|Home | About | Journals | Submit | Contact Us | Français|
Although angiogenesis and osteogenesis are critically linked, the importance of angiogenesis for stress fracture healing is unknown. In this study, mechanical loading was used to create a non-displaced stress fracture in the adult rat forelimb. Fumagillin, an anti-angiogenic agent, was used as the water soluble analogue TNP-470 (25 mg/kg) as well as incorporated into lipid-encapsulated αvβ3 integrin targeted nanoparticles (0.25 mg/kg). In the first experiment, TNP-470 was administered daily for 5 days following mechanical loading, and changes in gene expression, vascularity, and woven bone formation were quantified. Although no changes in vascularity were detected 3 days after loading, treatment-related downregulation of angiogenic (Pecam1) and osteogenic (Bsp, Osx) genes was observed at this early time point. On day 7, microCT imaging of loaded limbs revealed diminished woven bone formation in treated limbs compared to vehicle treated limbs. In the second experiment, αvβ3 integrin targeted fumagillin nanoparticles were administered as before, albeit with a 100-fold lower dose, and changes in vascularity and woven bone formation were determined. There were no treatment-related changes in vessel count or volume 3 days after loading, although fewer angiogenic (CD105 positive) blood vessels were present in treated limbs compared to vehicle treated limbs. This result manifested on day 7 as a reduction in total vascularity, as measured by histology (vessel count) and microCT (vessel volume). Similar to the first experiment, treated limbs had diminished woven bone formation on day 7 compared to vehicle treated limbs. These results indicate that angiogenesis is required for stress fracture healing, and may have implications for inducing rapid repair of stress fractures.
Osteogenesis, the process of new bone tissue formation, is strongly related to angiogenesis, the process by which new vessels extend from the existing vascular system . In fact, it has been shown that many osteogenic processes, including fracture healing [1, 2], distraction osteogenesis [3–5], skeletal development , and others [7, 8] are critically dependent on angiogenesis. Repetitive mechanical loading of the skeleton also stimulates osteogenesis, but the role of angiogenesis in this context is not well understood.
In general, mechanical loading at hyperphysiological strain levels leads to the formation of woven bone rather than lamellar bone . In the rat forelimb, a single bout of damaging, cyclic loading has been shown to reproducibly generate a non-displaced stress fracture that leads to woven bone formation at the ulnar mid-diaphysis [10, 11]. In this scenario, woven bone formation is associated with the upregulation of angiogenic genes (Vegf, Pecam1) and increased vascularity [12–14]. In addition, blood flow rate is increased at the site of woven bone formation 1 week after mechanical loading . These results suggest that angiogenesis may be a requirement for the woven bone formation response after a stress fracture.
Fumagillin, a hydrophobic mycotoxin produced by Aspergillus fumagatus, and TNP-470, a water soluble semisynthetic derivative of fumagillin, suppress angiogenesis by inhibition of methionine aminopeptidase 2 . In preclinical tumor models, TNP-470 has been shown to inhibit cancer growth and bone metastasis [17–19]. In addition, TNP-470 has been shown to prevent fracture healing  and bone formation during distraction osteogenesis  in animal studies. However, therapeutic systemic dosages of this drug can cause neurotoxicity [20, 21]. Targeted nanoparticle delivery permits lower overall dosages to be effective at specific sites of interest . The αvβ3 integrin has been shown to be a marker for blood vessels undergoing angiogenesis [23, 24], and its expression diminishes as blood vessels mature . Thus, by targeting the αvβ3 integrin, only neovasculature is affected whereas quiescent (mature) vessels are not. Recently, αvβ3 integrin targeted fumagillin nanoparticles have been shown to effectively inhibit blood vessel formation in tumors and atherosclerosis [22, 26–29].
The objective of this study was to determine if anti-angiogenic treatment affects stress fracture healing. To uncover the relationship of angiogenesis to osteogenesis in this process, changes in gene expression, woven bone formation, and vascularity were measured after damaging osteogenic mechanical loading in adult rats with and without anti-angiogenic treatment.
Male Fischer (F344/NHsd) rats were obtained at 13–14 weeks of age (Harlan) and housed under standard conditions until 18–22 weeks of age. The right forelimb of each rat was mechanically loaded using an established stress fracture (fatigue) protocol  known to stimulate an abundant woven bone formation response at the mid-diaphysis of the ulna. First, rats were anesthetized with isoflurane gas (1–3%). The right forelimb was axially compressed by placing the olecranon process and the flexed carpus into specially designed fixtures. A material testing system (Instron Electropuls 1000 or Dynamite 8841) was used to apply force and monitor displacement. A 0.3 N compressive pre-load was applied followed by a cyclic haversine waveform of 18 N at 2 Hz until an increase in peak displacement equal to 65% of the displacement to fracture (1.3 mm, relative to the 10th cycle) was achieved [12, 30]. The left forelimb was not mechanically loaded, and was used as an internal control. Following the procedure, rats were given an intramuscular injection of analgesic (0.05 mg/kg buprenorphine) and allowed unrestricted cage activity. Animals were euthanized by CO2 asphyxiation unless otherwise noted. All protocols were approved by our institution’s Animal Studies Committee.
In the first round of investigation, TNP-470 (25 mg/kg s.c.; a gift from Takeda Pharmaceutical), an agent known to inhibit angiogenesis [16, 17], or vehicle (40% (v/v) ethanol in phosphate buffered saline (PBS)), was administered daily for up to 5 days; the first injection was 1 hour prior to loading. Animals were euthanized 1 hour, 1 day, or 3 days after loading for gene expression analysis (n = 6 per group), 3 days after loading for vascular perfusion (n = 6 per group), and 7 days after loading for analysis of bone formation using microCT and dynamic histomorphometry (n = 6 per group).
In the second round of investigation, lipid-encapsulated αvβ3 integrin targeted fumagillin nanoparticles were used to inhibit angiogenesis. After loading, each rat was given a daily injection of nanoparticles (0.25 mg/kg i.v.) or vehicle (sterile water) for up to 5 days; the first injection was 1 hour prior to loading. The nanoparticles were directed to the αvβ3 integrin with a peptidomimetic vitronectin antagonist developed by Bristol-Myers Squibb Medical Imaging (U.S. patent 6,511,648 and related patents), as described previously . In general, nanoparticles were comprised of 20% (v/v) perfluorooctylbromide (Exfluor Research Corporation) and 2.0% (w/v) of a surfactant co-mixture, 1.7% (w/v) glycerin, and water for the balance. The surfactant commixture included 99 mol% highly purified egg yolk lecithin (Avanti Polar Lipids), and 0.1 mg/mL of the ανβ3 integrin antagonist conjugated to PEG2000-phosphatidylethanolamine (Kereos, Inc.). The surfactant components were dissolved in chloroform/methanol and dried in a 50 °C vacuum oven overnight. Nanoparticles were modified to include 0.1 mg/mL fumagillin (a gift from the National Cancer Institute) and 0.1 mg/mL Alexafluor594 (conjugated to lipid), which were substituted to the surfactant mixture at the expense of lecithin on an equimolar basis. Animals were euthanized 3 and 7 days after loading for vascular perfusion and immunohistochemistry (n = 7 or more per group), and 7 days after loading to assess bone formation using microCT imaging (n = 7 per group).
Changes in gene expression were assessed in treated and vehicle animals using quantitative real time PCR (qPCR) with established methods . Briefly, right and left ulnae (with surrounding periosteal tissue intact) were dissected and immediately placed into liquid nitrogen. A 5 mm section from the central ulna was pulverized, suspended in TRIzol (Invitrogen) and RNA extracted (Qiagen). RNA quantity (Nanodrop; 260/280 ratio average 2.01) and quality (Agilent bioanalyzer; RIN average 7.97) were assessed before making 1 µg cDNA (Superscript III, Invitrogen). qPCR was run (Applied Biosystems 7300) using validated primers  and Sybr® Green based detection. Five genes were evaluated, including the angiogenic markers vascular endothelial growth factor (Vegf) and platelet endothelial cell adhesion molecular (Pecam1), as well as the osteogenic markers bone morphogenetic protein 2 (Bmp2), osterix (Osx) and bone sialoprotein (Bsp). Measures of real-time PCR cycle to threshold were normalized to the expression of reference gene glyceraldehyde-3-phosphate dehydrogenase (Gapdh) for each ulna. To obtain a fold change comparison between experimental groups, Gapdh-normalized expression from each loaded ulna was divided by the normalized gene expression from the non-loaded contralateral control (2−ΔΔCt).
Vessel morphology was quantified using an established vascular perfusion technique [13, 31]. Animals were anesthetized using isoflurane gas (1–3%). After median sternotomy, an 18 gauge catheter was inserted into the aorta through the left ventricle and secured using adhesive (Loctite® 4471™, Henkel Technologies). After injection of 10 mL of Heparin Lock Flush (100 U/mL) to inhibit clotting, the animal was euthanized by exsanguination. The vasculature was then irrigated with approximately 100 mL of PBS at 37 °C. Finally, 60 mL of silicone rubber (Microfil® MV-122, Flow Tech Inc.) was injected and allowed to cure overnight at 4 °C. After curing, the forelimb was harvested and fixed in 10% buffered formalin (Fisher Scientific) for 16–24 hours. After imaging (see: MicroCT Imaging), forelimbs were decalcified (14% EDTA) and embedded in paraffin. Transverse 5 µm sections were cut 1 mm distal to the midpoint (site of maximal bone formation ) and stained with hematoxylin and eosin. Digital images of these sections were captured using bright field microscopy (Olympus BX-51) with a 10X objective. Perfused vessels in the expanded periosteum were segmented from surrounding tissue using a color threshold in Adobe Photoshop (Adobe Systems, Inc.). Using the segmented images, vessel count was calculated using ImageJ (NIH). Vessels were counted in the expanded periosteum, defined as the tissue located between the original cortical surface of the bone and the adjacent muscle.
Rats were given intraperitoneal injections of fluorescent bone formation markers immediately after loading (calcein green - 5 mg/kg, Sigma) and 5 days after loading (alizarin complexone - 30 mg/kg, Sigma). Two days later, forelimbs were harvested and fixed as above. After imaging (see: MicroCT Imaging), forelimbs were embedded in poly-(methyl methacrylate). Transverse thick sections were cut (SP 1600, Leica Microsystems) 1 mm distal to the midpoint, then polished to 100 µm and mounted on glass slides. Digital images of these sections were captured using fluorescence microscopy (Olympus IX-51) with fluorescein isothiocyanate (FITC) and tetramethylrhodamine isothiocyanate (TRITC) filters for calcein and alizarin, respectively. Images were analyzed in Bioquant for woven bone area.
Ex vivo micro computed tomography (µCT40, Scanco Medical AG) was used to analyze bone formation at the ulnar mid-diaphysis 7 days after mechanical loading. The central 8 mm of each ulna was scanned separately (45 kV, 177 µA, medium resolution, 200 msec integration, 16 µm voxel size). Scan slices were in the transverse plane (the forelimb was placed parallel to the z-axis of the scanner). Hand drawn contours (sigma = 1.2, support = 2, lower/upper threshold = 275/1000) were used to manually segment volumes of Microfil-perfused vessels (if applicable), newly formed woven bone, and original cortical bone using Scanco imaging software. In addition, the axial length along which new woven bone had formed (woven bone extent) was measured for each sample.
To evaluate the vasculature in the expanded periosteum at the ulnar mid-diaphysis using immunohistochemistry, transverse 5 µm sections of formalin-fixed, paraffin-embedded forelimbs were cut 1 mm distal to the midpoint. After deparaffination in xylene and rehydration in graded ethanol solutions, antigen retrieval was performed with either 2.1% boric acid at 55 °C overnight (CD105) or by microwaving slides in 0.01 M citrate buffer twice for 4 minutes (αSMA). Fifteen minute incubation in 3% H2O2 was used to block endogenous peroxidase activity, and sections were incubated in Ultra V Block (Thermo Scientific) to reduce nonspecific background staining. Following this, rabbit polyclonal CD105 antibody (sc-20632, Santa Cruz – 1:400 dilution) or mouse monoclonal anti-α-smooth muscle actin (A2547, Sigma – 1:1000 dilution) was applied using Ultra Clean Antibody Diluent (Thermo Scientific) at 4 °C overnight. Negative control slides were prepared by substituting normal goat serum for the primary antibody. To visualize binding, biotinylated goat anti-polyvalent (Thermo Scientific) was applied for 15 minutes followed by streptavidin peroxidase (Thermo Scientific) for 15 minutes and 1:4 dilution of VIP (Vector) for 10 minutes. The slides were then dehydrated, mounted, and examined with bright field microscopy. Positively stained vessels were counted in the expanded periosteum, defined as the tissue located between the original cortical surface of the bone and the adjacent muscle.
Data from the two rounds of experimentation were analyzed separately. Two-way analysis of variance (ANOVA) was used to compare across treatment groups and time points. Differences between individual treatment groups and time points were assessed using Fisher’s protected least significant difference test, while differences between loaded and non-loaded limbs were assessed using paired t-tests. In each case, p < 0.05 was considered significant.
In the first round of investigation, TNP-470 was used to inhibit angiogenesis following a stress fracture induced by mechanical loading. Both angiogenic (Vegf, Pecam1) and osteogenic (Bmp2, Osx, Bsp) genes were considered 1 hour, 1 day, and 3 days after loading for animals treated with TNP-470 and vehicle (Fig. 1). At 1 hour after loading, Bmp2, Pecam1 and Vegf were significantly upregulated in both TNP-470 and vehicle treated limbs compared to non-loaded limbs. Moreover, at 1 day and 3 days after loading, all five genes were significantly upregulated in both TNP-470 and vehicle treated loaded limbs compared to non-loaded limbs. However, in TNP-470 treated limbs the upregulation of Bsp, Osx, and Pecam1 was significantly less compared to vehicle treated limbs 3 days after loading, illustrating differential gene expression due to anti-angiogenic treatment.
Vasculature was perfused 3 days after loading and quantified using histology and microCT. Regardless of treatment, there were significant increases in vessel count and volume in loaded limbs compared to non-loaded limbs. Contrary to our expectation, treatment with TNP-470 did not decrease vascularity in loaded limbs compared to vehicle treatment by day 3 (not shown).
Bone formation was quantified using microCT and dynamic histomorphometry at day 7 (Fig. 2). Woven bone was evident in loaded limbs but absent in non-loaded limbs. In contrast to vascularity, woven bone extent (− 30%) and woven bone area (−60%) were significantly less in TNP-470 treated limbs compared to vehicle treated limbs. Because treatment-related reductions in bone formation and osteogenic and angiogenic gene expression were observed without measurable changes in vascularity at day 3, a second round of investigation was initiated with an additional vascular time point at day 7 and new outcome measurements.
In the second round of investigation, αvβ3 integrin targeted fumagillin nanoparticles were used to inhibit angiogenesis after osteogenic mechanical loading. Vasculature was perfused 3 and 7 days after loading and quantified using histology and microCT. Similar to treatment with TNP-470, nanoparticle treatment did not result in significant differences in vessel count or volume compared to vehicle treatment when evaluated 3 days after loading. However, 7 days after loading, nanoparticle treatment resulted in dramatic changes in vascularity with significantly fewer vessels (−30%) and significantly less vessel volume (−33%) compared to vehicle treatment (Fig. 3).
Bone formation was quantified using microCT imaging on day 7. Similar to TNP-470 treatment, treatment with fumagillin nanoparticles resulted in diminished bone formation, with significantly decreased woven bone extent (−30%) and less woven bone volume (−25%) in nanoparticle treated limbs compared to vehicle treated limbs (Fig. 4).
Immunohistochemistry was used to further assess vasculature by counting blood vessels stained positive for CD105 (a marker for angiogenic vessels) and αSMA (a marker for mature vessels). Loaded limbs had significantly more CD105 positive vessels than non-loaded limbs at day 3 (Fig. 5B). However, loaded limbs had significantly fewer (−50%) CD105 positive vessels in nanoparticle treated animals compared to vehicle treated animals at day 3. Similarly, loaded limbs had significantly more αSMA positive vessels than non-loaded limbs at day 7 (Fig. 5D). However, loaded limbs from nanoparticle treated animals had significantly fewer (−25%) αSMA positive vessels compared to vehicle at day 7. Thus, anti-angiogenic treatment resulted in fewer angiogenic vessels at day 3 which led to fewer mature vessels by day 7.
The goal of this study was to examine the effects of inhibiting angiogenesis after a non-displaced stress fracture generated by damaging mechanical loading. Two treatments were used to inhibit angiogenesis: TNP-470 and αvβ3 integrin targeted fumagillin nanoparticles. The results from vascular perfusion, microCT, dynamic histomorphometry, immunohistochemistry, and gene expression illustrate that anti-angiogenic treatment decreases vascularity, woven bone formation, and osteogenic and angiogenic gene expression at the site of stress fracture repair.
This study provides new insight into the timing of vascular changes during skeletal repair after a stress fracture from repetitive mechanical loading. A large increase in CD105 positive blood vessels was observed near the site of woven bone formation 3 days after loading. Because CD105 is a marker predominantly expressed in angiogenic endothelial cells , an increase in CD105 positive vessels should immediately precede an increase in vascularity. Indeed, vessel volume increased significantly in loaded limbs between days 3 and 7. Anti-angiogenic treatment almost completely blunted the loading-induced increase in angiogenic (CD105 positive) vessels on day 3, and subsequently diminished the number of mature (αSMA positive) vessels as well as the total vasculature (vessel count and volume) in loaded limbs on day 7. Since anti-angiogenic treatment does not result in significant differences in vascularity by day 3, our data indicate that limited angiogenesis occurs in the first 72 hours after loading. The gene expression data support this timeline. There were no treatment-related differences in gene expression in the first 24 hours after loading. However, Pecam1, Osx, and Bsp were significantly decreased in TNP-470 treated limbs on day 3, suggesting that anti-angiogenic treatment impaired angiogenic sprouting as well as bone formation capacity by day 3. These results indicate that significant vascular expansion from angiogenesis starts approximately 3 days after stress fracture.
The upregulation of angiogenic genes (Vegf, Pecam1) has been shown to immediately follow a stress fracture generated by damaging mechanical loading , and apparent vascularity is significantly greater in loaded limbs compared to non-loaded limbs 3 days after loading [12, 13]. Although the increase in apparent vascularity was originally taken as evidence of early angiogenesis, this study taken together with the previous data has lead to the hypothesis that the early response to damaging mechanical loading is the opening of dynamic vessel networks and marked vasodilation. These processes are independent from angiogenesis, which has been shown here to occur later, likely in response to the metabolic demand of increased bone mass. The process of inducing rapid patency of dormant vessels has been observed previously in studies of intrapulmonary arteriovenous anastomoses in humans and dogs [33, 34]. Although the exact mechanism for this phenomenon is not well understood, hypoxia may play a major role . In previous studies of stress fracture repair, hypoxia inducible factor 1α (Hif1α) was shown to be significantly increased as early as 1 hour after loading, and this significant increase persisted as long as 11 days after loading [12, 36]. These data indicate that the immediate response to a stress fracture generated by damaging mechanical loading is a local hypoxic environment, consistent with the proposed mechanism for rapidly opening dynamic vessel networks. However, Hif1α has also been shown to be regulated by mechanical stress [37–40], so more research is required to better understand this process, particularly in the context of loading-induced woven bone formation.
In this study, data from two angiogenic inhibition experiments was presented, one using TNP-470 and the other using αvβ3 integrin targeted fumagillin nanoparticles. Although these two anti-angiogenic treatments act through the same mechanism (methionine aminopeptidase 2 inhibition ), the compound is not identical (TNP-470 vs. fumagillin) and the dosage differs by two orders of magnitude (25 mg/kg vs. 0.25 mg/kg). Since αvβ3 antagonists alone have been shown to inhibit angiogenesis [23, 25], the αvβ3 integrin targeted nanoparticles may be preventing angiogenesis independent of fumagillin. Although the αvβ3 integrin is also essential for osteoclast function , the nanoparticles are constrained to the vascular system due to their size . The water soluble TNP-470 may be able to affect targets outside the blood supply, and has been shown to inhibit osteoclast resorption in vitro . However, osteoclasts do not appear in detectable numbers until the second week after stress fracture in this model [43, 44], so osteoclast-related effects are unlikely to have played a role in this study. Nonetheless, the lower drug dosage, additional mechanism of action, and vascular specificity make targeted nanoparticles preferable for future experimentation.
In previous studies, TNP-470 (30 mg/kg) completely prevented endochondral bone healing after closed femoral fracture in rats , as well as intramembranous bone formation occurring during gradual distraction osteogenesis . However, in the current study, angiogenic inhibition using either TNP-470 or targeted nanoparticles diminishes, but does not completely prevent, woven bone formation. The amount of woven bone formed after angiogenic inhibition was similar to the amount formed from less severe stress fractures (displacement in the range of 30% to 45% of fracture , rather than the 65% displacement used here). Although treatment likely did not completely eliminate angiogenic sprouting in the region of interest, the existing vascular system may have been able to support a limited amount of woven bone formation despite inhibited angiogenesis. This phenomenon has been indirectly observed in previous work, where no significant changes in vessel count were found despite limited woven bone formation occurring after 30% displacement to fracture . Therefore, these data suggest that angiogenesis is a response to the increased metabolic demand of newly formed woven bone, but may not be required for the initiation of woven bone formation. Additionally, the amount of angiogenesis driven by loading may have been underestimated in this model, since only periosteal vasculature was quantified. It is possible that intramedullary and intracortical angiogenesis may also be occurring, although given the small medullary cavity at the site of interest as well as the relatively early time points of this study (prior to intracortical resorption), any contribution from these sources is negligible. Nonetheless, these results clearly indicate that angiogenesis is necessary for a full osteogenic response to a stress fracture generated by damaging mechanical loading. Finally, the results of this study suggest that stress fracture healing could be accelerated by increasing vascular capacity or angiogenicity at the site of skeletal repair.
In summary, TNP-470 and αvβ3 targeted fumagillin nanoparticles were used to inhibit angiogenesis after a forelimb stress fracture from damaging mechanical loading in rats. There were no changes in vascularity 3 days after loading, but fewer angiogenic (CD105 positive) blood vessels were present at the site of bone formation in treated animals. Three days after loading, angiogenic (Pecam1) and osteogenic (Bsp, Osx) genes were less upregulated in treated limbs than vehicle limbs. Seven days after loading, treatment was associated with less vasculature (blood vessel count and volume) and less woven bone (woven bone extent, area, and volume). Taken together, these results demonstrate that vascular expansion by angiogenesis begins about 3 days after a stress fracture and is required for a full osteogenic response.
This study was funded by a grant from the National Institutes of Health (NIH R01 AR050211) and was performed at a facility supported by the Washington University Musculoskeletal Research Center (NIH P30 AR057235). In addition, we acknowledge additional nanomedicine research support from the NIH (HL113392, CA100623, CA154737, HL094470, AR056468, NS073457, CA136398) and the American Heart Association (0835426N and 11IRG5690011). Dr. Lanza is a scientific cofounder of Kereos, Inc, St. Louis, which has licensed angiogenesis-targeted perfluorocarbon nanotechnology intellectual property from Washington University/Barnes-Jewish Hospital for clinical development. The authors would like to thank Dr. Huiying Zhang for her assistance with immunohistochemistry and Ralph W. Fuhrhop for preparation of the nanoparticle formulations.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.