BDNF promoted osteoblast migration and fracture healing by up‐regulating integrin β1 via TrkB‐mediated ERK1/2 and AKT signalling

Abstract Brain‐derived neurotrophic factor (BDNF) has been reported to participate in fracture healing, whereas the mechanism is still unclear. Since osteoblast migration is important for fracture healing, investigating effects of BDNF on osteoblasts migration may help to reveal its mechanism. Here, MC3T3‐E1 cells were used in vitro while closed femur fracture mice were applied in vivo. Cells migration was assessed with Transwell assay. The protein expression was analysed by immunoblotting. X‐ray and Micro‐CT were performed at different time after fracture. Our results showed that BDNF promoted MC3T3‐E1 cells migration, integrin β1 expression and ERK1/2 and AKT phosphorylation. K252a, a specific inhibitor for TrkB, suppressed BDNF‐induced migration, integrin β1 expression and activation of ERK1/2 and AKT. PD98059 (an ERK1/2 inhibitor) and LY294002 (an AKT inhibitor) both inhibited BDNF‐induced migration and integrin β1 expression while integrin β1 blocking antibody only suppressed cell migration. X‐ray and Micro‐CT analyses showed that the adenoviral carried integrin β1 shRNA group had slower fracture healing at 7 and 21 days, but not 35 days compared to the control group. Thus, we proposed that BDNF stimulated MC3T3‐E1 cells migration by up‐regulating integrin β1 via TrkB mediated ERK1/2 and AKT signalling, and this may help to enhance the fracture healing.

coated with BDNF could enhance osteoblast differentiation from osteoporotic donors. 6 Bone cement filled with BDNF could significantly promote fracture healing in animal model. 7,8 However, the underlying mechanism of BDNF in fracture healing is largely unknown.
Fracture healing is a complex physiologic process that involves several cell types. Osteoblasts are the major bone-forming cells that play a crucial role in this process. 9 During fracture healing, osteoblast precursor cells must migrate into fracture sites from the bone marrow compartment, where they initiate the process of fracture repair including adherence, differentiation and deposition of the bone matrix. 10,11 Many studies have shown that BDNF plays important roles in the regulation of cell migration. [12][13][14][15] BDNF can promote young cardiac microvascular endothelial cells to migrate via the activation of the BDNF-TrkB-PI3K/AKT pathway after myocardial infarction. 15 In enhancing the generation of periodontal tissue, BDNF stimulates endothelial cell migration by a process involving TrkB/ ERK/integrin signalling. 14 In addition, BDNF also can potentiate EGF (epidermal growth factor)-induced migration of human foetal neural stem/progenitor cells via the PI3K/AKT pathway in neurogenesis. 12 Whether BDNF participates in fracture healing through regulating osteoblast migration remains unclear.
Integrins are transmembrane receptors composed of α subunits and β subunits assembled as heterodimers and are involved in cellcell and cell-extracellular matrix interaction and integrin β1 has been demonstrated as the major subunit in osteoblasts. 10,16 Previous study has proved that bone formation is impaired resulting from altering osteoblast function in transgenic mice with a dominant-negative integrin β1. 17 During fracture healing, Einhorn also demonstrates callus size is obviously diminished in α1β1 integrin-deficient mice. 18 However, the underlying mechanism is still poorly understood.
Cooperating with kinds of growth factors, integrins transduce signals from the environment into cells and often mediate cell adhesion and migration. 14,19-21 Xiong suggested TGFβ1 induced endometrial cancer cell adhesion and migration by up-regulating integrin αvβ3 while Furmento indicated granulocyte colony-stimulating factor (G-CSF) up-regulated integrin β1 and increased migration of human trophoblast Swan 71 cells. 19,20 For BDNF signalling, integrin β3 and β5 has already been proved to be up-regulated and participate in manipulating endothelial and chondrosarcoma cells migration, respectively. 14,21 Thus, we hypothesized that there must be an existing regulatory network involving BDNF and integrin β1 in osteoblasts migration during fracture healing.
In the present study, we sought to define the role of integrin β1 in BDNF-induced fracture healing and to identify the signalling pathway engaged by BDNF to mediate osteoblasts migration.
Quantification of the bands was performed using Quantity One densitometric analysis software (Bio-Rad Laboratories, Hercules, CA, USA).

| Transwell cell migration assay
Migration of MC3T3-E1 cells was determined by using the Costar Transwell System (8-μm pore size polycarbonate membrane inserts, Costar, Cambridge, MA). In brief, after different treatment, cells (1 × 10 3 cells/200 μL FBS-free medium) were seeded into the upper chamber and 600 μL complete medium was added into the lower chamber. After incubation at 37°C for 12 and 24 hours, non-migrated cells in the upper chamber were removed using cotton swab carefully and cells that went through the inserts were fixed with 4% paraformaldehyde, followed by staining in crystal violet. The number of migrated cells in five randomly chosen fields was counted under a fluorescence microscope (Nikon, Tokyo, Japan).

| Femur fracture animal model
Thirty adult male C57BL/6J mice were purchased from Shanghai SLAC Laboratory Animal Co., Ltd and were divided into two groups

| Histological examination
After micro-CT analysis, the fracture samples were decalcified using 15% EDTA at 25°C for 30 days. The decalcification solution was changed daily. The decalcified femora were embedded in paraffin after tissue processing and sectioned (5 mm The ROI in the histological analysis was 2 mm below and above the fracture line. Histomorphometry measurements were quantified using ImageJ software (National Institutes of Health, USA).

| Radiological examinations by X-ray
At 7, 21, and 35 days after the operation, mice were killed.
Radiographs of right femurs were acquired through an X-ray device (Faxitron MX20, USA). X-ray images were numerically scored independently by two senior orthopaedic surgeons as previously described: 0-no callus, 1-little to moderate callus, 2-profuse callus,

| Statistical analysis
All data were expressed as means ± standard deviation (SD) and analysed with the SPSS 20.0 software package (SPSS, USA). Statistical significance of differences between groups was calculated with an ANOVA or Student's t test. Significance was set at P < 0.05.

| BDNF promoted MC3T3-E1 cells migration and the expression of integrin β1
BDNF (50 ng/mL) was used to stimulate MC3T3-E1 cells for 0, 3, 6, 12 and 24 hours. As shown in Figure 1A, the expression of integrin β1 gradually increased and significant differences were found between 24 hours and other groups ( Figure 1B). To determine whether BDNF affect migratory potential of MC3T3-E1 cells, we performed Transwell assays for 12 and 24 hours. As shown in Figure 1B,C, BDNF played a confirmatory role in stimulating osteoblasts migration (P < 0.05).

| Integrin β1 effect on BDNF-induced MC3T3-E1 cells migration via TrkB receptor
Next, we determined whether the TrkB receptor and integrin β1 were involved in BDNF-mediated MC3T3-E1 cells migration.
Pre-treatment of cells with the TrkB receptor inhibitor K252a and integrin β1 blocking antibody (integrin β1 BL) obviously reduced BDNF-induced increases in cell migration ( Figure 3A,B). However, when MC3T3-E1 cells were stimulated directly by fibronectin, only integrin β1 BL could decrease the migration significantly ( Figure 3A,B).

| ERK1/2 and AKT activation were involved in BDNF-induced cell migration and integrin β1 expression in MC3T3-E1 cells
ERK1/2 and AKT is common downstream kinases in TrkB signalling. 2,3 To further confirm the involvement of ERK1/2 and AKT in BDNF effects on integrin β1 and MC3T3-E1 cells, the level of p-ERK1/2 thr202/tyr204 and p-AKT S473 was firstly detected. As shown in Figure 4A, the level of p-ERK1/2 thr202/tyr204 and p-AKT S473 was significantly increased and peaked at 30 minutes. Then, the specific inhibitors including K252a for TrkB, PD98059 for ERK1/2 and LY294002 for AKT were pretreated. All the three inhibitors could significantly suppress integrin β1 expression when successfully decreasing its corresponding kinase level ( Figure 4B).
K252a could inhibited the activation of ERK1/2 and AKT while PD98059 and LY294002 had no effects on TrkB ( Figure 4B).
Furthermore, the number of migrating cells was significantly decreased when pretreated with those three inhibitors separately ( Figure 4C,D). Taken together, it appears that the BDNF/TrkB axis acts through the ERK1/2 and AKT-dependent signalling pathway to enhance integrin β1 and cell migration in MC3T3-E1 cells.

| Comparison of X-ray imaging between Adintegrin β1 shRNA and Ad-NC group
Before making radiographic and histological analysis, we first checked the infection efficacy of adenovirus in vitro and in vivo.

MCET3-E1 cells were transfected and the green fluorescent view
showed that the greater than 95% infected ratio ( Figure 5A). The expressions of integrin β1 in the infected femoral callus were significantly down-regulated confirmed by Western blotting and immunohistochemistry ( Figure 5B-C).
At 7 days after operation, both Ad-integrin β1 shRNA and Ad-NC group showed disunion with clear fracture gaps and little periosteal callus ( Figure 5D). At 21 days after operation, the fracture lines were obscure in both the two groups, while the callus in Ad-NC group seemed denser and bigger than that in Adintegrin β1 shRNA group ( Figure 5D). At the 35 days time-points, the fracture line nearly disappeared and bony union was found in both the two groups ( Figure 5D). However, the X-ray score between the two groups showed no significant difference (data not shown).

F I G U R E 4 BDNF effects on MC3T3-E1 cells migration and integrin β1 expression through TrkB-mediated ERK1/2 and AKT activation:
A, MC3T3-E1 cells were treated with BDNF (50 ng/mL) for 0, 15, 30, 60 and 120 min. The level of p-ERK1/2 thr202/tyr204 and p-AKT S473 was significantly increased and peaked at 30 min. B, MC3T3-E1 cells were pretreated with K252a, PD98059 (25 μmol/L, an inhibitor of ERK1/2) and LY294002 (50 μmol/L, an inhibitor of AKT) 30 min prior to BDNF stimulation. All the three inhibitors could significantly suppress integrin β1 expression when successfully decreasing its corresponding kinase level. C, MC3T3-E1 cells were pretreated with K252a, PD98059 and LY294002 30 min prior to BDNF stimulation. The migration of MC3T3-E1 cells was determined by Transwell assay. D, The statistical analysis of C. Asterisks indicate statistical significance with control group (*P < 0.05). Each experiment was repeated three times

| Micro-CT and histological analysis of fracture healing and osteoblast migration between Ad-integrin β1 shRNA and Ad-NC group
At 7 and 21 days after operation, micro-CT scanning revealed a significantly lower BV/TV, Tb.Th, Tb.N and higher Tb.Sp in Ad-integrin β1 shRNA group compared to Ad-NC group (P < 0.05; Figure 6A-D). However, no significant differences in all of the above parameters between the two groups were found in 35 days after fracture (P > 0.05; Figure 6E,F).
As Runx2 was one of the specific markers of osteoblasts, it was detected by callus immunohistology to find the difference in the number of osteoblasts between Ad-integrin β1 and Ad-NC group at 7, 21 and 35 days after fracture. As shown in Figure 7, the expression of Runx2 at 7 and 21 days after fracture in Ad-integrin β1 group was significantly lower than that in Ad-NC group (P < 0.05) while no obvious difference was found at 35 days after fracture between the two groups (P > 0.05).

| D ISCUSS I ON
In the present study, we demonstrated that BDNF could promote the migration and integrin β1 expression of MC3T3-E1 cells. Scar bar = 50 μm. D, Representative X-ray films of fractured femurs. 7 d, the two groups both showed disunion with clear fracture gaps and little periosteal callus. 21 d, the fracture lines were obscure in both the two groups, while the callus in Ad-NC group seemed denser and bigger than that in Ad-integrin β1 shRNA group. 35 d, the fracture line nearly disappeared and bony union was found in both the two groups induced by fibronectin (Figures 1 and 3). These results indicated that BDNF could promote osteoblasts migration through TrkB-integrin β1 pathway. Then, we further investigated the role of two classical kinases ERK1/2 and AKT in BDNF/TrkB signalling. In consistent with our previous study, ERK1/2 and AKT also could be activated via BDNF-TrkB pathway in MC3T3-E1 cells. 24 Inhibition ERK1/2 or AKT could significantly reduce the migration of MC3T3-E1 cells as well as suppressing the expression of integrin β1 (Figure 4). At last, femur fracture mice also suggested that fracture healing would be perturbed especially in early and middles stage when integrin β1 shRNA was locally injected in fracture site.
Fracture repair is a complex, well-orchestrated physiologic process that involves the activation and recruitment of bone-forming cells, such as osteoblast precursor cells and osteoblasts. 9 In adulthood, bone remodelling and repair require osteogenic cells to reach the sites that need to be rebuilt, as a prerequisite for skeletal health and a failure of osteoblasts to reach the sites in need of bone formation may contribute to impaired fracture repair. 25  In time-course experiments, specific phosphorylation of ERK1/2 and AKT was observed. ERK1/2 phosphorylation was inhibited by pre-treatment with ERK1/2 inhibitor PD98059. However, treatment with AKT inhibitor LY249002 had no effect on ERK phosphorylation. Likewise, this induction of AKT phosphorylation was inhibited by pre-treatment with LY294002, but not by PD98059. Our results revealed that activation of ERK and AKT pathways occur separately from each other without any cross talk in BDNF/TrkB-mediated osteoblasts migration.
Integrins are α/β heterodimeric membrane receptors that regulate cellular migration, differentiation or proliferation via interacting with various extracellular matrix ligands such as fibronectin, vimentin and collagen. 16 Integrins involvement in growth factor signalling has long been recognized. 31 Our results showed that BDNF in- in bone formation or fracture healing was contrary in vivo. Ekholm demonstrated that callus size and cartilage synthesis was diminished in α1β1 integrin-deficient mice during bone fracture healing. 18 By constructing mature osteoblast specific osteocalcin-driven transgenic mice, Zimmerman found the impaired bone formation still existed at postnatal 90 days. 17 In the contrast, with using the same osteocalcin-driven transgenic mice, Shekaran demonstrated that integrin β1 deletion in mature osteoblasts did not affect bone density, biomechanics and fracture healing. 33 In our study, local injection of adenovirus, a widely used method in animal experiments, was applied. By comparing the two groups with X-ray and micro-CT,  Runx2, one of the specific markers of osteoblasts was used to reveal the difference of migrated osteoblasts between AD-integrin β1 shRNA and AD-NC group. The expression of Runx2 at 7 and 21 d after fracture in Ad-integrin β1 group was significantly lower than that in Ad-NC group (P < 0.05) while no obvious difference was found at 35 d after fracture between the two groups (P > 0.05). Scar bar = 50 μm In summary, our results suggested that BDNF contributed to MC3T3-E1 migration and increased the expression level of integrin β1 through activation of ERK1/2 and AKT signalling pathways. Adintegrin β1 shRNA local administration leaded to an impairment of early and middle-staged bone fracture healing stimulated by BDNF.
These results provided further understanding for the role of BDNF with extracellular matrix in osteoblasts migration and a novel potential therapeutic candidate for fracture healing.

ACK N OWLED G EM ENTS
This work was supported in part by the National Natural Science and QRX17050).

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