The clinical management of critical (>3 cm) segmental defects remains a major challenge for both amputation and limb salvage approaches, which continue to suffer from poor long-term outcomes.1, 2 Although devitalized cortical allografts lack osteogenic potential, they remain widely used for reconstructive surgery, because no synthetic alternatives with comparable biological and biomechanical properties have been developed. Unfortunately, the limited new bone formation and lack of remodeling associated with massive allograft healing are directly associated with the 23% to 43% clinical failure rate caused by nonunion (27% to 34%), late graft fracture (24% to 27%), and infection (9% to 16%).3 Thus, the development of an adjuvant therapy for patients recovering from reconstructive surgery for critical defects remains a high priority.
Based on its well-known anabolic effects on bone,4–6 recombinant parathyroid hormone (rPTH), either as the full-length (PTH1–84) peptide or N-terminal (PTH1–34) peptide, has been extensively studied in animal models7–13 and clinical trials14–16 of fracture healing with very promising results. We have also observed remarkable rPTH effects in a murine femoral allograft model, in which bone formation only occurs at the graft-host junction as a result of the host response to the necrotic allograft.17, 18 Whereas the increased bone formation in the rPTH-treated allografts was predicted, we were surprised by the remarkable decrease in fibrosis around the allografts, which lead to greater boney union between the graft and the host, and increased biomechanical properties.17, 18 Another unexpected finding was the apparent decrease in the vascularity of rPTH treated allografts (unpublished data), which contrasts data showing that increased angiogenesis from exogenous vascular endothelial growth factor A (VEGF-A) accelerates skeletal repair.19–22 However, recent studies have shown that whereas rPTH-induced bone formation is VEGF-A dependent, its anabolic effects occur without increasing bone vessel density by redistribution of smaller vessels closer to bone formation sites.13 Because elucidating the mechanism of decreased fibrosis has broad implications for all tissue engineering and regenerative medicine applications, we tested the hypothesis that rPTH therapy promotes small vessel angiogenesis and inhibits large vessel arteriogenesis during structural allograft healing in mice. The PTH/PTH-related protein receptor (PPR) has a central role in mediating the diverse actions of PTH on bone in vivo.23 It is evident that intermittent (once daily) exogenous rPTH administration has an anabolic effect on bone,24 whereas the continuous exposure to PTH, as in hyperparathyroidism, leads to hypercalcaemia and a net decrease in bone volume,25 which is referred to as its “catabolic effect.” In an attempt to elucidate these differential effects and mechanisms of PTH action in our femoral allograft model, we utilized Col1caPTHR transgenic mice that have a constitutively active PPR receptor under the Col1 promoter.26 These mice have a high trabecular bone mass with increased osteoclast activity as a result of increased osteoblast numbers and high bone turnover. Because this genetic model of PTH gain of function is specific to osteoblasts, this system allowed us to test if osteoblasts are the primary target of rPTH therapy during allograft healing.
Materials and Methods
Femoral autograft and allograft model
All experiments with live animals were performed on University of Rochester Committee for Animal Resources approved protocols. Femoral autograft and allograft surgeries were performed on mice, as previously described.20 Briefly, a 4-mm middiaphyseal osteotomy was created in the host mouse and then replaced with the same bone (live autograft), or with an acellular middiaphyseal femoral segment from an allogenic strain.
Eight-week-old female C57/BL6 mice underwent live autograft or allograft surgery. The allografted mice were randomized to treatment with daily intraperitoneal injections of teriparatide (40 µg/kg/day human PTH1–34, Forteo; Eli Lilly and Co., Indianapolis, IN, USA) (n = 5) or saline (placebo) (n = 5), as previously described.17 Mice were euthanized on days 7, 10, 14, 21, or 28 postoperatively (po) for gene expression studies, and additional mice were euthanized on days 21 and 28 for micro-CT and histology.
Eight-week-old Col1-caPTHR mice26 (n = 5) and their wild-type (WT) littermates (n = 5) in a FVB background received allografts or live isografts from Col1-caPTHR mice, and the grafted femurs were harvested after euthanasia on day 21 or 28 po for micro-CT and histology analyses.
Eight-week-old female C57/BL6 mice underwent allograft surgery. The allografted mice were randomized to treatment with biweekly intraperitoneal injections of L1-1027 (3 mg/kg/dose) (n = 5), a recombinant peptide-Fc fusion protein (peptibody) that selectively inhibits angiopoietin-2 (Ang-2) binding to its tyrosine kinase receptor (Tie2),28 AMGEN Inc., Thousand Oaks, CA) or saline (placebo) (n = 5). Mice were euthanized on day 21 po for micro-CT, histology, and histomorphometry.
X-rays, vascular perfusion, and micro-CT
Longitudinal plain radiographs were obtained using the Faxitron Cabinet X-Ray System (Faxitron X-Ray, Wheeling, IL, USA), as previously described.20 Animals were then euthanized using lead chromate intracardiac infusion, as previously described.29 Briefly, the mice were anesthetized and thoracotomy was performed to insert a 20-gauge angiocatheter into the left ventricle. The circulatory system was flushed with 20 ml of 0.9% saline and heparin (100 IU/mL), followed by 20 mL of 10% neutral buffered formalin. This procedure was immediately followed by injection of 4 mL lead base contrast agent MICROFIL MV-122 (Flow Tech, Inc., Carver, MA, USA) via the same route. The femurs were harvested, intramedullary pin removed, and the specimens were stored in 10% neutral buffered formalin until the initial micro-CT scan was performed using a 10.5 micron isotropic high-resolution micro-CT (VivaCT 40; Scanco Medical AG, Basserdorf, Switzerland). After the initial micro-CT scan, the specimens were then decalcified for 4 to 6 weeks in ethylene, diamine, tetra-acetic acid (EDTA). Following decalcification, the specimens were subjected to second micro-CT scan. Volumetric bone and vascular quantification was performed on the region of interest (ROI), which spanned the 4 mm graft and 1 mm of the host femur on both the proximal and distal graft junction, as previously described.29
Histology and pilot histomorphometry
After micro-CT, the femurs were dissected and processed for decalcified histology in paraffin, as previously described.29 At least three nonconsecutive 4-µm paraffin-embedded midsagittal sections of the ROI from each specimen were stained with either alcian blue/hematoxylin/orange G (AB/OG), or for tartrate resistant acid phosphatase (TRAP) and counterstained with safrin-O. Osteoclasts were defined as multinucleated cells (>3 nuclei) TRAP-positive cells on a bone surface in the ROI. Histomorphometry to quantify osteoclast and mast cell numbers, and mesenchymal tissue, cartilage, and bone area using Osteometrics Software (Decatur, GA, USA), was performed on three parallel section >0.2 mm, apart from each femur in which the mean values are the data for each mouse (n = 5).
Gene expression analyses
Grafted femurs were harvested; the 6 mm ROI, including the 4 mm of graft and 1 mm of host at each end, was homogenized using TissueLyser II (Qiagen, Hilden, Germany), and total RNA was extracted using the RNeasy minipurification kit (Qiagen, Valencia, CA, USA) according to manufacturer's protocol. The mRNA was reverse-transcribed using an iScript cDNA synthesis kit (Bio-Rad Lab, Hercules, CA, USA), and the cDNA was used as the template for real-time PCR analyses with the following gene-specific primers obtained from Integrated DNA Technologies (IDT, Coralville, IA, USA).
5′-CCA TGC TTG AGA TAG GAA CCA G-3′
5′-TTC AAG TCG GGA TGT TTG ATT T-3′
5′-AGC AGA TTT TGG ATC AGA CCA G-3′
5′-GCT CCT TCA TGG ACT GTA GCT G-3′
5′-CGG CTT AGT TCT CTG TGG AGT C-3′
5′-GGC ATC AGA CAC AAG AGG TAG G-3′
5′-AGA TGT GGA TCA GCA AGC AG-3′
5′-GCG CAA GTT AGG TTT TGT CA-3′
Each reaction was carried out in a final volume of 20 mL consisting of 0.1 mm primers, 1 mL Sybr Green PCR Super Mix (ABgene, Thermo Fisher Scientific Inc, Rochester, NY, USA), and 1 mL of the purified cDNA template. The samples were assayed in triplicate in a Rotor-Gene RG 3000 (Corbett Research, Sydney, AU). Mouse β-actin was used as the reference gene to normalize for differences in the amount of total RNA in each sample. For quantification, the relative level (mean ± SD) of each group normalized to β-actin was determined as the fold-change versus day 7 autograft.
Results are shown as the mean value plus or minus the standard deviation (±SD) of the mean. All statistical analysis was performed using Prism statistical package Version 4.0 (GraphPad, San Diego, CA, USA) with p values <0.05 being considered statistically significant. A one-way analysis of variance (ANOVA), Neuman-Keuls multiple comparison test, t test, and Mann-Whitney and Spearman's rank correlation coefficient was used for all statistical analyses. Significant differences in bridging new cortical bone formation were determined by Fisher's exact test.
Intermittent rPTH therapy promotes small vessel angiogenesis and inhibits large vessel arteriogenesis during femoral allograft healing
In order to establish the effects of intermittent rPTH treatment on the neovascularization of structural allografts after implantation, we performed a time course experiment in which mice received a femoral allograft surgery and were randomized to daily rPTH or placebo injections. The allografted femurs were harvested on days 7, 10, 14, 21, and 28 after surgery, and analyzed by micro-CT and histology versus a live autograft control group after radiographic confirmation of rPTH anabolic effects on bone formation (Supplemental Fig. 1 and 2). The results revealed that rPTH significantly affects allograft vascularity as early as day-14 po (Fig. 1). At day 28, rPTH significantly decreased total allograft vascularity compared with placebo (vascular volume = 0.58 ± 0.10 compared with 1.06 ± 0.19 mm3; mean vessel diameter = 0.08 ± 0.01 vs. 0.10 ± 0.01 mm; p <0.05). Review of the histology slides revealed a clear phenotypic difference between the sizes of lead-chromate perfused blood vessels, in that the vessels of the rPTH treated allografts appeared to be more abundant and markedly smaller versus the placebo (Fig. 1B–E). At day 28, rPTH significantly decreased total allograft vascularity, and the mean vessel diameter compared with the placebo (Fig. 1F, G). Contrary to the gross overall decrease in vascularity as assessed by micro-CT, pilot histomorphometry clarified this anomaly depicting a marked increase in the vascular parameters in rPTH-treated allografts versus placebo (Table 1).
Table 1. Vascular Histomorphometry of Allografts at 21 Days of Healing
Pilot (n = 5) histomorphometry was performed on the histology described in Fig. 1, in which the 6 mm mid-diaphyseal area of the allografted femurs were used as the ROI (4 mm of graft + 1 mm of host at both proximal and distal graft-host junctions). Mean ± SD. ap < 0.05 vs. placebo.
Region of interest (ROI) area (mm2)
13.85 ± 3.72
18.55 ± 0.93
74 ± 30
137 ± 39a
Vascular area (mm2)
0.32 ± 0.12
0.46 ± 0.12a
Vascular surface (mm)
17.17 ± 4.99
30.12 ± 7.84a
Vascular count per ROI area
5.41 ± 2.25
7.44 ± 2.17a
Vascular surface per ROI area
1.28 ± 0.41
1.63 ± 0.42a
To further confirm these differences, we quantified the number of smaller versus larger perfused vessels via micro-CT of allografts harvested on day 21 and 28 (Fig. 2), which shows that rPTH-treated allografts have significantly more small vessels (<90 µm), and significantly fewer large vessels (>90 µm) versus the placebo. Interestingly, this alteration in neovascularization corresponded with a dramatic alteration in the fibrotic tissue response versus the chondrogenic and osteogenic response in the placebo, versus rPTH treated allografts, respectively. Confirmatory histomorphometry revealed that the tissue composition of rPTH-treated allografts is remarkably similar to that of live autografts, in contrast with the fibrosis observed in placebo (Fig. 3A–E).
In order to elucidate the mechanism by which PTH signaling decreases large vessel arteriogenesis during allograft healing, we performed gene expression studies on femurs that received live autografts, or allografts followed by daily placebo or rPTH therapy for 7, 10, and 14 days. Although significant ∼2-fold increases in vegf-a and vegfr-r2 mRNA levels were detected in the rPTH treated allografts (Supplemental Fig. 3), the most remarkable findings revealed a reciprocal regulation of angiopoietin (ang) genes on day 7. We observed an 8-fold increase in ang-1, and a 35-fold decrease of ang-2 mRNA levels in healing allografts from rPTH treated mice verses the placebo (Fig. 3F–G). Because Ang-2 is known to be critically involved in remodeling small capillaries into large vascular networks, and Ang-1 is a known Ang-2 antagonist.30, 31 this novel finding suggests that rPTH inhibits large vessel arteriogenesis during allograft healing via reciprocal regulation of these factors.
Constitutive PTHR signaling in osteoblastic cells promotes small-vessel angiogenesis and inhibits large-vessel arteriogenesis during femoral allograft healing and extensive graft resorption
In order to confirm our observations on allograft vascularity, and further define the target cells of rPTH treatment in our model, we repeated the femoral allograft experiment using Col1-caPTHR transgenic mice and their nontransgenic littermates as controls. The micro-CT and histology results from these experiments were remarkably similar to that of the rPTH versus placebo treatments (Fig. 4). Specifically, micro-CT analyses displayed a similar decrease in vascularity of the allografts in Col1-caPTHR mice compared with their WT littermates on day 28 (vascular volume 0.36 ± 0.09 vs. 1.01 ± 0.31 mm3; vessel number 1.01 ± 0.11 vs. 1.26 ± 0.06 mm; p <0.05), and histology (Fig. 4B–E) revealed the same small versus large vessel ratio that was observed in placebo versus rPTH-treated mice. However, there were two remarkable differences between rPTH and Col1-caPTHR that could be attributed to intermittent versus continuous PTH signaling. The first is that both large and small blood vessel formation in Col1-caPTHR allografts are suppressed at 21 days, as the significant increase in small-vessel numbers (<90 µm) versus WT did not occur until day 28 (Fig. 5). The other major dissimilarity is that whereas we did not observe any significant differences between allograft bone volume and osteoclast numbers in rPTH versus placebo treated allografts (data not shown), X-rays revealed that femoral allografts in Col1-caPTHR mice are rapidly resorbed (Supplemental Fig. 4) as a result of a significant increase in osteoclast numbers (Table 2).
Table 2. Histomorphometry of Allografts into a Col1caPTHR and WT Host Mice at 21 Days of Healing
Pilot (n = 5) histomorphometry was performed on the histology described in Fig. 3, in which the 6 mm mid-diaphyseal area of the allografted femurs were used as the ROI (4 mm of graft + 1 mm of host at both proximal and distal graft-host junctions). Mean ± SD. ap < 0.05 vs. WT.
Region of interest area (mm2)
6.80 ± 0.89
5.362 ± 0.75
Host bone area (mm2)
2.11 ± 1.07
2.41 ± 0.68
Graft bone area (mm2)
0.78 ± 0.08
0.53 ± 0.20a
Osteoclast number/graft area (mm2)
5.53 ± 3.08
9.09 ± 5.49a
Osteoclast surface/graft surface (%)
22.13 ± 12.31
36.37 ± 21.97a
Intermittent rPTH treatment reduces mast cell numbers in healing allografts
Upon scrutiny of toluidine blue–stained histology sections to assess cartilage formation during allograft healing, we notice a paucity of mast cells (Supplemental Fig. 6) in immediate proximity to blood vessels in the transitional tissue adjacent to the allografts of rPTH-treated mice compared with placebo (Fig. 6A–F). Pilot histomorphometry showed that there appears to be a 2-fold decrease in mast cell numbers at the allograft-host junction on day 21 in rPTH versus placebo-treated mice, whereas no differences in mast cell numbers in the adjacent skeletal muscle were observed (Table 3).
Table 3. Mast Cell Histomorphometry of Allografts at 21 Days of Healing
Pilot (n = 5) histomorphometry was performed on the histology described in Fig. 6, in which the 6 mm mid-diaphyseal area of the allografted femurs were used as the ROI (4 mm of graft + 1 mm of host at both proximal and distal graft-host junctions). Mean ± SD. *p < 0.05 vs. placebo.
Region of interest (ROI) area (mm2)
76.87 ± 7.44
70.97 ± 20.98
Total mast cell count
170 ± 14
122 ± 17a
Mast cell/ROI area (mm2)
2.22 ± 0.07
1.72 ± 0.21a
Mast cells at graft-host junction
33 ± 9
14 ± 5a
Anti-Ang-2 treatment mimics rPTH effects on arteriogenesis, fibrosis, and mast cell recruitment during allograft healing
To test the functional importance of angiopoietin-2 during structural allograft healing, we performed femoral allograft surgery on mice randomized to placebo or (L1-10), a recombinant peptide-Fc fusion protein (peptibody) that selectively inhibits angiopoietin-2 (Ang-2) binding to its tyrosine kinase receptor (Tie2), and harvested the grafted femurs on day 21 for micro-CT analyses. As expected, anti-Ang2 (L1-10) therapy had no effect on small vessel formation, but significantly inhibited the formation of perfused vessels >180 µm in diameter (Fig. 7A). We also performed histology analyses on these samples and confirmed that anti-Ang2 treatment mimics PTH treatment in decreasing fibrosis, mast cell accumulation and arteriogenesis during allograft healing (Fig. 7B). Furthermore, we observed a significant increase in host-to-host bony union (5 out of 5 anti-Ang2 versus 0 out of 5 placebo; p <0.05 by Fisher's exact test), evidence by the presence of new bone that formed along the allograft surface (Fig. 7C). Because these outcome measures closely correlate with biomechanical healing of structural allografts,32 it suggests that inhibition of arteriogenesis, mast cell infiltration and fibrosis via anti-Ang2 treatment may be an effective adjuvant therapy after tissue reconstruction.
Despite the remarkable advances in reconstructive surgery and musculoskeletal tissue engineering to correct segmental defects over the last two decades, effective healing between the host and the implant remains a major challenge because of fibrosis at the interface. In the case of bone healing, which occurs through a highly regulated temporal process that proceeds from an initial inflammatory phase,33 blood and periosteum derived mesenchymal stem cells (MSC) at the fracture site must respond to biological and biomechanical cues to appropriately differentiate into chondrocytes and osteoblasts that will produce the bridging matrix. When this process is challenged by disease, infection, drugs, aging, or excessive tissue damage, MSC differentiation can be diverted to fibroblastic cells, leading to fibrosis at the interface and nonunion. To better understand the etiology of fibrosis we examined the effects of rPTH therapy on blood vessel formation and osseous integration during femoral allograft healing. The results showed that rPTH causes a significant increase in small vessel numbers, and decreased large vessel formation (Table 1; Fig. 1 and 2). Moreover, rPTH appeared to markedly reduced fibrosis around the allografts to levels observed in live autografts (Fig. 3), and decreased mast cell numbers at the graft-host junction (Table 3; Fig. 6). As these rPTH affects on allograft healing were recapitulated by femoral allografts in Col1caPTHR mice (Table 2; Fig. 4 and 5), our findings suggest that rPTH exerts its effects directly on osteoblasts at the site of bone repair. At the molecular level, our findings that rPTH reciprocally regulates angiopoietin-1 and angiopoietin-2 at day 7 of allograft healing (Fig. 3), and that anti-angiopoietin-2 peptibody (L1-10) treatment mimics rPTH effects on angiogenesis and fibrosis (Fig. 7), suggests that decreasing Ang-2 mediated arteriogenesis is a critical mechanism of action of rPTH therapy for bone repair. Given that the decreased arteriogenesis and fibrosis is associated with decreased mast cell numbers, it is tempting to speculate that the rPTH effects on angiogenesis function as a vascular sieve to limit profibrotic cell access to the site of tissue repair to allow tissue repair without scaring, and future studies to directly test this hypothesis are warranted.
Whereas angiogenesis is known to be a critical process during fracture repair,34 the nature of these new vessel (small capillaries vs. large vessels) and their proximity to newly forming bone has been investigated only recently.13 Gene expression studies in animal models have showed that angiogenic factors (Vegf-a, neuropilin, Ang, Vegfr-2, Tie-2 and Hif-1a) are immediately induced after fracture,35, 36 which is consistent with the known roles of angiogenesis in MSC recruitment, vascular invasion of the hypertrophy cartilage, and calcification of the soft callus, and intramembranous bone formation. However, it is also known that vascular regression is required for mesenchymal condensation and chondrogenesis, and that disruption of this hypoxic environment via VEGF-induced angiogenesis leads to decreased cartilage formation.37 Thus, our observation that rPTH limits the magnitude of vasculogenesis after structural allograft healing, to mimic that observed in the chondrogenic stages of autograft healing (Fig. 1 and 3), provides additional insight as to how this therapy stimulates cartilage formation versus fibrosis at the graft-host junction.
It has long been recognized that mast cells may play an important role in fracture healing.38 Histology studies of fractures in a rat model revealed that in the first two weeks, mast cells are found either in the vicinity of blood vessels or in the vascularized tissue proliferating into the cartilaginous portion of subperiosteal callus.39 This finding led to the view that mast cells are involved in digestion of extracellular matrix and angiogenesis in the early stages of fracture healing. However, mast cells are also known to be central mediators of chronic fibrosis via degranulation and release of fibroblast growth factors, TGFβ, and other factors that promote progressive sclerosis.40 Thus, our finding that rPTH therapy markedly decreases mast cell numbers in the transitional tissue between healing structural allografts and the adjacent skeletal muscle (Fig. 5) suggests another potential antifibrotic mechanism of this intervention. Whereas it is attractive for us to speculate that the decrease in mast cell numbers is related to the decreased large vessel formation in rPTH-treated allografts, future studies are needed to formally establish this relationship.
In order to identify the direct mechanism of rPTH-decreased arteriogenesis during allograft healing, we perform a broad gene-expression analysis of angiogenic factors, which focused our attention on ngiopoietins (data not shown). Our finding that Ang-2 is expressed 70-fold higher in allografts versus autografts on day 7, and that rPTH therapy reduces this expression level in allografts down to that observed in autografts, strongly implicates Ang-2 as a dominant mediator of arteriogenesis in our model (Fig. 3F and 3G). Interestingly, this rPTH-mediated decreased Ang-2 expression occurred with a reciprocal 8-fold increase in Ang-1 levels, which is somewhat surprising considering that Ang-1 is pro-angiogenic.41–43 However, it is noteworthy that other studies have shown that Ang-1 inhibits angiogenesis and vascular permeability.44–48 Thus, this rPTH-induced reciprocal regulation of Ang-1 and Ang-2 could be an important factor in the decreased vasculogenesis. More importantly, the demonstration that anti-Ang2 therapy selectively and significantly inhibits large-vessel (>0.18 mm) formation during allograft healing (Fig. 6A), and significantly enhances bony union by decreasing the amount of fibrosis and accelerating osteogenesis and chondrogenesis (Fig. 6B), confirms that Ang-2 is necessary and sufficient for arteriogenesis and fibrosis at the host-allograft interface in this model. Because a major clinical indication of structural allografting is reconstruction after bone tumor resection, and rPTH therapy is contraindicated in these patients, our finding that anti-Ang2 therapy achieves a similar outcome has obvious implications for this major unmet need for limb salvage in cancer patients. Additionally, although this research has focused on bone healing, the problem of postoperative fibrosis is common to all surgical procedures. Thus, future studies to evaluate the role of Ang-2 and the effects of rPTH treatment during soft tissue repair and regeneration are warranted.
All authors state that they have no conflicts of interest.
The authors would like to thank AMGEN Inc. for their gift of the L1-10 peptibody, Ryan Tierney and Sarah Mack for their assistance with the histology, and Michael Thullen for technical assistance with micro-CT analyses. This work was supported by research grants from the National Institutes of Health (DE019902, AR054041 and AR061307).
Authors' roles: Study design: ES, RD, LC, WT, RO, HA, and CX. Study conduct: RD and CX. Data collection: RD. Data analysis: RD, CX, LC, HA and ES. Data interpretation: ES, RD, LC, WT, RO, HA, and CX. Drafting manuscript: ES and RD. Revising manuscript content: ES, CX, LC, MZ, HA, and RO. Approving final version of manuscript: RD, CX, WT, LC, HA, MZ, RO, and ES. RD and ES take responsibility for the integrity of the data analysis.