Results

Distinct ability to induce an osteogenic marker, alkaline phosphatase (ALP), by 14 BMPs in the osteoblast progenitor C2C12 cells in vitro

In order to comprehensively analyze the distinct osteogenic activity of BMPs, we have recently constructed a panel of recombinant adenoviral vectors that express the 14 types of human BMPs, designated as AdBMPs.⁴⁵ As shown in Figure 1a, the level of transgene expression of the AdBMPs were in general comparable (ie, <2-fold difference among different BMPs), as demonstrated by RT-PCR analysis. These PCR products should be specifically derived from the adenoviral vector-mediated expression (rather than from the endogenous genes), as the 3′-end primer was derived from the SV40 poly-A cassette. Using these adenoviral vectors, we have recently demonstrated that BMPs displayed a distinct ability to induce osteoblast differentiation of mesenchymal progenitor cells in vitro.⁴⁵ In this study, we sought to determine the relative in vivo osteogenic activity of the 14 types of BMPs. We first tested the ability of individual AdBMPs to induce the earlier osteogenic marker alkaline phosphatase in the C2C12 line, which is myoblastic and can be trans-differentiated into osteoblast progenitors upon BMP stimulation. As shown in Figure 1b, ALP activity was remarkably induced by five of the 14 BMPs at four days after AdBMP infections. Among the five osteogenic BMPs, BMP-2, BMP-6, and BMP-9 induced the ALP activity to a much greater extent (approx. 169-, 215-, and 273-fold over the GFP control, respectively), while BMP-4 and BMP-7 increased ALP activity by 44- and 73-fold, respectively. These findings are consistent with our previous studies.⁴⁵ It should be pointed out that several BMPs (eg, BMP-10, and BMP-13) also induced a 2–3-fold increase of ALP activity over the basal level under the same assay conditions.

Orthotopic bone formation effectively induced by several but not all BMPs in athymic mice

We next sought to test the in vivo osteogenic effect of the 14 BMPs. In order to effectively assess the osteogenic ability of the BMPs, we used an orthotopic ossification animal model, in which C2C12 cells were first transduced with AdBMPs and subsequently implanted into the quadriceps muscles of athymic nude mice by intramuscular injection. Orthotopic ossification was assessed by X-ray radiography and histological evaluation at 3 and 5 weeks after injections. As illustrated in Figure 2, ossification was readily detected on X-ray radiographies from the animals injected with AdBMP-2, 6, 7, and 9-transduced C2C12 cells at 3 weeks (Figure 2a) and 5 weeks (Figure 2b). For BMP-6 and BMP-9, histologic examination at both time points (Figure 3a and b) revealed robust and highly mineralized woven bone with scattered osteoblast-rimming and occasional osteoclasts. Osteoid was also present. In addition, BMP-6 showed bone marrow elements with a range of hematopoietic cells. At the 3-week time point, BMP-2 was characterized by well-calcified foci without bone formation; however, at the 5-week time point, these foci had developed into mature bone. BMP-7, on the other hand, was shown to induce much weaker ossification. Interestingly, we failed to detect any signs of ossification in the animals injected with AdBMP-4-transduced C2C12 cells, which is surprising because we have demonstrated that BMP-4 is capable of inducing ALP activity in vitro (Figure 1b).⁴⁵ These results were reproducible in two additional rounds of animal studies using different batches of AdBMP-4 preparations, which were consistently shown to induce ALP activity in C2C12 cells in vitro (data not shown). Further, our RT-PCR analysis demonstrated that the expression level of BMP-4 was comparable with that of other BMPs, especially BMP-2, BMP-6, and BMP-9 under the same assay condition (Figure 1a). Currently, we are still searching for any satisfactory explanations for this discrepancy between BMP-4’s in vitro versus in vivo osteogenic activity. Nevertheless, all of the above findings were reproducible in three batches of independent experiments. The remaining samples had no evidence of ossification at 5 weeks. The injection site in these sections demonstrated exuberant granulation tissue and reparative changes.

Histology of BMP-induced bone formation

We next examined the histology of the recovered injection sites. Overall, the histology correlated well with the findings from X-ray radiography. At the 3-week time point, BMP-2, BMP-6, BMP-7, and BMP-9 demonstrated varying degrees of ossification. BMP-2 and BMP-7 were the least developed with small foci of woven bone (Figure 3a). Both BMP-6 and BMP-9, however, had multiple foci of immature woven trabecular bone. In addition, BMP-9 demonstrated focal cartilaginous differentiation. The bone in BMP-6 and BMP-9-treated animals formed a shell-like rim around a proliferating mass of spindle-shaped C2C12 cells. The 5-week samples demonstrated increased maturation with more mature osteoid matrix and trabecular bone-like structures with accentuation of the shell-like rim, especially in BMP-6 and BMP-9-treated animals. There was less extensive ossification in the BMP-2 sections. Bone marrow elements were present in the BMP-6 sections and chondrocytes and cartilaginous matrix were increased in the BMP-9 sections (Figure 3b). Interestingly, the injected C2C12 cells formed a desmoid-like cell mass in nonosteogenic BMP injections and the GFP control. Even in the animals injected with BMP-2, BMP-6, BMP-7, and BMP-9-transduced C2C12 cells, such cell mass was still visible, and multiple ossification centers were observed at the periphery of the cell mass. BMP-2-, BMP-6-, BMP-7-, and BMP-9-induced osteogenesis was further confirmed by Masson’s Trichrome staining (Figure 3c).

Antagonistic effect of BMP-3 on BMP-induced bone formation

We sought to investigate how the osteogenic BMPs were affected by BMP-3, a known negative regulator of bone formation, as BMP-3 knockout animals exhibited an increase in bone density.⁵² When C2C12 cells transduced by AdBMP-3 and one of the four osteogenic AdBMPs were coinjected intramuscularly for 3 weeks, BMP-2 and BMP-6-induced ossification was completely blocked by BMP-3, and most of the BMP-7-induced ossification was inhibited by BMP-3 (Figure 4a). However, BMP-3 coinjection did not exert any effect on BMP-9-induced calcification (Figure 4a), strongly suggesting that BMP-9 may exert its osteogenic activity via a distinct signaling mechanism. Similar results were obtained for the 5-week groups (data not shown). The histological findings were consistent with those from X-ray radiographic results (Figure 4b). The three samples without ossification demonstrated C2C12 cell proliferation with entrapped skeletal muscle while the BMP-3+BMP-9 sections had multiple foci of woven trabecular bone similar to BMP-9 injection alone.

Bone formation induced by direct intramuscular injection of AdBMPs

Recent studies suggest that skeletal muscles may harbor pluripotent mesenchymal stem cells, including osteoblast progenitors.^34, 44, 53 We next tested the osteoinductive activity of the 14 BMPs via direct intramuscular injection of AdBMPs. At the 3 and 5-week time points, we did not observe apparent ossification on X-ray radiography (data not shown). However, when the 5-week injection sites were examined histologically, various degrees of cartilaginous and/or osteoid matrix formation were observed in BMP-2, BMP-6, BMP-7, and BMP-9-injected animals (Figure 5). Samples derived from BMP-2 and, to a lesser extent, BMP-7 injection sites contained more cartilage-like chondrocyte-containing structure, while osteoid matrix and mature lamellar bone were present with evidence of bone marrow colonization and remodeling in BMP-6 and BMP-9-injected animals. Unlike in the experiments with C2C12 injections, direct intramuscular injections with AdBMPs (ie, the above-mentioned four osteogenic BMPs) induced more diffuse ossification. This may also explain why the calcification (ie, by BMP-9) was not readily detected by X-ray radiography. These findings also suggest that orthotopic osteogenesis induced by direct intramuscular injection with osteogenic AdBMPs may be less efficient than that induced by introduction of AdBMP-transduced osteoblast progenitor cells, implying that osteoblast progenitor cell-based gene therapy may be a more efficacious approach to bone regeneration, although it is possible that the reduced bone formation was resulted from the potentially less-efficient gene transfer associated with direct intramuscular injections than that with AdBMP-transduced C2C12 cells.

Successful bone regeneration mediated by biofactors could revolutionize the clinical management of musculoskeletal disorders, including fracture healing and spinal fusion.⁵⁴ Several biological factors, such as TGF, BMP, FGF, PDGF, IGF, and LMP-1, have been investigated for their potential use in bone regeneration and skeletal repair.^{55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69} BMPs have been shown to be the most promising, and clinical trials with recombinant BMP-2 and -7 are ongoing.^{54, 70, 71, 72, 73} These BMPs have shown varying degrees of success in the clinical setting and further study on their mechanisms of action and optimal formulations is required to optimize effectiveness of this strategy for promoting osteogenesis.

To the best of our knowledge, this reported study represents the first of its kind to evaluate the in vivo osteogenic activity of BMPs in a comprehensive fashion. The observation that BMP-2 exhibited osteogenic activity in our study is consistent with early data from human clinical trials.^25, 27 While BMP-7 exhibited apparent osteogenic activity, its ability to induce ossification was significantly less robust than that of BMP-2, BMP-6, and BMP-9. These findings mirror the moderate success of the rhBMP-7 (ie, OP-1) in a recent clinical trial.²⁶

It is intriguing, however, that BMP-6 and BMP-9 emerged as the most potent inducers of orthotopic ossification in vivo. Although considerable genetic and developmental studies have been conducted to elucidate the biological functions of BMP-6, its osteogenic activity has not been investigated to any significant degree in animal studies or clinical trials. BMP-6-deficient mice are largely unremarkable, with the exception of a defect in the sternum.⁷⁴ Its expression during embryogenesis is closely coupled with BMP-2, and the lack of noticeable defects in BMP-6-deficient mice may be due to functional compensation by BMP-2.⁷⁴ Nevertheless, our findings corroborate well with a recent study in which BMP-6 was shown to induce the most rapid tissue calcification when compared with BMP-2 or BMP-4 in an athymic nude rat model,⁷⁵ although for reasons to be determined AdBMP-4 reproducibly failed to induce orthotopic bone formation in this study.

BMP-9 is one of the least studied members of the BMP family. Originally identified from fetal mouse liver cDNA libraries, BMP-9 (a.k.a., GDF-2) is highly expressed in the developing mouse liver, and recombinant human BMP-9 (rhBMP-9) stimulates hepatocyte proliferation.⁷⁶ BMP-9 has also been shown to be a potent synergistic factor for hematopoietic progenitor cell generation and colony formation,⁷⁷ and may also play a role in the induction and maintenance of the neuronal cholinergic phenotype in the central nervous system.⁷⁸ In addition, it has recently been shown that BMP-9 exhibits an apparent osteoinductive effect in rat models.^21, 23, 79 However, the mechanisms underlying BMP-9-mediated osteogenic signaling remain to be defined. It is a very intriguing finding that BMP-9-mediated bone formation was not inhibited by BMP-3 in our studies. This result strongly suggests that BMP-9 may transduce a distinct osteogenic signaling pathway that is significantly different from that of BMP-2, BMP-6, and BMP-7. Through an expression profiling analysis, we have recently identified a group of downstream targets that may play an important role in the osteogenic BMP signaling pathway mediated by BMP-2, BMP-6, and BMP-9.⁸⁰ Interestingly, while BMP-2, BMP-6, and BMP-9 induced a very similar overall gene expression pattern, the clustering analysis revealed that BMP-2 and BMP-9 exhibited a more similarly related expression pattern.⁸⁰

In conclusion,

we have demonstrated the relative osteogenic ability of 14 BMPs and identified BMP-6 and BMP-9 (in addition to the currently used BMP-2 and BMP-7) as the most potent BMPs to induce orthotopic bone formation in vivo. Our results also suggest that the stem/progenitor cell-based ex vivo gene therapy may represent a more effective approach to bone regeneration. Future studies will focus on elucidating the major signaling differences among BMPs so that maximal synergy in bone formation can be achieved by combining BMPs that act through overlapping or converging signaling pathways. This line of investigation would help to elucidate the molecular mechanisms underlying bone formation and lead to the development of more efficacious approaches towards bone regeneration.

Recombinant adenoviral vectors expressing BMPs

The cDNA clones for human BMP-2, -3 (a.k.a., osteogenin), -4, -5, -6, -8 (a.k.a., OP-2), -9 (a.k.a., GDF-2), -10, -12 (a.k.a., GDF-7 or CDMP-3), and -13 (a.k.a., GDF6 or CDMP2) were kindly provided by the Genetics Institute (Cambridge, MA, USA). The coding sequences for BMP-7 (a.k.a., OP-1), -11 (a.k.a., GDF-11), -14 (a.k.a., GDF-5 or CDMP-1), and -15 (a.k.a., GDF-9) were PCR amplified from a human osteosarcoma cDNA library. The coding regions of the above BMPs were subcloned into pAdTrack-CMV, resulting in pAdTrack-BMPs; recombinant adenoviruses expressing BMPs (ie, AdBMPs) were subsequently generated as previously described. Details on vector constructions are available upon request.