Abstract
Bone‐lining tissues contain a population of resident macrophages termed osteomacs that interact with osteoblasts in vivo and control mineralization in vitro. The role of osteomacs in bone repair was investigated using a mouse tibial bone injury model that heals primarily through intramembranous ossification and progresses through all major phases of stabilized fracture repair. Immunohistochemical studies revealed that at least two macrophage populations, F4/80+Mac‐2−/lowTRACP− osteomacs and F4/80+Mac‐2hiTRACP− inflammatory macrophages, were present within the bone injury site and persisted throughout the healing time course. In vivo depletion of osteomacs/macrophages (either using the Mafia transgenic mouse model or clodronate liposome delivery) or osteoclasts (recombinant osteoprotegerin treatment) established that osteomacs were required for deposition of collagen type 1+ (CT1+) matrix and bone mineralization in the tibial injury model, as assessed by quantitative immunohistology and micro–computed tomography. Conversely, administration of the macrophage growth factor colony‐stimulating factor 1 (CSF‐1) increased the number of osteomacs/macrophages at the injury site significantly with a concurrent increase in new CT1+ matrix deposition and enhanced mineralization. This study establishes osteomacs as participants in intramembranous bone healing and as targets for primary anabolic bone therapies. © 2011 American Society for Bone and Mineral Research.
Introduction
The distinct processes of bone modeling and remodeling depend on the tightly regulated actions of three cell types: bone‐forming osteoblasts, mechanosensing osteocytes, and bone‐resorbing osteoclasts.1 Bone modeling, which occurs during development and is induced in adults by mechanical loading2 or injury,3 proceeds via uncoupled anabolic osteoblast bone formation and/or catabolic osteoclast bone resorption at distinct anatomic locations.4 In contrast, bone remodeling is achieved through quantitatively coupled sequential events of osteoclastic bone resorption and osteoblastic bone formation within the spatial confines of basic multicellular units (BMUs).5 The primary purpose of remodeling is to replace compromised bone in the adult skeleton while maintaining peak bone density and skeletal integrity.
Inadequate treatment options for improving and accelerating fracture repair and sustaining peak bone mass throughout life are a significant and increasing clinical problem.6 There are few anabolic bone therapies available to rebuild bone, and these treatments modulate the cellular and molecular machinery of the BMU so that the equilibrium within this unit is shifted to favor bone formation.7 This mechanism of action dictates that these anabolic therapies still depend at least in part on osteoclast production of a “coupling” factor.8 Given that the first‐line therapy for failed fracture repair and diseases of bone loss are antiresorptive agents (eg, the bisphosphonate drug class) and that many of these treatments have extended half‐lives, the efficacy of current anabolic therapies is likely to be compromised in individuals who have had prior antiresorptive therapy.9 There is therefore an unmet clinical need for effective anabolic agents that bypass the BMU machinery and target bone formation independent of bone resorption.7
The purest form of bone formation is intramembranous ossification, during which bone matrix is formed and deposited directly by osteoblasts with limited or unknown input from other cell lineages. Intramembranous ossification is poorly defined, and this knowledge gap limits the ability to specifically harness this process therapeutically. A recent development in understanding the complexity of bone biology is the appreciation that there is ongoing and dynamic interaction between the skeletal and immune systems.10 We have demonstrated that a population of tissue macrophages (osteomacs) resides within bone lining tissues in both mice11, 12 and humans.12 Osteomacs were localized to sites of bone modeling in vivo, where they formed a distinctive canopy structure over mature osteoblasts.12 Like many other tissue macrophage populations,13 osteomacs express the epidermal growth factor seven transmembrane (EGF‐TM7) protein EMR1/F4/80,11 which clearly distinguishes them from osteoclasts. Osteomacs enhanced osteoblast mineralization in vitro and were required for maintenance of mature osteoblasts in vivo.12 In combination with their anatomic location (within three cells of a bone surface and intercalated or associated with other bone lining cells), these observations suggest that osteomacs are likely to have a role in anabolic bone modeling.14
The development of macrophages and osteoclasts is controlled by macrophage colony‐stimulating factor 1 (CSF‐1),13 which signals through the CSF‐1 receptor. Animals that lack endogenous CSF‐1 (op/op mouse15 and toothless;tl/tl rat16) or the CSF‐1 receptor17 are osteopetrotic and have severe growth retardation. Mutation of the CSF‐1 receptor17 generates a similar but more severe phenotype that probably is attributable to the presence of an alternative CSF‐1 receptor ligand, interleukin 34 (IL‐34).18 We previously generated a transgenic mouse line in which the mouse csf1r promoter directs expression of an enhanced green fluorescent protein (eGFP) reporter.19 This transgene also highlighted the presence of eGFP‐expressing osteomacs on bone surfaces and as major contaminants of primary osteoblast cultures.12 In addition to CSF‐1, the formation of osteoclasts requires signaling from receptor activator of NF‐κB ligand (RANKL),20 and this signal can be blocked by osteoprotegerin.21
In this study, we investigated the contribution of osteomacs to intramembranous ossification using a mouse model of bone healing.22 We manipulated the contribution of macrophages to bone formation by eliminating either csf1r‐expressing cells or phagocytic macrophages. Osteoclast‐mediated contributions to the bone‐formation process were also assessed using in vivo osteoprotegerin (OPG) treatment. Finally, the therapeutic potential of manipulating osteomacs/macrophages to facilitate bone formation was investigated through the administration of recombinant CSF‐1.
Materials and Methods
Animals
Macrophage fas‐induced apoptosis (Mafia) transgenic mice23 and C57Bl/6 control animals were obtained from institutional or commercial (Animal Resources Center, Canning Vale, WA, USA) breeding colonies. Mafia mice, which express a ligand‐inducible Fas‐dependent suicide receptor under the control of the myeloid‐specific csf1r promoter,19 were a generous gift from Professor D Cohen (Department of Microbiology, Immunology and Molecular Genetics, University of Kentucky, Lexington, KY, USA). The University of Queensland Molecular Biosciences and Health Sciences Ethics Committees approved all protocols involving animals.
Tibial bone injury model
Eleven‐ to 12‐week‐old mice were anesthetized, and hair was removed from the left hind limb. An incision was made in the skin over the medial aspect of the proximal tibia. Soft tissue was cleared from the distal end of the tibial crest, and a hole (0.8 mm in diameter) that penetrated through both lateral cortices and the intervening medulla was created in the bone using a 21‐gauge needle attached to an electric drill.22
Depletion of macrophages during bone healing using the Mafia transgenic mouse model
Mafia transgenic mice were treated with vehicle control or the osteomac/macrophage‐depleting ligand AP20187 (a generous gift from ARIAD Pharmaceuticals, Inc., Cambridge, MA, USA, www.ariad.com/regulationkits) via either direct injection into the tibial injury site at the time of surgery (n = 6/group) or intravenous retro‐orbital delivery at 3 days after surgery (n = 3/group). AP20187 ligand was prepared as described previously.12
Depletion of macrophages during bone healing using clodronate liposome treatment
Clondronate (Cl2MDP, a gift of Roche Diagnostics GmbH, Mannheim, Germany) was encapsulated in liposomes containing phosphatidylcholine (Lipoid GmbH, Ludwigshafen, Germany) and cholesterol (Sigma‐Aldrich, St Louis, MO, USA) as described previously.24 C57Bl/6 mice were treated with an intradefect injection (100 µL) of control PBS liposomes or osteomac/macrophage‐depleting clodronate liposomes (equal to 7 mg/mL of clodronate in suspension) at the time of surgery (n = 9/group), followed by daily intraperitoneal injections (10 µL/g).
OPG treatment during bone healing
C57Bl/6 mice were administered OPG (Pepro‐Tech, Rocky Hill, NJ, USA) or appropriate vehicle via an intradefect injection at the time of surgery plus subcutaneous injections every second day (2, 4, 6, and 8 days after surgery; n = 6/group on day 7 and n = 4/group on day 9). OPG was diluted in saline to a concentration of 0.5 mg/mL and delivered at a final dose of 1 mg/kg.
CSF‐1 treatment during bone healing
C57Bl/6 mice were administered mouse recombinant CSF‐1 (Pepro‐Tech) or appropriate vehicle intradefect at the time of surgery (n = 5/group) plus one subcutaneous injection 2 days after surgery. CSF‐1 was diluted in saline to a concentration of 0.1 mg/mL, delivering a final bolus dose of 10 µg/mouse.
Tissue collection
Left hind limbs were dissected and fixed overnight in 4% paraformaldehyde (PFA; Sigma, St Louis, MO, USA) at 4°C. Bones for histologic analysis were decalcified for at least 2 weeks in 14% ethylenediaminetetraacetic acid (EDTA; Sigma), pH 7.2. Once decalcified, all specimens were processed for paraffin embedding using a Shandon Pathcenter Processor (Thermo Electron Corporation, Waltham, MA, USA). Bones for micro–computed tomographic (uCT) analysis were placed in 69% ethanol following fixation and shipped to Zürich.
Flow cytometric analysis
Bone marrow was harvested from the contralateral limb, and 1 × 106 bone marrow cells were blocked with 0.5% FBS/PBS before incubation with anti‐F4/80‐RPE alone or in combination with anti‐CD11b‐Alexa 647 (AbD Serotec, Kidlington, Oxford, UK) or relevant directly labeled rat IgG2b isotype control antibodies (AbD Serotec). Cells then were washed twice, fixed using 4% PFA (Sigma), and examined by flow cytometry on a BD LSR II Analyser (BD Biosciences Pharmingen, San Jose, CA, USA) with 50,000 events analyzed. Data were analyzed using the Weasel software (Walter and Eliza Hall, Institute of Medical Research, Victoria, Australia).
Immunohistochemistry
Immunohistochemistry (IHC) was performed on deparaffinized and rehydrated sections, as described previously,12 with specific primary antibodies: F4/80 (rat anti‐mouse; AbD Serotec), collagen type 1 (rabbit anti‐mouse; US Biological, Swampscott, MA, USA), osteocalcin (OCN; rabbit anti‐mouse; Alexis Biochemicals, San Diego, CA, USA), and Mac‐2 (rat anti‐mouse; Alexis Biochemicals), or relevant isotype control antibodies [normal rat IgG2b (AbD Serotec), rat IgG2a (BD Bioscience Pharmingen), and normal rabbit IgG (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA)]. All sections were counterstained using Mayer's hematoxylin (Sigma‐Aldrich) and mounted using permanent mounting medium (Thermo Fisher Scientific, Waltham, MA, USA). Tissue staining was viewed using a Nikon Eclipse 80i microscope with a Nikon D5‐Ri1 camera (Nikon, Tokyo, Japan) and NIS‐Elements imaging software (Nikon).
TRACP staining
Dewaxed slides were stained for tartrate‐resistant acid phosphatase (TRACP), as described previously.25
Serum TRACP‐5b assay
Blood was isolated from anesthetized OPG‐ and vehicle‐treated mice via cardiac puncture at 7 days after treatment using a 1 mL syringe and 27‐gauge needle. Blood samples were allowed to clot at 37°C and centrifuged at 1500 rpm for 5 minutes, and the serum supernatant was collected and stored at –20°C. Levels of serum TRACP isoform 5b (TRACP‐5b) expressed by bone‐resorbing osteoclasts26 were assayed using the MouseTRAP Assay TRACP‐5b ELISA kit (Immunodiagnostic Systems, Boldon, Tyne and Wear, UK) according to the manufacturer's instructions.
Quantitative immunohistology
The percent area of CT1+ new woven bone matrix was quantified in 5‐µm sections of the injury site IHC stained for CT1 expression. Three independent measurements were collected for each sample at three sectional depths at least 50 µm apart. Multiple images (×20 magnification) representing the entire injury site were collected for each section and subsequently tiled together in Adobe Photoshop (Adobe Systems, San Jose, CA, USA). The images then were cropped and imported into ImageJ image analysis software (National Institutes of Health, Bethesda, MD, USA) to allow quantification of the percent area of CT1+ woven bone matrix within the intramedullar region of the injury site. Quantification of macrophage number within the injury site was carried out on 5‐µm serial sections stained for either F4/80 or Mac‐2 expression. Three representative images (×40 magnification) were taken within the intramedullar injury zone for each sample at two sectional depths at least 50 µm apart. The number of F4/80+ and Mac‐2+ cells was quantified by counting each positively stained cell in each field of view (FOV). Quantification of total number of TRACP+ cells and the average surface area of each single TRACP+ cell within the injury site was performed at two sectional depths for each sample at least 50 µm apart. Images were collected (×40 magnification) and tiled to create a collaged image representing the entire intramedullar injury zone. A representative region of interest (ROI) within the intramedullar injury zone containing woven bone was cropped (area of ROI was the same for each sample) from the tiled image. The number of TRACP+ cells and the average surface area (µm2) of each TRACP+ cell were quantified using ImageJ image analysis software. Quantifications of the percent area of F4/80, Mac‐2, and TRACP staining within the injury site or growth plate were performed at two sectional depths for each sample at least 50 µm apart. Images were collected (×10 magnification) and tiled to create a collaged image representing either the entire intramedullar injury zone or the entire growth plate spongiosa at at least 200 µm in length, with measurement initiating from the base of the hypotrophic chondrocyte zone. The areas of positive staining for F4/80 and Mac‐2 expression or TRACP activity within the defined and cropped ROIs were quantified using ImageJ software.
Micro–computed tomography (µCT)
The µCT imaging system (µCT 40, Scanco Medical AG, Brüttisellen, Switzerland) used in this study was equipped with a 5‐µm focal spot X‐ray tube as a source and a 2D charge coupled device (CCD) coupled with a thin scintillator as a detector. The long axis of the tibia was oriented perpendicular to the rotation axis of the scanner. The integration time for capturing the projections was set to 200 ms, and projections were taken two times and then averaged (frame averaging). Scans were performed at an isotropic nominal resolution of 20 µm (medium resolution mode). A cylindrical ROI with the dimensions of the primary injury site was calculated based on the diameter of the drill needle (0.8 mm) and the width of the bone at the injury site (ranged between 0.8 and 1.4 mm) and then placed in the digital image data to select the injury area. This volume of interest included both the intercortical and intramedullar injury site zones. A constrained 3D Gaussian filter (σ = 0.8; support of one voxel) was used to partly suppress the noise in the volumes, and the mineralized tissue was segmented from soft tissues by a global thresholding procedure.27 The bone volume density (BV/TV) then was determined within the cylindrical injury site ROI. The measurements and analyses were performed according to the guidelines for assessment of bone microstructure in rodents using µCT.28
Statistical analysis
Statistically significant differences were determined using unpaired t tests with one‐ or two‐tailed distributions using PRISM 5 (GraphPad software, La Jolla, CA, USA). A value of p < 0.05 was deemed statistically significant. In all cases, data are represented as mean ± standard error of mean (SEM).
Results
Osteomacs are present during multiple phases of intramembranous bone healing in vivo
To examine the distribution of macrophages and, in particular, osteomacs during intramembranous ossification, a tibial bone injury model was used.22, 25 We initially confirmed histologically that this model heals through the standard phases of stabilized fracture healing, including an inflammation phase (days 1 to 3; data not shown), anabolic bone‐modeling phase (days 4 to 7; Supplemental Fig. S1A, C, and E), and a catabolic modeling/remodeling phase (days 8 to 12; Supplemental Fig. S1G), as has been reported previously.22 IHC analysis focused on early (days 4 to 5) and late (day 7) anabolic healing and catabolic modeling (day 9) phases. To simplify anatomic descriptions, different areas of the injury site were subdivided into the following regions: periosteal, intramedullar, intercortical, and peripheral injury zones (Fig. 1A). Supplemental Fig. S1 contains sagittal cross‐sectional images of the entire injury site across this time course stained for CT1 or F4/80 expression and clearly shows the bone‐healing process and the overall distribution of F4/80+ macrophages in all regions of the injury site. We have focused largely on the intramedullar injury zone because this region is the principal site of de novo woven bone deposition.
Osteomac distribution during intramembranous bone healing in vivo. (A) Schematic illustrating the different areas of bone healing: peripheral (green box), intramedullar (red box), intercortical (purple box), and periosteal (black box) injury zones. (B–M) IHC and histologic staining of tibial injury sites for anti‐CT1 (B, E, H, and K), anti‐F4/80 (C, F, I, and L), and TRACP (D, G, J, and M) in 12‐week‐old C57Bl/6 mice at 4 (B–D), 5 (E–G), 7 (H–J), and 9 (K–M) days after surgery (n = 5). B) At 4 days after surgery, areas of CT1+ matrix/osteoid were present within the intramedullar injury zone (brown). (C) Serial section stained for F4/80 expression, at a higher magnification (boxed area in B), showing F4/80+ macrophages (brown) intercalated among the CT1+ matrix/osteoid. (D) TRACP+ osteoclast‐like cells were present primarily in the peripheral injury zone (arrow). (E) At 5 days after surgery, areas of new CT1+ matrix deposition (arrow) were present throughout the intramedullar and intercortical injury zones (marked area). (F) Serial section of panel E stained for F4/80 expression (brown) at a higher magnification (boxed area in E) illustrating F4/80+ cells interlaced between areas of CT1+ matrix (arrows). (G) Minimal TRACP activity in the intramedullar zone (crosshatch). TRACP+ osteoclast‐like cells are present primarily in the peripheral injury zone on fragments of original bone (arrow). (H) At 7 days after surgery, CT1+ woven bone (brown) bridged the injury site along with fragments of original cortical bone (crosshatch). (I) Higher‐power magnification of boxed area of panel H stained for F4/80 expression demonstrating numerous F4/80+ macrophages (brown) interlaced throughout the woven bone. (J) Few TRACP+ mononuclear cells (arrows) present, indicating an absence of mature osteoclasts. (K) On 9 days after surgery, areas of CT1+ woven bone have been removed from the central core of the injury site. (L) Serial section stained for F4/80 expression at higher magnification (boxed area in K) illustrating that F4/80+ macrophages (brown) are intercalated throughout the remaining woven bone. (M) TRACP+ osteoclast‐like cells were present throughout the residual woven bone. Original magnification for panels B, E, H, and K ×10 and for C, D, F, G, I, J, L, and M ×20.
By 4 days after surgery, the hematoma was being replaced by cellular granulation tissue in the intramedullar, intercortical, and peripheral injury zones (Fig. 1B and Supplemental Fig. S1A, GT). Localization of CT1 expression demonstrated that original CT1+ bone fragments generated during surgery were present predominantly within the peripheral injury zone (Fig. 1B, arrowhead). Within the intramedullar zone, small foci of de novo CT1+ extracellular matrix deposition were observed (Fig. 1B, arrows). At 4 days after surgery, F4/80+ mature macrophages were located predominantly in the peripheral injury zone (Supplemental Fig. S1B, brown) and in regions where CT1+ matrix was deposited (Fig. 1C). In contrast, few TRACP+ osteoclast‐like cells were detectable within the intramedullar zone of the injury site (data not shown). TRACP+ cells were restricted primarily to the peripheral injury zone on original bone fragments generated during surgery (Fig. 1D, arrow).
By 5 days after surgery, CT1+ deposition was detectable throughout the intramedullar region and extended into the peripheral and intracortical injury zones (Fig. 1E and Supplemental Fig. S1C). Compared with day 4, this matrix had matured to resemble woven osteoid, including embedded osteoblast‐osteocyte‐like cells. F4/80+ mature macrophages were interlaced throughout the CT1+ matrix (Fig. 1F, arrows, and Supplemental Fig. 1D). By contrast, few cells within the intramedullar mineralized woven bone bridge were TRACP+ (Fig. 1G, crosshatch), with TRACP+ cells seen primarily only in the peripheral injury zone (Fig. 1G, arrow).
By 7 days after surgery, extensive CT1+ matrix deposition was detected within the intramedullar and intercortical injury zones, completely bridging the injury site (Fig. 1H, and Supplemental Fig. S1E). µCT scanning confirmed that this CT1+ matrix was mature mineralized bone. Toluidine blue staining did not detect any proteoglycan‐rich cartilaginous material, indicating that this bone was formed via intramembranous ossification (data not shown). F4/80+ mature macrophages were interlaced throughout the mineralized CT1+ woven bone bridge (Fig. 1I and Supplemental Fig. 1F). Distribution of TRACP+ cells (Fig. 1J, arrows) closely resembled observations on day 5 after surgery, as described earlier.
At 9 days after surgery, the distribution of CT1+ woven bone within the injury site indicated that catabolic modeling/remodeling of the bone bridge had been initiated. A cleared channel was evident within the core of the injury site parallel to the longitudinal axis of the cortical bone. This channel contained a large blood vessel that traversed from the bone marrow through the peripheral and intramedullar injury zones (Fig. 1K, BV and arrows, and Supplemental Fig. S1G) and extended into bone marrow on the opposite side. Normal bone marrow tissue also had invaded into the channel in place of the woven bone that was present 2 days prior. Similar to results seen on day 7, F4/80+ mature macrophages were intercalated throughout the remaining woven bone matrix present within the peripheral, intramedullar, and intercortical injury zones (Fig. 1L and Supplemental Fig. S1H). By 9 days after surgery, a large number of TRACP+ multinucleated osteoclast‐like cells (Fig. 1M, arrows) were dispersed throughout the woven bone. Overall, these observations confirmed that F4/80+ tissue macrophages were present during all stages of bone repair specifically associated with CT1+ matrix production from the earliest stages of matrix deposition to late stages of bone mineralization. Osteoclasts were detected within the intramedullar zone only during the catabolic modeling stage.
Osteomacs are intimately associated with osteoblasts on new woven bone surfaces and do not express TRACP
During physiologic bone formation, osteomacs are in direct contact with CT1+OCN osteoblasts, forming an intriguing canopy‐like structure over these cells on bone surfaces.12 We therefore sought evidence of a similar canopy structure over osteoblasts during bone healing. Many of the F4/80+ macrophages within the intramedullar zone were located within three cell diameters of the matrix‐osteoid‐bone surface on day 5 (Fig. 2A), day 7 (Fig. 2C), and day 9 after surgery (data not shown). Staining in serial sections demonstrated that on days 5 and 7, these F4/80+ osteomacs (Fig. 2A, C, arrows) were intimately associated with mature OCN+ osteoblasts (Fig. 2B, D, arrowheads), forming the distinctive canopy structure described previously.12 Thus osteomacs are in direct contact with matrix‐producing and ‐mineralizing osteoblasts throughout the anabolic bone‐healing response.
F4/80+/TRACP− osteomacs form a canopy over osteoblasts on new bone surfaces. (A–G) IHC and histologic staining of tibial injury sites in 12‐week‐old C57Bl/6 mice stained for anti‐F4/80 (A, C), anti‐OCN (B, D), and TRACP and F4/80 double staining (E–G). Images in panels A through D are serial sections with landmarks denoted as crosshatches. F4/80 expression (A, C, brown) illustrating F4/80+ osteomacs forming a canopy structure (arrows) over OCN+ (B, D, arrowheads) osteoblasts sitting on new woven bone surfaces on both day 5 (A, B) and day 7 (C, D) after surgery. (E) At 7 days after surgery, IHC double staining for TRACP/F4/80 expression illustrating a large number of F4/80+ osteomacs (E, arrows) and a small number of TRACP+ cells present among the newly deposited woven bone. (F) IHC double staining at 9 days after surgery illustrating TRACP+ multinucleated osteoclasts present on woven bone surfaces (arrowheads), as well as F4/80+ macrophages and osteomacs (arrows). (G) Rare colocalization of TRACP and F4/80 expression illustrating a TRACP+ intracellular vesicle (arrow) present in an F4/80+ macrophage. Original magnification for A–F ×40 and for G ×60.
Under physiologic conditions, osteomacs are TRACP− (data not shown). As described earlier, few TRACP+ multinucleated osteoclast‐like cells were present within the intramedullar injury zone during the anabolic bone‐modeling phase (days 4 to 7). However, a limited number of TRACP+ cells were present in the peripheral injury zone. At 7 days after surgery in the peripheral injury zone, F4/80 and TRACP double staining confirmed that F4/80+ osteomacs forming a canopy over cuboidal osteoblast‐like cells on new bone surfaces were TRACP− (Fig. 2E, arrows) and were a distinct population from F4/80−TRAP+ multinucleated osteoclast‐like cells (Fig. 2E, arrowhead).
By 9 days after surgery, TRACP+ multinucleated osteoclasts (Fig. 2F, arrowheads), both F4/80+ mature macrophages (F4/80‐expressing cells greater than three cell diameters from a bone surface) and F4/80+ osteomacs (Fig. 2F, arrows) were present throughout the new woven bone remaining at the injury site. TRACP+ multinucleated osteoclasts did not coexpress F4/80 (Fig. 2F, arrowheads), and F4/80+ osteomacs were TRACP− (Fig. 2F, arrows). Overlap of TRACP and F4/80 expression was not detected in any multinucleated cells within the injury site at any of the time points examined. In the rare instances where staining was colocalized, it occurred in F4/80+ macrophages, with the TRACP activity restricted to intracellular vesicles (Fig. 2G, arrow) in a pattern suggesting that the enzyme was phagocytosed rather than endogenously expressed by these cells. These observations clearly indicated that osteomacs and osteoclasts are distinct myeloid populations within the injury site and support the idea that osteomacs are undertaking a specific functional role at this site that is independent of their undetermined potential to serve as an osteoclast precursor during bone repair.
Osteomacs are the injury‐associated macrophage cells that are enriched in areas of bone matrix deposition during bone healing
Macrophages have been detected previously during fracture healing29, 30 and were inferred to be important for early inflammatory events that precede repair.31, 32 To assess the inflammatory status of macrophages in healing bone, we performed IHC staining in serial sections using anti‐F4/80 and anti‐Mac‐2 antibodies. Mac‐2 detects galectin‐3, which has been considered an inflammatory macrophage marker, especially in wound repair and fibrosis.33 Mac‐2 is also detected in osteoclasts and their progenitors.34 At 5 days after surgery, F4/80+ macrophages were distributed throughout all areas of the injury site (Fig. 3B). Mac‐2‐expressing cells were concentrated in the peripheral injury zone (Fig. 3C) with only small clusters of Mac‐2+ cells within the intramedullar zone. Direct comparison of F4/80‐ and Mac‐2‐staining patterns within serial sections allowed clear distinction of at least two macrophage populations in the injury zone: F4/80+Mac‐2−/low osteomacs forming canopy over cuboidal osteoblasts on bone surfaces (Fig. 3D–F, arrows) and F4/80+Mac‐2hi inflammatory macrophages (Fig. 3E, F, arrowheads). Similar to day 5 observations described earlier, by 7 days after surgery, there are significantly more F4/80+ cells within the intramedullar injury zone than Mac‐2+ cells (Fig. 3G–I, p < 0.0001), indicating that the number of osteomacs (F4/80+Mac‐2−/low) within this zone is greater than the number of inflammatory macrophages (F4/80+Mac‐2hi). F4/80+Mac‐2−/low osteomacs can be seen forming a canopy over osteoblasts on new bone surfaces (Fig. 3G, H, arrows) with only small clusters of F4/80+Mac‐2hi inflammatory macrophages (Fig. 3G, H, circled area).
Two macrophage subsets are present during intramembranous bone healing. (A–H) Tibial injury sites from 12‐week‐old C57Bl/6 mice at 5 (A–F) and 7 (G, H) days after surgery. Images in panels A–C and G and H are near serial sections with landmarks denoted by crosshatches. (A) CT1+ matrix/osteoid is deposited at the injury site (brown). (B) F4/80 expression (brown) illustrating F4/80+ mature tissue macrophage distribution throughout the entire injury site. (C) Mac‐2 expression (brown) illustrating Mac‐2+ inflammatory cells located predominantly in the peripheral injury zone. Higher‐magnification images of the peripheral injury zone (boxed areas), clearly distinguishing the presence of and distribution of both F4/80+Mac‐2−/low osteomacs on a bone surface (D–F, arrows) and F4/80+Mac‐2hi inflammatory macrophages (E, F, arrowheads). (G, H) Representative images taken within the intramedullary injury zone 7 days after surgery demonstrating the presence of F4/80+Mac‐2−/low osteomacs (arrows), as well as small clusters of F4/80+Mac‐2hi inflammatory macrophages (circled area). (I) Quantification of F4/80+ and Mac‐2+ cells within the intramedullar injury zone 7 days after surgery demonstrating significantly more F4/80+ cells/FOV compared with Mac‐2+ cells. Original magnification for A–C ×10 and for D–H ×40. Statistically significant differences were calculated using unpaired t tests between treatments. ap < 0.0001.
Osteomac/macrophage depletion in Mafia transgenic mice at the time of tibial injury surgery significantly suppressed woven bone deposition and mineralization during bone healing in vivo
Systemic depletion of macrophages, and in particular, osteomacs, using Mafia transgenic mice ablates osteoblast bone‐forming surfaces at sites of physiologic endocortical modeling.12 We investigated whether this dependence was recapitulated during bone healing. Macrophage depletion during bone healing using Mafia transgenic mice was carried out via an intradefect injection of vehicle or AP20187 ligand at the time of surgery. Flow cytometric analysis confirmed a significant reduction in F4/80+ mature bone marrow macrophages (average 46.83% reduction, p = 0.01) within the contralateral limb bone marrow 7 days after injection in AP20187 ligand compared with vehicle treated mice.
In vehicle‐treated Mafia mice, F4/80+ osteomacs (Fig. 4A, arrow) were intercalated throughout the CT1+ new woven bone that fully bridged the injury site (Fig. 4B, brown). AP20187 ligand–treated Mafia mice had a marked reduction in F4/80+ cells throughout the entire injury site (Fig. 4C) and a dramatic suppression in CT1+ new woven bone deposition (Fig. 4D, brown). Small spicules of CT1+ new woven bone were present in the peripheral and intercortical injury zones in AP20187 ligand–treated mice (Fig. 4D, arrow), but the repair response failed to achieve injury site bridging. Suppression of new bone formation was also noted in AP20187 ligand treatment groups on days 5 and 9 after surgery (data not shown). Quantitative immunohistology was employed to quantify the percent area of CT1+ woven bone within the intramedullar region of the injury site. In vehicle‐treated mice, CT1+ woven bone constituted up to 40% of the intramedullar injury zone, whereas it was less than 20% in AP20187 ligand–treated mice (Fig. 4E). AP20187 ligand treatment also resulted in a significant suppression of new mineralized bone deposition detected using µCT (Fig. 4F). 3D reconstructed images of the proximal tibia were generated, and a cylindrical region representative of the drill diameter and bone width was overlaid at the injury site, and mineralized matrix within this region was pseudocolored in red (Fig. 4G). In vehicle‐treated Mafia mice, the injury site was completely bridged by new mineralized bone (red), whereas in AP20187 ligand–treated mice, new mineralized bone was restricted to the margins of the injury site (Fig. 4G). These observations demonstrate that AP20187 ligand treatment in Mafia mice significantly impaired new woven bone deposition and mineralization during bone healing.
Osteomac/macrophage depletion in Mafia mice significantly represses new bone formation and mineralization in vivo. (A–D) Day 7 tibial injury sites stained with anti‐F4/80 or anti‐CT1 antibody in 12‐week‐old Mafia mice treated with vehicle or AP20187 ligand at the time of surgery. Images in panels A through D are serial sections with landmarks denoted with crosshatches. In vehicle‐treated mice, F4/80+ osteomacs (A, arrow) are interlaced throughout CT1+ new woven bone (B, brown). In AP20187 ligand–treated mice, small numbers of F4/80+ osteomacs and mature tissue macrophages (C), together with conservative CT1+ woven bone deposition (D, arrow), were noted within the injury site. Quantitative immunohistology of CT1+ woven bone deposition (E) and µCT quantification of bone mineral density (F) within the injury site at 7 days after surgery showed a significant reduction in bone deposition in response to AP20187 ligand treatment. (G) 3D reconstructed µCT images of the injury site with new mineralized bone demarked in red. (H–K) Day 7 tibial injury sites stained with anti‐F4/80 or anti‐CT1 antibody in 12‐week‐old Mafia mice treated with vehicle or AP20187 ligand 3 days after surgery. In vehicle‐treated Mafia mice, F4/80+ osteomacs (H, arrows) were intercalated throughout new CT1+ woven bone (I, brown). (J) In delayed AP20187 ligand–treated mice at a central depth (intramedullar injury zone within the middle region of the injury), a mild reduction in F4/80+ cells was observed. (K) CT1 expression also was reduced after delayed AP20187 ligand treatment. (L–N) Quantification of CT1+ woven bone deposition in vehicle‐ and AP20187 ligand–treated Mafia mice at shallow (L), moderate (M), and midline (N) depths. Results demonstrated a significant decrease in CT1+ woven bone deposition at the moderate (M) and midline (N) depths in AP20187 ligand– compared with vehicle‐treated mice. Statistically significant differences were calculated using unpaired t tests between treatments. ap = 0.0001; bp = 0.036; cp = 0.018; dp = 0.031. Original magnification for A–D and H–K ×10.
Delayed depletion of osteomacs/macrophages in Mafia mice after the primary inflammatory healing phase resulted in suppression of new bone formation in the tibial injury model
AP20187 ligand systemic delivery in Mafia mice (either a single high dose or daily 10 mg/kg doses) did not significantly deplete macrophages until at least 72 hours.23 Therefore, it is likely that local delivery of AP20187 ligand, which depletes all mature macrophage populations, at the time of surgery had the potential to prevent macrophage contributions to early inflammatory events needed for subsequent healing. To differentiate between the contribution of macrophages to early inflammatory healing processes versus later anabolic bone modeling events, AP20187 ligand administration was delayed until 3 days after surgery and delivered by a single systemic injection. Delayed delivery of vehicle to Mafia mice had no effect on bone healing 7 days after surgery with F4/80+ osteomacs (Fig. 4H, arrows) intercalated throughout CT1+ woven bone (Fig. 4I) that completely bridged the intramedullar injury zone. Flow cytometry confirmed that delayed systemic delivery of AP20187 ligand in Mafia mice resulted in significant reduction in F4/80+ mature bone marrow macrophages (average 54.29% reduction, p = 0.03) within the contralateral limb bone marrow. IHC analysis demonstrated that delayed AP20187 ligand treatment resulted in a mild reduction in F4/80+ osteomacs and macrophages within the injury site (Fig. 4J, brown). Shallow sections close to the endocortical perimeter of the injury site (representing the border between the intramedullar and peripheral injury zones) detected areas of robust CT1+ matrix deposition (Supplemental Fig. 2B). However, the amount of new CT1+ matrix/osteoid decreased progressively as sectioning approached the core of the injury site. In sections representing the center of the injury site in AP20187 ligand–treated mice, only small foci of CT1+ matrix were present (Fig. 4K, arrow). Quantitative immunohistologic techniques measured the percent of CT1+ woven bone deposition within the intramedullar region of the injury site in delayed vehicle‐ and AP20187 ligand–treated mice 7 days after surgery. Three independent sectional depths per sample were analyzed, representing multiple depths through the injury site. This analysis demonstrated a significant reduction in area of CT1+ matrix in AP20187‐treated mice at two of the three sectional depths (Fig. 4L–N). These results indicate that osteomacs are required throughout the mineralization process in bone healing.
Osteomac/macrophage depletion via clodronate liposome administration significantly suppressed new bone formation in vivo
To validate observations made using the Mafia transgenic model, clodronate‐loaded liposomes24 were used as an alternative model for in vivo macrophage depletion. Flow cytometric analysis showed a significant reduction in F4/80+ mature bone marrow macrophages (average 25.5% reduction, p = 0.0002) in contralateral limb bone marrow. IHC analysis for F4/80 and CT1 expression confirmed a reduction in F4/80+ osteomacs/macrophages at the injury site with a subsequent repression of CT1+ matrix deposition (Fig. 5A–D). Quantitative µCT analysis confirmed a significant repression of mineralized bone deposition within the injury site in mice treated with clodronate‐loaded compared with PBS‐loaded liposomes (Fig. 5E). Thus two independent macrophage depletion strategies had the same striking effect in reducing new bone formation.
Osteomac/macrophage depletion via clodronate liposome treatment significantly represses new bone formation and mineralization in vivo. (A–D) Day 7 tibial injury sites stained with anti‐F4/80 or anti‐CT1 antibody in 12‐week‐old C57Bl/6 mice that received an intradefect injection of either control PBS liposomes (A, B) or osteomac/macrophage‐depleting clodronate liposomes (C, D) at the time of surgery, followed by daily intraperitoneal injections. In PBS liposome–treated mice, F4/80+ osteomacs (A, brown, arrows) were intercalated throughout new CT1+ woven bone (B, brown) that filled the injury site. Clodronate liposome treatment resulted in a reduction in F4/80+ cells (C) and reduced CT1+ woven bone deposition compared with vehicle‐treated mice (D). (E) µCT quantification of bone mineral density demonstrated a significant repression of mineralized bone deposition in clodronate liposome–treated mice. Statistically significant differences were calculated using unpaired t tests between treatments. ap = 0.0267. Original magnification for A and C ×20 and for B and D ×10.
Osteoclasts are not required for intramembranous woven bone deposition during bone healing in the tibial injury model
Owing to the close lineage relationship of macrophages and osteoclasts,35 there are no reliable in vivo macrophage depletion strategies that do not also compromise osteoclasts36 and potentially osteoclast precursors given that most myeloid cells have the potential to differentiate into these cells at least in vitro.35, 37 In contrast, osteoclasts can be specifically targeted in vivo using OPG treatment,21 which blocks the osteoclastogenic actions of RANKL. In vehicle‐treated mice at 7 days after surgery, only rare scattered TRACP+ multinucleated osteoclast‐like cells were present in the peripheral injury zone (Fig. 6A, arrow). In OPG‐treated mice at 7 days after surgery, only rare TRACP+ mononuclear cells were present in the peripheral injury zone (Fig. 6B, arrow). In vehicle‐treated mice at 9 days after surgery, TRACP+ multinucleated osteoclast‐like cells were common on woven bone surfaces in the intramedullar injury zone (Fig. 6C, arrow). In OPG‐treated mice, there was a significant reduction in both the number (Fig. 6D, E) and size [average 34% reduction in cell surface area (µm2), p = 0.04] of TRACP+ cells in the intramedullar injury zone, which was not surprising because many of the remaining TRACP+ cells were mononuclear. TRACP‐5b assays also confirmed a significant reduction in serum TRACP levels on both day 7 (Fig. 6F) and day 9 (Fig. 6G) after surgery. Flow cytometry and IHC staining for F4/80 expression demonstrated that OPG treatment had no effect on osteomac numbers or their distribution during bone healing (data not shown). Taken together, these observations confirm that OPG treatment successfully and specifically reduced osteoclast number and activity. Quantitative immunohistology demonstrated no statistically significant difference in percent area of intramedullar CT1+ woven bone deposition in vehicle‐ and OPG‐treated mice at 7 days after surgery (Fig. 6H–J). By 9 days after surgery, catabolic modeling of the woven bone bridge had begun in vehicle‐treated mice (Fig. 6K). The woven bone bridge persisted in OPG‐treated animals (Fig. 6L), and consequently, the percent area of CT1+ woven bone within the intramedullar region was significantly greater in OPG‐ compared with vehicle‐treated mice (Fig. 6M). Hence OPG treatment reduced osteoclast number within the injury site and inhibited osteoclast‐mediated bone resorption across a 9‐day time course but did not compromise intramembranous woven bone deposition.
OPG treatment prevents removal, not deposition, of new woven bone during bone healing in vivo. (A–D) Tibial injury sites from 12‐week‐old vehicle‐ or OPG‐treated C57Bl/6 mice stained histologically for TRACP activity (magenta). (A) In vehicle‐treated mice at 7 days after surgery, few TRACP+ multinucleated osteoclast‐like cells (arrow) were present in the peripheral injury zone. (B) TRACP+ mononuclear cells (arrow) were present in the peripheral injury zone following OPG treatment. At 9 days after surgery, vehicle‐treated mice had large TRACP+ multinucleated osteoclast‐like cells (C, arrow). OPG treatment resulted in a predominance of TRACP+ mononuclear cells (D, arrow). (E) Quantification of total number of TRACP+ cells/ROI in the intramedullar injury zone confirmed a significant reduction in TRACP+ cells after OPG treatment. (F, G) ELISA quantification of serum TRACP‐5b (U/L) showed a significant reduction in TRACP‐5b activity on both day 7 and day 9. CT1+ woven bone bridged the injury site in (H) vehicle‐ and (I) OPG‐treated mice 7 days after surgery. (J) Quantitative immunohistology demonstrated no difference in CT1+ woven bone deposition 7 days after surgery after OPG treatment. At 9 days after surgery, catabolic modeling of the injury site had begun in vehicle‐treated mice (K), but in OPG‐treated mice, the CT1+ woven bone bridge persisted (L). (M) Significantly more woven bone remained in OPG‐ compared with vehicle‐treated mice 9 days after surgery. Statistically significant differences were calculated using unpaired t tests between treatments. ap < 0.001; bp = 0.02; cp = 0.009; dp = 0.001. Original magnification for A–D ×40 and for H, I, K, and L ×4.
CSF‐1 treatment accelerated bone healing in the tibial injury model
CSF‐1 is required for the proliferation and differentiation of mononuclear phagocyte progenitor cells to monocytes, macrophages, and osteoclasts.13, 38 To determine whether CSF‐1 treatment increased osteomac numbers in vivo and/or enhanced bone healing, mice were treated with recombinant CSF‐1 during bone healing in the tibial injury model. The CSF‐1 treatment regime significantly increased the number of F4/80+ cells (quantified as percent area of F4/80 staining within the intramedullar zone; Fig. 7A–C). CSF‐1 did not significantly alter Mac‐2+ cell numbers within the injury site (Fig. 7D–F), indicating that CSF‐1 preferentially expanded resident and/or recruited noninflammatory macrophage populations, including osteomacs, over recruitment of inflammatory macrophages. Additionally, there was no significant difference in the small number of TRACP+ cells or the percent area of TRACP activity in the surgical limb growth plate (Fig. 7G–I). Local expansion of osteomac/macrophage populations via CSF‐1 treatment resulted in a concurrent and significant increase in new CT1+ matrix deposition (Fig. 7J–L) and a 20% increase in new mineralized matrix detected by quantitative µCT (Fig. 7M). We conclude that the CSF‐1 treatment regime led to an increase in macrophage/osteomac recruitment, proliferation, and/or differentiation within the tibial injury site and that this subsequently accelerated CT1+ matrix deposition and mineralization.
CSF‐1 treatment enhanced bone healing in vivo. Tibial injury sites of 12‐week‐old C57Bl/6 mice at 4 days after surgery stained with anti‐F4/80 (A, B), anti‐Mac‐2 (D, E), TRACP activity (G, H), or anti‐CT1 antibodies (J, K). Mice received an intradefect injection of vehicle or CSF‐1 at the time of surgery, followed by a subcutaneous injection on day 2. (A) In vehicle‐treated mice, F4/80+ mature tissue macrophages (brown) are present throughout the injury site predominantly within the intramedullar and peripheral injury zones. (B) CSF‐1 treatment increased F4/80+ mature tissue macrophages (brown) throughout the injury site. (C) A significant increase in percent area of F4/80 staining in the intramedullar injury zone in CSF‐1‐treated mice compared with vehicle‐treated mice. Minimal Mac‐2+ cells were present within the injury site of vehicle‐treated mice (D, arrow). CSF‐1 treatment did not significantly increase the amount Mac‐2+ cells in the intramedullary injury zone (E, arrows, and F). Minimal TRACP+ mononuclear cells were present within the injury site of (G, arrow) vehicle‐ and (H, arrow) CSF‐1‐treated mice. (I) There was no significant difference in the percent area of TRACP activity staining at the growth plate in vehicle‐ and CSF‐1‐treated mice. Compared with vehicle‐treated mice (J), CSF‐1 treatment resulted in an increase in CT1+ matrix/osteoid deposition (K, arrow). (L) A significant increase in percent area of CT1+ matrix deposition at the injury site (n = 5) in CSF‐1‐ compared with vehicle‐treated mice. (M) µCT demonstrated a trend toward elevated mineralized bone deposition in CSF‐1‐treated mice (n = 4). Statistically significant differences were calculated using unpaired t tests between treatments. ap = 0.002; bp = 0.026. Original magnification for A, B, D, and E ×10; for G and H ×20; and for J and K ×10.
Discussion
Macrophages make important contributions to healing in many systems,39, 40 including during fracture repair.31, 32 Such effects are commonly attributed to macrophages driving early inflammatory events. While these classic inflammatory contributions are undoubtedly important, resident tissue macrophages that have acquired specific adaptations to their local microenvironment41 also can participate during tissue repair.42, 43 This study revealed that both inflammatory macrophages and osteomacs, specialized resident bone macrophages, were located within bone injury sites during all major phases of stabilized fracture repair. Osteomacs were specifically associated with sites of intramembranous bone deposition, forming an organized canopy structure over matrix‐producing and mineralizing osteoblasts. In vivo models of osteomac/macrophage or osteoclast depletion established that osteomacs, as opposed to either early inflammatory macrophage activities or osteoclasts, were pivotal in optimal bone formation during repair of the tibial injury model.
There is a close lineage relationship between macrophages and osteoclasts35 and considerable cellular plasticity among the myeloid lineage.44 Herein we clearly show that osteomacs are distinct from osteoclasts and that the latter were not required for bone formation in the tibial injury model used. Additional support indicating that osteomacs are distinct from osteoclast precursors is provided by a recent study identifying the immediate osteoclast precursor as a CSF‐1R+RANK+ and F4/80− quiescent cell.45 These data support the idea that osteomac and osteoclast precursor cells have diverged from a common progenitor earlier in the myeloid lineage and matured along independent pathways to perform different functional roles. In this study we have clearly demonstrated the presence of a mature macrophage population, which we have named osteomacs, that was associated with intramembranous bone healing in this injury model and is undertaking a proanabolic functional role unrelated to the ability to serve as an osteoclast precursor.
Our observation that osteomacs enhance intramembranous ossification dictates that this bone modeling mechanism now should be considered as a multi‐lineage event. The involvement of osteomacs in intramembranous ossification is analogous to the proposed “coupling” role of osteoclasts during bone remodeling.8, 14 Intramembranous ossification does not involve resorption of either a cartilage or bone template prior to bone formation, and it is therefore logical that a nonresorptive mature myeloid cell has evolved as the cellular source of trophic and/or anabolic factor(s) to aid osteoblast recruitment, maturation, and/or function.
A remaining question is whether osteomacs specifically contribute to all bone‐formation events that occur during development, homeostasis, healing, and in response to mechanical loading. We previously demonstrated in mice that osteomacs were located within the growth plate spongiosa during postnatal bone growth12 and that they are also associated with BMUs.14, 46 Similarly, F4/80+ macrophages have been shown to be present in invading vascular canals during formation of the primary and secondary ossification centers in mouse long bone development.47 In addition, depletion of macrophages in osteoarthritic models reduced osteophyte formation dramatically.48 Interestingly, a recent publication reported that preventing the recruitment of inflammatory macrophages in a CCR2−/− mouse nonunion fracture model resulted in reduced endochondral bone formation but did not impede intramembranous bone formation.49 This study demonstrated that at least some of the macrophages involved in endochondral bone formation are a recruited cell population, but a remaining question is whether resident macrophages also play a role in endochondral bone formation during fracture healing. This study also reinforces our observations that early inflammatory macrophage contributions do not enhance intramembranous bone formation. Overall, these observations support a role for osteomacs in endochondral ossification, bone remodeling, and ectopic bone deposition.
Bone formation still occurs in mice and rats with germ‐line mutations resulting in loss of CSF‐1 or its receptor and gross deficiencies in most macrophage populations.15–17 Thus osteomac/osteoclast contributions are not absolutely required for osteoblast‐mediated matrix deposition and mineralization. Nonetheless, the bones in these animals are highly abnormal and poorly mineralized, and there is severe growth retardation.50 The absence of osteoclasts is undoubtedly a major contributor to the osteopetrotic phenotype. However, while osteoclastogenesis is reestablished at approximately 10 weeks of age in op/op mice, the osteoblast bone formation rate remains suppressed.50 This suggests that ongoing deficiencies in macrophages contribute to persistent abnormal bone cell dynamics in these genetically modified animals.
A functional role for osteomacs in intramembranous ossification provides a novel candidate cellular target for the development of anabolic bone therapies. Proof‐of‐concept evidence was provided here through acceleration of anabolic bone repair in response to CSF‐1 treatment in the tibial injury model. A previous study using daily postfracture delivery of CSF‐1 over an extended time course (14 days) in rabbits concluded that osteoclasts were the primary cellular target of CSF‐1, leading to improved fracture healing.51 In that study, CSF‐1 did not alter osteoclast number in the peak anabolic phase of fracture healing (4 weeks after surgery) but lead to prolonged significant elevation of osteoclasts 8 weeks after surgery. Based on these observations, it was concluded that the prolonged elevation in osteoclast number was the underlying mechanism resulting in increased area of mineralized callus. However, CSF‐1 treatment significantly increased mineralized callus area at both 4 and 8 weeks after fracture despite no significant difference in osteoclast or osteoblast number at the earlier time point.51 This study did not examine the potential for CSF‐1 to promote macrophage number and/or function. In our study, we examined and ruled out the possibility that enhanced bone formation induced by CSF‐1 treatment in the tibial bone injury mouse model was mediated through increased osteoclast numbers. Instead, we demonstrated that CSF‐1 increased osteomac/macrophage numbers within the injury site. These data cannot definitively rule out that CSF‐1 treatment increased local production of an osteoblast‐coupling factor by osteoclasts. However, there were very few TRACP+ multinucleated osteoclasts associated with regions of CT1+ matrix deposition during the anabolic bone‐modeling phase of this model. Other studies in animal models of fracture have inhibited osteoclast number and function either via OPG,52 denosumab,53 or bisphosphonate treatment53, 54 and concluded that osteoclasts were not required for normal callus formation but were important during hard callus remodeling. Macrophages are a major source of factors known to promote bone deposition/mineralization,55–58 and at least some of these can be induced by CSF‐1.56 The substantial numbers of osteomacs associated with the bone surface has not been recognized, and as such, macrophage production of these factors has not been considered in studies investigating bone formation.
Taken together, our data demonstrate for the first time that osteomacs are an integral cellular component of osteal tissues during bone healing and implicate osteomacs as critical participants in intramembranous ossification during bone repair. The identification of this novel osteoimmunologic mechanism expands our understanding of the substantial contributions the immune system has on bone/callus outcomes during fracture repair59, 60 and provides a new research avenue for manipulating bone formation for therapeutic applications. Identification of osteomac‐derived anabolic factors ultimately may lead to new approaches to improve fracture healing and reverse skeletal damage after pathologic bone loss.
Disclosures
All the authors state that they have no conflicts of interest.
Acknowledgements
LJ Raggatt and AR Pettit contributed equally to this work.
We thank Professor Don Cohen for the Mafia mice, as well as the Institute for Molecular Bioscience and Herston Medical Research Center animal houses for their technical assistance. This work was supported by NHMRC project grants (455941 and 631484) to ARP and LJR, an NHMRC Dora Lush Postgraduate Scholarship (409914) to KAA, and a NHMRC CDA award (519 744) to ARP.
