|Home | About | Journals | Submit | Contact Us | Français|
While the osteoinductive activity of recombinant bone morphogenetic protein 7 (BMP7) is well established, evaluation of the role of endogenous BMP7 in bone formation and fracture healing has been hampered by perinatal lethality in BMP7 knockout mice. Here we employ conditional deletion of BMP7 from the embryonic limb prior to the onset of skeletogenesis to create limb bones lacking BMP7. We find that the absence of locally produced BMP7 has no effect on postnatal limb growth, articular cartilage formation, maintenance of bone mass, or fracture healing. Our data suggest that other BMPs present in adult bone are sufficient to compensate for the absence of BMP7.
Bone morphogenetic proteins (BMPs), multifunctional signaling molecules belonging to the transforming growth factor (TGF)-β gene superfamily were identified by their ability to induce de novo bone formation at non-bone sites.1,2 This unique osteoinductive property is due to activation of the canonical BMP signaling pathway in mesenchymal cells and, to date, all BMPs that activate canonical BMP signaling have been shown to possess osteoinductive activity.3,4 The ability of BMPs to induce new osteogenesis has led to their development as therapeutic agents for bone regeneration. In the repair setting, exogenously supplied osteogenic BMPs are equipotent in their ability to enhance bone repair.5
The ability to remove individual BMP genes has also provided information on the specific roles that endogenous BMPs perform during embryonic development and in postnatal life. The wide range of phenotypes that result from loss of specific BMP genes highlights the complex and often unpredictable functions that individual BMPs have in normal skeletal physiology. For example, loss of function analysis of Gdf5 and Gdf6 has shown that these signals are required for proper joint morphogenesis during skeletal development, and the localization patterns of Gdf5 and Gdf6 are consistent with the loss of function phenotype observed.6,7 In contrast, BMP6 is highly expressed by growth plate chondrocytes but is not required for embryonic skeletal development, postnatal bone growth, or maintenance of bone mass, as loss of BMP6 has no impact on the skeleton.8,9 While BMPs 2 and 4 have been localized in the developing limb, and are synthesized by chondrocytes and osteoblasts, targeted deletion of BMP4 from the early limb is without serious consequences to endochondral ossification, bone mass regulation, or fracture healing, while loss of BMP2 results in spontaneous fracture and inability to initiate fracture repair.10–12 That the presence of other osteogenic BMPs does not appear to compensate for loss of BMP2, but is sufficient to make up for loss of BMP4, reinforces the idea that it is difficult to predict if functional redundancy will occur for BMPs that colocalize within individual skeletal elements.
Information on the function of BMP7 in skeletal biology has been hampered by the perinatal lethality that occurs from global loss of BMP7 due to kidney failure.13–15 At birth, BMP7 null mice often have hind limb polydactyly but little else in the way of skeletal anomalies that would identify a novel role for BMP7 in bone formation, leaving the role of BMP7 in adult bone homeostasis and fracture repair to be determined. Here, we examine the function of locally produced BMP7 in the postnatal skeleton, using a floxed allele of the BMP7 gene, that allows us to remove BMP7 from the limb skeleton in a spatially and temporally restricted manner. We find that BMP7 is not required for articular cartilage formation, postnatal appendicular skeletal growth, or the maintenance of bone mass. Limb bones lacking BMP7 are of equivalent size to control (C) littermates, display no apparent histological or radiographic anomalies up to 52 weeks of age, and undergo normal fracture repair. Since BMP activity is required for normal skeletal physiology, our data suggest that other BMPs present in adult bone are sufficient to compensate for the absence of BMP7.
Animal protocols were approved by the Harvard Medical School IACUC prior to initiation of these experiments.
Mice carrying a floxed allele of BMP7 were bred to Prx1::cre mice to obtain limb-specific conditional BMP7 knockout mice (denoted here as BMP7cko).16,17 Littermates not expressing cre recombinase were designated as control mice. Mouse genotypes were identified by PCR.
Staining of limbs by alizarin red S and alcian blue was carried out as described by McLeod (1980).18 Briefly, newborn mice were euthanized, and skin, viscera, adipose tissue were removed. Samples (n = 4 per genotype) were placed in 95% ethanol for 5 days, acetone for 2 days, stained for 3 days at 37°C with alizarin red S and alcian blue, and cleared with 1% KOH. For X-rays (n = 4 per genotype per time point), mice were anesthetized, placed on X-ray film (HR-G, Fujifilm, Greenwood, SC) and X-rayed using a Micro50 Microfocus Imaging at 50 kV for 120 s.
Unilateral fractures were produced in the right femurs of 10–12-week-old mice using a scaled-down version of the fracture apparatus previously described (n = 6 per genotype; Bonnarens and Einhorn, 1984; Cho et al., 2002).19,20 Mice were anesthetized and X-rayed to assess healing at 3, 10, and 20 days postfracture. After X-ray at each time point, a group of mice was sacrificed for histology.
WISH was performed using the protocol by Brent et al. (2003).21 Samples for histology (n = 2 per genotype per time point) were fixed in 4% paraformaldehyde, decalcified, and embedded in paraffin. Five-micron sections were collected and stained with toluidine blue (0.1%) using standard procedures.
Bones (n = 4 per time point) were harvested from control mice and total RNA was extracted using the RNeasy kit (Qiagen, Valencia, CA). Q-PCR was performed on pooled cDNA using the LightCycler 480 Real-Time PCR system (Roch Applied Sciences, Indianapolis, IN). BMP expression levels were normalized to levels of β-actin. Relative amounts of mRNA were calculated as previously described (Livak and Schmittgen, 2001; Niikura et al., 2006).22,23 PCR primers used for these experiments were previously validated (Tsuji et al., 2006).10
We used the well-characterized Prx1 limb enhancer to conditionally inactivate BMP7 in the limb bud prior to the onset of endochondral ossification. This transgene expresses cre very early in limb development, resulting in complete recombination of floxed alleles at early limb bud stages (E9.5–E10.5).17 We verified the ability of Prx1::cre to recombine the conditional BMP7 allele using WISH for BMP7. As seen in Figure 1, BMP7 is highly expressed throughout limb mesenchyme at E 10.5 and this expression is lost in the presence of the Prx1::cre transgene. Loss of BMP7 early in limb development is without significant impact on limb patterning or endochondral ossification, as BMP7cko mice have normal limb skeletons at birth (Fig. 1). Mice with limbs deficient in BMP7 activity in both mesoderm and ectoderm, through global inactivation of BMP7, have been previously described to have no defects in the formation of appendicular skeletal elements.13–15 Our data confirm these findings and allow us to conclude that BMP7 is not required for early limb skeletogenesis.
We followed the status of limb bones lacking BMP7 by performing X-rays of BMP7cko mice and control littermates at 4, 12, 24, and 52 weeks after birth, and confirmed our X-ray data with histological examination of bones at 16 weeks and 52 weeks. Using this approach, we found no evidence that lack of locally produced BMP7 affected the growth and maturation of limb bones (Fig. 2). At each time point analyzed, X-rays of BMP7cko mice were indistinguishable from those of controls. Closer examination of bone structure of femurs in both 16- and 52-week-old BMP7cko mice showed well-formed articular cartilage, normal growth plates, regularly spaced osteocytes, and bone marrow populated by hematopoietic cells and stromal components (Fig. 3). As the size of the BMP7cko limbs was equal to that of controls, we conclude that locally produced BMP7 is dispensable for postnatal bone growth and the viability of growth plate chondrocytes. Since we did not observe any abnormalities in the limb joints of BMP7cko mice out to 1 year of age by X-ray, it is likely the presence of other BMPs is sufficient for maintenance of articular chondrocytes in postnatal bone.
To analyze the bone repair capacity of adult mice in the absence of local production of BMP7, we created transverse fractures in the femora of 10–12-week-old BMP7cko mice and control littermates, and evaluated healing by X-ray and histology. The healing response we observed was identical between controls and those lacking BMP7. In both control and BMP7cko mice, the periosteum adjacent to the fracture site appeared to undergo significant proliferation at 3 days after fracture (Fig. 4, red arrows); at day 10 after fracture, there was exuberant callus present that was partially mineralized (Fig. 4, yellow stars); and at 20 days after fracture, bridging callus was undergoing replacement by bone (Fig. 4, green arrows). From these data, we conclude that fracture healing proceeds normally without bone-derived BMP7, and both the temporal sequence of events and the robustness of the response are unaffected by the absence of BMP7 made in bone.
Previous studies have suggested that BMP7 and BMP4 may cooperate to influence bone homeostasis, as both are expressed in similar patterns in ventroposterior mesoderm during skeletal development and have parallel expression patterns during fracture healing.20,24 Availability of floxed alleles for both BMP4 and BMP7 allowed us to create mice in which BMP4 and BMP7 are both removed from the early limb mesoderm (denoted here as BMP4;BMP7cko), and to examine the effect of loss of BMP4 and BMP7 on limb skeletogenesis. We found that the absence of BMP4 and BMP7 together results in the loss of specific skeletal elements but does not block endochondral ossification in the remaining skeletal structures (Fig. 5). In the forelimb (Fig. 5, top panels), lack of BMP4 together with BMP7 leads to about a 20% reduction in the size of the humerus and impairs development of the humeral notch, but does not interfere with joint morphogenesis or with the normal development of the remainder of the forelimb bones. When viewed by X-ray, 3 weeks after birth, the humerus of the BMP4;BMP7cko mice remain smaller and misshapen. In the hind limb (Fig. 5, bottom panels), loss of both BMP4 and BMP7 has greater impact on individual skeletal elements: The fibula is completely absent and the proximal portion of the femur fails to develop, so that, at 3 weeks of age, the distal portion of the femur is reasonably well formed and in the correct location, but the proximal end of the femur is missing, as evidenced by a large gap between the femur and its point of articulation in the hip socket. These data are consistent with the idea that combinations of osteogenic BMPs provide the bone inductive signal at specific skeletal sites.
To examine the contributions of BMPs 2, 4, and 7 to the endogenous BMP activity present in adult bone, we measured the relative levels of BMPs 2, 4, and 7 produced by individual skeletal elements from weeks 4–15 after birth. As shown in Figure 6, transcripts for BMPs 2, 4, and 7 can be found in each skeletal element examined. In contrast to BMP2, BMP7 transcripts appear to drop to almost undetectable levels in humerus, radius, femur, tibia, and fibula at the onset of skeletal maturity.
Unlike prior studies in which global loss of BMP7 resulted in perinatal lethality due to kidney agenesis, availability of floxed BMP7 mice allowed us to study the requirement for locally produced BMP7 in postnatal skeletal growth, accrual of bone mass, and the ability to heal fractures. As expected, removal of BMP7 from early limb mesoderm prior to the onset of skeletogenesis did not hinder skeletal development as the size, shape, and temporal progression of formation of individual limb skeletal elements was unaffected by the absence of BMP7. These data are in agreement with previous studies of Luo et al. (1995) and Dudley et al. (1995), who reported that global deletion of BMP7 resulted in minor abnormalities in the hind limbs that were most frequently manifested by the presence of a sixth preaxial digit.13,14 While perinatal lethality due to failure of kidney development did not allow for examination of the effects of lack of BMP7 in the postnatal skeleton, mice in which BMP7 was inadvertently inactivated by insertion of a Bcl2 transgene survive to P12.15 These mice have delayed mineralization of several limb skeletal elements at birth, but this delay was quickly ameliorated so that when they died at P12, mineralization appeared normal. Bcl2tg mice do show significant defects in formation of vertebral bodies that do not normalize with time. As BMP7cko mice express BMP7 in the axial skeleton, our studies do not clarify the requirement for BMP7 in the morphogenesis of these structures.
Our finding that mice lacking both limb skeletal BMP4 and limb skeletal BMP7 exhibit patterning anomalies in several bones, and have seriously compromised skeletal function after birth, substantiates the idea that BMP4 and BMP7 may have overlapping functions during skeletal development, and supports the general concept that there is functional redundancy at skeletal sites where BMPs colocalize.24 Overlap in expression of individual BMPs may be a way to ensure proper BMP signal strength in specific skeletal elements.
While our finding that BMP7 produced within bone is not necessary for postnatal skeletal growth, bone mass accrual and fracture repair is unexpected it is not unprecedented when compared to the requirements for other osteogenic BMPs in these processes. Mice lacking BMP4 or BMP6 exhibit normal skeletal growth, maintain bone mass, and successfully undergo fracture repair.9,11 Detailed information on BMP expression during the fracture healing process indicates that endogenous BMP7 levels are virtually unchanged until after callus formation is complete.25 The rise in BMP7 expression that takes place during healing is paralleled by a rise in BMP4. As both molecules are produced by osteoblasts that replace callus tissue, they may be able to substitute for each other during healing, as has been shown in other tissues where BMPs 4 and 7 are coexpressed.26 The increases in BMP7 and BMP4 found during fracture repair have also been shown to be dependent upon the initiation of healing in response to BMP2, as in the absence of BMP2, the levels of BMP7 and BMP4 do not increase.10 BMP7 levels within bone are also unchanged during distraction osteogenesis, a further indicator that locally produced BMP7 may not be involved in the initiation of the repair response.27 Our results documenting the relatively low level of BMP7 production in adult bone are consistent with those of Edgar et al. (2007), who reported that BMP7 was not expressed in the marrow stromal compartment of adult bone, and with those of Sengle et al. (2008), who reported that the pro region of BMP7 targets the protein to the ECM where it may be stored in inactive form and largely unavailable to participate in repair.28,29 Additional studies by Kwong et al. (2008) examined the expression of BMPs in nonunion tissue, and found that patients with nonunion had decreased levels of BMP2 but no change in levels of BMP7.30 Based on our studies, we conclude that production of BMP7 by bone cells is not required for skeletal homeostasis. The newly identified roles for BMP7 in brown fat regulation, adipogenesis, and energy expenditure, and the already established role for BMP7 in kidney homeostasis, suggest that endogenous BMP7 is not a primary mediator of bone formation.31–34
This study was funded by a grant from the Musculoskeletal Transplant Foundation to V. Rosen.