Evaluation of Osteogenic Cell Differentiation in Response to Bone Morphogenetic Protein or Demineralized Bone Matrix in a Critical Sized Defect Model Using GFP Reporter Mice Farhang Alaee,1 Seung-Hyun Hong,2 Alex G. Dukas,1 Michael J. Pensak,1 David W. Rowe,3 Jay R. Lieberman4 1 Department of Orthopaedic Surgery, New England Musculoskeletal Institute, University of Connecticut Health Center, Farmington, Connecticut 06030, 2Computer Science and Engineering Department, University of Connecticut, Storrs, Connecticut 06269, 3Department of Reconstructive Sciences, School of Dental Medicine, University of Connecticut Health Center, Farmington, Connecticut 06030, 4Department of Orthopaedic Surgery, Keck School of Medicine at USC, Los Angeles, California 90033
Received 16 December 2013; accepted 8 May 2014 Published online 2 June 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jor.22657
ABSTRACT: We evaluated the osteoprogenitor response to rhBMP-2 and DBM in a transgenic mouse critical sized defect. The mice expressed Col3.6GFPtopaz (a pre-osteoblastic marker), Col2.3GFPemerald (an osteoblastic marker) and a-smooth muscle actin (a-SMACherry, a pericyte/myofibroblast marker). We assessed defect healing at various time points using radiographs, frozen, and conventional histologic analyses. GFP signal in regions of interest corresponding to the areas of new bone formation was quantified using a novel computer assisted algorithm. All defects treated with rhBMP-2 healed. In contrast, the majority of the defects in the DBM (27/30) and control (28/30) groups did not heal. Quantitation of pre-osteoblasts demonstrated a maximal response (% GFPþ cells/ TV) in the Col3.6GFPtopaz mice at day 7 (7.2% 6.0, p < 0.05 compared to days 14, 21, 28, and 56). The maximal response of the Col2.3GFP cells was seen at days 14 (8.04% 5.0) and 21 (8.31% 4.32), p < 0.05. In contrast, DBM and control groups showed a limited osteogenic response at all time points. In conclusion, we demonstrated that the BMP and DBM induce vastly different osteogenic responses which should influence their clinical application as bone graft substitutes. ß 2014 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 32:1120–1128, 2014. Keywords: femoral defect; rhBMP-2; demineralized bone matrix; bone histomorphometry; GFP reporters
The development of bone repair strategies that are consistently successful requires an understanding of the influence of graft materials on host cells. Orthopaedic surgeons have been using bone graft substitutes such as recombinant human bone morphogenetic protein (rhBMP) and demineralized bone matrix (DBM) to enhance bone repair. RhBMP-2 has been proven to be an osteoinductive agent in both preclinical models1–5 and clinical trials.6,7 DBM is derived from human cadaver bone and contains several BMPs.8,9 DBMs are osteoconductive and have demonstrated osteoinductive activity in animal models, but osteoinductive activity has not been definitively demonstrated in humans.10–12 There is confusion amongst surgeons regarding the biological activity of the rhBMP-2 compared to DBM, because there is varying concentrations of BMPs in DBM. Our goal is to elucidate the interaction of these graft materials and host cells in a critical-sized defect to provide data that can be used to hopefully provide guidance with respect to the appropriate use of these graft materials in clinical situations. We used novel transgenic mice that contain GFP reporter constructs that facilitate the identification of the host response to these grafts. Novel GFP reporter transgenic mice have been developed to assess the timing of osteoblast differentiation. We used three reporters, based on Col1a1 (Col3.6GFPtopaz and Col2.3eGFP) promoter frag-
Conflict of interest: None. Grant sponsor: Musculoskeletal Transplant Foundation. Correspondence to: Jay R. Lieberman (T: þ1-323-442-8117; F: þ1-323-442-7621; E-mail: [email protected]) # 2014 Orthopaedic Research Society. Published by Wiley Periodicals, Inc.
1120
JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
ments and a-Smooth Muscle Actin (a-SMACherry). These reporters have facilitated the tracking of the differentiation of osteoblast lineage cells.13–17 In vitro and in vivo experiments have established that strong Col3.6GFP signal is found in newly formed osteoblasts14,18 while the Col2.3GFP marker is expressed in differentiated osteoblasts and osteocytes.14,17 Ushiku et al.19 used a double transgenic mouse (Col3.6-blue and osteocalcin-GFP-topaz) in a closed tibial fracture model to study the recruitment of the osteoprogenitors from periosteum. The a-SMACherry indicates cells with a pericyte/myofibroblast-like phenotype that have the potential to differentiate into functional osteoblasts.16 The progression of the a-SMA positive progenitor cells to mature osteoblasts during fracture healing has been documented.20 The purpose of our study was to evaluate the osteoprogenitor response to rhBMP-2 and DBM and evaluate the pattern of cell activation in a mouse critical sized femoral defect model. We hypothesized that there were critical differences in progenitor cell response between rhBMP-2 and DBM that should influence the clinical application of these agents. We used a novel computerized algorithm to quantitate the osteogenic response to the rhBMP-2 and DBM.
MATERIALS AND METHODS Animal Studies All animal studies were performed after approval by the Animal Care Committee in our institution. Critical sized femoral defects (2 mm) were created in 14- to 16-week-old male BL/6 mice according to a previously published method.21 We used non-transgenic C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME) for conventional histologic evaluation or lineage-specific transgenic mice expressing
EVALUATION OF OSTEOGENIC RESPONSE TO BMP-2 AND DBM
Col3.6GFPtopaz, Col2.3eGFP and a-SMACherry for preparation of frozen embedded samples to evaluate tissue fluorescence. The following groups were included in the study: Group I, rhBMP-2, 5 mg on a piece of collagen sponge (Medtronic, Minneapolis, MN), Group II, DBM (Musculoskeletal Transplant Foundation, Edison, NJ) and Group III, empty defect (control group). There were six mice per transgene per treatment per time point. Twenty-four hours before the sacrifice either xylenol orange (XO red label, to Col2.3eGFP and Col 3.6GFPtpz mice), or calcein (Green label, to a-SMA-Cherry mice) were injected intraperitoneally to mark the areas of the new bone formation while providing the color contrast with the native fluorescent cells. The Col2.3eGFP and Col3.6GFPtpz mice were sacrificed at various time points after surgery (4, 7, 14, 21, 28, and 56 days). The a-SMA-Cherry mice were sacrificed at 4, 7, and 14 days after surgery because preliminary data (not shown) demonstrated no activity beyond the 14 days time point. Radiographic Evaluation Radiographs at the 4- and 8-week time points were assessed by three independent observers and the defect healing was scored from 0 to 5 based on a previously published protocol which had been originally developed to evaluate bone repair in a rat critical-sized defect model.22 Briefly, a score of 0 was given if there was no healing, 0–25% healing was given 1, 25–50% was 2, 50–75% was 3, 75–99% was 4 and complete healing was 5. A kappa statistic was calculated to assess the inter-observer variability among the three observers. Histologic Evaluation At the time of sacrifice the operated leg was harvested and the external fixator was removed. The samples were assessed using two different histologic protocols including regular (decalcified, non-frozen) or fluorescent (non-decalcified, frozen) sections. Non-Decalcified Frozen Histology A previously established protocol was used to perform the frozen sections.23 After initial fluorescent imaging, selected samples were stained with a modified tartrate-resistant acid phosphatase (TRAP) stain and the fluorescent substrate ELF-97 that gave a distinct yellow fluorescence to the osteoclasts.24 Decalcified Standard Histology Formalin fixation at room temperature for 1 day, and decalcification in 10% EDTA for 2 weeks were performed. Sagittal sections through the defect area were mounted on glass slides and were stored at room temperature until use. Standard protocols were used for both H&E and conventional TRAP staining of the samples.22,25 Histomorphometric analysis of the TRAP stained sections (the number of osteoclasts per square millimeter of bone) was performed on an Olympus system (Olympus, Melville, NY) using the histomorphometric analysis software (Osteomeasure, Osteometrics, Decatur, GA) as previously described.26,27 Four rectangular areas were selected at each corner of the defect and the average number of osteoclasts per slide was used in the Statistical Analysis Section. Fluorescent Imaging Photographs of the frozen sectioned samples were taken using a Zeiss Imager Z1 microscope (Carl Zeiss, Thornwood,
1121
NY) and Axio Vision Rel.4.7 software (Carl Zeiss). Specific fluorescent filters (Chroma Technology, Bellows Falls, VT), based on a previously published protocol19,23 were used to obtain separate images that were subsequently merged to produce a final image. A differential interference contrast (DIC) image was acquired at the same time as the endogenous fluorescence imaging for final recompilation. A scanned composite image is generated which is created after combining the individual grayscale images from each channel. Histomorphometric Analysis of the GFP Expression Regions of interest (ROI) layers were created on top of the scanned composite images using Adobe Photoshop software. In the BMP-2 treated defects with bridging bone 3 distinct regions of interest (ROIs) were marked. ROI 1 represented the defect area, ROI 2 and 3 represented the superior and inferior bridge, respectively. Analysis of the bridge thickness and GFP signal density was determined on the aggregate data from ROI 2 and 3. Since there was no bony bridge formation in the DBM and control animals, the region of interest only included the defect area (ROI1). Within each ROI, the volume ratio of GFP positive cells over the total bone volume (GFPþ cell/TV) was measured and calculated using computer-automated bone histomorphometric techniques (automatic segmentation and automatic morphological transformation) introduced by Hong et al.28 and Parfitt et al.29 Segmentation of the signal was performed by Otsu’s thresholding method30 for bone and GFP signals separately. Otsu’s method is based on the histogram of the gray intensities of the signal. This method provides a threshold that separates grayscale pixels into black and white signal groups. It finds an optimum threshold intensity value that minimizes the standard deviation of the pixels that fall into either black or white segmented groups, while maximizing the inter standard deviation between the two groups. The white pixels represent the signal that was intended to be captured in the grayscale image. The volume ratio is a ratio of the total GFP positive cell volume (area occupied by GFP positive cells within a given ROI) over the total volume (area of ROI). Area is represented by the number of pixels in each segmented image. To determine the area of the GFP positive cells, the computer algorithm written in Matlab software simply counts the number of white pixels in the segmented (black and white) image that is obtained from the grayscale GFP signal channel. The average thickness of the bony bridge in the rhBMP-2 treated animals was measured at 14-, 21-, 28-, and 56-day time points. In order to measure the thickness of the bridge, the computerized algorithm determined the surface norm vectors along the bony bridge. Each surface norm vector defines the direction which is perpendicular to the longitudinal axis of the bony bridge at surface intervals measuring 10 pixels. Subsequently sequential lines were drawn parallel to the direction of the surface norm vectors at 5 pixel intervals. These lines were extended to pass through the bridging bony region in ROI 2 and 3 (Supplementary Fig. S1). The thickness of the bridge was measured as the length of the perpendicular lines from one side of the bone surface to the other side of the bony bridge. The average value of the thicknesses of these lines represented the final thickness of the specific bony bridge. Statistical Analysis Data were expressed as mean SD, and group comparisons were made between the average X-ray scores of rhBMP-2, JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
1122
ALAEE ET AL.
DBM, and control groups as well as average GFP positive volume ratios and average osteoclast numbers in various time points of sacrifice using an ANOVA with significance at p < 0.05. (Fig. 1).
RESULTS Radiographic Assessment of Healing Radiographic analysis showed that all rhBMP-2 treated animals demonstrated bone formation across the defect at day 14 and were considered healed at 28 days. Healing was achieved by forming a bony bridge over the defect. There was unanimous observer agreement that 30 out of 30 rhBMP-2 treated defects were healed at the 28- and 56-day time points. The radiographic score was 5.0 in all defects at days 28 and 56. In contrast, the majority of the defects treated with DBM and the control defects were not healed at 56 days. Complete healing (score 5) was seen in 3/30 animals in DBM and 2/30 controls at 28 and 56 days. Greater than 75% healing (score 4 and 5) of the defect was observed in 5/30 in the DBM and 3/30 of the control animals at 28 and 56 days. The mean radiographic healing score of rhBMP-2 treated animals (5.0 0.0) was significantly higher than the DBM (2.78 1.0) or control (2.40 1.0) groups, (p < 0.05 for both comparisons). There was no statistical difference in radiographic healing with DBM compared to controls (p > 0.05; Fig. 2). Plain Histology Two different patterns of healing were observed in plain histologic evaluation of the samples. In rhBMP-2 treated animals a hypercellular area of cells was observed in the defect at 7 days after surgery. At 14 days post-operatively abundant formation of woven
Figure 2. Radiographic data representative X-rays showing complete healing of the BMP-2 (A) treated defects as opposed to incomplete healing by the DBM (B) and Control (C) animals which formed bony caps on either side of the defect (marked areas).
bone was observed to bridge the defect. Gradual thinning of the bony bridge was seen at later time points. At 56 days the BMP-2 had healed the defect, but the new bone was present as a thin layer of cortical bone (Supplementary Fig. S2). TRAP staining showed the presence of many TRAP positive mononuclear cells in the area of the defect at the 7-day time point. No evidence of osteoid formation or multinucleated osteoclasts was present at this time point (Fig. 3). At 14 days multinucleated osteoclasts were present
Figure 1. Regions of interest outline of the regions of interest for histomorphometric evaluation of the transgenic cell activation, ROI 1 represented the defect area, ROI 2 and 3 represented the superior and inferior bridge, respectively. JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
EVALUATION OF OSTEOGENIC RESPONSE TO BMP-2 AND DBM
1123
Figure 3. Activation of pre-osteoclasts at 7 days after implantation of rhBMP-2 in the defect A, B, and C are 10 images that show hypercellular areas of abundant small TRAPþ mononuclear cells (pre-osteoclasts).
throughout the areas of the new bone formation (Supplementary Fig. S3). The highest average number of TRAP positive cells per mm in the defect area was observed at 7 days in the rhBMP-2 treated group (14.7 7.3 OC/mm2) compared to the later time points. The osteoclast numbers at 7 days were also significantly higher (p < 0.05, ANOVA) in rhBMP-2 treated animals compared to DBM and controls. The OC numbers in days 21–56 specimens were not statistically different between the three groups (Fig. 4). The vast majority of the defects in the DBM treated (27/30) and control (28/30) groups had not healed at
Figure 4. Analysis of the osteoclast numbers. The average number of osteoclasts (OC) per mm of bone perimeter at the defect site was quantified to study the effect of treatment with rhBMP-2 or DBM on osteoclasts. Data obtained from conventional TRAP stained specimens, n ¼ 3 per treatment per time point. p < 0.05 significant compared to defects treated with rhBMP-2 at later time points and also compared to the DBM and control defects at 7 days.
28- and 56-day time points. These defects attempted to heal by forming proximal and distal bony caps that advanced towards the center of the defect from each side, but failed to establish bony continuity (Fig. 2). Osteoclastic activation as evidenced by the presence of TRAP positive cells was observed in the areas of new bone formation including the advancing caps and the bone that was being laid down as a periosteal reaction along the original cortices outside the defect. There was a trend toward higher osteoclast numbers at 14 days in the DBM treated defects compared to the other time points. This was also true about the osteoclast numbers of DBM treated defects at 14 days when compared with the rhBMP-2 and control groups at 14-day time point. However, the difference was not statistically significant in these comparisons (p > 0.05; Fig. 4). Histologic Evaluation of Frozen Sections The bony bridge in rhBMP-2 treated animals was thick and disorganized at day 14, but gradually became noticeably thinner from days 21 to 56 (Fig. 5). The mean bridge thickness (ROI 2 þ 3) was significantly higher at 14 days (144.5 15.7 mM) compared to 21 days (70.1 8.3 mM), 28 days (37.8 3.4 mM), and 56 days (73.4 7.7 mM; p < 0.05, Fig. 6). Modified TRAP staining with ELF-97 revealed the appearance of the TRAP positive cells at day 7. At day 14 the osteoclasts were located in the inner border of the bridge (Supplementary Fig. S4) and were thought to be responsible for bone resorption after day 14. Although the bridge gradually thinned out, a cortical shell remained intact as far out as day 56 and the defects were considered healed because of the presence JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
1124
ALAEE ET AL.
Figure 5. RhBMP-2 induced bridge formation. Representative fluorescent images obtained from Col2.3eGFP and Col 3.6GFPtpz mice which were treated with rhBMP-2 in their defect. There was no cellular activity in the defect at 4 days post-operatively. At 7 days Col3.6þ and Col2.3þ cells were seen around the defect site. Mineralization was first seen on day 14 with presence of abundant thick amorphous woven bone. The bony bridge underwent gradual thinning in the later time points.
of the bony bridge across the defect (Fig. 5). In the DBM treated and control groups, the vast majority of the defects (27/30 in the DBM and 28/30 in the control group) had not healed by 56 days. Modified TRAP staining using ELF-97 showed osteoclast presence along the remodeling bone surfaces in the bony caps and the new bone formed on top of the original cortices.
Figure 6. RhBMP-2 induced bridge thickness. Data collected from Col2.3eGFP and Col 3.6GFPtpz mice which were treated with rhBMP-2 in their defect. Mean bridge thickness (ROI 2 þ 3) was compared in different time points. p < 0.05 compared to all other time points. JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
Patterns of Activation of the Osteoprogenitor Cells in RhBMP-2, DBM, and Control Group RhBMP-2 Treatment At 4 days there were no GFP positive cells in the defect area. At 7 days after surgery Col3.6 positive cells (6/6 animals) began to form a non-mineralized bridge over the defects and represented the early osteoblastic differentiation that form the future bony bridge. In the Col 2.3 animals GFP positive cells started to appear around the defect area in rhBMP-2 treated defects on day 7 (Figs. 7 and 8), but they were strongly present at days 14 and 21. In both the Col 3.6 and 2.3 animals the thick boney bridge noted on days 14 and 21 contained a large number of GFP positive cells which were actively producing osteoid as evidenced by areas of XO staining adjacent to the GFP positive cells. Col 3.6 and Col 2.3 positive cells were present in the bony bridge at days 14– 56. As it continued to thin out by day 56, there was a diminished presence of these cells in the bridge (Fig. 5). The a-SMA positive pericyte/myofibroblast-like cells strongly appeared in response to rhBMP-2 in 2/6 animals on day 7. There was increased signal in the surrounding musculature (Fig. 7, panels a–c). Minimal SMA signal was detected in the area of bony bridge at 14 days. DBM Treatment and Control Groups No significant cell activation was observed at 4 days in either group. The osteogenic differentiation (Col3.6
EVALUATION OF OSTEOGENIC RESPONSE TO BMP-2 AND DBM
1125
Figure 7. Two millimeters defects at 7 days post-operatively from the a-SMACherry (a, d, g), Col3.6GFPtopaz (b, e, h) and Col2.3eGFP (c, f, i) transgenic mice, RhBMP-2 (a–c), or DBM (e–g) were placed in the defects of the experimental groups and the controls (h–j) received no materials in the defect. At 7 days there is still no bone formation. Presence of the a-SMACherry and Col3.6GFPtopaz cells are evident in the defect site (a and b) and Col2.3eGFP positive cells are starting to appear in response to rhBMP-2 (c). These cells are being organized to form the future bony bridge. The source of the recruitment appears to be from the surrounding periosteum and perhaps the skeletal muscle. In the DBM and control samples, there is no significant recruitment of osteoprogenitor cells (d–j). A scant endosteal or periosteal reaction can be appreciated (d, f, h, i). There is no contribution from the surrounding muscle to the healing process. The majority of these defects failed to eventually heal.
Figure 8. Two millimeter defects at 7 days post-operatively from the Col2.3eGFP transgenic mice treated with Rh-BMP-2. A, B, and C showing GFP positive cells from periosteum and surrounding muscle tissue being organized to form the future bony bridge. JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
1126
ALAEE ET AL.
and Col2.3 animals) was limited to the periosteal region at the 7-day time point and no bony bridge was formed at later time points. In both the Col3.6 and Col2.3 animals GFP positive cells were present in the area of the bony caps from days 14 until 56 (Supplementary Fig. S5). In the a-SMACherry mice, positive cells were not detected in significant numbers at these time points (Fig. 7, panel d–i). Quantitation of the Osteogenic Differentiation in RhBMP2, DBM, and Control Groups RhBMP-2 Treatment The highest contribution of the Col3.6 cells in formation of the healing bridge was at day 7 (7.2% 6.0 vs. 0.44% 0.2 at 28 days and 1.67% 1.5 at 56 days, p < 0.05) using the density of the GFP signal in the area of the future bony bridge (ROI 2 and 3). The Col2.3 positive cells had their highest presence in the bony bridge (ROI 2 þ 3) on day 14 (8.04% 5.0) and day 21 (8.31% 4.32) versus 0.44% 0.26 at 28 days and 2.3% 1.44 at 56 days (p < 0.05, Table 1). The histomorphometric analysis of the Col3.6 and Col2.3 signal density in the defect area (ROI 1) also showed the highest expression of the Col 3.6 at day 7 and for the Col 2.3 positive cells at days 14 and 21 (Table 1). Although SMA positive pericyte/myofibroblast-like cells were detected in response to rhBMP-2 at day 7 in two out of six mice, the quantitation of the aSMACherry in ROI 2 þ 3 did not reveal any statistical differences between the 4-, 7-, and 14-day time points. DBM Treatment and Control Groups The histomorphometric analysis of the signal density in the defect area (ROI 1) in the Col3.6, Col2.3 and the a-SMACherry mice did not reveal any significant differences at any time points (Table 1).
DISCUSSION We demonstrated superior bone healing with rhBMP-2 in a critical sized femoral defect compared to DBM. The rhBMP-2 healed the defect by formation of a bony
bridge across the defect site. In contrast in the vast majority of the DBM and control animals the bone healing took place by formation of a bony cap on either side of the defect, but the defects did not completely heal. We were able to quantify a significant difference in the osteogenic response to rhBMP-2 compared to the DBM using a novel computer algorithm for the first time. We believe that this quantitative method is compatible with the histological observations and has the potential to be used as a measurement tool for the cellular response in various treatments to promote bone repair. Our quantitative results showed that the early osteoblasts (col 3.6 positive cells) were mostly detected at 7 days whereas the mature osteoblastic cells (Col2.3 positive cells) were most abundant at 14 and 21 days. This finding is consistent with prior studies where aSMA and Col 3.6 positive cells were noted to be expressed in pre-osteoblasts.14,15 The a-SMA positive pericyte/myofibroblast-like cells were only detectable at 7 days post-operatively and their presence was only observed in two out of six animals. The reason for the undetectable SMA signal in the other four animals is not known, but may be due to variability in animal response, a short duration of signal detectability or variations in signal strength. As the bony bridge underwent remodeling, the presence of Col3.6 and Col2.3 positive cells were seen up until 56 days. In the rhBMP-2 treated animals osteoblastic differentiation was very active in the periosteum, but no significant osteoprogenitor cellular contribution was noted from the bone marrow in any specimen. The bony bridge formed by the rhBMP-2 was thick and disorganized at day 14, but gradually became noticeably thinner from days 21 to 56. TRAP staining revealed high numbers of the mononuclear osteoclast precursors at day 7 which subsequently turned into active polynuclear osteoclasts at day 14. Although the bridge gradually thinned out, a cortical shell remained intact as far out as day 56 and the defects were considered healed because of the presence of the bony bridge.
Table 1. Col3.6GFPtopaz and Col2.3eGFP Measured at 7, 14, 21, 28, and 56 Days Time Point (Days)
Col 3.6 RhBMP-2
GFPþ cell/TV defect (ROI 1) 7 2.22 1.33 14 2.67 3.31 21 0.69 0.48 28 0.11 0.08 56 0.21 0.26 GFPþ cell/TV Bridge (ROI 2 þ 3) 7 7.2 6.02 14 4.07 2.72 21 2.52 1.45 28 0.44 0.26 56 1.67 1.48
Col 3.6 DBM
Col 3.6 Control
Col 2.3 RhBMP-2
Col 2.3 DBM
Col 2.3 Control
1.00 0.68 0.84 0.76 1.4 0.66 0.59 0.38 1.2 1.26
1.74 2.86 0.87 1.23 1.48 2.09 0.37 0.65 0.70 0.75
1.43 0.70 3.77 1.79 2.53 2.81 0.04 0.03 1.05 1.71
1.43 1.38 2.33 1.14 2.10 0.96 1.50 0.60 1.87 0.96
0.83 0.89 0.84 1.24 5.80 7.5 1.27 0.76 0.94 1.05
NA NA NA NA NA
NA NA NA NA NA
5.31 3.15 8.04 5.06 8.31 4.32 0.44 0.23 2.30 1.44
NA NA NA NA NA
NA NA NA NA NA
p < 0.05 compared to other time points in the same group. Data expressed as mean SD.
JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
EVALUATION OF OSTEOGENIC RESPONSE TO BMP-2 AND DBM
In various animal models, DBM has been noted to be osteoinductive,31,32 which is thought to be associated with the concentration of the BMPs in a specific DBM.33 In our study, the use of DBM in the femoral defects elicited a fundamentally different progenitor cell response compared to the rhBMP-2. DBM did not have the inductive capacity of rhBMP-2 which was evident by the lack of osteogenic response. The osteogenic caps at either end of the bone which seemed to originate from the periosteal layer consistently failed to make the connection between the two ends of the defect. In this respect, the pattern of bone repair in the mice treated with DBM and the controls was clearly different than BMP-2. A limitation of our study is that we did not characterize our DBM with regards to different growth factor concentrations and it is possible that the concentration of the BMPs in this particular DBM was not sufficient to induce bone formation. Although the exact source of the osteogenic cells in response to rhBMP-2 could not be determined by using the current fluorescent markers, there was a significant difference between the rhBMP-2 treated animals and DBM or controls with respect to the patterns of cell recruitment. Our findings strongly suggest that DBMs induce a minimal biologic response in bone defects and should not be used for the same indications or as an alternative to the rhBMP-2. Although BMP-2 is the most potent osteoinductive agent available for clinical use today, there have been concerns regarding inconsistent healing in humans and the quality of bone formed with treatment with rhBMP-2.34 BMP-2 not only stimulates osteoblastic differentiation but also can induce osteoclastogenesis through increased RANKL in the osteoblasts35 and this may influence the quality and quantity of bone repair. In our study the rhBMP-2 healed the defects, but the thinning of the bone was a concern. TRAP staining revealed a prominent osteoclast response at days 7 and 14 which could have a negative impact on the quality of bone. The combined therapy with rhBMP-2 and an osteoclast inhibitor (zoledronic acid or alendronate) has been noted to improve quality of bone repair.20,31,36 These findings are consistent with our analysis that early and robust osteoclast activity may inhibit bone repair. It is not clear if this early osteoclastic response to BMP is related to the dose of BMP or the kinetics of BMP release from the collagen sponge. In conclusion, rhBMP-2 and DBM induced vastly different osteogenic responses in a mouse critical sized femoral defect model. It is essential that clinicians understand the different biologic properties of these agents when choosing therapies to treat bone repair problems.
ACKNOWLEDGEMENT
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
This study was supported by a grant from Musculoskeletal Transplant Foundation to J.R.L.
18.
REFERENCES
19.
1. Murakami N, Saito N, Horiuchi H, et al. 2002. Repair of segmental defects in rabbit humeri with titanium fiber mesh
1127
cylinders containing recombinant human bone morphogenetic protein-2 (rhBMP-2) and a synthetic polymer. J Biomed Mater Res 62:169–174. Vogelin E, Jones NF, Huang JI, et al. 2005. Healing of a critical-sized defect in the rat femur with use of a vascularized periosteal flap, a biodegradable matrix, and bone morphogenetic protein. J Bone Joint Surg Am 87:1323–1331. Boyce AS, Reveal G, Scheid DK, et al. 2009. Canine investigation of rhBMP-2, autogenous bone graft, and rhBMP-2 with autogenous bone graft for the healing of a large segmental tibial defect. J Orthop Trauma 23:685–692. Gerhart TN, Kirker-Head CA, Kriz MJ, et al. 1993. Healing segmental femoral defects in sheep using recombinant human bone morphogenetic protein. Clin Orthop Relat Res 263:317–326. Seeherman HJ, Li XJ, Smith E, et al. 2012. rhBMP-2/ calcium phosphate matrix induces bone formation while limiting transient bone resorption in a nonhuman primate core defect model. J Bone Joint Surg Am 94:1765–1776. Starman JS, Bosse MJ, Cates CA, et al. 2012. Recombinant human bone morphogenetic protein-2 use in the off-label treatment of nonunions and acute fractures: a retrospective review. J Trauma Acute Care Surg 72:676–681. Govender S, Csimma C, Genant HK, et al. 2002. Recombinant human bone morphogenetic protein-2 for treatment of open tibial fractures: a prospective, controlled, randomized study of four hundred and fifty patients. J Bone Joint Surg Am 84-A:2123–2134. Martin GJ Jr, Boden SD, Titus L, et al. 1999. New formulations of demineralized bone matrix as a more effective graft alternative in experimental posterolateral lumbar spine arthrodesis. Spine (Phila Pa 1976) 24:637–645. Morone MA, Boden SD. 1998. Experimental posterolateral lumbar spinal fusion with a demineralized bone matrix gel. Spine (Phila Pa 1976) 23:159–167. Wang JC, Alanay A, Mark D, et al. 2007. A comparison of commercially available demineralized bone matrix for spinal fusion. Eur Spine J 16:1233–1240. Huang YZ, Cai JQ, Xue J, et al. 2012. The poor osteoinductive capability of human acellular bone matrix. Int J Artif Organs 35:1061–1069. Barradas AM, Yuan H, van Blitterswijk CA, et al. Osteoinductive biomaterials: current knowledge of properties, experimental models and biological mechanisms. Eur Cell Mater 21:407–429; discussion 429. Bilic-Curcic I, Kalajzic Z, Wang L, et al. 2005. Origins of endothelial and osteogenic cells in the subcutaneous collagen gel implant. Bone 37:678–687. Kalajzic I, Kalajzic Z, Kaliterna M, et al. 2002. Use of type I collagen green fluorescent protein transgenes to identify subpopulations of cells at different stages of the osteoblast lineage. J Bone Miner Res 17:15–25. Kalajzic I, Kalajzic Z, Wang L, et al. 2007. Pericyte/ myofibroblast phenotype of osteoprogenitor cell. J Musculoskelet Neuronal Interact 7:320–322. Kalajzic Z, Li H, Wang LP, et al. 2008. Use of an alphasmooth muscle actin GFP reporter to identify an osteoprogenitor population. Bone 43:501–510. Marijanovic I, Jiang X, Kronenberg MS, et al. 2003. Dual reporter transgene driven by 2.3Col1a1 promoter is active in differentiated osteoblasts. Croat Med J 44:412–417. Bilic-Curcic I, Kronenberg M, Jiang X, et al. 2005. Visualizing levels of osteoblast differentiation by a two-color promoter-GFP strategy: type I collagen-GFPcyan and osteocalcinGFPtpz. Genesis 43:87–98. Ushiku C, Adams DJ, Jiang X, et al. Long bone fracture repair in mice harboring GFP reporters for cells within the osteoblastic lineage. J Orthop Res 28:1338–1347. JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
1128
ALAEE ET AL.
20. Schindeler A, Birke O, Yu NY, et al. Distal tibial fracture repair in a neurofibromatosis type 1-deficient mouse treated with recombinant bone morphogenetic protein and a bisphosphonate. J Bone Joint Surg Br 93:1134–1139. 21. Alaee F, Sugiyama O, Virk MS, et al. 2014. Suicide gene approach using a dual-expression lentiviral vector to enhance the safety of ex vivo gene therapy for bone repair. Gene Ther 21:139–147. 22. Lieberman JR, Daluiski A, Stevenson S, et al. 1999. The effect of regional gene therapy with bone morphogenetic protein-2-producing bone-marrow cells on the repair of segmental femoral defects in rats. J Bone Joint Surg Am 81:905–917. 23. Jiang X, Kalajzic Z, Maye P, et al. 2005. Histological analysis of GFP expression in murine bone. J Histochem Cytochem 53:593–602. 24. Filgueira L. 2004. Fluorescence-based staining for tartrateresistant acidic phosphatase (TRAP) in osteoclasts combined with other fluorescent dyes and protocols. J Histochem Cytochem 52:411–414. 25. Peterson B, Zhang J, Iglesias R, et al. 2005. Healing of critically sized femoral defects, using genetically modified mesenchymal stem cells from human adipose tissue. Tissue Eng 11:120–129. 26. Feeley BT, Gamradt SC, Hsu WK, et al. 2005. Influence of BMPs on the formation of osteoblastic lesions in metastatic prostate cancer. J Bone Miner Res 20:2189–2199. 27. Hsu WK, Virk MS, Feeley BT, et al. 2008. Characterization of osteolytic, osteoblastic, and mixed lesions in a prostate cancer mouse model using 18F-FDG and 18F-fluoride PET/ CT. J Nucl Med 49:414–421. 28. Hong SH, Jiang X, Chen L, et al. 2012. Computer-automated static, dynamic and cellular bone histomorphometry. J Tissue Sci Eng S1:004.
JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
29. Parfitt AM, Drezner MK, Glorieux FH, et al. 1987. Bone histomorphometry: standardization of nomenclature, symbols, and units. Report of the ASBMR Histomorphometry Nomenclature Committee. J Bone Miner Res 2:595–610. 30. Otsu N. 1978. A threshold selection method from gray-scale histogram. IEEE Trans Syst Man Cybern 8:62–66. 31. Vandermeer JS, Kamiya N, Aya-ay J, et al. Local administration of ibandronate and bone morphogenetic protein-2 after ischemic osteonecrosis of the immature femoral head: a combined therapy that stimulates bone formation and decreases femoral head deformity. J Bone Joint Surg Am 93:905–913. 32. Arrington ED, Smith WJ, Chambers HG, et al. 1996. Complications of iliac crest bone graft harvesting. Clin Orthop Relat Res 329:300–309. 33. Gruskin E, Doll BA, Futrell FW, et al. Demineralized bone matrix in bone repair: history and use. Adv Drug Deliv Rev 64:1063–1077. 34. Fu R, Selph S, McDonagh M, et al. 2013. Effectiveness and harms of recombinant human bone morphogenetic protein-2 in spine fusion: a systematic review and meta-analysis. Ann Intern Med 158:890–902. 35. Granholm S, Henning P, Lindholm C, et al. Osteoclast progenitor cells present in significant amounts in mouse calvarial osteoblast isolations and osteoclastogenesis increased by BMP-2. Bone 52:83–92. 36. Doi Y, Miyazaki M, Yoshiiwa T, et al. Manipulation of the anabolic and catabolic responses with BMP-2 and zoledronic acid in a rat femoral fracture model. Bone 49:777–782.
SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at the publisher’s web-site.
Received 16 December 2013; accepted 8 May 2014 Published online 2 June 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jor.22657
ABSTRACT: We evaluated the osteoprogenitor response to rhBMP-2 and DBM in a transgenic mouse critical sized defect. The mice expressed Col3.6GFPtopaz (a pre-osteoblastic marker), Col2.3GFPemerald (an osteoblastic marker) and a-smooth muscle actin (a-SMACherry, a pericyte/myofibroblast marker). We assessed defect healing at various time points using radiographs, frozen, and conventional histologic analyses. GFP signal in regions of interest corresponding to the areas of new bone formation was quantified using a novel computer assisted algorithm. All defects treated with rhBMP-2 healed. In contrast, the majority of the defects in the DBM (27/30) and control (28/30) groups did not heal. Quantitation of pre-osteoblasts demonstrated a maximal response (% GFPþ cells/ TV) in the Col3.6GFPtopaz mice at day 7 (7.2% 6.0, p < 0.05 compared to days 14, 21, 28, and 56). The maximal response of the Col2.3GFP cells was seen at days 14 (8.04% 5.0) and 21 (8.31% 4.32), p < 0.05. In contrast, DBM and control groups showed a limited osteogenic response at all time points. In conclusion, we demonstrated that the BMP and DBM induce vastly different osteogenic responses which should influence their clinical application as bone graft substitutes. ß 2014 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 32:1120–1128, 2014. Keywords: femoral defect; rhBMP-2; demineralized bone matrix; bone histomorphometry; GFP reporters
The development of bone repair strategies that are consistently successful requires an understanding of the influence of graft materials on host cells. Orthopaedic surgeons have been using bone graft substitutes such as recombinant human bone morphogenetic protein (rhBMP) and demineralized bone matrix (DBM) to enhance bone repair. RhBMP-2 has been proven to be an osteoinductive agent in both preclinical models1–5 and clinical trials.6,7 DBM is derived from human cadaver bone and contains several BMPs.8,9 DBMs are osteoconductive and have demonstrated osteoinductive activity in animal models, but osteoinductive activity has not been definitively demonstrated in humans.10–12 There is confusion amongst surgeons regarding the biological activity of the rhBMP-2 compared to DBM, because there is varying concentrations of BMPs in DBM. Our goal is to elucidate the interaction of these graft materials and host cells in a critical-sized defect to provide data that can be used to hopefully provide guidance with respect to the appropriate use of these graft materials in clinical situations. We used novel transgenic mice that contain GFP reporter constructs that facilitate the identification of the host response to these grafts. Novel GFP reporter transgenic mice have been developed to assess the timing of osteoblast differentiation. We used three reporters, based on Col1a1 (Col3.6GFPtopaz and Col2.3eGFP) promoter frag-
Conflict of interest: None. Grant sponsor: Musculoskeletal Transplant Foundation. Correspondence to: Jay R. Lieberman (T: þ1-323-442-8117; F: þ1-323-442-7621; E-mail: [email protected]) # 2014 Orthopaedic Research Society. Published by Wiley Periodicals, Inc.
1120
JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
ments and a-Smooth Muscle Actin (a-SMACherry). These reporters have facilitated the tracking of the differentiation of osteoblast lineage cells.13–17 In vitro and in vivo experiments have established that strong Col3.6GFP signal is found in newly formed osteoblasts14,18 while the Col2.3GFP marker is expressed in differentiated osteoblasts and osteocytes.14,17 Ushiku et al.19 used a double transgenic mouse (Col3.6-blue and osteocalcin-GFP-topaz) in a closed tibial fracture model to study the recruitment of the osteoprogenitors from periosteum. The a-SMACherry indicates cells with a pericyte/myofibroblast-like phenotype that have the potential to differentiate into functional osteoblasts.16 The progression of the a-SMA positive progenitor cells to mature osteoblasts during fracture healing has been documented.20 The purpose of our study was to evaluate the osteoprogenitor response to rhBMP-2 and DBM and evaluate the pattern of cell activation in a mouse critical sized femoral defect model. We hypothesized that there were critical differences in progenitor cell response between rhBMP-2 and DBM that should influence the clinical application of these agents. We used a novel computerized algorithm to quantitate the osteogenic response to the rhBMP-2 and DBM.
MATERIALS AND METHODS Animal Studies All animal studies were performed after approval by the Animal Care Committee in our institution. Critical sized femoral defects (2 mm) were created in 14- to 16-week-old male BL/6 mice according to a previously published method.21 We used non-transgenic C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME) for conventional histologic evaluation or lineage-specific transgenic mice expressing
EVALUATION OF OSTEOGENIC RESPONSE TO BMP-2 AND DBM
Col3.6GFPtopaz, Col2.3eGFP and a-SMACherry for preparation of frozen embedded samples to evaluate tissue fluorescence. The following groups were included in the study: Group I, rhBMP-2, 5 mg on a piece of collagen sponge (Medtronic, Minneapolis, MN), Group II, DBM (Musculoskeletal Transplant Foundation, Edison, NJ) and Group III, empty defect (control group). There were six mice per transgene per treatment per time point. Twenty-four hours before the sacrifice either xylenol orange (XO red label, to Col2.3eGFP and Col 3.6GFPtpz mice), or calcein (Green label, to a-SMA-Cherry mice) were injected intraperitoneally to mark the areas of the new bone formation while providing the color contrast with the native fluorescent cells. The Col2.3eGFP and Col3.6GFPtpz mice were sacrificed at various time points after surgery (4, 7, 14, 21, 28, and 56 days). The a-SMA-Cherry mice were sacrificed at 4, 7, and 14 days after surgery because preliminary data (not shown) demonstrated no activity beyond the 14 days time point. Radiographic Evaluation Radiographs at the 4- and 8-week time points were assessed by three independent observers and the defect healing was scored from 0 to 5 based on a previously published protocol which had been originally developed to evaluate bone repair in a rat critical-sized defect model.22 Briefly, a score of 0 was given if there was no healing, 0–25% healing was given 1, 25–50% was 2, 50–75% was 3, 75–99% was 4 and complete healing was 5. A kappa statistic was calculated to assess the inter-observer variability among the three observers. Histologic Evaluation At the time of sacrifice the operated leg was harvested and the external fixator was removed. The samples were assessed using two different histologic protocols including regular (decalcified, non-frozen) or fluorescent (non-decalcified, frozen) sections. Non-Decalcified Frozen Histology A previously established protocol was used to perform the frozen sections.23 After initial fluorescent imaging, selected samples were stained with a modified tartrate-resistant acid phosphatase (TRAP) stain and the fluorescent substrate ELF-97 that gave a distinct yellow fluorescence to the osteoclasts.24 Decalcified Standard Histology Formalin fixation at room temperature for 1 day, and decalcification in 10% EDTA for 2 weeks were performed. Sagittal sections through the defect area were mounted on glass slides and were stored at room temperature until use. Standard protocols were used for both H&E and conventional TRAP staining of the samples.22,25 Histomorphometric analysis of the TRAP stained sections (the number of osteoclasts per square millimeter of bone) was performed on an Olympus system (Olympus, Melville, NY) using the histomorphometric analysis software (Osteomeasure, Osteometrics, Decatur, GA) as previously described.26,27 Four rectangular areas were selected at each corner of the defect and the average number of osteoclasts per slide was used in the Statistical Analysis Section. Fluorescent Imaging Photographs of the frozen sectioned samples were taken using a Zeiss Imager Z1 microscope (Carl Zeiss, Thornwood,
1121
NY) and Axio Vision Rel.4.7 software (Carl Zeiss). Specific fluorescent filters (Chroma Technology, Bellows Falls, VT), based on a previously published protocol19,23 were used to obtain separate images that were subsequently merged to produce a final image. A differential interference contrast (DIC) image was acquired at the same time as the endogenous fluorescence imaging for final recompilation. A scanned composite image is generated which is created after combining the individual grayscale images from each channel. Histomorphometric Analysis of the GFP Expression Regions of interest (ROI) layers were created on top of the scanned composite images using Adobe Photoshop software. In the BMP-2 treated defects with bridging bone 3 distinct regions of interest (ROIs) were marked. ROI 1 represented the defect area, ROI 2 and 3 represented the superior and inferior bridge, respectively. Analysis of the bridge thickness and GFP signal density was determined on the aggregate data from ROI 2 and 3. Since there was no bony bridge formation in the DBM and control animals, the region of interest only included the defect area (ROI1). Within each ROI, the volume ratio of GFP positive cells over the total bone volume (GFPþ cell/TV) was measured and calculated using computer-automated bone histomorphometric techniques (automatic segmentation and automatic morphological transformation) introduced by Hong et al.28 and Parfitt et al.29 Segmentation of the signal was performed by Otsu’s thresholding method30 for bone and GFP signals separately. Otsu’s method is based on the histogram of the gray intensities of the signal. This method provides a threshold that separates grayscale pixels into black and white signal groups. It finds an optimum threshold intensity value that minimizes the standard deviation of the pixels that fall into either black or white segmented groups, while maximizing the inter standard deviation between the two groups. The white pixels represent the signal that was intended to be captured in the grayscale image. The volume ratio is a ratio of the total GFP positive cell volume (area occupied by GFP positive cells within a given ROI) over the total volume (area of ROI). Area is represented by the number of pixels in each segmented image. To determine the area of the GFP positive cells, the computer algorithm written in Matlab software simply counts the number of white pixels in the segmented (black and white) image that is obtained from the grayscale GFP signal channel. The average thickness of the bony bridge in the rhBMP-2 treated animals was measured at 14-, 21-, 28-, and 56-day time points. In order to measure the thickness of the bridge, the computerized algorithm determined the surface norm vectors along the bony bridge. Each surface norm vector defines the direction which is perpendicular to the longitudinal axis of the bony bridge at surface intervals measuring 10 pixels. Subsequently sequential lines were drawn parallel to the direction of the surface norm vectors at 5 pixel intervals. These lines were extended to pass through the bridging bony region in ROI 2 and 3 (Supplementary Fig. S1). The thickness of the bridge was measured as the length of the perpendicular lines from one side of the bone surface to the other side of the bony bridge. The average value of the thicknesses of these lines represented the final thickness of the specific bony bridge. Statistical Analysis Data were expressed as mean SD, and group comparisons were made between the average X-ray scores of rhBMP-2, JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
1122
ALAEE ET AL.
DBM, and control groups as well as average GFP positive volume ratios and average osteoclast numbers in various time points of sacrifice using an ANOVA with significance at p < 0.05. (Fig. 1).
RESULTS Radiographic Assessment of Healing Radiographic analysis showed that all rhBMP-2 treated animals demonstrated bone formation across the defect at day 14 and were considered healed at 28 days. Healing was achieved by forming a bony bridge over the defect. There was unanimous observer agreement that 30 out of 30 rhBMP-2 treated defects were healed at the 28- and 56-day time points. The radiographic score was 5.0 in all defects at days 28 and 56. In contrast, the majority of the defects treated with DBM and the control defects were not healed at 56 days. Complete healing (score 5) was seen in 3/30 animals in DBM and 2/30 controls at 28 and 56 days. Greater than 75% healing (score 4 and 5) of the defect was observed in 5/30 in the DBM and 3/30 of the control animals at 28 and 56 days. The mean radiographic healing score of rhBMP-2 treated animals (5.0 0.0) was significantly higher than the DBM (2.78 1.0) or control (2.40 1.0) groups, (p < 0.05 for both comparisons). There was no statistical difference in radiographic healing with DBM compared to controls (p > 0.05; Fig. 2). Plain Histology Two different patterns of healing were observed in plain histologic evaluation of the samples. In rhBMP-2 treated animals a hypercellular area of cells was observed in the defect at 7 days after surgery. At 14 days post-operatively abundant formation of woven
Figure 2. Radiographic data representative X-rays showing complete healing of the BMP-2 (A) treated defects as opposed to incomplete healing by the DBM (B) and Control (C) animals which formed bony caps on either side of the defect (marked areas).
bone was observed to bridge the defect. Gradual thinning of the bony bridge was seen at later time points. At 56 days the BMP-2 had healed the defect, but the new bone was present as a thin layer of cortical bone (Supplementary Fig. S2). TRAP staining showed the presence of many TRAP positive mononuclear cells in the area of the defect at the 7-day time point. No evidence of osteoid formation or multinucleated osteoclasts was present at this time point (Fig. 3). At 14 days multinucleated osteoclasts were present
Figure 1. Regions of interest outline of the regions of interest for histomorphometric evaluation of the transgenic cell activation, ROI 1 represented the defect area, ROI 2 and 3 represented the superior and inferior bridge, respectively. JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
EVALUATION OF OSTEOGENIC RESPONSE TO BMP-2 AND DBM
1123
Figure 3. Activation of pre-osteoclasts at 7 days after implantation of rhBMP-2 in the defect A, B, and C are 10 images that show hypercellular areas of abundant small TRAPþ mononuclear cells (pre-osteoclasts).
throughout the areas of the new bone formation (Supplementary Fig. S3). The highest average number of TRAP positive cells per mm in the defect area was observed at 7 days in the rhBMP-2 treated group (14.7 7.3 OC/mm2) compared to the later time points. The osteoclast numbers at 7 days were also significantly higher (p < 0.05, ANOVA) in rhBMP-2 treated animals compared to DBM and controls. The OC numbers in days 21–56 specimens were not statistically different between the three groups (Fig. 4). The vast majority of the defects in the DBM treated (27/30) and control (28/30) groups had not healed at
Figure 4. Analysis of the osteoclast numbers. The average number of osteoclasts (OC) per mm of bone perimeter at the defect site was quantified to study the effect of treatment with rhBMP-2 or DBM on osteoclasts. Data obtained from conventional TRAP stained specimens, n ¼ 3 per treatment per time point. p < 0.05 significant compared to defects treated with rhBMP-2 at later time points and also compared to the DBM and control defects at 7 days.
28- and 56-day time points. These defects attempted to heal by forming proximal and distal bony caps that advanced towards the center of the defect from each side, but failed to establish bony continuity (Fig. 2). Osteoclastic activation as evidenced by the presence of TRAP positive cells was observed in the areas of new bone formation including the advancing caps and the bone that was being laid down as a periosteal reaction along the original cortices outside the defect. There was a trend toward higher osteoclast numbers at 14 days in the DBM treated defects compared to the other time points. This was also true about the osteoclast numbers of DBM treated defects at 14 days when compared with the rhBMP-2 and control groups at 14-day time point. However, the difference was not statistically significant in these comparisons (p > 0.05; Fig. 4). Histologic Evaluation of Frozen Sections The bony bridge in rhBMP-2 treated animals was thick and disorganized at day 14, but gradually became noticeably thinner from days 21 to 56 (Fig. 5). The mean bridge thickness (ROI 2 þ 3) was significantly higher at 14 days (144.5 15.7 mM) compared to 21 days (70.1 8.3 mM), 28 days (37.8 3.4 mM), and 56 days (73.4 7.7 mM; p < 0.05, Fig. 6). Modified TRAP staining with ELF-97 revealed the appearance of the TRAP positive cells at day 7. At day 14 the osteoclasts were located in the inner border of the bridge (Supplementary Fig. S4) and were thought to be responsible for bone resorption after day 14. Although the bridge gradually thinned out, a cortical shell remained intact as far out as day 56 and the defects were considered healed because of the presence JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
1124
ALAEE ET AL.
Figure 5. RhBMP-2 induced bridge formation. Representative fluorescent images obtained from Col2.3eGFP and Col 3.6GFPtpz mice which were treated with rhBMP-2 in their defect. There was no cellular activity in the defect at 4 days post-operatively. At 7 days Col3.6þ and Col2.3þ cells were seen around the defect site. Mineralization was first seen on day 14 with presence of abundant thick amorphous woven bone. The bony bridge underwent gradual thinning in the later time points.
of the bony bridge across the defect (Fig. 5). In the DBM treated and control groups, the vast majority of the defects (27/30 in the DBM and 28/30 in the control group) had not healed by 56 days. Modified TRAP staining using ELF-97 showed osteoclast presence along the remodeling bone surfaces in the bony caps and the new bone formed on top of the original cortices.
Figure 6. RhBMP-2 induced bridge thickness. Data collected from Col2.3eGFP and Col 3.6GFPtpz mice which were treated with rhBMP-2 in their defect. Mean bridge thickness (ROI 2 þ 3) was compared in different time points. p < 0.05 compared to all other time points. JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
Patterns of Activation of the Osteoprogenitor Cells in RhBMP-2, DBM, and Control Group RhBMP-2 Treatment At 4 days there were no GFP positive cells in the defect area. At 7 days after surgery Col3.6 positive cells (6/6 animals) began to form a non-mineralized bridge over the defects and represented the early osteoblastic differentiation that form the future bony bridge. In the Col 2.3 animals GFP positive cells started to appear around the defect area in rhBMP-2 treated defects on day 7 (Figs. 7 and 8), but they were strongly present at days 14 and 21. In both the Col 3.6 and 2.3 animals the thick boney bridge noted on days 14 and 21 contained a large number of GFP positive cells which were actively producing osteoid as evidenced by areas of XO staining adjacent to the GFP positive cells. Col 3.6 and Col 2.3 positive cells were present in the bony bridge at days 14– 56. As it continued to thin out by day 56, there was a diminished presence of these cells in the bridge (Fig. 5). The a-SMA positive pericyte/myofibroblast-like cells strongly appeared in response to rhBMP-2 in 2/6 animals on day 7. There was increased signal in the surrounding musculature (Fig. 7, panels a–c). Minimal SMA signal was detected in the area of bony bridge at 14 days. DBM Treatment and Control Groups No significant cell activation was observed at 4 days in either group. The osteogenic differentiation (Col3.6
EVALUATION OF OSTEOGENIC RESPONSE TO BMP-2 AND DBM
1125
Figure 7. Two millimeters defects at 7 days post-operatively from the a-SMACherry (a, d, g), Col3.6GFPtopaz (b, e, h) and Col2.3eGFP (c, f, i) transgenic mice, RhBMP-2 (a–c), or DBM (e–g) were placed in the defects of the experimental groups and the controls (h–j) received no materials in the defect. At 7 days there is still no bone formation. Presence of the a-SMACherry and Col3.6GFPtopaz cells are evident in the defect site (a and b) and Col2.3eGFP positive cells are starting to appear in response to rhBMP-2 (c). These cells are being organized to form the future bony bridge. The source of the recruitment appears to be from the surrounding periosteum and perhaps the skeletal muscle. In the DBM and control samples, there is no significant recruitment of osteoprogenitor cells (d–j). A scant endosteal or periosteal reaction can be appreciated (d, f, h, i). There is no contribution from the surrounding muscle to the healing process. The majority of these defects failed to eventually heal.
Figure 8. Two millimeter defects at 7 days post-operatively from the Col2.3eGFP transgenic mice treated with Rh-BMP-2. A, B, and C showing GFP positive cells from periosteum and surrounding muscle tissue being organized to form the future bony bridge. JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
1126
ALAEE ET AL.
and Col2.3 animals) was limited to the periosteal region at the 7-day time point and no bony bridge was formed at later time points. In both the Col3.6 and Col2.3 animals GFP positive cells were present in the area of the bony caps from days 14 until 56 (Supplementary Fig. S5). In the a-SMACherry mice, positive cells were not detected in significant numbers at these time points (Fig. 7, panel d–i). Quantitation of the Osteogenic Differentiation in RhBMP2, DBM, and Control Groups RhBMP-2 Treatment The highest contribution of the Col3.6 cells in formation of the healing bridge was at day 7 (7.2% 6.0 vs. 0.44% 0.2 at 28 days and 1.67% 1.5 at 56 days, p < 0.05) using the density of the GFP signal in the area of the future bony bridge (ROI 2 and 3). The Col2.3 positive cells had their highest presence in the bony bridge (ROI 2 þ 3) on day 14 (8.04% 5.0) and day 21 (8.31% 4.32) versus 0.44% 0.26 at 28 days and 2.3% 1.44 at 56 days (p < 0.05, Table 1). The histomorphometric analysis of the Col3.6 and Col2.3 signal density in the defect area (ROI 1) also showed the highest expression of the Col 3.6 at day 7 and for the Col 2.3 positive cells at days 14 and 21 (Table 1). Although SMA positive pericyte/myofibroblast-like cells were detected in response to rhBMP-2 at day 7 in two out of six mice, the quantitation of the aSMACherry in ROI 2 þ 3 did not reveal any statistical differences between the 4-, 7-, and 14-day time points. DBM Treatment and Control Groups The histomorphometric analysis of the signal density in the defect area (ROI 1) in the Col3.6, Col2.3 and the a-SMACherry mice did not reveal any significant differences at any time points (Table 1).
DISCUSSION We demonstrated superior bone healing with rhBMP-2 in a critical sized femoral defect compared to DBM. The rhBMP-2 healed the defect by formation of a bony
bridge across the defect site. In contrast in the vast majority of the DBM and control animals the bone healing took place by formation of a bony cap on either side of the defect, but the defects did not completely heal. We were able to quantify a significant difference in the osteogenic response to rhBMP-2 compared to the DBM using a novel computer algorithm for the first time. We believe that this quantitative method is compatible with the histological observations and has the potential to be used as a measurement tool for the cellular response in various treatments to promote bone repair. Our quantitative results showed that the early osteoblasts (col 3.6 positive cells) were mostly detected at 7 days whereas the mature osteoblastic cells (Col2.3 positive cells) were most abundant at 14 and 21 days. This finding is consistent with prior studies where aSMA and Col 3.6 positive cells were noted to be expressed in pre-osteoblasts.14,15 The a-SMA positive pericyte/myofibroblast-like cells were only detectable at 7 days post-operatively and their presence was only observed in two out of six animals. The reason for the undetectable SMA signal in the other four animals is not known, but may be due to variability in animal response, a short duration of signal detectability or variations in signal strength. As the bony bridge underwent remodeling, the presence of Col3.6 and Col2.3 positive cells were seen up until 56 days. In the rhBMP-2 treated animals osteoblastic differentiation was very active in the periosteum, but no significant osteoprogenitor cellular contribution was noted from the bone marrow in any specimen. The bony bridge formed by the rhBMP-2 was thick and disorganized at day 14, but gradually became noticeably thinner from days 21 to 56. TRAP staining revealed high numbers of the mononuclear osteoclast precursors at day 7 which subsequently turned into active polynuclear osteoclasts at day 14. Although the bridge gradually thinned out, a cortical shell remained intact as far out as day 56 and the defects were considered healed because of the presence of the bony bridge.
Table 1. Col3.6GFPtopaz and Col2.3eGFP Measured at 7, 14, 21, 28, and 56 Days Time Point (Days)
Col 3.6 RhBMP-2
GFPþ cell/TV defect (ROI 1) 7 2.22 1.33 14 2.67 3.31 21 0.69 0.48 28 0.11 0.08 56 0.21 0.26 GFPþ cell/TV Bridge (ROI 2 þ 3) 7 7.2 6.02 14 4.07 2.72 21 2.52 1.45 28 0.44 0.26 56 1.67 1.48
Col 3.6 DBM
Col 3.6 Control
Col 2.3 RhBMP-2
Col 2.3 DBM
Col 2.3 Control
1.00 0.68 0.84 0.76 1.4 0.66 0.59 0.38 1.2 1.26
1.74 2.86 0.87 1.23 1.48 2.09 0.37 0.65 0.70 0.75
1.43 0.70 3.77 1.79 2.53 2.81 0.04 0.03 1.05 1.71
1.43 1.38 2.33 1.14 2.10 0.96 1.50 0.60 1.87 0.96
0.83 0.89 0.84 1.24 5.80 7.5 1.27 0.76 0.94 1.05
NA NA NA NA NA
NA NA NA NA NA
5.31 3.15 8.04 5.06 8.31 4.32 0.44 0.23 2.30 1.44
NA NA NA NA NA
NA NA NA NA NA
p < 0.05 compared to other time points in the same group. Data expressed as mean SD.
JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
EVALUATION OF OSTEOGENIC RESPONSE TO BMP-2 AND DBM
In various animal models, DBM has been noted to be osteoinductive,31,32 which is thought to be associated with the concentration of the BMPs in a specific DBM.33 In our study, the use of DBM in the femoral defects elicited a fundamentally different progenitor cell response compared to the rhBMP-2. DBM did not have the inductive capacity of rhBMP-2 which was evident by the lack of osteogenic response. The osteogenic caps at either end of the bone which seemed to originate from the periosteal layer consistently failed to make the connection between the two ends of the defect. In this respect, the pattern of bone repair in the mice treated with DBM and the controls was clearly different than BMP-2. A limitation of our study is that we did not characterize our DBM with regards to different growth factor concentrations and it is possible that the concentration of the BMPs in this particular DBM was not sufficient to induce bone formation. Although the exact source of the osteogenic cells in response to rhBMP-2 could not be determined by using the current fluorescent markers, there was a significant difference between the rhBMP-2 treated animals and DBM or controls with respect to the patterns of cell recruitment. Our findings strongly suggest that DBMs induce a minimal biologic response in bone defects and should not be used for the same indications or as an alternative to the rhBMP-2. Although BMP-2 is the most potent osteoinductive agent available for clinical use today, there have been concerns regarding inconsistent healing in humans and the quality of bone formed with treatment with rhBMP-2.34 BMP-2 not only stimulates osteoblastic differentiation but also can induce osteoclastogenesis through increased RANKL in the osteoblasts35 and this may influence the quality and quantity of bone repair. In our study the rhBMP-2 healed the defects, but the thinning of the bone was a concern. TRAP staining revealed a prominent osteoclast response at days 7 and 14 which could have a negative impact on the quality of bone. The combined therapy with rhBMP-2 and an osteoclast inhibitor (zoledronic acid or alendronate) has been noted to improve quality of bone repair.20,31,36 These findings are consistent with our analysis that early and robust osteoclast activity may inhibit bone repair. It is not clear if this early osteoclastic response to BMP is related to the dose of BMP or the kinetics of BMP release from the collagen sponge. In conclusion, rhBMP-2 and DBM induced vastly different osteogenic responses in a mouse critical sized femoral defect model. It is essential that clinicians understand the different biologic properties of these agents when choosing therapies to treat bone repair problems.
ACKNOWLEDGEMENT
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
This study was supported by a grant from Musculoskeletal Transplant Foundation to J.R.L.
18.
REFERENCES
19.
1. Murakami N, Saito N, Horiuchi H, et al. 2002. Repair of segmental defects in rabbit humeri with titanium fiber mesh
1127
cylinders containing recombinant human bone morphogenetic protein-2 (rhBMP-2) and a synthetic polymer. J Biomed Mater Res 62:169–174. Vogelin E, Jones NF, Huang JI, et al. 2005. Healing of a critical-sized defect in the rat femur with use of a vascularized periosteal flap, a biodegradable matrix, and bone morphogenetic protein. J Bone Joint Surg Am 87:1323–1331. Boyce AS, Reveal G, Scheid DK, et al. 2009. Canine investigation of rhBMP-2, autogenous bone graft, and rhBMP-2 with autogenous bone graft for the healing of a large segmental tibial defect. J Orthop Trauma 23:685–692. Gerhart TN, Kirker-Head CA, Kriz MJ, et al. 1993. Healing segmental femoral defects in sheep using recombinant human bone morphogenetic protein. Clin Orthop Relat Res 263:317–326. Seeherman HJ, Li XJ, Smith E, et al. 2012. rhBMP-2/ calcium phosphate matrix induces bone formation while limiting transient bone resorption in a nonhuman primate core defect model. J Bone Joint Surg Am 94:1765–1776. Starman JS, Bosse MJ, Cates CA, et al. 2012. Recombinant human bone morphogenetic protein-2 use in the off-label treatment of nonunions and acute fractures: a retrospective review. J Trauma Acute Care Surg 72:676–681. Govender S, Csimma C, Genant HK, et al. 2002. Recombinant human bone morphogenetic protein-2 for treatment of open tibial fractures: a prospective, controlled, randomized study of four hundred and fifty patients. J Bone Joint Surg Am 84-A:2123–2134. Martin GJ Jr, Boden SD, Titus L, et al. 1999. New formulations of demineralized bone matrix as a more effective graft alternative in experimental posterolateral lumbar spine arthrodesis. Spine (Phila Pa 1976) 24:637–645. Morone MA, Boden SD. 1998. Experimental posterolateral lumbar spinal fusion with a demineralized bone matrix gel. Spine (Phila Pa 1976) 23:159–167. Wang JC, Alanay A, Mark D, et al. 2007. A comparison of commercially available demineralized bone matrix for spinal fusion. Eur Spine J 16:1233–1240. Huang YZ, Cai JQ, Xue J, et al. 2012. The poor osteoinductive capability of human acellular bone matrix. Int J Artif Organs 35:1061–1069. Barradas AM, Yuan H, van Blitterswijk CA, et al. Osteoinductive biomaterials: current knowledge of properties, experimental models and biological mechanisms. Eur Cell Mater 21:407–429; discussion 429. Bilic-Curcic I, Kalajzic Z, Wang L, et al. 2005. Origins of endothelial and osteogenic cells in the subcutaneous collagen gel implant. Bone 37:678–687. Kalajzic I, Kalajzic Z, Kaliterna M, et al. 2002. Use of type I collagen green fluorescent protein transgenes to identify subpopulations of cells at different stages of the osteoblast lineage. J Bone Miner Res 17:15–25. Kalajzic I, Kalajzic Z, Wang L, et al. 2007. Pericyte/ myofibroblast phenotype of osteoprogenitor cell. J Musculoskelet Neuronal Interact 7:320–322. Kalajzic Z, Li H, Wang LP, et al. 2008. Use of an alphasmooth muscle actin GFP reporter to identify an osteoprogenitor population. Bone 43:501–510. Marijanovic I, Jiang X, Kronenberg MS, et al. 2003. Dual reporter transgene driven by 2.3Col1a1 promoter is active in differentiated osteoblasts. Croat Med J 44:412–417. Bilic-Curcic I, Kronenberg M, Jiang X, et al. 2005. Visualizing levels of osteoblast differentiation by a two-color promoter-GFP strategy: type I collagen-GFPcyan and osteocalcinGFPtpz. Genesis 43:87–98. Ushiku C, Adams DJ, Jiang X, et al. Long bone fracture repair in mice harboring GFP reporters for cells within the osteoblastic lineage. J Orthop Res 28:1338–1347. JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
1128
ALAEE ET AL.
20. Schindeler A, Birke O, Yu NY, et al. Distal tibial fracture repair in a neurofibromatosis type 1-deficient mouse treated with recombinant bone morphogenetic protein and a bisphosphonate. J Bone Joint Surg Br 93:1134–1139. 21. Alaee F, Sugiyama O, Virk MS, et al. 2014. Suicide gene approach using a dual-expression lentiviral vector to enhance the safety of ex vivo gene therapy for bone repair. Gene Ther 21:139–147. 22. Lieberman JR, Daluiski A, Stevenson S, et al. 1999. The effect of regional gene therapy with bone morphogenetic protein-2-producing bone-marrow cells on the repair of segmental femoral defects in rats. J Bone Joint Surg Am 81:905–917. 23. Jiang X, Kalajzic Z, Maye P, et al. 2005. Histological analysis of GFP expression in murine bone. J Histochem Cytochem 53:593–602. 24. Filgueira L. 2004. Fluorescence-based staining for tartrateresistant acidic phosphatase (TRAP) in osteoclasts combined with other fluorescent dyes and protocols. J Histochem Cytochem 52:411–414. 25. Peterson B, Zhang J, Iglesias R, et al. 2005. Healing of critically sized femoral defects, using genetically modified mesenchymal stem cells from human adipose tissue. Tissue Eng 11:120–129. 26. Feeley BT, Gamradt SC, Hsu WK, et al. 2005. Influence of BMPs on the formation of osteoblastic lesions in metastatic prostate cancer. J Bone Miner Res 20:2189–2199. 27. Hsu WK, Virk MS, Feeley BT, et al. 2008. Characterization of osteolytic, osteoblastic, and mixed lesions in a prostate cancer mouse model using 18F-FDG and 18F-fluoride PET/ CT. J Nucl Med 49:414–421. 28. Hong SH, Jiang X, Chen L, et al. 2012. Computer-automated static, dynamic and cellular bone histomorphometry. J Tissue Sci Eng S1:004.
JOURNAL OF ORTHOPAEDIC RESEARCH SEPTEMBER 2014
29. Parfitt AM, Drezner MK, Glorieux FH, et al. 1987. Bone histomorphometry: standardization of nomenclature, symbols, and units. Report of the ASBMR Histomorphometry Nomenclature Committee. J Bone Miner Res 2:595–610. 30. Otsu N. 1978. A threshold selection method from gray-scale histogram. IEEE Trans Syst Man Cybern 8:62–66. 31. Vandermeer JS, Kamiya N, Aya-ay J, et al. Local administration of ibandronate and bone morphogenetic protein-2 after ischemic osteonecrosis of the immature femoral head: a combined therapy that stimulates bone formation and decreases femoral head deformity. J Bone Joint Surg Am 93:905–913. 32. Arrington ED, Smith WJ, Chambers HG, et al. 1996. Complications of iliac crest bone graft harvesting. Clin Orthop Relat Res 329:300–309. 33. Gruskin E, Doll BA, Futrell FW, et al. Demineralized bone matrix in bone repair: history and use. Adv Drug Deliv Rev 64:1063–1077. 34. Fu R, Selph S, McDonagh M, et al. 2013. Effectiveness and harms of recombinant human bone morphogenetic protein-2 in spine fusion: a systematic review and meta-analysis. Ann Intern Med 158:890–902. 35. Granholm S, Henning P, Lindholm C, et al. Osteoclast progenitor cells present in significant amounts in mouse calvarial osteoblast isolations and osteoclastogenesis increased by BMP-2. Bone 52:83–92. 36. Doi Y, Miyazaki M, Yoshiiwa T, et al. Manipulation of the anabolic and catabolic responses with BMP-2 and zoledronic acid in a rat femoral fracture model. Bone 49:777–782.
SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at the publisher’s web-site.
