Endogenous FGF-2 levels impact FGF-2/BMP-2 growth factor delivery dosing in aged murine calvarial bone defects

Abstract

Bone repair in elderly mice has been shown to be improved or negatively impacted by supplementing the highly osteogenic bone morphogenetic protein-2 (BMP-2) with fibroblast growth factor-2 (FGF-2). To better predict the outcome of FGF-2 supplementation, we investigated whether endogenous levels of FGF-2 play a role in optimal dosing of FGF-2 for augmenting BMP-2 activity in elderly mice. In vivo calvarial bone defect studies in Fgf2 knockout mice with wildtype controls were conducted with the growth factors delivered in a highly localized manner from a biomimetic calcium phosphate/polyelectrolyte multilayer coating applied to a bone graft substitute. Endogenous FGF-2 levels were measured in old mice vs young and found to decrease with age. Optimal dosing for improving bone defect repair correlated with levels of endogenous FGF-2, with a larger dose of FGF-2 required to have a positive effect on bone healing in the Fgf2 knockout mice. The same dose in wildtype old mice, with higher levels of FGF-2, promoted chondrogenesis and increased osteoclast activity with less overall bone formation. The results suggest a personalized medicine approach, based on a knowledge of endogenous levels of FGF-2, should guide FGF-2 supplementation in order to avoid provoking excessive bone resorption and cartilage formation, both of which inhibited calvarial bone repair.

Introduction

Aging, as well as disease, leads to a reduction in bone healing capacity due to age-related changes in a myriad of processes including osteoprogenitor proliferation ^1,2 and the tightly regulated balance between osteoblast and osteoclast activity ^3,4. Extended bone healing time increases inactivity of an older person, compounding the negative effects of a bone injury ⁵. We previously demonstrated a powerful strategy for overcoming reduction in bone healing due to age is through the local delivery of growth factors from biomaterials implanted into the bone defect ⁶.

Growth factors can be adsorbed, embedded or chemically linked into materials for use in orthopedic or maxillocraniofacial repair and then implanted to deliver locally osteogenic growth factors directly into a bony defect ^7–14. We developed a growth factor delivery system in the form of a layered coating made of biomimetic calcium phosphate and polyelectrolyte multilayers that can be applied to any porous scaffold ¹⁵. Our growth factor delivery studies have focused on understanding the effects of the combined delivery of fibroblast growth factor-2 (FGF-2), also known as basic fibroblast growth factor (bFGF), and bone morphogenetic protein 2 (BMP-2) on aged bone repair ^6,16.

The effective doses of FGF-2 used alone or in combination with BMP-2 for bone repair have varied by orders of magnitude from nanograms to hundreds of micrograms ^16–21. Using low doses of FGF-2 to augment a powerful osteogenic growth factor like BMP-2, rather than simply increasing the dose of BMP-2 to improve bone healing, is a valuable approach because higher doses of BMP-2 are associated with undesirable outcomes such as ectopic mineralization, inflammation and resorption ^22,23. It’s not known, a priori, what dose of FGF-2 is required to optimize bone repair, although existing studies provide some starting points. In our experience more than five different animal studies were required to hone in the correct doses for a given age of mouse ^6,24 which was expensive and time consuming, particularly when studying responses in aged mice, which each cost upwards of 100x what a young mouse costs. Attrition in old mice colonies also complicates these studies. Although maintenance of FGF-2 activity within the scaffold is a fundamental requirement, that is only a first step in dose optimization. It is known that when a certain FGF-2 dosing threshold is exceeded, negative effects on bone repair processes are observed; e.g. over promotion of osteoclast formation ^25–27, increased bone resorption ²⁸ and inhibition of osteoblast differentiation ^29–31. In order to avoid multiple failed regenerative outcomes when supplementing BMP-2 with FGF-2, there is a critical need to better understand the variables at play that provide upper limits to dosing of FGF-2 in order to maximize bone repair efficacy.

This study set out to answer these questions - 1. Is it possible to administer a dose of FGF-2 combined with BMP-2 from a biomaterial coating on a bone graft substitute that will improve the impaired bone repair in mice genetically manipulated to have no FGF-2 expression? 2. Do endogenous levels of FGF-2 play an important role in dose selection of FGF-2 for augmenting BMP-2 in elderly mice? The questions were answered by in vivo calvarial bone defect studies in Fgf2 knockout mice, previously shown to have no gene expression for Fgf2 ³², as compared to normal wildtype mice. Localized growth factor delivery was accomplished by a biomaterial coating we have previously shown can deliver FGF-2 and BMP-2 to increase bone repair in young mice ^15,24. Given the significant medical need for improving bone healing in the elderly, the studies were conducted in mice that were aged sufficiently to represent a murine equivalent of a 75 year old human ³³.

Materials and Methods

Preparation of the disk scaffold for calvarial defect implantation

An overview of the method used to prepare the growth factor loaded disk scaffolds for implantation in the mouse calvarial defects is shown in Figure 1 A. A sheet of commercially available bone graft substitute, collagen/hydroxyapatite (Col-HA) (Healos^™, DePuy, Raynham, MA) was cut into 1 mm thick × 3.5 mm diameter disks and UV-sterilized for 10 min on each side. A biomimetic calcium phosphate/ polyelectrolyte multilayer (bCaP/PEM) coating which enhances drug delivery, as described by us previously ^15,24, was applied to the pre-cut disks as shown schematically in Fig. 1 A. The growth factors were applied both prior to the bCaP-PEM coating and after bCaP-PEM by adsorption to mimic the embedded nature of growth factors in bone. Lyophilized BMP-2 (E. coli expressed non-glycosylated recombinant human BMP-2) Biologics Corp (Cat# CY115) and carrier free FGF-2 (recombinant human FGF basic, 223-FB/CF) from R&D Systems, Minneapolis, MN were used for these studies. FGF-2, 12.5 ng in 1.25 μL, was mixed with 5 μg of BMP-2 (2.5 μL) to get a total of 3.75 μL volume and pipetted on the Col-HA prior to bCaP/PEM coating. For the BMP-2 only control group, the same volume of BMP-2 was used but diluted with additional saline. This relatively small volume of growth factor solution was selected because it wicked uniformly throughout the Col-HA disk scaffold without excess fluid loss. Growth factor adsorption was allowed to occur for 1 hr at room temperature with disks kept in a humid environment, prior to the deposition of the biomimetic bCaP-PEM coating.

Figure 1.

Overview of the biomimetic calcium phosphate/polyelectrolyte multilayer (bCaP/PEM) coating process. (A) Application of the various layers of the bCaP/PEM coating on the bone graft substitute (Col/HA) disks for mouse calvarial bone defect implantation. (B) Schematic cross-section of the coatings and growth factors tested in this study showing the location of growth factors on either side of the bCaP barrier layer.

A PEM layer at the outer surface allows for immediate release of the growth factors, while a bCaP layer serves as a barrier layer to prevent immediate release of the factors adsorbed directly to the base scaffold. Previous publications have confirmed the retention of FGF-2 and BMP-2 growth factor activity for this bCaP and PEM coating process ^15,17. A two-step bCaP coating process was followed²⁴ with a disk of bone graft substitute first immersed in 1 mL of a five times concentrated simulated body fluid (SBF) overnight at 37 °C with gentle rocking to form a uniform amorphous calcium phosphate layer. This initial amorphous layer of calcium phosphate encloses the growth factors and serves to stabilize and prevent their degradation during the second bCaP step which occurs at 50°C for 24 hrs. The second SBF solution had slightly varied ion concentrations that promoted the formation of a crystalline carbonated apatite layer ²⁴. After the two step bCaP process was complete, an exponentially growing polyelectrolyte multilayer (PEM) film was applied through a manual layer-by-layer approach with the disk scaffolds repetitively submerged in oppositely charged polyelectrolyte solutions followed by 3 saline rinses. Solution volumes were 300 μL of 1 mg/mL poly-L-Glutamic acid (PLGlut) or 1 mg/mL poly-L-Lysine (PLLys) in saline (Sigma, St. Louis, MO). Each bCaP coated scaffold received 10 min immersions in the polyelectrolyte solutions or saline rinse while in the upper compartment of centrifugal filter units (Ref# UFC30DV0S, Merck Milliopore Ltd). Removal of the solutions from the pores of the scaffolds was accomplished by centrifuging at 1000 rpms for 10 seconds after each adsorption or rinse step. The process was repeated over two days for a total of 21 bilayers which is sufficient to act as a growth factor reservoir³⁴. Lastly, a mixture of 12.5 ng of FGF-2 with 5 μg of BMP-2 in 3.75 uL,or BMP-2 only, was absorbed into the bCaP-PEM coated scaffolds and allowed to bind for 1 h at room temperature. Scaffolds were rinsed in 300 μl of cell culture medium (DMEM No. 11995, Gibco BRL, Invitrogen) without any additional supplements prior to implantation in the mouse bone calvarial defects. A schematic of the completed disk implants and their layered structure with embedded growth factors is shown in Figure 1 B.

To determine if there was a loss of the growth factor doses placed on the base scaffold during the two bCaP coating steps or while during rinsing in DMEM prior to implantation (for 3–5 hrs), ELISA testing was done on the various solutions the scaffolds were immersed in, i.e. the first and second SBF solutions and also from the DMEM holding solution. BMP-2 and FGF-2 enzyme-linked immunosorbent assays were performed using commercially available kits (BMP-2 ELISA Kit, Antigenix America, Huntington Station, NY and FGF-2 R&D Systems, Minneapolis, MN RHF913CKX2). It was found that 70% of the FGF-2 and 80% of BMP-2 remained adsorbed within the coated scaffold. Specifically, for the FGF-2 and BMP-2 combination, 61% of FGF-2 and 67% of BMP-2 remained in the inner layer, and 84% of FGF-2 and 97% of BMP-2 remained in the outer layer bCaP/PEM deposition and storage. Based on the known losses during processing, growth factor solutions were increased to achieve the desired doses on the calvarial bone implants: 18 ng FGF-2 and 8.2 μg BMP-2 or 8.2 ug of BMP-2 only.

In vitro growth factor release studies

FGF-2 and BMP-2 release from the bCaP-PEM coated scaffolds was determined by an in vitro release study conducted over 10 days. The scaffolds with coatings and growth factors were prepared as described above and then placed in 2 ml of cell culture medium (DMEM No. 11995, Gibco BRL, Invitrogen) and incubated at 37°C. Samples were collected every 24 hrs with complete replacement of the medium. The samples were frozen immediately and only thawed once prior to analysis to avoid degradation. Released BMP-2 and FGF-2 in the supernatants were analyzed with ELISA as above. Each assay was performed with strict adherence to the manufacturer’s instructions.

mRNA isolation and quantitative PCR analysis to assess endogenous levels of Fgf2

Generation of the Fgf2 knockout mice (Fgf2−/−) was previously described in detail including proving the lack of FGF-2, a bone phenotype, and impaired bone formation compared with their wild type littermates (Fgf2+/+) ^35–37. For these studies, male and female heterozygote female Fgf2+/− were bred in-house and female offspring were aged for 18–33 months to generate the group of old Fgf2−/− knockouts and Fgf2+/+ wildtype (WT) female mice. These mice are considered aged equivalently to a ≥ 75 year old human according to the National Institutes of Health (NIH) - National Institute on Aging (NIA) Rodent Colony website ³⁸ and Jackson Laboratories website ³⁹. Mice aged 4–6 months are young adult mice. Total RNA was extracted from tibiae harvested from 3-month-old and 24-month-old Fgf2+/+ mice (N=3). For real-time quantitative reverse transcription PCR analysis, the RNA to cDNA EcoDry Premix Kit (Clontech Inc., Takara Bio, Mountain View, CA) was used to reverse transcribe the RNA to cDNA. iTaq Universal SYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA) and a MyiQ instrument (Bio-Rad Laboratories) were used for quantitative PCR (qPCR). The relative change in mRNA level was normalized to the mRNA level of β-actin, which served as an internal reference for each sample. The primer sequences for the genes of interest are shown in Table 1.

Table 1.

Primers used in quantitative Real Time-PCR.

Gene	Forward	Reverse
Fgf2	5’-GTCACGGAAATACTCCAGTTGGT-3’	5’-CCCGTTTTGGATCCGAGTTT-3’
β-actin	5’-ATGGCTGGGGTGTTGAAGGT-3’	5’-ATCTGGCACCACACCTTCTACAA-3’

Western blot analysis

Protein was extracted from whole tibiae harvested from 4-month-old and 20-month-old Fgf2+/+ WT female mice. Protein extracts were harvested in RIPA buffer (Cell Signaling, MA, USA), supplemented with protease inhibitors 1mM PMSF + 1X Protease inhibitor cocktail (Cell Signaling, MA, USA). Protein concentration was assayed with bicinchoninic acid assay protein assay reagent (Thermo Fisher Scientific). Equal amounts of protein were fractioned by SDS-PAGE (Mini-Protean ^® TGXTM Gel, Bio-Rad, CA, USA) and transferred onto a PVDF membrane (Bio-Rad, CA, USA). To perform immunoblotting, membranes were blocked for one hour with 5% non-fat dry milk, and then incubated overnight at 4°C with anti-FGF2 (1:1000, BD Biosciences, CA, USA). Membranes were then incubated with appropriate secondary antibody (Amersham Bioscience, NJ, USA) at room temperature for 1 h. Blots were developed with Supersignal West Dura Extended Duration Substrate (Thermo Scientific). Finally, blots were re-probed with Actin antibody (Santa Cruz, CA, USA). Quantification by densitometric analysis of digitized autoradiograms with NIH Image J software was performed.

Mouse Calvarial Defect Model

The study protocol was approved by the Animal Care and Use Committee at the University of Connecticut Health prior to initiating the studies. NIH guidelines for the care and use of laboratory animals (NIH Publication #85–23 Rev. 1985) have been observed. Each Fgf2+/+ or Fgf2−/− mouse was anaesthesized with isoflurane and received a single 3.5 mm diameter critical-sized calvarial bone defect in the center of the right parietal bone, while avoiding the cranial suture which can inhibit bone formation. The disk scaffold was placed in the calvarial defect, and then the periosteum and soft tissues were closed by interrupted sutures (5–0 Vicryl sutures, Ethicon, Somerville, NJ). Experimental procedures were conducted following an Institutional Animal Care and Use Committee (IACUC) approved protocol. There were three groups in the study: control scaffolds without growth factors, BMP-2 only or FGF-2 + BMP-2. Due to the unanticipated high rate of attrition of these transgenic mice as they were aged, the number of old Fgf2−/− and Fgf2+/+ mice per group were lower than planned, but were still sufficient to detect statistically significant differences. For the BMP-2 only groups there were three mice Fgf2+/+ mice and five Fgf2−/− mice. There were four mice in the negative control groups and four mice in the FGF-2 + BMP-2 treatment groups for both Fgf2+/+ and Fgf2−/−. Mice were given 0.08 mg/kg buprenorphine via intramuscular injection for pain management every 8–14 hrs apart for 72 hrs post-surgery. All mice received peritoneal cavity injections of 3 μg/g calcein and alizarin complexone calcium labels on 7 day and 1 day, respectively, prior to euthanasia at 4 wks to label active mineralization. Mice were euthanized with carbon dioxide and calvarial bones were harvested for non-decalcified tissue analysis.

Micro-Computed Tomography and Histological Analysis

Mouse calvarial bones were fixed in 10% neutral buffered formalin solution for 24 hours at 4 °C prior to conebeam micro-focus X-ray computed tomography (micro-CT). The volume of new mineralized bone within the calvarial bone defect sites was quantified using with a μCT40 (Scanco Medical AG, Bassersdorf, Switzerland) as described previously ⁶. Serial tomographic images were acquired at 55 kV and 145 μA, collecting 1000 projections per rotation at 300 ms integration time on fixed tissue samples prior to decalcification for histology. Three-dimensional 16- bit grayscale images were reconstructed using standard convolution back-projection algorithms with Shepp and Logan filtering, and rendered within a 16.4 mm field of view at a discrete density of 244,141 voxels/mm3 (isometric 16-μm voxels). Segmentation of bone from scaffold and marrow and soft tissue was performed in conjunction with a constrained Gaussian filter to reduce noise.

Calvarial bones were sectioned through the center of the defect and prepared for cryohistological analysis as described previously ²⁴. After a section was imaged for native fluorescent signals (scaffold and bone mineral stains), the cover slide was removed and then the same section was processed for additional staining. The sections were stained for tartrate-resistant acid phosphatase staining (TRAP) activity using ELF97 (Life Tech, E6589) as the fluorescent substrate (yellow) and imaged. The same sections were next stained for alkaline phosphatase (AP) activity using fast red substrate (Sigma, #F8764-5G) and DAPI (Mol Probes #D-1306) and re-imaged. Toluidine Blue was the last stain applied (Toluidine Blue, Sigma, #T3260). Imaging of the sections was performed with the AxioImager Z1 microscope (Carl Zeiss, Thornwood, NY). Fluorescence images of the entire scaffold area were quantitatively analyzed for “% area of expression” of Calcein, Alizarin Complexone, ALP and TRAP by using the program ImageJ (www.imagej.nih.gov).

Statistical Analysis

Statistical analysis was performed using one-way ANOVA with a Bonferroni post-test in GraphPad Prism to conduct multiple comparisons and significance set at p ≤0.05. Data was plotted as mean ± standard deviation, except as indicated for the ELISA testing which was mean ± standard error.

Results

Age associated changes in endogenous Fgf2 expression

By quantitative RT-PCR analysis, a significant (p<0.05) age-related decrease in Fgf2 mRNA was found between extracted total RNA from whole tibiae of young (3 month) and the aged (24 month) WT female mice (Figure 2 A). Western blotting demonstrated a significant age-related decrease in FGF-2 protein in the tibiae of old WT mice compared to young mice (Figure 2 B). It was unknown if endogenous levels of FGF-2 varied between young and old WT mice and this new finding supports the strategy of supplementing old mice with FGF-2 as a means to improve bone repair outcomes.

Figure 2.

Age associated changes in endogenous FGF-2 in whole tibiae harvested from young or aged Fgf2+/+ WT female mice (N=3). (A) Fgf2 gene expression. (B) FGF-2 protein levels.

Calvarial bone defect repair in Fgf2−/− knockout mice vs. wildtype Fgf2+/+ mice

Dosing aged Fgf2−/− knockout mice with 18 ng of FGF-2 and 8.2 μg of BMP-2 delivered by the bCaP/PEM coating on a bone graft substitute evoked a positive effect on calvarial bone defect repair. As seen from the micro-CT images and quantified data of the Fgf2−/− samples, new bone volume was increased significantly from FGF-2 and BMP-2 relative to the BMP-2 only group (p = 0.0485) (Figure 3 A, Figure 3 B). In the wildtype mice, Fgf2+/+, the addition of FGF-2 did not increase new bone formation relative to BMP-2 alone (Figure 3 C, Figure 3 D). A negligible amount of new bone formation seen in the negative control bCaP-PEM coated scaffolds without growth factors in both types of mice: Fgf2−/− and Fgf2+/+ mice (Figure 3 A and Figure 3 C, p < 0.0001).

Figure 3.

Bone defect repair quantified by microCT analysis. (A) Representative 3-dimensional micro-CT reconstructions (top row) and 2-dimensional micro-CT scans (bottom row) of Fgf2−/−mouse skulls with calvarial defects harvested after four weeks. (B) Quantification of new bone volume in Fgf2−/− mice. (C) Micro-CT images for the Fgf2+/+ mice with 3-dimensional micro-CT reconstructions (top row) and 2-dimensional micro-CT scans (bottom row). (D) Quantification of new bone volume in Fgf2+/+ mice. Mean ± SD. *p = 0.0485, **p = 0.0066, ***p = 0.0006 and ****p < 0.0001.

Histological evaluation of FGF-2 effects on calvarial bone defects treated with BMP-2

In the toluidine blue (TB) stained histological sections, bone formation was observed throughout the defects in the growth factor treated groups of the Fgf2−/− mice, but not the control scaffold only groups (Figure 4). The lack of bone formation in the scaffold only groups confirms the critical sized nature of the bone defect and the lack of osteoinductivity of the scaffold. In the Fgf2+/+ group, new bone formation was visualized in the BMP-2 only group with less in the FGF-2+BMP-2 group. Interestingly, in Fgf2+/+ mice treated with FGF-2+BMP-2, chondrocytes, surrounded by the TB-stained polysaccharide matrix, were present in the bone defect near the original defect edges (Figure 4 B). The images at highest magnification indicate the morphology of the chondrocytes is consistent with that of hypertrophic chondrocytes. Chondrocytes were not seen in response to BMP-2 only treatment.

Figure 4.

Light micrographs of toluidine blue stained sections of the mouse calvarial bone defects at 4 weeks. (A) Fgf2−/− mice. (B) Fgf2+/+ mice. New bone formation marked with yellow asterics or the letter B. Areas marked with C =cartilage. Black arrows indicate the edges of the original bone defect. The images in the second row of each panel are higher magnification images of the black boxed areas in the top row. A further magnified image is shown at the end of each row to more clearly see the bone and cartilage formed in the defect. Scale bars = 200 μm.

Timed administration of calcein and alizarin complexone, was used to assess differences in active calcification at the bone osteoid interface of old mice, with and without endogenous FGF-2, or FGF-2 supplementation. In the BMP-2 and FGF-2+BMP-2 treatment groups of both Fgf2−/− and Fgf2+/+ mice, discrete layers of calcein labeling (green) and alizarin labeling (red) were observed, indicating that bone mineralization was actively in process at 7 days prior to euthanasia and at 1 day prior to euthanasia (Figure 5 A and C). This was not observed in the scaffold only group. The Fgf2−/− mice had significantly increased area of calcein and alizarin staining in the FGF-2+BMP-2 treated mice as compared to BMP-2 only indicating more active bone formation in the group supplemented with FGF-2 (Figure 5 B and D). This correlates with the increased bone volume observed by micro-CT analysis. In the Fgf2+/+ mice there were no significant differences for calcein and alizarin staining in response to FGF-2+BMP-2 vs BMP-2 only. Addition of FGF-2 to the Fgf2−/− mice increased the amount of alizarin and calcein staining to a comparable level to the WT Fgf2+/+ mice treated with BMP-2 only.

Figure 5.

Mineral labelling study results. (A) Fgf2−/− images and (B) quantified data, mean ± SD. # = p < 0.0001 for calcein FGF-2 and BMP-2 relative to BMP-2 only. $ = p = 0.0181 for alizarin FGF-2+BMP-2 relative to BMP-2 only. * = p<0.05 for calcein vs alizarin. (C) Fgf2+/+ images and (D) quantified data, mean ± SD. No statistical difference between the BMP-2 and the FGF-2+BMP-2 groups. All BMP-2 or FGF-2 and BMP-2 groups in both Fgf2+/+ and Fgf2−/− were individually significantly higher (p< 0.05) than the scaffold only. White arrows in the images indicate the edges of the original bone defect. The images in the second row of each panel are higher magnifications of the white boxed areas in the top row. Scale bar = 200 μm.

To further verify if a dose of FGF-2 capable of increasing bone regeneration in the Fgf2−/− mice had been delivered, the bone formation and bone resorption activity within the bone defect areas were analyzed by ALP and TRAP staining (Figure 6 and Figure 7, respectively, with quantification in Figure 8). Positive ALP staining indicates the presence of osteoblasts actively depositing bone matrix and was enhanced relative to the osteoclast-associated TRAP staining in the FGF-2 treated Fgf2−/− mice (Figure 6 and 8 A). In the Fgf2+/+ group there was a non-significant increase in ALP staining with FGF-2 administration (Figure 6 and 8 B) relative to BMP-2 only, but any benefit of this increased ALP staining was offset by the significantly increased TRAP staining (p = 0.0041). Increased bone resorption in the FGF-2 supplemented Fgf2+/+ group its increase relative to the BMP-2 only group led to a larger ratio of TRAP to ALP, indicating that balanced bone formation/resorption was disturbed by an excessive FGF-2. The result was less total new bone formation (Fig. 3D). Little to no TRAP activity was observed in the scaffold only controls in either mouse type indicating the scaffold itself did not stimulate resorption.

Figure 6.

Images of alkaline phosphatase stained calvarial bone defect sections. (A) Fgf2−/− mice and (B) Fgf2+/+ mice. The ALP stain is pink and nuclei are stained with blue DAPI. White arrows in the images indicate the edges of the original bone defect. The images in the second rows of both (A) and (B) are higher magnification images of the white boxed areas in the top row. Scale bar = 200 μm. Quantified data for ALP expression in (C) Fgf2−/− mice and (D) Fgf2+/+.

Figure 7.

Images of tartrate-resistant acid phosphatase (TRAP) stained sections of calvarial bone after 4 weeks. The TRAP stain is yellow. (A) Fgf2−/− mice and (B) Fgf2+/+ mice. White arrows in the images indicate the edges of the original bone defect. The images in the second row of both (A) and (B) are higher magnification images of the white boxed areas in the top row. Scale bar = 200 μm. Quantified data in (C) for Fgf2−/− mice and (D) for Fgf2+/+ mice.

Figure 8.

ALP/TRAP staining ratios for (A) Fgf2−/− mice and (B) Fgf2+/+ mice. Percent positive area relative to total defect area shown for each stain, in stacked bar graph form, with ALP in pink and TRAP in yellow. Mean ± SD. ** = p = 0.0041 for TRAP staining only, not ALP.

In vitro FGF-2 and BMP-2 release from the bCaP/PEM coated scaffold

The cumulative in vitro release of FGF-2 from the coated scaffold, without cells present, was approximately 1% of the 18 ng dose applied (Figure 9 A). The concentration of FGF-2 measured in the supernatant was highest on day 1, then reduced to an undetectable level after day 3 (Figure 9 B). BMP-2 release, loaded at 8.2 μg, had a sustained release profile and reached a maximum of 80% at 10 days (Figure 9 A). The concentration of BMP-2 in the release supernatant was highest on day 1 (2 μg/ml) and then reduced to a steady 1 μg/ml each day between day 2 and day 5, and then further reduced to a low, but detectable level, for the remaining 5 days (Figure 9 B).

Figure 9.

In vitro BMP-2 and FGF-2 release into medium without cells from the bCaP-PEM coated scaffold. (A) Release data expressed as % cumulative release. (B) The same data expressed as concentration of growth factors in release media as a function of time. The values are presented as mean ± standard deviation (errors bars are small and within the data point markers).

Discussion

This study shows that it is possible to administer a low dose of FGF-2 by a biomaterial to improve the efficacy of BMP-2 in bone defect repair of aged Fgf2 knockout mice. The knockout mice are depleted of FGF-2, an essential growth factor involved in bone formation and repair ³⁵. By delivering FGF-2/BMP-2 from a bCaP/PEM coating applied to a Col/HA porous scaffold implanted into the calvarial bone defect, a low dose of FGF-2 (18 ng), delivered in combination with BMP-2 (8.2 μg), increased new bone volume in the Fgf2−/− mice (Figure 3). Calcium labelling studies further verified the microCT results of increased mineralization within the calvarial defects treated with FGF-2 and BMP-2 (Figure 5 B,D). Balanced osteoblast/osteoclast activity as measured by ALP/TRAP staining was observed in the Fgf2−/− mice (Figure 8A). This result supports previous in vitro experiments wherein bone marrow stromal cell cultures from long bones of young and old Fgf2−/− mice had decreased osteogenic potential that was rescued by treatment with FGF-2 as measured by increase ALP colony formation and ALP activity⁴⁰.

To answer our second question if endogenous FGF-2 levels play a role in dose selection of FGF-2 for augmenting BMP-2 in elderly mice, an animal model with groups of different levels of endogenous FGF-2 was required. Previous studies had already verified the lack of FGF-2 in Fgf2−/− knockout mice³⁵, but the levels in the aged Fgf2+/+ wildtype mice used in this study were unknown. We previously observed an age-related decrease of FGF-2 levels in human mesenchyme derived progenitors cells from hand bones⁴¹ and we hoped the murine model would echo that age-related decline. We found that aged Fgf-2+/+ wildtype mice had decreased FGF-2 levels relative to young mice (Figure 2), a finding which broadly supports the strategy of supplementing older mice with FGF-2 as a way to rejuvenate bone healing. There is no FGF-2 in the knockout mice so the comparison between wildtype and knockout was another way of testing the impact of endogenous levels on FGF-2 supplementation to bone repair by BMP-2. The evidence for a dependence on FGF-2 endogenous levels was that the FGF-2/BMP-2 dose that enhanced calvarial defect healing in the aged Fgf2−/− knockout mice failed to increase new bone formation in the aged Fgf2+/+ wildtype mice (micro-CT analysis, Figure 3). FGF-2 addition to the Fgf2+/+ mice did not boost osteoblast activity relative to BMP-2 as measured by mineral labelling (Figure 5). The dose of 18 ng dose of FGF-2 was not only ineffectual, but actually excessive in aged Fgf2+/+ mice, as seen by the increased osteoclast activity (TRAP staining, Figure 7 and 8 B) stimulated by the administered FGF-2. Previous literature supports the role of FGF-2 in enhancing formation of osteoclasts in vitro through increased RANKL ^25,42.

Our previous publications have reported the inhibition of bone formation by adding a low dose of FGF-2 to BMP-2 in young mice, but not old ⁶. The FGF-2 dose in that study was 3x lower than the dose used here and had a positive effect on calvarial defect repair in old wildtype BALB/c mice. Inhibition of in vitro osteogenesis in young, but not old, osteoprogenitor cultures due to FGF-2 addition with BMP-2 was also observed in our previous studies using young and old murine calvarial and long bone osteoprogenitor cultures ¹⁶. The new data in the present study regarding the reduction of FGF-2 expression in bones of old mice relative to young mice helps to explain those results. Interestingly, the innovation of a biomaterial coating capable of providing sequential, staged release of a low dose of FGF-2 followed by BMP-2 was able to improve bone formation in young mice ^16,17. Timing and duration of FGF-2 delivery is another important variable to consider in therapeutic applications of FGF-2.

Additional evidence of a relationship between endogenous FGF-2 levels and bone repair in response to FGF-2 therapy has been reported²⁸. Lau et al. reported that systemic FGF-2 gene therapy administration of cells genetically manipulated prior to implantation in mice had distinctly negative effects on only the caudal vertebra ²⁸. They stated that a limited number of MSCs able to participate in the repair was part of the problem and showed that Fgf2 mRNA was higher in caudal vertebra where FGF-2 gene therapy was not effective. The present study expands on that finding to applicability in a calvarial defect model in aged mice and use of a biomaterial to accomplish FGF-2 therapy. A shortcoming of our work was the low number of animals in each group associated with unexpected attrition of the mouse colony during the aging process exacerbated by the genetic manipulation. Despite this, the effects of the FGF-2 dose delivered in this study were distinctive enough to be observed with statistical significance to answer the study questions.

The unexpected stimulation of chondrogenesis within the bone defect of wildtype Fgf2+/+ mice capable of producing their own FGF-2 (Figure 4) is likely explained by the mitogenic role of FGF-2. FGF-2 has been previously shown to enhance in vitro chondrogenesis of mesenchymal stem cells expanded in medium supplemented with FGF-2 ^43–45. Our data demonstrates that a high FGF-2 dose delivered locally in vivo will also promote chondrogenesis of endogenous MSCs participating in bone defect repair. This outcome was undesirable in the Fgf2+/+ mice because it caused a reduction of new bone formed at the 4 wk time point relative to BMP-2 only mice that did not receive FGF-2. The overall goal of our research is to accelerate bone regeneration in elderly as rapidly as possible because extended periods of inactivity are detrimental for the elderly person’s health, ⁵ and FGF-2 induced chondrogenesis delayed bone repair. Alternatively, inciting endochondral ossification with supplementation of FGF-2 could be an effective strategy for osteochondral repair involving both cartilage and bone.

The low level of FGF-2 release into the supernatant seen in the in vitro release studies (Figure 9) combined with a demonstrated activity of this scaffold in vivo mimics how growth factors, despite being extracellular matrix-bound, can influence regenerative outcomes. The low level of release does not imply a lack of growth factor activity based on demonstrated activity in the present studies. An even lower dose of FGF-2 was shown to improve in vivo calvarial defect studies in young mice in previous studies with the same bCaP/PEM scaffold, but required the two growth factors to be separated so that FGF-2 delivery in PEM occurred first, followed by BMP-2 located below the bCaP/PEM ²⁴. That study confirmed that despite no detectable release of FGF-2, a proliferative response by osteoprogenitors coming in direct or close contact with the FGF-2 coated scaffold occurred ¹³. In vitro studies conducted earlier also reported increased osteoprogenitor proliferation over six days when they were cultured directly on bCaP/PEM coatings loaded with FGF-2⁴⁶. Since FGF-2 degrades quickly⁴⁷, the matrix-bound approach is a valuable strategy to maximize effects from low doses and to reduce possible side effects from widely dispersing growth factors⁴⁸.

In summary, this study shows how bone repair in aged Fgf2−/− knockout mice can be enhanced by delivering FGF-2 by a biomaterial delivery system in a sustained, matrix-bound manner from a bCaP/PEM coating applied to a porous bone graft. The efficacy of BMP-2 was enhanced by the addition of FGF-2 in the aged Fgf2−/− knockout mice. This study is also the first report of the decline of FGF-2 levels with age in the normal wildtype Fgf2+/+ mice which supports a future clinical strategy of supplementing BMP-2 with a low dose of FGF-2 to accelerate bone healing in elderly patients. Most importantly, these studies show that endogenous levels of FGF-2 play an important role in dose selection of FGF-2. Endogenous levels of FGF-2 put constraints on the maximum dose that can be used therapeutically in order to avoid triggering excessive osteoclast activity or chondrogenesis that will impair calvarial bone defect healing.