TGF‐β/Smad2 signalling regulates enchondral bone formation of Gli1+ periosteal cells during fracture healing

Abstract Objectives Most bone fracture heals through enchondral bone formation that relies on the involvement of periosteal progenitor cells. However, the identity of periosteal progenitor cells and the regulatory mechanism of their proliferation and differentiation remain unclear. The aim of this study was to investigate whether Gli1‐CreERT2 can identify a population of murine periosteal progenitor cells and the role of TGF‐β signalling in periosteal progenitor cells on fracture healing. Materials and methods Double heterozygous Gli1‐CreERT2;Rosa26‐tdTomatoflox/wt mice were sacrificed at different time points for tracing the fate of Gli1+ cells in both intact and fracture bone. Gli1‐CreERT2‐mediated Tgfbr2 knockout (Gli1‐CreERT2;Tgfbr2flox/flox) mice were subjected to fracture surgery. At 4, 7, 10, 14 and 21 days post‐surgery, tibia samples were harvested for tissue analyses including μCT, histology, real‐time PCR and immunofluorescence staining. Results Through cell lineage‐tracing experiments, we have revealed that Gli1‐CreERT2 can be used to identify a subpopulation of periosteal progenitor cells in vivo that persistently reside in periosteum and contribute to osteochondral elements during fracture repair. During the healing process, TGF‐β signalling is continually activated in the reparative Gli1+ periosteal cells. Conditional knockout of Tgfbr2 in these cells leads to a delayed and impaired enchondral bone formation, at least partially due to the reduced proliferation and chondrogenic and osteogenic differentiation of Gli1+ periosteal cells. Conclusions TGF‐β signalling plays an essential role on fracture repair via regulating enchondral bone formation process of Gli1+ periosteal cells.


| INTRODUC TI ON
Bone has a high regenerative capacity that enables most fractures healed in a native form and function. 1 This reparative nature of bone relies mainly on the existence of local active progenitor cells. 2,3 Fracture healing is a complex process that undergoes three major biologically distinct but overlapping phases including haematoma, fracture callus formation and bone remodeling. 4 Progenitor cells make differential contributions to each phase, such as recruitment and proliferation at the initial haematoma phase and chondrogenic and osteogenic differentiation at subsequent phases. 5 Although the importance of progenitor cells to fracture healing have been well documented, the identity and regulatory mechanism of progenitor cells are still largely unknown.
Several potential sources of skeletal progenitor cells are proposed for bone regeneration, including bone marrow, 6 periosteum, 7 endosteum, 8 adjacent soft tissue 9,10 and vascular walls. 11 Recent findings highlight the importance of progenitor cells within periosteum since they can give rise directly to cartilage and bone during the healing process. 1,5,12 Removal of the periosteum tissue leads to clinical delayed union or nonunion of fractures with no fracture callus formation. 13 Over the last decade, with development of lineage-tracing technology, some periosteal markers, such as Prx1, 14 Sox9, 15 aSMA 16 and CTSK 17 have been identified in mice. Nevertheless, it still needs to vigorously investigate the promising progenitor cell populations for better defining the contribution of periosteal progenitor cells to fracture healing. Gli1 is a mediator of Hedgehog signalling that controls bone development. 18 Previous studies have revealed that Gli1 + cells within the craniofacial sutures 19 and growth plate 20,21 have the progenitor properties, and more remarkably, they largely contribute to fracture callus 20 and heterotopic bone formation. 22 Here, we seek to further determine whether Gli1 can identify a population of periosteal progenitor cells during fracture healing.
Amongst numerous growth factors and cytokines, transforming growth factor β (TGF-β) is one of the most important factors in regulation of fracture healing. 23,24 Clinical evidence shows a rapid elevation of TGF-β serum responding to fracture in patients. 25 Patients with low TGF-β level are tending to have delayed union or nonunion. 26,27 TGF-β regulates bone regeneration mainly via the Smaddependent canonical pathway. 28 After TGF-β ligand binding to type II receptor (TGF-βRII), phosphorylated Smad2 in turn is translocated into the nucleus and activates the downstream target genes which are responsible for cell proliferation, cell differentiation and extracellular matrix production. 24,29 Currently, the role of TGF-β/Smad2 signalling in periosteal progenitor cells remains unclear in the context of fracture repair.
In the present study, we hypothesize that TGF-β/Smad2 signalling can regulate the reparative response of Gli1 + periosteal cells for murine fracture healing. By tracing the fate of Gli1-expressing lineage cells in both intact and fracture tibiae in mice, we have demonstrated that Gli1 identifies a population of periosteal cells in vivo that persistently resides in periosteum tissue and also can give rise to chondrocytes and osteoblasts during fracture healing process. Furthermore, by utilizing Gli1-Cre-mediated Tgfbr2 inducible knockout mice, we have revealed that inhibition of TGF-β/ Smad2 signalling in Gli1 + periosteal cells negatively affects their proliferation, chondrogenic and osteogenic differentiation, therefore resulting in an impaired endochondral bone formation in fracture healing.

| Animals
Gli1-CreER T2 mice, Rosa26-tdTomato flox/flox mice and Tgfbr2 flox/ flox mice were obtained from Jackson Laboratory. For lineagetracing experiments, a double heterozygous Gli1-CreER T2 ;Rosa26-tdTomato flox/wt (Tomato Gli1ER ) mice were generated, and tamoxifen (1 mg/10 g body weight/day, diluted in corn oil) was injected intraperitoneally into 1-month-old mice for 3 consecutive days. To investigate the role of TGF-β signalling in Gli1 + periosteal cells in fracture healing, Gli1-CreER T2 ;Tgfbr2 flox/flox (Tgfbr2 Gli1ER ) mice and Gli1-CreER T2 ;Tgfbr2 flox/flox ; Rosa26-tdTomato flox/wt (Tgfbr2 Gli1ER ;ROS A tdTomato ) mice were generated following with 3 consecutive intraperitoneal injections of tamoxifen at 1 month of age, or mice were subcutaneously injected with TGF-β neutralizing antibody (5 mg/kg body weight) at the fracture site once every 2 days starting immediately after fracture. The specific information of transgenic mice were provide in Table 1. Both male and female mice were used in lineage-tracing studies, but only males were subjected to fracture surgery to avoid sex-dependent difference. All animal experiments were approved by the Animal Ethics Committee of Zhejiang Chinese Medical University (LZ12H27001).

| Tibial fracture model
An open transverse tibial fracture model was established unilaterally in the male mice as previously described. 30,31 Briefly, an incision of 1 cm was made along the surface of tibial crest after mice were LY16H270010 and LY18H270004; Youth Foundation of Zhejiang Chinese Medical University, Grant/Award Number: KC201932 Conclusions: TGF-β signalling plays an essential role on fracture repair via regulating enchondral bone formation process of Gli1 + periosteal cells.
anesthetized by intraperitoneal injection of pentobarbital (60 mg/kg body weight). Medial to the patellar tendon, a 26-gauge needle was inserted into the tibial intramedullary cavity through the tibial platform. The needle was removed followed by a transverse cut with a NO.11 surgical blade at the midpoint of the tibia. The transverse fracture was then fixed again by the needle. Mice were sacrificed at 4, 7, 10, 14 and 35 days post-fracture, and tibia samples were harvested for further analysis.
To determine the importance of periosteum to bone repair, we removed 0.1 mm periosteum tissue on the fractured tibia. Briefly, after the transverse fracture, the antero-and posterior-lateral periosteum was striped off by the NO.11 surgical blade. Mice were sacrificed at 4 and 14 days post-fracture for phenotypical analyses.

| CidU administration
Tomato Gli1ER mice received the artificial nucleoside chlorodeoxyuridine (CidU; Sigma; St. Louis, USA) immediately after fracture surgery via subcutaneous injection once at a concentration of 10 mg/ mL followed by oral administration for another 3 days at a concentration of 1 mg/mL. 32 Tibia samples were harvested next day for in vivo cell proliferation analysis.

| Histology and histomorphometry
Tibia samples were processed for 3-μm-thick paraffin section or 10-μm-thick frozen section. The sections were stained with DAPI staining for cell lineage-tracing or Alcian Blue Hematoxylin (ABH)/ Orange G for histological analysis. 33

| Immunofluorescence assay
Immunofluorescence (IF) assay were performed on the frozen sections according to the previously established procedures. 33  After incubation with secondary antibodies for 20 minutes, tissue sections were counter-stained with 4',6-diamidino-2-phenylindole (DAPI). Fluorescent quantitative analysis was calculated from three mice (one representative section per mouse) using Image-PRo Plus software.

| Quantitative gene expression analysis
Fracture callus including 1 mm adjacent bone tissue on either side of the fracture line were collected for real-time PCR analysis as previously described. 30,31 Primer sequences for target genes are provided in Table 2.

| Statistical analysis
Statistical analyses including one-way ANOVA followed by Tukey's test and unpaired Student's t tests were performed with the software of sPss 20.0. * P < .05 was considered statistically significant.

| Postnatal Gli1 + cells persistently reside in periosteum and contribute to fracture callus formation
To  Figure 3D,E, black arrows) and fracture callus formation were remarkedly decreased ( Figure 3F, 3g, red arrows). Altogether, these findings indicated that Gli1 + periosteal cells were essential to normal fracture healing, and Gli1 + cells residing in other locations could not be recruited to repair fracture.

| Continuous activation of TGF-β/ Smad2 signalling in Gli1 + periosteal cells during fracture healing
To evaluate the expression of TGF-β/Smad2 signalling in Gli1 + periosteal cells during fracture healing, IF assay was performed in the

| Local application of TGF-β1 neutralizing antibody results in a delayed and impaired endochondral bone formation in fractured mice
We analysed the essential role of TGF-β1 in fractured microenviron-

| Deletion of Tgfbr2 in Gli1 + periosteal cells leads to a delayed endochondral bone formation in fractured mice
In order to determine the effects of TGF-β/Smad2 signalling on Histological analyses also revealed a delayed and impaired endochondral bone formation in Tgfbr2 Gli1ER mice. Compared to F I G U R E 2 Gli1 + periosteal cells undergo proliferation and differentiation into chondrocytes, osteoblasts and osteocytes during fracture healing. Tomato Gli1ER mice induced with tamoxifen at 1 mo of age were subjected to the fracture surgery at 10 wk of age and sacrificed 4, 7, 10, 14 and 35 d later. A-E, Representative immunofluorescence images of fractured tibiae from each time point. F-J, High magnification images of local fracture sites in (A-E), respectively. (f-j) ABH stained images were an adjacent section to (F-J) respectively, indicating the components that Gli1 + cells proliferated and differentiated at each time point. (A, F, f) Gli1 + cells largely expanded on the periosteal surface closed to the fracture site at day 4. Yellow arrows: expanded periosteum. B-D, G-I, g-i, Gli1 + cells differentiated into chondrocytes, osteoblasts and osteocytes to form fracture callus at days 7-14. Red arrows: chondrogenic differentiated Gli1 + cells. Green arrows: osteogenic differentiated Gli1 + cells. (E, J, j) Gli1 + cells were presented in the newly formed periosteum at day 35. K, Schematic experimental design for data in (L). L, Gli1 + periosteal cells both near to and away from the fracture sites highly expressed immunofluorescence signal of Cidu (green) at day 4 post-fracture. Red: tdTomato + cells, blue: nuclear staining by DAPI. Scale bars: 1000 µm F I G U R E 3 Periosteal-derived Gli1 + cells are essential for fracture healing. Tomato Gli1ER mice induced with tamoxifen at 1 mo of age were subjected to tibia fracture surgery combined with removing antero-and posterior-lateral periosteum at 10-wk-old. A, Representative threedimensional (3D) µCT images showed a distinct fracture line (red arrow) at day 14 in the periosteum removed mice. B, C, Quantitative μCT analysis of bone volume and BV/TV for fracture callus tissues at day 14. D-G, No Gli1 + cells were presented in the periosteum removed side at days 4 and 14 post-fracture. (d-g) ABH staining of an adjacent section to (D-G), respectively. The periosteum removed mice presented a deficiency of periosteal expansion at day 4 (black arrows) and almost no bone callus formation (red arrows) at day 14 the control mice, Tgfbr2 Gli1ER mice presented a reduced perios-

| Deletion of Tgfbr2 in Gli1 + periosteal cells inhibits proliferation and differentiation of Gli1 + periosteal cells into chondrocytes and osteoblasts during healing process
We then examined the proliferation and differentiation of Gli1 + periosteal cells in Tgfbr2 Gli1ER mice. Gli1-CreER T2 ;Tgfbr2 flox/flox ;Rosa26-tdTomato flox/wt (Tgfbr2 Gli1ER ;ROSA tdTomato ) mice were generated to label Gli1 + cells with tdTomato red fluorescence and at the same time to delete Tgfbr2 in Gli1 + cells. At day 4 after fracture, the percentage of Gli1 + ;CidU + periosteal cells in Tgfbr2 Gli1ER ;ROSA tdTomato mice was much less than that in Tomato Gli1ER mice, indicating that the proliferative Gli1 + periosteal cells was reduced by loss of TGF-β pathway ( Figure 8A). Immunostaining analysis demonstrated that

| DISCUSS IONS
Periosteum is the tissue that makes a major cellular contribution to both cartilage and bone formation during fracture healing process, and absence of periosteum leads to impaired fracture healing and even fracture nonunion. 1,3,5,13 The progenitor cells isolated from the periosteum show higher regenerative capacity compared to bone marrow mesenchymal stem cells and adipose-derived mesenchymal cells; therefore, they are considered as ideal candidates for tissue engineering applications. 34,35 Previous studies have revealed that periosteum transplantation can successfully heal bone defects nonunion and in animal models. 36,37 However, the identify of the progenitor cells within periosteum is not well defined.

F I G U R E 6
Deletion of Tgfbr2 in Gli1 + periosteal cells leads to a delayed and impaired enchondral bone formation in fractured mice. Tgfbr2 Gli1ER mice induced with tamoxifen at 1 mo of age were subjected to tibial fracture surgery at 10 wk of age. A, Representative µCT images showed the distinct fracture lines at day 14  to determine whether TGF-β1 can be used in clinic to treat fracture patients.
In summary, Gli1 can identify a population of periosteal progenitor cells in juvenile mice. TGF-β/Smad2 signalling in Gli1 + periosteal F I G U R E 7 Deletion of Tgfbr2 in Gli1 + periosteal cells down-regulates expressions of chondrocyte-and osteoblast-specific marker genes in callus tissues. Total RNA was extracted from callus tissues (n = 3) of Tgfbr2 Gli1ER mice at different time points. A, Expression of Tgfbr2 was decreased at day 7-21. B, Expression of Col2a1 was decreased at day 7. C, Expression of Col10a1 was decreased at day 7 and 10, but increased at day 14. D, E, Expression of Runx2 and osteocalin was decreased at day 10 and 14, but increased at day 21 cells is essential to the cell proliferation as well as chondrocyte and osteoblast differentiation in fracture healing.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
Ping-er Wang and Hongting Jin contributed to study conception

F I G U R E 8
Deletion of Tgfbr2 in Gli1 + periosteal cells inhibits proliferation and differentiation of Gli1 + periosteal cells into chondrocytes and osteoblasts during fracture healing. Tgfbr2 Gli1ER ;ROSA tdTomato mice induced with tamoxifen at 1 mo of age were subjected to fracture surgery at 10 wk of age. A, Percentage of tdTomato + ; Cidu + over tdTomato + cells. B, Percentage of tdTomato + chondrocytes surrounded by green fluorescence signal of Col-II over tdTomato + cells. C, Percentage of tdTomato + ; Cidu + over tdTomato + cells. Scale bars: 1000 µm

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.