Chondrocyte‐to‐osteoblast transformation in mandibular fracture repair

Abstract The majority of fracture research has been conducted using long bone fracture models, with significantly less research into the mechanisms driving craniofacial repair. However, craniofacial bones differ from long bones in both their developmental mechanism and embryonic origin. Thus, it is possible that their healing mechanisms could differ. In this study we utilize stabilized and unstabilized mandible fracture models to investigate the pathways regulating repair. Whereas fully stable trephine defects in the ramus form bone directly, mechanical motion within a transverse fracture across the same anatomical location promoted robust cartilage formation before boney remodeling. Literature investigating long bone fractures show chondrocytes are a direct precursor of osteoblasts during endochondral repair. Lineage tracing with Aggrecan‐CreERT2::Ai9 tdTomato mice demonstrated that mandibular callus chondrocytes also directly contribute to the formation of new bone. Furthermore, immunohistochemistry revealed that chondrocytes located at the chondro‐osseous junction expressed Sox2, suggesting that plasticity of these chondrocytes may facilitate this chondrocyte‐to‐osteoblast transformation. Based on the direct role chondrocytes play in bone repair, we tested the efficacy of cartilage grafts in healing critical‐sized mandibular defects. Whereas empty defects remained unbridged and filled with fibrous tissue, cartilage engraftment produced bony‐bridging and robust marrow cavity formation, indicating healthy vascularization of the newly formed bone. Engrafted cartilage directly contributed to new bone formation since a significant portion of the newly formed bone was graft/donor‐derived. Taken together these data demonstrate the important role of chondrocyte‐to‐osteoblast transformation during mandibular endochondral repair and the therapeutic promise of using cartilage as a tissue graft to heal craniofacial defects.


| INTRODUCTION
Craniofacial bones differ from appendicular bones in both their embryonic origin and developmental mechanism. 1 Appendicular bones are derived from lateral plate mesoderm through the process of endochondral ossification. This process begins with the differentiation of mesenchymal stem cells into chondrocytes that form a cartilage template in the approximate size and shape of the future bone. As the skeleton develops, the cartilage anlage is eventually converted to bone. 2 In contrast, craniofacial bones develop from the cranial neural crest through the process of intramembranous ossification. Unlike endochondral ossification, this process does not involve a cartilage intermediate. Rather, osteochondral progenitors directly differentiate into osteoblasts to form bone. 1 The majority of mechanistic fracture research has been focused on long bone fractures in the appendicular skeleton. Given the significant developmental difference between craniofacial and long bones, it is possible that these bones differ in their healing mechanisms as well. To optimize current treatments and develop novel therapies for craniofacial fractures, improving our mechanistic understanding of specific fracture types is critical.
Long bone fractures heal through the combination of intramembranous and endochondral ossification, with local mechanical stability generally believed to be the fundamental factor in driving the mechanism of healing. [3][4][5] A relatively high amount of mechanical motion between the fractured bone ends promotes cartilage formation and endochondral repair, whereas rigid stability promotes direct bone formation through intramembranous ossification. 6 Recent genetic evidence from long bone fracture studies demonstrate that chondrocytes within the provisional fracture callus matrix directly contribute to the formation of bone by transforming into osteoblasts. [7][8][9][10][11][12][13][14] There is limited research into the molecular mechanisms by which chondrocytes transform into osteoblasts, but recent data suggest hypertrophic chondrocytes gain cellular plasticity in response to the invading vasculature through the expression of programs associated with stemness (Sox2, Oct4, Nanog), and then undergo osteogenesis through activation of the Wnt/β-catenin signaling pathway. 7,8,11,13,14 Importantly, these new studies update our mechanistic understanding of long-bone fracture repair and can be used to develop novel approaches to stimulate fracture repair by promoting the process of endochondral ossification. 15 While much less frequently studied, the mandible is one of the most commonly fractured craniofacial bones, sustaining up to 70% of all maxillofacial fractures. 16 Its daily functional requirement and esthetic role necessitate the development of new and improved therapies. Although the mandible develops via intramembranous ossification, it has been shown to heal similarly to long bone fractures with a combination of intramembranous and endochondral ossification. 17 In this study we aimed to study the role of chondrocyte-to-osteoblast transformation during mandible fracture repair by utilizing two mandible fracture models: a stabilized trephine defect model that heals via intramembranous ossification, and an unstabilized osteotomy that heals via endochondral ossification.
In addition to building new knowledge about the molecular and cellular mechanisms of mandible fracture repair, this study also aimed to explore endochondral routes of promoting mandibular bone regeneration. Autologous bone grafts are the current standard of care for improving healing in clinical cases of malunion or in large bone defects in both long bone and mandibular fractures. While clinical outcomes of autologous bone grafts are generally favorable, they are associated with significantly increased risk to the patient.
Autograft procedures typically require a second surgical site, increased surgical time, increased blood loss and are associated with high rates of donor site morbidity. 11,18 Alternatively, bone allografts are readily available, but they have significantly reduced bioactivity and a high rate of clinical failure due to poor osteointegration and osteonecrosis of the graft. 19 Thus, the discovery and development of alternative graft tissues and materials would be clinically beneficial.
Unlike bone, cartilage is an avascular tissue that is not only capable of surviving without a vascular supply but is also responsible for recruiting blood vessel invasion for the newly formed bone during endochondral ossification. 2 8,11 Mandible fractures with different levels of stability were generated in adult (10-16 week) male mice by creating either (1) an unstable osteotomy through the right mandibular ramus from the anterior border of the coronoid process to the anterior border of the angular process ( Figure 1A) or (2) by drilling a 1-mm diameter hole to form a trephine defect through the right mandibular ramus ( Figure 1B). In the unstable model, no bone was resected and fractured bone ends were allowed to oppose one another. No external or internal fixation was used to stabilize this model to reduce irritation of the surrounding muscle and to promote robust cartilage callus formation. Animals were randomly assigned to fracture group, and the fracture site location was confirmed via radiography. Animals were revived using an atipamezole reversal agent (6 mg/kg) and were given subcutaneous buprenorphine F I G U R E 1 Fracture stability mediates mandibular ossification mechanisms. X-ray images of an unstable transverse osteotomy (A) and stable 1 mm trephine defect (B) confirm the proper anatomical location of fracture site (red arrow) within the right mandibular ramus of C57BL/6 mice. Mandibles were imaged at 7 days postfracture (N = 5/group). HBQ (C, E, G, I, K, M) and Saf-O (D, F, H, J, L, N) standard histology staining of transverse sections reveals that mandibles with unstable osteotomies (C-H) develop a robust cartilage callus and primarily heal through endochondral ossification, whereas mandibles with stable trephine defects (I-N) develop little to no cartilage and primarily heal through intramembranous ossification. Small amounts of cartilage were found adjacent to the fracture site in the stabilized fracture model at 10 days postfracture, usually ventral and posterior to the defect (K, yellow arrowhead). Cartilage was only observed in the stable fracture model at day 10 postfracture and was most-likely due to motion derived from residual bone chips left behind from drilling the trephine defect. Stereological analysis of callus tissue composition confirmed that the absolute volume of cartilage was significantly less in stable than in unstable fractures at

| Stereology
Quantification of callus size and composition (cartilage, bone, other) was determined using an Olympus CAST system and software by Visiopharm according to established methodologies. 20 HBQ stained serial sections through the entire fracture site, spaced 300 µm apart were analyzed. The fracture callus was outlined using low magnification (×20, ×2 objective with ×10 ocular magnification) to determine the region of interest. Twenty percent of this region was quantified using automated uniform random sampling to meet or exceed the basic principles of stereology. 20 Cell and tissue identity within each random sampling domain was determined at high magnification (×200, ×20 objective with ×10 ocular magnification) according to histological staining patterns and cell morphology. Absolute volumes of specific tissues (e.g., bone or cartilage) were determined using the Cavalieri formula and these absolute volumes were used to determine the relative percent composition of tissues within the fracture callus. 20 Total fracture callus volume was measured as callus points.
The callus was divided into the central callus versus the transition zone using the cellular morphology and regional demarcations described previously. 8

| Lineage tracing
Aggrecan-Cre ERT2 and Ai9 tdTomato HZE mice were bred so that their progeny were heterozygous for Cre and tdTomato reporter alleles. Genotype was confirmed through gel electrophoresis according to Jackson Lab's genotyping protocols. Unstable mandible fractures were created as above. Cre recombination was induced with daily injections of Tamoxifen (75 mg/kg) from days 6 to 10 postfracture. 8 This recombination period aligned with the peak of the soft, cartilaginous phase of mandibular fracture repair. Samples were harvested 14 days postfracture, fixed and decalcified as above, cryoembedded, and sectioned (10 μm). Samples were mounted in VectaShield with DAPI (H-1200; Vector) and visualized using an epifluorescence microscope (N = 5).

| Cartilage grafts
Critical-sized defects were created in the right mandibular ramus of immunocompromised Nu/J mice either by creating two parallel osteotomies to form a 1.5 mm fracture gap or by drilling a 2-mm diameter hole to create a trephine defect through the right mandibular ramus. Defects were filled with cartilage harvested from tibia fracture calli of eGFP or LacZ reporter mice 7 days postfracture as done previoulsy. 11   brightness was adjusted to include all possible pixels within the region of interest. Hue and saturation were adjusted to best approximate the HBQ staining patterns of bone, cartilage, and marrow for each sample. Pixel measurements for total callus, bone, cartilage, and marrow were recorded for each image. To confirm the accuracy of these measurements, bone, cartilage, and marrow pixels were summed and compared to the total callus pixels for each image. On average the sum was equivalent to 94% of the total callus pixel count.
Raw pixel counts were used to calculate % total pixels (N = 2 images/ animal). Donor versus host-derived bone was determined via X-gal staining as done previously. 11 X-gal staining was used to detect β-galactosidase on frozen sections post-fixed in 0.2% glutaraldehyde for 15 min, and then exposed to X-gal staining solution overnight at 37°C.

| Statistics
Statistical significance was determined using Prism software. For analyses involving more than two treatment groups, one-way analysis of variance followed by Tukey's nonparametric multiple comparisons test was performed. For analyses involving only two treatment groups, either the Mann-Whitney nonparametric or the unpaired, parametric t test was performed. p ≤ .05 was designated as statistically significant. The bone marker, OC, was expressed most strongly in the hypertrophic chondrocytes at the Transition Zone with more lower, but distinct staining found in the central cartilage callus (Figure 3E-G).
A statistically significant difference in regional OC was not detected upon cell count, likely due to the protocol used to count positively staining cells irrespective of staining intensity.

| Cartilage grafts promote vascularized bone regeneration in critical-sized mandibular defects
Due to the important role of chondrocytes during mandibular endochondral repair, we tested the ability of cartilage grafts to heal critical-sized mandibular defects using two different models: (1) an unstable 1.5 mm defect created by making two parallel osteotomies and (2) a stable 2-mm diameter trephine defect. Defects were confirmed to be critical-sized since no bony bridging was observed in empty defect controls for either model ( Figure S1). HBQ histology revealed that engrafted cartilage began to integrate with the sur- However, as nearly the entire span of newly formed bone within the defect site was X-gal positive and X-gal staining was limited to within the donor region, this suggests not only that cartilage graft chondrocytes contributed the majority of newly formed bone but also that the X-gal staining was highly specific.

| DISCUSSION
Despite differences in embryonic origin and developmental mechanism, our study demonstrates that mandibular fractures proceed through conserved mechanisms of repair as seen in long bones. One of the basic tenants of long-bone fracture repair is that mechanical motion regulates the mechanism of healing, such that motion promotes cartilage formation and endochondral ossification whereas rigid stabilization results in direct bone formation through intramembranous ossification. 21 In this study we confirm that motion within the mandible plays a similar, dominant role in directing the mechanism of bone formation during fracture repair (Figure 1). Although small regions of cartilage were sometimes observed in our stable fracture samples, cartilage formation always occurred outside of the trephine defect and was associated with the presence bone chips. This correlates to tibia fracture studies that show an absence of cartilage with absolute ridged fixation resulting in a model of intramembranous repair. 21 One observation from our study was that stabilized fractures healed faster than those that were unstabilized. Whereas complete bony-bridging was observed in stable fractures at 10 days postfracture, unstable fractures completely bridged at 21 days postfracture. The fracture models used in this study differed in both size and shape (Figure 1). Thus, the difference in healing time is likely an artifact of the different fracture designs rather than the result of an intrinsic difference in the rate of healing between endochondral and intramembranous ossification. Indeed, Thompson et al. 21 demonstrated using unstabilized versus stabilized transverse tibia fracture models that the time required to achieve bony bridging is identical when bones heal through endochondral versus intramembranous ossification. One limitation of our study is that we did not evaluate the quality of regenerated bone. The degree of bone repair was solely determined by the degree of bone formation and bony bridging through histology.
There has been a centuries-long debate regarding the fate of callus chondrocytes during endochondral repair. In the early 1800s cartilage was believed to transform into bone. [22][23][24] However, in the mid-1800s Muller and Sharpy reported that all chondrocytes ultimately undergo apoptosis and that bone is derived from a separate population of osteoprogenitors. 22 The development of modern murine genetics has enabled accurate lineage tracing and the determination of cell fate. Recent genetic evidence now demonstrates that although some callus chondrocytes undergo apoptosis to make room for the marrow cavity, a significant portion of chondrocytes transform into osteoblasts and directly contribute to the formation of new bone. [8][9][10]12,13 This has been confirmed using a variety of Cre lines and genetically labelled transplants in the contexts of long-bone development, growth, and repair as well as mandibular condylar development. [8][9][10][11][12][13][14] Here we demonstrated via Aggrecan-Cre ERT2 lineage tracing that mandibular callus chondrocytes undergo a similar chondrocyte-to-osteoblast transformation during mandibular fracture repair and that callus chondrocytes significantly contribute to the formation of new bone (Figure 2). Limited work has been done to investigate the mechanisms that regulate chondrocyte-to-osteoblast transformation. Previous work in long bone fracture repair demonstrates that hypertrophic chondrocytes undergoing chondrocyte-to-osteoblast transformation at the Transition Zone express genes traditionally associated with stem cell pluripotency (Sox2, Oct4, Nanog) as well as classic osteogenic F I G U R E 3 Hypertrophic chondrocytes at the mandibular fracture callus transition zone express stem cell and osteogenic markers. C57BL/6 mandibles were given unstable osteotomies and harvested 10 days postfracture. HBQ (A, B) histology shows a robust TZ (black dotted line) region within the fracture callus, marked by newly formed blood vessels (*) of the invading vasculature at the chondroosseous junction. Immunohistochemical analysis demonstrates that hypertrophic chondrocytes at the TZ express the stem cell marker Sox 2(D) as well as the classic osteogenic marker Osteocalcin(F). Chondrocytes at the center of the cartilage callus and distant from the TZ (boxed region in A) either do not express these genes or express at a significantly lower level (C, E). Quantification of % positively staining cells (G) reveals a statistically significantly difference in staining patterns between these two regions for Sox2. However, a statistically significant difference was not noted for OC. This is likely due to the inclusion of cells with even faint staining as positive. Qualitative comparison of the two regions demonstrates a consistent difference in the strength of staining. HBQ: cartilage is blue, bone is red. N = 5/antibody. Scale = 200 µm. HBQ, hall brunt quadruple; OC, osteocalcin; TZ, transition zone [Color figure can be viewed at wileyonlinelibrary.com] genes (Runx2, Osx, OP, OC, Col1), suggesting that chondrocytes may undergo stem cell-like plasticity before differentiating into osteoblasts. 8,11 Sox2 and β-catenin have been shown to play critical roles in regulating chondrocyte transformation. Hu et al. 8 25 Although autologous bone grafts have had significant clinical success, they require a second surgical site and are accompanied by notable costs including donor site morbidity and limited supply. 11,18 Bone allografts have been developed as an alternative graft material and are readily available; however, they have significantly reduced bioactivity and a high rate of clinical failure due to poor osteointegration and osteonecrosis of the graft. 19 As an avascular tissue, cartilage is not only capable of surviving without a vascular supply but is also responsible for recruiting blood vessel invasion for the newly formed bone during endochondral ossification. 2,3 Thus, cartilage is not only better suited than bone to survive within poorly vascularized areas of tissue injury but its use also parallels the natural healing process of endochondral repair.
There has been increasing interest in the development of cartilage-based therapies. 11,[26][27][28][29] Bahney et al. 11 demonstrated that cartilage grafts heal critical-sized tibial defects and that a significant portion of newly formed bone is donor-derived. Furthermore, they demonstrated that human bone marrow-derived mesenchymal stem cells (hMSCs) can be differentiated and cultured into viable cartilage grafts for transplant. 11 Other studies have also demonstrated the efficacy of using cartilage to engineer bone. [27][28][29] Scotti et al. 29 demonstrated that subcutaneous implantation of hypertrophic cartilage derived from hMSCs gives rise to an ectopic "bone organ" with a structure and functionality comparable to that of native bones. Dang et al. 28 reported that critical-sized calvarial defects can heal through endochondral ossification when treated with hMSCs pushed toward chondrogenesis through the controlled release of transforming growth factor β1 and bone morphogenetic protein-2 delivered via bioactive microparticles. And Cunniffee et al. 27 showed that decellularized hypertrophic cartilage freeze-dried to generate porous scaffolds can produce complete bony bridging of critical-sized femoral defects. Our data support that cartilage is an effective tissue graft to promote bone regeneration in the mandible in both stabilized and unstabilized fracture contexts (Figure 4). Although bony bridging was achieved using both models, stabilized critical-sized defects were much better at retaining the cartilage graft within the defect than unstabilized de- Identification of host versus donor tissues demonstrates that healing occurs through the transformation of engrafted chondrocytes to osteoblasts that form the new bone. It is unlikely that the contribution of donor cells to newly formed bone was due to contamination of cartilage grafts by a separate stem cell population. Our lab has performed extensive analysis on the composition of tissues harvested using this methodology. 11 Previous analysis of grafts reveals nearly pure cartilage. In situ hybridization and qPCR analysis reveals significantly higher expression of cartilage markers (Sox9, Col2, Col10) and nearly nonexistent expression of the osteoblast marker OC compared to total callus. 11 Furthermore, the role of engrafted chondrocytes in contributing to new bone formation is corroborated by our lineage tracing analysis, which demonstrates the contribution of Aggrecan-expressing chondrocytes to osteoblast and bone-lining populations.
Our data demonstrate that the ability of cartilage to heal a critical-sized defect is not limited by the endogenous healing pathway and is effective for vascular bone regeneration. Importantly, since complete mandibular healing was attained through the use of tibia-derived cartilage, our data indicates that the embryonic origin or mode of tissue development does not significantly impact the efficacy of graft tissues (Figure 4).
Although craniofacial bones and long bones have significant developmental differences, our study supports that these bones heal via conserved mechanisms that reflect recent advances in our understanding of long bone fracture repair. Notably, during endochondral fracture repair, both sets of bones rely on the significant contribution of chondrocytes for the formation of new bone through chondrocyte-to-osteoblast transformation. This data provides important insight into the enhancement of craniofacial fracture therapies, such as the use of cartilage as a graft tissue, and it supports the application of findings from appendicular fracture studies to the treatment of craniofacial fractures.