Impaired Angiogenesis, Early Callus Formation, and Late Stage Remodeling in Fracture Healing of Osteopontin-Deficient Mice


  • Craig L Duvall,

    1. Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, Atlanta, Georgia, USA
    Search for more papers by this author
  • W Robert Taylor,

    1. Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, Atlanta, Georgia, USA
    2. Division of Cardiology, Department of Medicine, Emory University, Atlanta, Georgia, USA
    3. Veterans Affairs Medical Center, Atlanta, Georgia, USA
    Search for more papers by this author
  • Daiana Weiss,

    1. Division of Cardiology, Department of Medicine, Emory University, Atlanta, Georgia, USA
    Search for more papers by this author
  • Abigail M Wojtowicz,

    1. Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, Atlanta, Georgia, USA
    Search for more papers by this author
  • Robert E Guldberg PhD

    Corresponding author
    1. Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, Atlanta, Georgia, USA
    2. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA
    • 315 Ferst Drive, Institute for Bioengineering and Bioscience, Atlanta, GA 30332, USA
    Search for more papers by this author

  • The authors state that they have no conflicts of interest.

  • This work was supported by NIH Grants R01AR051336, RO1HL70531, PO1HL58000, the Georgia Tech/Emory Center for the Engineering of Living Tissues (GTEC) NSF Grant EEC-9731643, and the NSF Graduate Research Fellowship Program (CLD).


OPN is an ECM protein with diverse localization and functionality. The role of OPN during fracture healing was examined using wildtype and OPN−/− mice. Results showed that OPN plays an important role in regulation of angiogenesis, callus formation, and mechanical strength in early stages of healing and facilitates late stage bone remodeling and ECM organization.

Introduction: Osteopontin (OPN) is an extracellular matrix (ECM) protein with diverse localization and functionality that has been reported to play a regulatory role in both angiogenesis and osteoclastic bone remodeling, two vital processes for normal bone healing.

Materials and Methods: Bone repair in wildtype and OPN−/− mice was studied using a femoral fracture model. μCT was used for quantitative angiographic measurements at 7 and 14 days and to assess callus size and mineralization at 7, 14, 28, and 56 days. Biomechanical testing was performed on intact bones and on fracture specimens at 14, 28, and 56 days. Histology and quantitative RT-PCR were used to evaluate cellular functions related to ECM formation and bone remodeling.

Results: OPN deficiency was validated in the OPN−/− mice, which generally displayed normal levels of related ECM proteins. Intact OPN−/− bones displayed increased elastic modulus but decreased strength and ductility. Fracture neovascularization was reduced at 7 but not 14 days in OPN−/− mice. OPN−/− mice exhibited smaller fracture calluses at 7 and 14 days, as well as lower maximum torque and work to failure. At 28 days, OPN−/− mice had normal callus size but a persistent reduction in maximum torque and work to failure. Osteoclast differentiation occurred normally, but mature osteoclasts displayed reduced functionality, decreasing late stage remodeling in OPN−/− mice. Thus, at 56 days, OPN−/− fractures possessed increased callus volume, increased mechanical stiffness, and altered collagen fiber organization.

Conclusions: This study showed multiple, stage-dependent roles of OPN during fracture healing. We conclude that OPN deficiency alters the functionality of multiple cell types, resulting in delayed early vascularization, altered matrix organization and late remodeling, and reduced biomechanical properties. These findings contribute to an improved understanding of the role of OPN in vivo and provide new insight into mechanistic control of vascularization and bone regeneration during fracture repair.


Bone fracture healing involves a well-characterized cascade of events that includes hematoma formation, inflammation, soft cartilaginous callus formation, neovascularization, osteoblastic callus mineralization, and osteoclastic remodeling of the hard callus back to mature lamellar bone.(1) This complex sequence of biological processes is orchestrated by a variety of growth factors and matricellular proteins that regulate differentiation, chemotaxis, and haptotaxis of the cell types responsible for mediating these events. For example, the family of bone morphogenic proteins (BMPs) has been thoroughly characterized for induction of osteogenesis at fracture sites and is used clinically to promote bone regeneration.(2) In addition, Street et al.(3) have shown that VEGF, a known promoter of angiogenesis, stimulates neovascularization and promotes fracture healing, whereas treatment with angiogenesis inhibitors that specifically target vascular cells blocks fracture healing and produces atrophic nonunions.(4) Furthermore, as reviewed by Alford and Hankenson,(5) extracellular matrix (ECM) proteins, while not always vital for normal embryonic bone development, play an important role in mediating cellular function and serve as important modulators of bone regeneration. Further understanding of the intricate interplay between this diverse set of cell types, endogenous factors, and matrix proteins is a necessity for the development of improved therapeutic strategies for promoting bone repair.

Osteopontin (OPN) is one of the major noncollagenous ECM proteins in bone. It has been studied based on its role in a diverse range of biological processes including inflammation, immunity, angiogenesis, wound repair, tumor formation and metastasis, cellular survival and migration, and osteoclastic bone remodeling. Although now known to also be present in a number of nonmineralized tissues, OPN was originally cloned from bone where it has been shown to have a functional role in controlling mineralization and remodeling. OPN inhibits mineral crystal formation and growth,(6,7) and OPN-deficient mice have been found to possess increased mineral content and crystallinity.(8) In addition, OPN is known to be a major ligand for CD44 on bone cells, which is important in mediating osteoclast recruitment and function.(9–11) The in vivo importance of OPN in bone remodeling has been further shown by the finding that OPN-deficient mice undergo significantly less bone resorption than wildtype mice in response to reduction in mechanical loading, ovariectomy, stimulation by PTH, or administration of a high-phosphate diet.(12–15)

Whereas many researchers have studied the role of OPN in bone, it is also important for normal arterial physiology(16) and is produced by the primary cell types involved in blood vessel growth and remodeling: monocytes/macrophages, endothelial cells, and smooth muscle cells.(17) In addition to its interactions with CD44, OPN also contains an arginine-glycine-aspartate (RGD) motif, which allows it to interact with the integrin family of cell-adhesion molecules including α4β1(18) and αvβ3.(19) Previous in vitro studies have shown that cellular interactions with OPN mediated through these integrins regulate a wide variety of cellular functions relevant to angiogenesis including vascular cell adhesion and spreading,(20) macrophage adhesion and migration,(21) endothelial cell migration,(19) endothelial cell survival,(22) chemotactic response of smooth muscle cells,(20) and smooth muscle cell migration.(23) Other in vivo studies have also implicated OPN in neovascularization and remodeling of the vessel wall. For example, OPN mRNA has been shown to be locally upregulated at the site of ischemia-induced retinal neovascularization in mice,(24) and OPN deficiency has been found to decrease angiogenesis around ectopic bone implants and to diminish arterial remodeling.(16,25) Based on these observations, it is likely that the vascular effects of OPN are also relevant to vessel formation and maturation during fracture healing.

The role of OPN in a diverse set of processes including macrophage function, angiogenesis, ECM mineralization, and osteoclastic bone remodeling suggests that OPN may be important during multiple stages of fracture healing. Previously, in situ hybridization studies have noted OPN expression in the fracture callus by osteoprogenitors in woven bone, hypertrophic chondrocytes in the cartilage to bone transitional region, and in the osteocytes, osteoblasts, and osteoclasts of the hard callus.(1,26,27) However, the specific function of OPN during bone healing has not been previously determined. In this study, we hypothesized that OPN deficiency alters neovascularization, mineralization, remodeling, and the restoration of mechanical properties during fracture healing. To test this hypothesis, we studied fracture repair in wildtype and OPN−/− mice.



Male wildtype C57BL/6 mice were purchased from the Jackson Laboratory. OPN-deficient mice were originally received from Dr Lucy Liaw of the Maine Medical Center,(28) and they were subsequently backcrossed 10 generations onto the C57BL/6 background. All animals were fed a standard chow and had free access to water. All protocols were approved by the Institutional Animal Care and Use Committee and done in accordance with the federal guidelines on the principles for the care and use of animals in research.

Intact bone biomechanics

The baseline biomechanical consequence of OPN deficiency in intact bone was assessed by testing femora from 10-week-old wildtype (n = 5) and OPN−/− (n = 7) mice. Soft tissues were removed, and the bones were wrapped in PBS-soaked gauze and frozen at –20°C. The left femora from these mice were tested in three-point bending, and the right femora were testing in torsion. All bones were thawed in PBS at room temperature for 3 h before mechanical testing. During thawing, the specimens were imaged using the VivaCT 40 μCT imaging system (Scanco Medical) at a voxel size of 21 μm. The μCT images were used to determine moment of inertia within the mid-diaphyseal region, which was used to calculate bone material properties.

For three-point bending tests, specimens were loaded onto a three-point bending setup with a 6.2-mm distance between lower supports. An 858 Mini Bionix II testing system (MTS) was used to load the femora to failure at a rate of 0.05 mm/s with the anterior side in tension and posterior side in compression. Maximum load, elastic modulus, work to failure, and postyield displacement were determined from the recorded force-displacement data.

For torsional testing, we designed fixtures and a custom potting apparatus that allowed us to reproducibly align and pot the femora in Wood's metal with a gauge length of 6.5 mm. After the femora were potted, they were loaded in torsion at a rate of 3°/s until failure using an ELF 3200 testing system (Bose). Using the recorded test data, maximum torque, shear modulus, work to failure, and postyield rotation were determined.

Bone fracture healing model

A well-established, unilateral femoral fracture model(29) was used to study bone repair in 10-week-old male wildtype and OPN−/− mice. Animals were anesthetized by intraperitoneal injection of xylazine (10 mg/kg) and ketamine (80 mg/kg). All hair was removed from the surgical site, and the area was cleansed with sterile water followed by betadine. A 25-gauge needle was inserted in a retrograde manner into the intramedullary canal of the right femur. Subsequently, a mid-diaphyseal fracture was created in this leg, and the contralateral leg was left intact. The mice were allowed to recover on a heated pad, and, after awakening, they were returned to their cages and allowed to ambulate freely. On death, any animals that displayed intramedullary pin displacement, fractures that were not transverse, or fractures not in the mid-diaphyseal region were removed from the study.

μCT analysis of fracture site neovascularization

Recently developed quantitative μCT-based methods(30) were used for evaluation of fracture callus vascularity. Mice were killed using carbon dioxide inhalation at 7 (wildtype: n = 10, OPN−/−: n = 10) or 14 days (wildtype: n = 9, OPN−/−: n = 6) after surgery. The vasculature of the mice was sequentially perfused at physiologic pressure using heparinized (100 units/ml) normal saline, 10% neutral buffered formalin, and again with heparinized saline. The vascular system was injected with a radiopaque, lead chromate-based contrast agent (Flow Tech), which was allowed to polymerize for 24 h at 4°C. The fractured femora were isolated from the surrounding musculature under a dissecting microscope, and the intramedullary pins were carefully removed. The femora were stored at 4°C for 48 h in 10% neutral buffered formalin, soaked 48 h in a formic acid–based solution for decalcification of the mineralized bone, washed thoroughly using water, and placed in 10% neutral buffered formalin until imaging. The decalcification procedure is a vital step that allows radiodensity-based segmentation of the contrast-filled vessels from the surrounding mineralized tissues and therefore facilitates calculation of vascular morphology parameters. The specimens were imaged at a 10.5-μm isotropic voxel size, and 2D tomograms were reconstructed. The volume of interest (VOI) was defined by drawing contour lines along the borders of the callus in the 2D slices. Then, within this VOI, the 2D images were globally thresholded based on X-ray attenuation and used to render binarized 3D images of the radiopaque, contrast-filled vascular network segmented from the surrounding tissues. These 3D images were evaluated for vessel volume, volume fraction, and average diameter.

μCT analysis of fracture callus formation and mineralization

Mice were killed postoperatively for μCT imaging at 7 (wildtype: n = 7, OPN−/−: n = 6), 14 (wildtype: n = 7, OPN−/−: n = 6), 28 (wildtype: n = 5, OPN−/−: n = 8), and 56 (wildtype: n = 7, OPN−/−: n = 6) days. The femora were removed and dissected free from surrounding musculature under a dissecting microscope. The intramedullary pins were removed, and the bones were wrapped in PBS-soaked gauze and frozen at –20°C. On removal from the freezer, the bones were placed in PBS to thaw at room temperature.

While thawing in PBS, the specimens were imaged using μCT at a voxel size of 21 μm. The newly formed fracture callus tissue was spatially segmented from the native cortical bone in the 2D tomograms. Before generation of 3D images of the mineralized callus, we scanned a set of hydroxyapatite (HA) phantoms (0–800 mg HA) to define our mineralization threshold. Based on the precedent set in a similar fracture healing study,(3) we defined the fracture callus mineralization threshold (295 mg HA/cm3) as 50% of the mineral density that we used to segment intact cortical bone. Using this threshold, 3D images of the mineralized callus were rendered, and total volume, percent mineralization, and average mineral density were measured on the digitally extracted callus tissue.

Fracture specimen biomechanical testing

Specimens were thawed in PBS at room temperature for a total of 3 h before mechanical testing, during which time μCT imaging was completed as discussed above. For three-point bending tests, a pilot set of 28-day postfracture specimens were loaded onto a three-point bending setup as described for intact bone biomechanics (wildtype: n = 8, OPN−/−: n = 8). Ultimate load, work to maximum load, and displacement at maximum load were determined from the recorded force-displacement data.

For torsional mechanical testing, the fracture calluses were loaded to failure using the same testing setup described for intact bone biomechanics. Stiffness, yield point, maximum torque, work to failure, rotation at failure, and postyield rotation were determined for the fracture specimens. Specimens collected at the earliest time-point (7 days) were too fragile to test, so this analysis was only completed on specimens 14, 28, and 56 days after fracture (14 days: wildtype, n = 7; OPN−/−, n = 6; 28 days: wildtype, n = 5; OPN−/−: n = 8; 56 days: wildtype, n = 7; OPN−/−, n = 6).

Real-time RT-PCR

RNA was isolated from the mid-diaphysis of intact bones and from fracture callus tissues at 3, 7, and 14 days after surgery. The tissue was snap frozen and homogenized in QIAzol lysis reagent (Qiagen). RNA was purified using a commercial kit (Qiagen) and reverse transcribed into cDNA using the SuperScript III First Strand Synthesis System (Invitrogen), which was subsequently purified using a commercially available kit (Qiagen). Primers were designed using the ABI Primer Express software (Table 1), and SYBR Green intercalating dye (Applied Biosystems) was used to perform real time PCR with the ABI Prism 7700 Sequence Detection System (Applied Biosystems). Standards for each gene were amplified from cDNA and purified. Standard concentrations were determined using spectrophotometric measurement at 260 nm, and standards were serially diluted to an appropriate range of concentrations. Transcript concentration in template cDNA solutions was quantified from the linear standard curve and expressed as 10−18 moles of transcripts/μg of total RNA.

Table Table 1.. Oligonucleotides for Quantitative RT-PCR
original image

Histological analysis

Mice were perfused with heparinized 0.9% normal saline, followed by 10% neutral buffered formalin. Tissues were decalcified using a formic acid based agent (Cal-Ex II; Fisher Scientific), embedded in paraffin, and cut into 5-μmthick longitudinal sections. Images shown are representative of sections taken from n ≤ 4 different animals. Immuno- staining was done with antibodies to OPN (Immuno-Biological Laboratories), fibronectin (Chemicon), and bone sialoprotein (University of Iowa Developmental Studies Hybridoma Bank). Primary antibodies were detected using an avidin-biotin-alkaline phosphatase method from a commercially available kit (Vector Laboratories), and the sections were counterstained with hematoxylin. Using commercially available kits (Sigma-Aldrich), safranin O staining was completed for detection of cartilage, and multinucleated osteoclasts were identified based on staining for TRACP.

Picrosirius red staining was done using standard methods.(31) Briefly, specimens were deparaffinized, rehydrated, stained for 1 h in 0.1% sirius red F3B in saturated picric acid, washed in 0.5% acetic acid, dehydrated, cleared in xylene, and mounted for imaging using planar polarized light microscopy. Samples were aligned −45° from the transmission axis of the polarizing filters to maximize brightness of the images from each sample and provide a consistent assessment. Using this technique, one can qualitatively assess collagen fiber organization based on polarization color.(32)

Measuring total collagen content

For analysis of total collagen content, the hydroxyproline assay(33) was performed on fracture calluses. Briefly, the tissue was lyophilized and dry weight measurements were recorded. The samples were decalcified in 10% formic acid, diced into small pieces, digested in proteinase K, hydrolyzed in 6N HCl at 110°C, and assayed for hydroxyproline content using a standard colorimetric assay.(33)

Statistical analysis

All data are presented as mean ± SE. Statistical analyses were performed using Minitab software. ANOVA was used to model the effect of genotype on all response variables. P < 0.05 was interpreted as significant in all analyses.


Intact bone biomechanics

To determine the baseline biomechanical phenotype in intact bone, we tested two matched sets of intact wildtype and OPN−/− femora in torsion and in three-point bending. We found consistent results for the two testing methods, confirming that OPN−/− bones possessed higher elastic/shear modulus, lower maximum load/torque, and reduced work to failure and postyield deflection/rotation (Table 2).

Table Table 2.. Comparison of Wildtype (n = 5 for Both Testing Setups) and OPN-Deficient (n = 7 for Both Testing Setups) Intact Bone Properties
original image

Expression and immunolocalization of ECM proteins in OPN−/− mice

Quantitative RT-PCR measurement of OPN expression in wildtype and OPN−/− intact and fractured femora at 3, 7, and 14 days after surgery showed increased OPN expression in the setting of fracture healing in the wildtype animals. In addition, it validated the complete absence of OPN mRNA in the knockout animals.

Immunohistochemical staining of 28-day postsurgery tissue sections taken from the fracture callus corroborated this finding, showing the lack of OPN at the protein level in the knockout animals and the localization of OPN in the fractured wildtype bones (Fig. 1).

Figure Figure 1.

Validation of OPN deficiency and evidence that compensatory changes in gene expression or protein localization of bone sialoprotein and fibronectin during fracture healing in OPN−/− mice are largely absent. (A) Quantitative RT-PCR measurement of OPN, BSP, and fibronectin expression in OPN-deficient and wildtype (WT) intact (No Fx) and fracture healing (Fx) bone samples 3, 7, and 14 days after surgery expressed as 10−18 moles of transcripts/μg of RNA. (B) Immunohistochemistry photomicrographs of the fracture sites from wildtype and OPN deficient mice identifying OPN at 28 days after fracture and BSP and fibronectin at 14 days after fracture. These data show increased OPN expression in the setting of fracture healing and the absence of OPN mRNA in the knockout animals, in addition to the lack of OPN at the protein level in the OPN−/− mice and the localization of OPN in the wildtypes. BSP and fibronectin were upregulated in the setting of fracture healing, and, other than a modest decrease in BSP expression in OPN−/− fractures at 14 days, no alterations were seen in gene expression or protein quantity or localization of these ECM proteins in the fracture callus caused by OPN deficiency. ap < 0.05.

It has been hypothesized that the relatively mild bone phenotype seen in OPN−/− mice without external stimulation could be caused by compensatory activity of other, similar ECM proteins. Here we measured the gene expression and protein localization of fibronectin and bone sialoprotein (BSP), two bone ECM proteins that possess similar recognition sequences and properties to OPN. These genes displayed intense upregulation in fractures relative to intact bones for both genotypes, but, other than a significant but modest deficiency in BSP expression at 14 days, we found that OPN-deficient mice displayed no aberrant expression of these genes relative to the wildtype animals. Using immunohistochemistry, we further observed apparently normal localization of these related proteins within the calluses of OPN−/− mice at 14 days after fracture (Fig. 1).

μCT analysis of fracture site neovascularization

We used μCT imaging of contrast perfused, decalcified specimens for visualization and quantification of vascular growth within the fracture callus (Fig. 2A). Significantly reduced vessel volume and vessel volume fraction, along with a trend toward decreased average vessel diameter (p = 0.071), were detected in the fracture calluses of OPN−/− mice compared with wildtypes at 7 days after surgery (Fig. 2B). At 14 days after surgery, there was a significant reduction in average vessel diameter in OPN−/− mice, but significant differences for other vascular parameters were no longer present (Fig. 2C).

Figure Figure 2.

Reduced early stage neovascularization in OPN-deficient mice was recovered at later time-points. (A) 7- and 14-day postsurgery μCT images visibly showed reduced early neovascularization in OPN-deficient mice that seemed to be recovered by 14 days. (B) At 7 days, this angiogenic deficiency was found to be significant in quantitative 3D image analyses of vascular volume and volume fraction (wildtype: n = 10, OPN−/−: n = 10). (C) At 14 days after surgery, μCT images and quantitative analysis revealed reduced average vessel diameter but an overall recovery of the early vascular defect seen in the OPN-deficient mice (wildtype: n = 9, OPN−/−: n = 6). Scale = 1 mm. ap < 0.05, bp < 0.005, and cp < 0.0005.

μCT analysis of fracture callus formation and mineralization

μCT imaging was used to measure volume, percent mineralization, and average mineral density within the newly formed fracture callus that was outside the periphery of the pre-existing cortical bone structure (Fig. 3A). The OPN−/− mice displayed a significant reduction in callus volume relative to the wildtypes at days 7 (−27%) and 14 (−30%). No differences were seen in callus volume between genotypes at 28 days, and callus volume was significantly increased (46%) in the OPN-deficient mice compared with wildtypes at 56 days after fracture (Fig. 3B). The OPN−/− mice displayed accelerated early mineral formation and had significantly higher percent callus mineralization at 14 days. However, no differences were seen in percent callus mineralization at other time-points (Fig. 3B). Last, the average density of the mineralized portion of the fracture callus was significantly lower in the OPN-deficient mice at 28 days, and this trend persisted but was no longer significant at 56 days (Fig. 3B).

Figure Figure 3.

OPN-deficient mice displayed reduced early stage callus size, increased early mineralization, and delayed late stage remodeling. (A) μCT imaging captured the time-course of mineralized callus formation and remodeling in the wildtype and OPN−/− mice. (B) Quantitative image analysis revealed that OPN−/− mice possessed reduced callus size 7 and 14 days after surgery but increased callus volume 56 days after surgery. OPN−/− mice also displayed increased percent callus mineralization at 14 days. Analysis of average mineral density within the newly formed callus found that this parameter increased significantly over time after fracture, and it was found that OPN deficiency resulted in a small but significant decrease in average mineral density at 28 days after fracture. 7-day (wildtype: n = 7, OPN−/−: n = 6), 14-day (wildtype: n = 7, OPN−/−: n = 6), 28-day (wildtype: n = 5, OPN−/−: n = 8), and 56-day (wildtype: n = 7, OPN−/−: n = 6). Scale = 1 mm. ap < 0.05, bp < 0.005.

Fracture specimen biomechanical testing

In initial three-point bending tests done at 28 days after surgery, fractured OPN−/− femora displayed reduced deflection and work to ultimate load relative to wildtypes (data not shown). However, in this pilot experiment, we found that the three-point bending setup resulted in artifacts from fracture callus indentation by the middle support. This affected the initial portion of the tests and made it difficult to accurately determine the stiffness and yield point, so, as a result, we switched to the torsional testing modality for biomechanical assessment. Using torsional testing, we found that OPN−/− fractures possessed significantly reduced maximum torque at 14 and 28 days, but this difference did not persist at 56 days after fracture (Fig. 4A). OPN−/− calluses required significantly less work to failure at 14 and 28 days after surgery (consistent with our pilot study using three-point bending) but were similar to the wildtypes at 56 days (Fig. 4B). Furthermore, OPN-deficient mice tended to have reduced postyield deformation at all time-points, but ANOVA analysis did not find genotype to be a significant predictor (p = 0.054) of this parameter (Fig. 4C). Finally, stiffness was similar between genotypes at 14 and 28 days, but it was significantly higher in OPN−/− mice at 56 days after fracture, consistent with the observed increase in callus size at that time-point (Fig. 4D).

Figure Figure 4.

Altered fracture callus mechanics in OPN−/− mice. OPN−/− mice had significantly reduced (A) maximum torque and (B) work to failure at 14 and 28 days after fracture. The OPN-deficient mice also showed a strong trend toward (C) decreased overall postyield deformation (p = 0.054) as determined by ANOVA analysis and displayed (D) increased stiffness at 56 days. 14-day (wildtype: n = 7, OPN−/−: n = 6), 28-day (wildtype: n = 5, OPN−/−: n = 8), 56-day (wildtype: n = 7, OPN−/−: n = 6). ap < 0.05.

Gene expression and histology

Cartilage callus formation

Because of the reduction in fracture callus size at early time-points, we hypothesized that the OPN-deficient mice had defective development and hypertrophy of chondrocytes for formation of the early cartilaginous callus. To test this hypothesis, we measured relative gene expression of type II and type X collagen, early and late markers of chondrocyte differentiation, respectively. Quantitative RT-PCR measurement revealed that these genes were upregulated by 7 days after fracture, and there seemed to be a slight lag in chondrogenesis in the OPN-deficient mice, which tended to have stronger persistence of cartilage markers at 14 days. However, there was no significant difference in expression between genotypes (Fig. 5A), indicating that OPN deficiency does not significantly hinder chondrocyte formation and maturation at the fracture site. Safranin O staining revealed similar staining patterns for the presence of cartilage in fracture callus sections from the two genotypes at 14 days after fracture, confirming this finding (Fig. 6A).

Figure Figure 5.

Quantitative RT-PCR measurement of markers of chondrocyte formation, bone remodeling, type I collagen production, and matrix cross-linking. Quantitative RT-PCR measurement of (A) collagen II and collagen X, markers of chondrocyte formation and maturation, did not reveal differences in expression of these genes in the OPN-deficient mice, indicating chondrogenesis was not altered in the absence of OPN. Gene expression analyses also showed OPN-deficient mice had no alteration in the expression of (B) OPG and RANKL, regulators of osteoclast maturation, or (C) cathepsin K and TRACP, markers of osteoclast activity, suggesting that osteoclast differentiation is not altered in the absence of OPN. Expression of (D) type I collagen and lysyl oxidase was also normal in OPN-deficient mice. All data are expressed as 10−18 moles of transcripts/μg of RNA.

Figure Figure 6.

Histological analysis of callus cartilage content, osteoclast activity, and collagen organization. (A) Safranin O (SO) staining 14 days after fracture confirmed normal cartilaginous callus formation at the fracture site in OPN-deficient mice. (B) Staining for TRACP activity showed similar osteoclast number and activity in the OPN−/− mice compared with wildtypes at 28 days after fracture. (C) Whereas type I collagen gene expression was similar between genotypes, polarized light microscopy of picrosirius red (PSR) stained sections revealed a qualitative difference in polarization colors, suggesting altered collagen organization in OPN−/− mice at 56 days after fracture.

Bone remodeling

Next, because of the larger residual fracture callus seen in OPN-deficient mice at the latest time-point, we hypothesized that OPN plays a critical role in osteoclastogenesis. To test this hypothesis, we first measured expression of RANKL and OPG, two of the most prominent proteins involved in regulation of osteoclast formation.(34) In addition, we measured expression of TRACP and cathepsin K, two enzymes produced by mature osteoclasts to digest the mineralized bone matrix during remodeling. Quantitative RT-PCR showed no significant differences in expression of these markers of osteoclastogenesis between genotypes (Figs. 5B and 5C). Furthermore, histological staining for TRACP activity indicated that the OPN-deficient mice displayed similar levels of TRACP+ osteoclasts as their wildtype counterparts at 28 days after fracture (Fig. 6B). These data indicate that there is no defect in the basic machinery required for mature osteoclast formation and activity in the OPN−/− mice.

Type I collagen content and organization

OPN has been postulated to bind to other ECM proteins and to be important in matrix reorganization after injury.(28) Therefore, we measured gene expression of type I collagen, the most abundant bone ECM protein, and lysyl oxidase (LOX), an enzyme that cross-links collagen fibers.

Results indicated that these genes undergo rapid upregulation in the setting of fracture healing, but no differential expression between genotypes was detected (Fig. 5D). In addition, we measured total collagen content at the protein level in 56-day postsurgery fracture callus specimens as determined by the hydroxyproline assay. At this late time-point, the bony callus is primarily composed of type I collagen, but a well-known limitation of this assay is its inability to distinguish between different collagen types. Based on this assumption, results indicated that total collagen content (collagen weight/total dry weight) in the fracture callus was not different between genotypes with an average of 12.7 ± 0.3% for wildtypes and 12.3 ± 0.3% for OPN−/− mice. Next, we used picrosirius red staining combined with polarized light microscopy for qualitative determination of collagen fiber organization. This imaging technique suggested that, whereas there is no difference in collagen content, the newly formed bone in the OPN-deficient mice contained abnormal collagen organization relative to their wildtype counterparts at 56 days after fracture (Fig. 6C).


In this study, we compared wildtype and OPN−/− mouse femoral fracture callus formation, neovascularization, mineralization, and mechanical properties. OPN deficiency was found to significantly alter but not prevent bone regeneration and remodeling of fractures in mice. No compensatory overexpression of other ECM components was found, suggesting that there may be redundant mechanisms that allow fracture healing to occur (albeit delayed) in the absence of OPN because of common cellular binding sites (i.e., RGD sequence) and functional overlap between OPN and other ECM proteins. Specifically, our data indicate that the presence of OPN is essential for normal early callus formation, neovascularization, and biomechanical strength and ductility. Additionally, OPN deficiency was found to delay the time-course of remodeling of the fracture callus during the later stages of healing. Last, whereas ECM content seemed generally unchanged, abnormal collagen organization was observed within the remodeling calluses of OPN−/− mice.

At early stages of healing, OPN was found to play an important role in callus formation and to have a significant but transient effect on neovascularization. The reduction in early vascular volume within the fracture callus at 7 days was recovered by 14 days. However, a significant decrease in average vessel diameter persisted, indicating that OPN may also play a role in vessel maturation. This may be related to the stimulatory role that OPN has on smooth muscle cell migration and chemotaxis,(20,23) or it could be a consequence of the OPN−/− mouse macrophage phenotype, which has been linked to abnormal vascular remodeling.(16) Relevant to the latter hypothesis, OPN deficiency has been found to result in normal macrophage numbers but decreased levels of macrophage activation in skin wounds, and intradermal injection of exogenous OPN has been found to stimulate macrophage infiltration.(21,28) The prior study was consistent with our findings, which indicated normal macrophage numbers (data not shown) but evidence that there may be defective functionality of these cells at the site of injury in the OPN−/− mice. Therefore, in addition to the direct effects of OPN on endothelial and smooth muscle cell survival and migration, a reduced inflammatory response likely has secondary effects on callus neovascularization, vessel maturation, and ECM formation caused by decreased reorganization of damaged tissue and cytokine production during the early stages of fracture healing.

One interesting observation from this study is that, whereas the OPN−/− mice displayed a reduction in callus volume at 1 and 2 weeks after fracture, they possessed 46% larger fracture calluses at 8 weeks. The observed time-dependent phenotype provides further evidence for the multifunctionality of OPN during fracture healing. The increased residual fracture callus present in the OPN−/− mice at 56 days is likely caused by the well-documented role that OPN has in mediating activity of mature osteoclasts during bone remodeling.(9–11) This defect, as was shown here, is not a result of inhibition of osteoclast differentiation or production of proteolytic enzymes. Previous studies have also shown that osteoclastogenesis is not inhibited in the setting of OPN deficiency, and, in fact, a compensatory increase in osteoclast number has been reported in the bones of OPN−/− mice under basal conditions in some studies.(9,13) As elegantly shown by Chellaiah et al.,(9) the reduction in functionality of OPN-deficient osteoclasts can be attributed to decreased motility caused by a lack of cell surface expression of CD44 and activation of the αvβ3 integrin.

In the aforementioned study, Chellaiah et al. were the first to link OPN−/− osteoclast dysfunctionality to an observable in vivo bone phenotype in OPN-deficient animals at baseline. Although neither Rittling et al.(11) or Liaw et al.(28) found any abnormal bone phenotype on initial development of the OPN knockout mouse, this group more recently reported larger femoral moment of inertia and, as determined by four-point bending, increased elastic modulus, ultimate moment, and energy to failure in OPN−/− mice.(9) In this study, as verified using independent three-point bending and torsional mechanical testing modalities, we found results that are partially contradictory to these previously published data. Similar to Chellaiah et al., we observed an increase in the elastic modulus of OPN−/− bone. However, the OPN-deficient mice used for our study did not display increased femoral moment of inertia, and we found decreased maximum load and energy to failure relative to bones from wildtype mice. In addition, we found the OPN−/− mice to have an −40% decrease in postyield deformation, a parameter that was not reported in previous studies. Taken together, these biomechanics data indicate that OPN deficiency increases the material stiffness of the bone tissue and also causes it to be more brittle. A few possible explanations exist for the discrepancy between our data and the previously published results. The slightly older mice used in the Chellaiah et al. publication could have displayed a phenotype that develops as the mice age caused by a mild, but prolonged defect in osteoclast functionality. These divergent results could also be a function of the difference in genetic background. The previous study was completed with mice on a hybrid background, whereas the mice used here were backcrossed onto a C57BL/6 genetic background. The potential relevance of this cannot be ignored considering the significantly dissimilar baseline bone phenotype between background strains(35) and the different susceptibility of each mouse strain to a variety of pathologies including osteoporotic bone loss, femoral artery blockage, and atherosclerosis.(36–39) The final difference of note is that the mice used in our study were derived from the OPN-deficient mouse population developed by Liaw et al., rather than the Rittling et al. mouse. The two different OPN knockouts were derived using slightly different methodologies, and the Rittling et al. mouse has been found to express a small but likely nonfunctional fragment of the OPN protein.(11,28)

Significant functional effects of OPN deficiency were also evident in the biomechanical tests on the fracture callus specimens. The physical significance of altered callus size was evident in the torsional tests, which showed that the smaller OPN−/− calluses at 14 days after fracture possessed significantly reduced maximum torque and work to failure. Interestingly, significantly reduced mechanical properties (maximum torque, work to failure) were also found in the OPN deficient mice at 28 days after fracture, but no differences were seen in callus size or percent mineralization between genotypes at this endpoint. These data, in addition to the strong overall trend toward decreased postyield behavior in the OPN−/− mice, indicate the existence of fracture callus material property differences as a result of OPN deficiency. The observed brittle material behavior in the OPN−/− fracture specimens corroborates the observation of decreased ductility in OPN−/− intact bone, and this consistent decrease in bone quality may be the result of altered organization of the ECM within the callus of the OPN−/− mice. The biological relevance of the callus volume was also evident at 56 days, at which point the OPN−/− specimens possessed significantly increased stiffness. It should be noted, however, that at 56 days, the geometry of the fracture callus had undergone significant remodeling in both genotypes and was relatively small compared with callus size measurements at earlier time-points. However, even at 56 days, it seemed that neither group had achieved a remodeling equilibrium or completely regained the mechanical properties of intact bone. This incomplete restoration of bone quality can be readily explained by the fact that, whereas the bone geometry is approaching baseline, mineral density of the fracture callus has only reached −60% of the mineral density of normal intact bone at this endpoint (data not shown).

Despite alterations in size and biomechanical strength of the fracture callus, other than a modest change found in BSP mRNA levels, measurement of gene expression, immunohistochemical protein localization, and quantitative measurement of collagen content do not indicate compensatory changes in other ECM proteins in OPN−/− mice. However, while no differences were seen in total collagen content, picrosirius red staining suggested abnormal collagen organization within remodeling OPN−/− fracture calluses, which could be a result of alterations in collagen fibrillogenesis or turnover in OPN-deficient mice. This interpretation agrees with previous observations that OPN deficiency results in altered collagen fibril diameter and organization in skin wounds(28) and increased collagen maturity in bone.(8) Considering these observations, in conjunction with the fact that OPN has a high affinity for binding collagen,(40) it is possible that the absence of an OPN–collagen interaction could result in aberrant fibril organization and contribute to the reduced mechanical integrity detected here for both intact and fractured OPN−/− bones. Altered collagen organization may have a detrimental effect on the ability of the bone tissue to dissipate energy as mineralized fibrils begin to move relative to each other during postyield deformation of the tissue. Consistent with this interpretation, Hansma et al. have recently hypothesized that the mechanical behavior of bone is greatly influenced by nonfibrillar matrix proteins that form sacrificial bonds between mineralized collagen fibers. They specifically suggest that OPN and related proteins may be important components of the “glue” that resists separation of collagen fibers and propagation of cracks during mechanical failure.(41,42) This recent work may provide a mechanistic explanation for both the alteration in collagen organization and the observed brittle behavior in the bones of the OPN−/− mice.

In conclusion, OPN deficiency significantly alters several stages of the bone healing process but does not prevent bone union. More specifically, this study indicates that OPN plays multiple roles during early callus formation, neovascularization, and late stage remodeling. In addition, our extensive biomechanical analyses suggest that OPN functions to enhance bone strength and its ability to dissipate energy before failure. The fact that no alterations in content of other ECM proteins were found suggests that the phenotype could be driven by altered matrix turnover and collagen organization, which seems to span multiple organ systems. Therefore, we believe that a unifying link between the vascular and bone related defects found here may be related to diminished collagenous matrix organization and remodeling. Hypothetically, this defect could decrease the ability of cells to bind to ECM and respond appropriately to environmental cues in the absence of OPN. However, future studies are necessary to more definitively determine whether diminished ECM integrity is a primary effect of OPN deficiency or secondary to altered cellular functionality in the absence of OPN. These findings contribute to an improved understanding of the role of OPN in vivo and provide new insight into bone biomechanics and mechanistic control of vascularization and bone regeneration during fracture repair.


This work was supported by NIH Grants R01AR051336, RO1HL70531, PO1HL58000, the Georgia Tech/Emory Center for the Engineering of Living Tissues NSF Grant EEC-9731643, and the NSF Graduate Research Fellowship Program (CLD). The authors thank Tracey Couse for help with histology, Galen Robertson for assistance in design of three-point bending fixtures, Kenneth Dupont for contributions to design of the torsional mechanical testing protocol, and Angela Lin for advice on μCT analysis. The BSP antibody was received from the Developmental Systems Hybridoma Bank at the University of Iowa.