduction, the cells and the conditioned media were sampled for BMP4 bioassay and Western blot analysis.(11) The β-gal staining was performed using the β-Gal Staining Kit (Invitrogen) according to the instructions provided by the manufacturer. The transduced cells were expanded for 2-4 weeks to obtain the number of cells required for use in the animal experiments.

ALP and bromodeoxyuridine double-labeling

After BMP4 stimulation of the PP6 cells cultured in chamber slides for a designated period of time, bromodeoxyuridine (BrdU; Amersham) was added to the culture medium at a final concentration of 10 μM. After culturing for an additional 24 h, the cells were collected for ALP staining as described previously. The ALP-stained cells were then subjected to BrdU immunostaining. In brief, the cells were fixed in methanol at −20°C for 20 minutes. The cellular DNA was then denatured by incubation with 2 M HCl for 2 h. After quenching the endogenous peroxidase with 0.3% H2O2 for 5 minutes, the cells were blocked with 2% normal horse serum for 30 minutes and sequentially incubated with 1:100 diluted biotinylated anti-BrdU monoclonal antibodies (Zymed Lab) for 90 minutes, ABC reagent (Vector Laboratories) for 30 minutes, and DAB substrate (Vector Laboratories) for 5 minutes, with PBS washing between each step. The cells were counterstained with eosin and mounted. The percentages of ALP+ and BrdU+ cells were determined by counting the number of respective cells in six representative fields on each slide.

Animal experiments

All animal experiments were conducted with the approval of the Animal Research and Care Committee of the Children's Hospital of Pittsburgh (Protocol 02/01). For skull defect assay, 36 Fischer 344 rats (male, 9 weeks old) were divided randomly into four designated groups. Under anesthesia, a 9-mm-diameter defect was created in the parietal bone without breaching the dura. A 9-mm-diameter Gelfoam disk (Pharmacia and Upjohn), impregnated with 2,000,000 transduced cells, was implanted into the defect, and the wound was closed with suture. The dose we used (2 × 106 cells/disk) is the optimal one according to the results of our pilot experiment, which showed that lower doses led to less bone formation, whereas higher doses did not significantly increase bone formation. Bone healing was monitored radiographically and histologically at designated time points after surgery. The bone density detected by radiography was estimated using Northern Eclipse software, version 6.0 (Empix Imaging). The statistical differences between groups were analyzed using a two-tailed Student's t-test performed with Microsoft Excel software (Microsoft).

For the cell survival assay, 10 Fischer 344 rats were randomly divided into two groups. Group 1 rats were implanted with PP1-BMP4 cells in the right thigh and PP1-LacZ in the left thigh, whereas group 2 animals received PP6-BMP4 cells in the right thigh and PP6-LacZ cells in the left thigh. The cells were seeded onto 9-mm Gelfoam disks and implanted subcutaneously using a sterile surgical procedure. At designated time points after cell implantation, the rats were killed, and the disks were dissected, snap frozen, and stored at −80°C for histological analysis.

Histological analysis

After fixation in 14% neutralized formaldehyde for more than 48 h, the skull samples were decalcified in 10% EDTA (pH 7.0) for 3 weeks. The decalcified tissues were embedded and sectioned to a thickness of 6 μm. The tissue sections were stained with hematoxylin and eosin (H&E) and visualized under microscopy. For the cell survival assay, the cryosections of implanted tissue were fixed in 0.5% glutaraldehyde, stained with the β-Gal Staining Kit (Invitrogen), and counterstained with eosin. The number of β-gal+ cells was counted on six representative fields on each section with five sections counted in each group.

Cell proliferation and apoptosis assays

To compare the cell proliferation rate of PP1 and PP6 cells transduced with retroBMP4 or retroLacZ, the cells were seeded into T175 flasks precoated with bovine collagen type I (5 × 105 cells/flask). After 7 days in culture, the cells in the flasks were trypsinized and counted. To determine the effects of BMP4 protein stimulation on cell proliferation or apoptosis, the cells were treated with recombinant human BMP4 (rhBMP4; 50 ng/ml) with or without noggin (200 ng/ml; R & D), transforming growth factor-β2 (TGF-β2; 10 ng/ml; Sigma), or vehicle (bovine serum albumin [BSA]) for 24 h. Cell proliferation was determined using the CellTiter 96 Aqueous One Solution Cell Proliferation Assay (Promega) in accordance with the protocol recommended by the manufacturer. Apoptotic cells were detected using the In situ Cell Death Detection, POD kit (Roche Molecular Biochemicals) according to the instructions provided by the manufacturer. More than 500 cells in each quadruplicated sample were counted to determine the percentage of apoptotic cells in each group. Statistical differences in cell proliferation were analyzed by a two-tailed Student's t-test; statistical differences in apoptosis were evaluated using a χ2 test.

RESULTS

Disparate responses of different populations of muscle-derived cells to BMP2 and BMP4 stimulation

Rat primary muscle-derived cells were isolated using enzymatic digestion and separated into six populations based on their ability to adhere to collagen-coated flasks. Three representative populations—PP1, PP3, and PP6—were selected for further testing. Rat muscle PP1 cells are mostly fibroblast-like cells (Fig. 1A): 70-80% of the cells tested positive for vimentin, a marker of fibroblasts (data not shown). This is similar to the composition of PP1 muscle cells isolated from mice.(17-20) The rat PP6 cell population is highly enriched for myogenic cells, evidenced by the formation of myotubes after cell culturing in differentiation medium (data not shown), which is normal medium with 0.5% horse serum.(20) After stimulation with BMP2 and BMP4 for 5 days, PP1 and PP3 cells did not undergo detectable osteogenic differentiation, as indicated by the lack of a significant increase in the number of cells expressing ALP, a marker of early osteogenic differentiation (Fig. 1A), compared with the control. In contrast, PP6 cells responded strongly to BMP2 and BMP4 stimulation with a marked increase in the number of ALP+ cells (Fig. 1A). Similar patterns of response were observed with serial doses of BMP4 (50, 100, and 200 ng/ml) and different stimulation periods (3, 5, and 8 days). Additionally, RT-PCR results indicated that the expression of the osteogenic markers ALP and PTH-R also increased in PP6 cells after BMP4 stimulation (Fig. 1B). These findings indicate that rat PP1 and PP3 cell populations contain few myogenic cells and have little osteogenic potential. In contrast, the rat PP6 cell population is enriched with myogenic cells and has high osteogenic potential.

Figure FIG. 1..

Rat muscle-derived PP6 cells, unlike PP1 and PP3 cells, underwent osteogenic differentiation after BMP2 or BMP4 stimulation. (A) BMP2 and BMP4 stimulation induced osteogenic differentiation of the PP6 cells, as indicated by a marked increase in the number of cells expressing the early osteogenic marker ALP. In contrast, the PP1 and PP3 cells did not undergo detectable osteogenic differentiation after BMP2 or BMP4 stimulation, as indicated by no significant change in the number of ALP+ cells. In the control groups, cells were treated only with the vehicle (0.2% BSA). Bar: 20 μm. (B) RT-PCR detection of osteogenic mRNA expression in PP6 cells after BMP4 stimulation for 5 or 8 days (5d and 8d). The control cells were cultured for 8 days in the normal medium containing only the vehicle. PTH-R, parathyroid hormone receptor; β-act, β-actin.

Retroviral transduction of rat muscle-derived cells

The ideal cellular vehicle for ex vivo gene therapy should be susceptible to transduction by a retroviral vector, the most commonly used vector for ex vivo gene therapy. When transduced with a retroviral vector expressing the LacZ reporter gene (retroLacZ), more than 90% of the rat muscle-derived cells were transduced, as indicated by the expression of β-galactosidase (β-gal). There was no significant difference in susceptibility to retroviral transduction among the three populations of rat primary muscle-derived cells that were tested (PP1, PP3, and PP6; Fig. 2A).

Figure FIG. 2..

Transduction of rat muscle-derived cells with a retroviral vector and expression of LacZ or BMP4 in the transduced cells. (A) β-Gal staining after transduction with retroLacZ reveals that the three different populations of rat muscle-derived cells (PP1, PP3, and PP6) were equally susceptible to retroviral transduction. (B) Western blot analysis showed BMP4 expression in the PP1 and PP6 cells transduced with retroBMP4, but not in the cells transduced with retroLacZ. Each lane was filled with cell lysate equivalent to 105 cells except for the last lane, which was filled with 10 ng rhBMP4 to serve as the positive control. Bar: 20 μm.

In light of the aforementioned initial characterization of the rat muscle-derived cells, we selected two distinct populations (PP1 and PP6) for further testing. We began by examining the level of BMP4 expression exhibited by these cells after transduction with a retroviral vector expressing BMP4 (retroBMP4). Transduced PP1 and PP6 cells secreted biologically active BMP4 at a level of 50 ± 10 ng/106 and 45 ± 10 ng/106 cells/24 h, respectively. In contrast, cells transduced with retroLacZ (PP1LacZ, PP6LacZ) did not secrete detectable levels of BMP4. These results were confirmed by Western blot analysis (Fig. 2B).

Healing of critical-sized bone defects elicited by rat muscle-derived cells expressing BMP4

To determine the capacity of the transduced cells to elicit healing of rat critical-sized calvarial defects, we implanted an identical number (2 × 106 cells) of either PP1 or PP6 cells transduced with either retroBMP4 or retroLacZ (as a control) into the defects. Our pilot experiment indicated that this dose of cells is optimal for inducing bone healing. The degree of bone healing was monitored radiographically and histologically at various time points after cell implantation. The defects implanted with PP1 cells expressing BMP4 (PP1-B4) showed evidence of more advanced healing 7 weeks after implantation in all four of the four animals tested (Fig. 3A). In contrast, no bone healing occurred in the defects implanted with PP6 cells transduced with BMP4 (PP6-B4; Fig. 3A); nor was any healing observed in the defects implanted with either PP1 or PP6 cells transduced with the control vector, retroLacZ (PP1-LacZ or PP6-LacZ; Fig. 3A).

Figure FIG. 3..

Rat muscle-derived PP1 cells transduced with a retroviral vector expressing BMP4, in contrast to transduced PP6 cells, elicited healing of critical-sized calvarial defects. (A) Radiography indicates healing of the bone defects implanted with PP1 cells expressing BMP4 at 7 and 14 weeks after implantation, but not in the defects implanted with the control group of PP1 cells expressing LacZ. PP6 cells expressing BMP4 had not healed the defects when examined at 7 or 14 weeks after implantation; nor had the PP6 cells expressing LacZ healed the defects when observed at the same time points. (B) Quantitative analysis shows that the relative bone density in the defects implanted with the transduced PP1 cells expressing BMP4 increased between 7 and 14 weeks. In contrast, the bone density in the defects treated with the other cell types did not increase during this time frame. *p < 0.01, n = 4.

It is possible that bone healing was simply delayed (rather than absent) in the defects implanted with the PP6-B4 cells. To rule out this possibility, five more animals in each group were followed for another 7 weeks. The bone density increased by ∼23% in the defects implanted with the PP1-B4 cells (Fig. 3), indicating continuous mineral deposition in the regenerated bone from 7 to 14 weeks after implantation. At 14 weeks after implantation, there was still no detectable bone healing in the defects implanted with the PP6-B4 cells or in the defects implanted with the control cells (Fig. 3A). Quantitative analysis of the bone healing process revealed a significant difference between the healing group (PP1-BMP4 or P1B4) and the nonhealing groups (PP6-BMP4 or P6B4, PP1-LacZ or P1Lac, and PP6-LacZ or P6Lac; Fig. 3B).

The radiographic evidence of bone healing was confirmed by histological analysis. Bony bridging of the defects had occurred by 7 weeks after implantation in the defects implanted with PP1-B4 cells (Fig. 4). However, no bone healing occurred in any of the other groups (Fig. 4). These results show that PP1 cells transduced with a retroviral vector expressing BMP4 can elicit more advanced healing of critical-sized defects, whereas PP6 cells transduced with the same vector cannot.

Figure FIG. 4..

Histological evidence of bone healing in the skull defects. H&E staining of the decalcified skull tissue reveals more advanced bone healing of the critical-sized defect implanted with PP1 cells expressing BMP4. Fibrotic tissue filled the defects implanted with PP6 cells expressing BMP4, control PP1 cells expressing LacZ, and control PP6 cells expressing LacZ. Arrows indicate the edges of the bone defects. Bar: 200 μm.

BMP4 expression hampered proliferation of the retrovirally transduced PP6 cells but not the transduced PP1 cells

To elucidate the mechanisms responsible for the markedly different abilities of the retroBMP4-transduced PP1 and PP6 cells to heal bone defects, we explored additional differences between these two transduced cell populations. Besides the expected difference in the number of ALP+ cells (the PP1-B4 cell population contained far fewer ALP+ cells than did the PP6-B4 cell population), the growth kinetics of these cell populations were also very different: the PP6-B4 cells proliferated more slowly than the PP1-B4 cells. Indeed, careful cell growth analysis revealed that the PP1-B4 cells proliferated at a rate similar to PP1-Lac cells, whereas the PP6-B4 cells proliferated at a significantly slower rate than PP6-Lac cells (Fig. 5). The average doubling times were 65 h for PP1-B4 cells, 62 h for PP1-Lac cells, 134 h for PP6-B4 cells, and 70 h for PP6-Lac cells. These results indicate that BMP4 expression in the transduced cells hampered the proliferation of the primary rat muscle-derived PP6 cells but not that of the PP1 cells.

Figure FIG. 5..

BMP4 expression impaired the proliferative capacity of the rat muscle-derived PP6 cells, but not that of the PP1 cells. After retroviral transduction, PP6 cells expressing BMP4 (PP6BMP4) proliferated at a markedly lower rate than did PP6 cells expressing LacZ (PP6LacZ). In contrast, PP1 cells expressing BMP4 (PP1BMP4) proliferated at a rate similar to that of control PP1 cells expressing LacZ (PP1LacZ). n = 6 in each group.

BMP4 inhibited the proliferation of the rat muscle-derived PP6 cells but not the PP1 cells

The impeded proliferation of the PP6 cells after transduction with retroBMP4 may have resulted from either the direct autocrine effect of BMP4 protein or from the combined effect of both retroviral transduction and BMP4 expression. To explore these two possibilities, we determined the acute effect of BMP4 treatment on the proliferation of nontransduced PP1 and PP6 cells. BMP4 treatment for 24 h led to significant inhibition of PP6 cell proliferation (Fig. 6B), but not of PP1 cell proliferation (Fig. 6A). PP6 cell proliferation continued to be inhibited for up to 7 days (the end of the experiment), as long as fresh BMP4 was added to the culture medium every day (data not shown). The inhibitory effect on rat muscle-derived PP6 cells seems to be specific to BMP4; TGF-β2, another member of the TGF-β superfamily, did not exert a similar effect (Fig. 6B). Such specificity was confirmed by the ability of the BMP4 antagonist noggin to block the inhibitory effect of BMP4 on PP6 cell proliferation (Fig. 6B).

Figure FIG. 6..

BMP4 inhibited the proliferation of the PP6 cells but not that of the PP1 cells. (A) PP1 cell proliferation was not affected by treatment with recombinant human BMP4 (50 ng/ml), TGF-β2 (10 ng/ml), or the vehicle. The number of viable cells correlated positively with OD490. (B) Proliferation of the PP6 cells was inhibited by BMP4 but not by TGF-β2. The inhibitory effect of BMP4 on the PP6 cells could be alleviated through administration of the BMP4 antagonist noggin. n = 8 in each group.

BMP4 inhibits proliferation of osteogenically differentiated PP6 cells, but not undifferentiated PP6 cells

Two mechanisms may have accounted for the inhibition of PP6 cell proliferation by BMP4: (1) direct inhibition of PP6 cell proliferation by BMP4 regardless of the cells' differentiation status or (2) osteogenic differentiation of PP6 cells leading to impaired cell proliferation. To distinguish between these two possibilities and to directly investigate the relationship between osteogenic differentiation and cell proliferation, we used the ALP-BrdU double-labeling technique to simultaneously detect differentiated cells and proliferating cells in the PP6 cell population after stimulation with different doses of BMP4 for 5 days. The PP6 cells responded to BMP4 treatment by exhibiting a dose-dependent increase in the number of cells expressing ALP (Figs. 7A and 7B). As expected based on the second possibility, most of the ALP+ cells were not proliferating, as indicated by the absence of BrdU labeling (Fig. 7A). Accordingly, most of the proliferating cells were BrdU+ and ALP (Figs. 7A and 7C). Notably, the percentage of proliferating, ALP cells did not decrease with increasing concentrations of BMP4. These results indicate that BMP4 did not inhibit proliferation of the PP6 cells that had not undergone osteogenic differentiation. Thus, the global reduction in the proliferation of the PP6 cells after BMP4 treatment was attributable primarily to the inability of the cells to proliferate after osteogenic differentiation.

Figure FIG. 7..

BMP4 inhibited proliferation of osteogenically differentiated PP6 cells, but not of undifferentiated PP6 cells. (A) ALP and BrdU double-labeling shows that the ALP+ cells (stained purple) were not proliferating, as indicated by the absence of BrdU-labeling in the nuclei. In contrast, many ALP cells were proliferating, as demonstrated by the presence of BrdU-labeled nuclei (stained brown, indicated by arrows). Serial doses (50, 100, 200 ng/ml) of BMP4 were used in the experiments. (B) A dose-dependent osteogenic response of PP6 cells to BMP4 stimulation was observed. (C) Quantitative analysis shows that most of the ALP+ cells were not proliferating. In contrast, a high percentage of the ALP cells were proliferating. The response pattern was independent of the BMP4 dosages. Bar: 10 μm.

BMP4 induced apoptosis of the PP6 cells but not the PP1 cells

Recently, BMP2 has been demonstrated to induce apoptosis of osteoblasts.(21) We thus hypothesized that BMP4 might exert a similar pro-apoptotic effect on rat muscle-derived PP6 cells, which display osteogenic potential. To investigate this possibility, we used the TUNEL assay to detect apoptotic cells in nontransduced PP1 and PP6 cell populations after treatment with BMP4, TGF-β2, or vehicle (BSA). Our results indicate that BMP4 treatment induced significantly more apoptosis in the PP6 cells than in the PP1 cells (Fig. 8). In contrast, TGF-β2 treatment did not lead to significant apoptosis of either the PP1 cells or the PP6 cells (Fig. 8). These findings indicate that the rat PP6 cells, but not the PP1 cells, are susceptible to the pro-apoptotic effects of BMP4.

Figure FIG. 8..

BMP4 induced apoptosis of rat muscle-derived PP6 cells but not of PP1 cells. (A) TUNEL assay shows apoptotic cells with brown-stained nuclei (indicated by arrows) after treatment of the PP6 cells with BMP4. Apoptotic cells were not observed after treatment of the PP1 cells with BMP4 or after treatment of either the PP1 or PP6 population of cells with TGF-β2 or BSA vehicle (control). (B) Quantitative analysis reveals a significant increase in the number of apoptotic cells after BMP4 treatment of the PP6 cells but not of the PP1 cells. TGF-β2 treatment did not induce significant apoptosis in either the PP1 or the PP6 cell populations. *p < 0.01, n = 4. Bar: 10 μm.

BMP4 expression impeded in vivo survival of the PP6 cells but not the PP1 cells

To determine whether the in vitro effects of BMP4 on cell differentiation, proliferation, and apoptosis reflect in vivo situations, we investigated the impact of BMP4 expression on the in vivo survival of the transduced PP1 and PP6 cells. The identification of the implanted cells was made possible by labeling these cells with a retroviral vector expressing β-gal. The efficiency of transduction was similar between the two groups of cells: ∼70-80% of the cells were β-gal+ after transduction (data not shown). The PP1-B4 cells persisted for up to 4 weeks after implantation, which is a period of time comparable with that displayed by the control cells (PP1-Lac; Table 1). In contrast, the PP6-B4 cells were detectable for only up to 2 weeks after implantation, which is a significantly shorter time of survival than displayed by the control cells (PP6-Lac; Table 1). These results indicate that BMP4 expression did not adversely affect the in vivo survival of the PP1 cells but significantly shortened the persistence of the PP6 cells at the implanted sites compared with the survival of control, β-gal-expressing cells.

Table Table 1.. In Vivo Survival of Transduced Muscle-Derived Cells
original image

DISCUSSION

This study revealed that the osteogenic potential of rat muscle-derived cells is comparable with that of mouse muscle-derived cells.(9, 17) The early preplate population (PP1) from both species contains few myogenic cells or osteoprogenitor cells. In contrast, the later preplate cell population (PP6) contains abundant numbers of such cells. The co-existence of the myogenic and osteogenic cells in muscle-derived cell populations suggests that it is the myogenic cells that display osteogenic potential. Indeed, previous research has demonstrated that mouse myogenic cells (e.g., satellite cells) are osteogenic, as evidenced by their expression of ALP and other osteogenic markers after stimulation with BMP2.(22, 23) Muscle-derived stem cells (MDSCs), a small subpopulation of cells existing within the mouse muscle PP6 cells, also have exhibited osteogenic potential.(9) However, the exact nature of the rat osteogenic progenitor cells isolated from skeletal muscle (i.e., whether they are subpopulations of the satellite cells or are derived from the muscle-derived stem cells) remains to be determined.

Osteogenic cells versus non-osteogenic cells as cellular vehicles for BMP2 or BMP4 delivery in ex vivo gene therapy

Osteogenic cells have been identified as potentially ideal cellular vehicles for BMP-based gene therapy to improve bone healing because of the dual roles played by these cells—the production of bone growth factor (i.e., BMP) and the production of bone matrix through cell differentiation into osteoblasts. However, the possible advantages of using primary osteogenic cells rather than non-osteogenic cells have not been determined. Intriguingly, the findings from our study contradict the general supposition. The non-osteogenic cells used in our study healed critical-sized bone defects more effectively than did the osteogenic cells after transduction of both cell populations with a retroviral vector expressing BMP4. This outcome is not limited to osteogenic cells derived from skeletal muscle. Results from our subsequent studies, in which we compared non-osteogenic muscle PP1 cells with osteogenic bone marrow stromal cells isolated from the same rats, also indicated the superior ability of the non-osteogenic cells to heal segmental bone defects when used in ex vivo gene therapy (unpublished data). Notably, most of the osteogenic cells used in previous studies to induce bone formation or bone healing through BMP2 or BMP4 gene delivery were cell lines or stem cell clones.(9, 12-15) Such cell lines and cell clones differ markedly from the primary osteogenic cells that we isolated from muscle or bone marrow in several important respects, particularly in terms of their proliferation capacity after BMP2- or BMP4-induced osteogenic differentiation (as detailed below). Furthermore, in studies evaluating the use of primary bone marrow stromal cells to deliver osteogenic BMP and induce bone formation,(7, 11) it is difficult to determine whether the resultant induction of bone formation should be attributed to the osteogenic or the non-osteogenic subpopulation of cells, because only a fraction of the bone marrow stromal cells are osteogenic precursors.

Non-osteogenic PP1 cells expressing BMP4 may not contribute significantly to osteogenic lineages in vivo. We detected a very small number of osteocytes (<1%) that might have derived from the implanted PP1 cells at the bone-forming site. Thus, the primary function of retroBMP4-transduced PP1 cells seemed to be BMP4 expression, which attracted host cells to participate in the process of bone formation and bone healing. This finding underscores the importance of host osteoprogenitor cells in promoting bone healing in our animal model. The use of PP1 cells genetically engineered to express BMP4 alone may not be sufficient to elicit complete bone healing when the supply of host osteoprogenitor cells is limited, such as in some cases of fracture nonunion.

Converse relationship between osteogenic differentiation and cell proliferation

The association between osteogenic differentiation and exiting the cell cycle remains controversial. Our finding that BMP4 expression hampered the proliferation of PP6 cells but not that of PP1 cells suggests that there is an association between osteogenic differentiation and inhibition of cell proliferation, similar to that observed with osteoblasts.(24) The main differences between these two cell populations are their myogenic and osteogenic potential, as indicated by their distinctive abilities to form myotubes and their disparate responses to BMP4 or BMP2 stimulation. Indeed, our ALP-BrdU double-labeling technique clearly demonstrated that the PP6 cells that differentiated into osteogenic cells after BMP4 stimulation were markedly less proliferative than the undifferentiated PP6 cells. Furthermore, our research using rat primary bone marrow stromal cells has generated similar results: osteogenic differentiation induced by BMP4 expression inhibited the proliferation of the stromal cells (unpublished data).

The mechanism coupling osteogenic differentiation and the inhibition of muscle osteoprogenitor cell proliferation is not yet fully understood. This process may involve molecular pathways similar to the one that regulates cell cycle arrest during the differentiation of osteoblastic cell lines. That pathway involves the induction of the cyclin-dependent kinase inhibitor p21, whereas the expression of fibroblast growth factor receptor 3 (FGFR3) plays a role in the induction of osteogenic markers such as ALP.(25) The coordinated expression of both p21 and FGFR3 involves changes in Helix-Loop-Helix transcription factors, including an increase in E2A and a decrease in Twist and Ids.(25)

The association between osteogenic differentiation and the inhibition of cell proliferation that is observed in primary cell cultures may not fully apply to osteogenic cell lines, which have been used successfully to deliver the BMP2 gene to induce bone formation or bone healing.(14, 15) Transduced MDSCs expressing BMP2 or BMP4 also have been used successfully in our previous studies to elicit bone formation and bone healing.(9, 12, 13) The discrepancy between the behavior of primary cells and that of stable cell lines or stem cell clones may be caused primarily by the robust proliferation displayed by the latter cell types, a feature that enables the cell lines or stem cell clones to maintain relatively high (yet still decreased) levels of proliferation despite the impact of osteogenic differentiation. Second, the ability of these stem cells to self-renew(26) may enable replenishment of the dividing cell pool, thus compensating for any differentiation-related decrease in cell proliferation and thereby enabling the proliferative potential of the cells to be maintained. Our findings support the first possibility. ALP-BrdU double-labeling revealed significantly higher percentages of proliferating (BrdU+) osteogenic (ALP+) cells in the MDSC population than in the primary myogenic cell population after BMP4 stimulation of both cell types (unpublished data). Although stem cells that preserve both their high proliferation ability and their high potential for multilineage differentiation after transduction may represent the ultimate cell source for ex vivo gene therapy, it is still worthwhile to study other more readily available primary cells because of their immediate availability for use in clinical applications.

BMP4 treatment induces apoptosis of osteoprogenitor cells

The impaired proliferation of BMP4-transduced PP6 cells also could be attributable to the susceptibility of these cells to the pro-apoptotic effects of BMP4 on osteogenic cells. Apoptosis is a physiological process that occurs during skeletal development.(27, 28) Some osteotropic hormones that regulate osteoblastic cell proliferation and differentiation also have been found to control osteoblast apoptosis.(29, 30) Indeed, there is an association between the differentiation of osteoblasts into mature bone cells and the occurrence of programmed cell death.(31) Our previous research has demonstrated the occurrence of apoptosis during bone formation induced by MDSCs expressing BMP4.(12) BMP2 also has been found to promote apoptosis of primary human calvarial osteoblasts and immortalized human neonatal calvarial osteoblasts.(21) To our knowledge, our study is the first to show a pro-apoptotic effect of BMP4 on muscle-derived osteoprogenitor cells. It remains unclear whether osteogenic differentiation is associated with this increase in apoptosis.

Cell proliferation potential is a major determinant of the bone healing ability of muscle-derived cells expressing BMP4

This study showed that the in vitro proliferation potential of transduced rat muscle-derived cells expressing BMP4, but not their osteogenic potential, correlated strongly with the cells' ability to elicit bone healing. This finding suggests that cells displaying a better in vitro proliferation potential may survive longer in vivo, and thus produce higher levels of BMP4 in situ. In contrast, cells with poor proliferation potential or poor viability under the influence of BMP4 may fail to generate an amount of BMP4 sufficient to trigger the cascade of bone regeneration, which primarily involves recruitment of host mesenchymal stem cells to the bone formation site.(12, 13) Indeed, our findings indicated that the rat muscle-derived PP1 cells expressing BMP4 survived twice as long at the site of implantation as the PP6 cells expressing BMP4.

It is also possible that PP6 cells expressing BMP4 express a higher level of BMP4 antagonists (e.g., noggin) in a manner analogous to the expression profile of BMP4-stimulated osteoblasts.(32) These BMP4 antagonists neutralize the osteogenic effects of BMP4, and thereby inhibit bone formation. If this were the case, we would expect to observe a lower level of osteogenic differentiation of the transduced PP6 cells. On the contrary, the transduced PP6 cells expressing BMP4 consistently displayed a high level of osteogenic differentiation. Furthermore, the dose-dependent increase in osteogenic differentiation of the PP6 cells after BMP4 stimulation argues against the notion that inducible BMP4 antagonists significantly blocked BMP4 activity on these cells.

The PP6 cells also could respond to BMP4 stimulation by producing angiogenic inhibitors, which would impair bone healing. This possibility is unlikely because we previously have found that transduced MDSCs expressing BMP4 induced bone formation and bone healing.(12) Regardless of the actual mechanism, our results show that cells that respond adversely to BMP4 stimulation are not suitable for the delivery of BMP4 using ex vivo gene therapy.

In summary, this study highlights the importance of selecting the proper cellular vehicle for delivery of a particular therapeutic gene to achieve optimal gene therapeutic outcome. The ideal cellular vehicle should not respond adversely to the autocrine effects of the transgene. In this regard, rat muscle-derived early preplate cells (PP1) are a good choice of cellular vehicle for use in osteogenic BMP-based gene therapy to improve bone healing. In addition to their abundance and ease of isolation, the PP1 cells are suitable for osteogenic protein delivery because of their ability to resist the differentiating and pro-apoptotic effects sometimes conferred by BMP4 stimulation and thus to maintain good proliferation potential in vitro and prolonged survival in vivo. This study also demonstrated that cellular behavior in response to a therapeutic gene—including cell differentiation, cell proliferation, and cell survival—can be used to predict the suitability of using a particular type of cell as a vehicle in effective ex vivo gene therapy.

Acknowledgements

We thank Ryan Sauder for excellent editorial assistance. This work was supported in part by National Institutes of Health Grant 1 R01 DE13420-01 to JH, as well as Pittsburgh Tissue Engineering Initiative (PTEI) Grants and the Albert B. Ferguson, Jr., M.D. Orthopaedic Fund of The Pittsburgh Foundation to HP and JH. The Growth and Development Laboratory also receives financial support from the William F. and Jean W. Donaldson Chair at the Children's Hospital of Pittsburgh and the Henry J. Mankin Endowed Chair for Orthopaedic Research at the University of Pittsburgh.

Ancillary