Novel complexities regarding BMPs and fracture healing

Since their discovery as an osteoinductive activity in the 1960s and their subsequent cloning in the 1980s, the bone morphogenetic proteins (BMPs) are among the most prominent molecules in the bone field, albeit their physiologic functions are not restricted to the skeleton.1–3 Molecularly, with the exception of BMP-1, BMPs belong to the transforming growth factor β (TGF-β) superfamily, and their cellular response is mostly mediated through Smad-dependent signaling pathways.4 Functionally, BMPs promote the most important steps of endochondral ossification, such as mesenchymal cell condensation, chondrogenesis, and subsequent bone formation.5 This explains why there have been several attempts by many investigators to use recombinant humans BMPs (rhBMPs) to promote skeletal regeneration in human patients.6 Following a large number of preclinical experiments in various animal models, rhBMP-2 and rhBMP-7 [also termed osteogenic protein 1 (OP-1)) have been selected as therapeutic agents and are now approved for clinical use in states of severely compromised fracture healing, such as open fractures of long bones, nonunions, and spinal fusion.6–9 However, most of the clinical studies reported so far have only shown that the positive effect of rhBMP administration is comparable with that of autologous bone grafting, albeit not superior. Taken together, these findings raised the important question, Why are the impressive results seen in animal models so difficult to reproduce in human patients?6, 10–13

Given the reasons for these apparent inconsistencies are still unclear, it is commonly accepted that there is a need to optimize the methods of rhBMP delivery and to deepen our understanding regarding the effects of rhBMP administration at the cellular and molecular levels.14 In this regard, it is of hallmark clinical importance that Minear and colleagues have reported in the current issue of JBMR that the response of murine skeletal progenitor cells to recombinant BMP-2 (rBMP-2) in a cortical defect model is completely different in the endosteal and periostal compartments.15 Moreover, since both types of responses are mediated through an inhibition of β-catenin-dependent Wnt signaling, it appears not only that there is molecular crosstalk between two of the most important pathways regulating bone formation but also that canonical Wnt sigaling controls a different set of genes depending on the location of the skeletal progenitor cells. Taken together, these data provide important novel insights into the influence of BMP and Wnt signaling on skeletal repair. In addition, they offer a molecular explanation for conflicting results regarding the failure of BMPs to induce osteogenic differentiation of human bone marrow stromal cells.16, 17

To achieve their goals, the authors have employed a murine monocortical tibial defect model and inserted a sponge soaked with rBMP-2 into the marrow space, thus applying the same delivery method that is used commonly in human patients with skeletal injuries. Compared with PBS-treated control samples, the rBMP-2 administration resulted in a massive increase in cartilage callus volume 6 days after surgery. One week later, this tissue was fully replaced by a large bony callus that was located exclusively in the extracortical space. While these results basically confirmed expectations, it came as a surprise that the response of rBMP-2 administration was completely different in the bone marrow cavity. In fact, while the PBS-treated animals displayed robust osteogenesis within the bone marrow, there was no bone formation in this area when the animals received rBMP-2. This was confirmed by a decreased expression of osteoblast markers following rBMP-2 administration, whereas the number of osteoclasts increased compared with PBS-treated animals. Moreover, although rBMP-2 treatment resulted in an enhancement of cellular proliferation in the injured periostum, it had the opposite effect in the endosteal region, where PCNA⁺ cells were found only in PBS-treated animals.15

To understand the molecular basis of these site-specific rBMP-2 effects, the authors performed the same experiments with axin2-lacZ transgenic mice, a model designed to monitor β-catenin-dependent Wnt signaling.18 Here they observed that the number of Xgal⁺ cells in the bone marrow cavity was significantly decreased following rBMP-2 administration 3 and 4 days after surgery, thus suggesting that impaired canonical Wnt signaling is mediating the arrest of endosteal osteoblast differentiation caused by rBMP-2. Interestingly, the number of Xgal⁺ cells also was decreased by rBMP-2 treatment in the extracortical region, which was surprising, given the different cellular responses of these cells in terms of proliferation and chondrogenic differentiation. However, since Sox9, the key transcriptional regulator of chondrogenesis is negatively regulated by Wnt signaling,19 the effect of rBMP-2 on periostal cells was readily explained by an increased expression of Sox9 compared with PBS-treated samples. Taken together, these experiments provide evidence for the existence of molecular differences between endosteal and periostal skeletal progenitor cells that are relevant for our understanding of bone formation in general and fracture healing in particular.15 However, since the primary focus of the work by Minear and colleagues was to obtain deeper insights into the modulation of skeletal repair by BMPs, the major conlusions from this study should be seen in the context of the still incomplete understanding regarding the conflicting outcomes of clinical studies, where rhBMPs have been used for bone-specific regenerative strategies.

In this regard, the most relevant result is probably that rBMP-2 treatment fully abrogates bone formation in the marrow cavity, which is in line with in vitro studies demonstrating that BMP-2 does not enhance osteogenesis of human bone marrow precursor cells.15–17 Given the fact that the poor osteogenic potential of these cells is one of the major limitations for autologous bone-grafting strategies, there is in fact a direct clinical relevance of these findings because they suggest that other strategies, for instance, an activation of Wnt signaling, might be preferable to BMP administration.20, 21 In addition, the results of Minear and colleagues also show that in the extracortical space, as well as in the surrounding muscle, rBMP-2 treatment did promote endochondral ossification, as expected, thereby demonstrating that there are not only detrimental but also beneficial effects of BMP administration on skeletal repair. Most important, however, the influence of BMPs clearly depends on the location of the responding skeletal progenitor cells, and the same is the case for their response to impaired canonical Wnt signaling. Based on these arguments, although their relevance has so far been demonstrated only in mice, it will be important in future experiments to look more carefully at the effects of rhBMP administration, especially at the different responses in the endosteal and cortical regions.

Thus, although the findings reported by Minear and colleagues have led to important insights into the role of BMPs in skeletal repair, they also have raised several novel questions that have to be addressed in future studies. First, it remains to be established whether the BMP response in the bone marrow is indeed explained by an intrinsic property of skeletal progenitor cells or if it depends on the interaction with other cell types. In this regard, it is important to emphasize that the bone marrow environment is different from the periostal compartment and that additional factors, such as hypoxia or compromised vascularity, also may contribute to the site-specific differences reported by Minear and colleagues. Second, since the authors have performed their experiments only with BMP-2, it remains to be established whether the same effects are mediated by BMP-7. In fact, it will be of hallmark importance to address the question of whether BMP-7 also represses osteogenesis in the marrow cavity because it is obvious that different members of the same protein family do not necessarily influence their target cells in the exact same way. Third, since BMP administration is commonly achieved by placing a BMP-soaked sponge in the area of skeletal injury, the BMP concentration decreases gradually with increasing distance from the releasing source. Thus, if BMPs promote endochondral ossification specifically in the periosteal compartment but repress bone formation in the marrow cavity, the remaining problems of efficacy may be solved simply by placing the sponge at a different location, thereby increasing its distance to the endosteal compartment. And fourth, it would be interesting to evaluate how BMP administration would affect fracture healing in the metaphyseal region and whether there are also molecular differences between endosteal and periostal cells in this location. Since studies on metapyhseal fracture healing cannot be performed in rodents, however, this will require development of large animal models, which is important, because the majority of osteoporotic fractures in human patients are metaphyseal and not diaphyseal fractures.22

Disclosures

All the authors state that they have no conflicts of interest.