Mice

All procedures were performed by accordance with the “Guide for the Care and Use of Laboratory Animals” and in compliance with the relevant ethics policy and the requirements of the Ethics Committee of Southern University of Science and Technology (certificate SUSTCJY20190427). All mice were maintained in a C57BL/6J background. Specific pathogen-free (SPF) twenty-week­old female C57BL/6J mice were purchased from Byneng Biotechnology Co., Ltd. (Guangzhou, China). LysM-Cre mice were generously provided by Professor Ling Guo (Cancer Hospital Chinese Academy of Medical Sciences Shenzhen Hospital). ROSA 26 mice (No. 007914) and POSTN fl/fl mice (CKOCMP-50706-Postn-B6J-VA, Cyagen Biosciences, Inc.) were purchased from Jackson Laboratory. ROSA mice were crossed with LysM-Cre mice to obtain LysM-Cre-tdTomato (tdTomato) mice. To generate LysM-Cre; POSTN−/− (cKO) mice, POSTN fl/fl mice were mated with LysM-Cre mice to obtain LysM-Cre; POSTN fl/+ mice, which were then mated with POSTN fl/fl mice. To confirm knockout efficiency, littermate analysis was performed by investigators blinded to the mouse genotype. All mice were bred under SPF conditions and weighed every two weeks.

Tibial defect model

Mice bone defect repair was modeled by creating a monocortical defect on the anteromedial surface of the tibia. In brief, mice were anesthetized via inhalation of 2.5% isoflurane and subcutaneously injected with meloxicam. The skin and muscles on the surface of the mouse tibia were peeled off to expose the tibia, and a hole with a diameter of 1 mm was drilled on the anteromedial surface of the tibia using a high-speed drill. The drilling site was flushed with saline to remove bone debris. Then the surrounding muscles and skin were sutured, and erythromycin ointment was applied. Mice were placed on a heating plate until they were awake and able to move normally.

Mechanical loading

On postsurgical days (PSDs) 5, 6, 7, and 8, mechanical loading was applied to the left tibia of the mouse using an electromechanical testing system (ElectroForce 3200 system, TA Instruments, USA). The force applied to wild-type (WT) and tdTomato mice was 6 N peak, 2 Hz sinusoidal, 120 cycles per day. Forces of 2 N, 3 N, 4 N, 5 N, 6 N, and 8 N peak, 2 Hz sinusoidal, 120 cycles/day forces were applied to the tibia of cKO mice, resulting in the same micro-strain in cKO mice as in WT mice. The right tibia served as an unloaded control. All mice were sacrificed on PSD 10.

Immunofluorescence staining

Fresh tibiae were fixed in 4% paraformaldehyde at 4°C for 4 h and decalcified in 0.5 mol/L EDTA at 4 °C for 24 h. The tibiae were cryoprotected in 20% sucrose solution at 4°C for 24 h, then embedded in a gelatin-based medium and stored at −80 °C. Tibiae were cryosectioned into 80 μm thick tissue slices along the long axis of the tibia, perpendicular to the plane of the defect for the creation of longitudinal sections with a cryostat (Leica CM1950, Weztlar, Germany). The samples were stained with primary antibodies against OSX (1:200; ab22552, Abcam, Cambridge, UK), F4/80 (1:200; ab6640, Abcam, Cambridge, UK), periostin (1:200; AF2955, R&D Systems, Minneapolis, MN, USA), CD31 Alex Fluor 488 (1:200; FAB3628G, R&D Systems, Minneapolis, MN, USA), iNOS (1:200; ab3523, Abcam, Cambridge, UK), CD206 (1:200; PA5-46994, Thermo Fisher Scientific, Waltham, MA, USA), Sca-1 (1:200; 710952, Thermo Fisher Scientific, Waltham, MA, USA), Ki-67 (1:200; 710952, Abcam, Cambridge, UK) and endomucin (1:200; sc­65495, Santa Cruz Biotechnology, USA) followed by Alexa Fluor secondary antibodies from donkey (1:400; Thermo Fisher Scientific, Waltham, MA, USA). The slides were mounted with DAPI Fluoromout­G (0100-20, SouthernBiotech, Birmingham, AL, USA) and sealed with coverslips.

Confocal and two-photon microscopy imaging

Three-dimensional fluorescent images were acquired with a 20× objective lens of ZEISS LSM 980 confocal microscope (Germany). The Z­stacks (40 μm thick) were taken at a size of 1 024 × 1 024 pixels, x­y resolution of 0.624 mm, and z­step of 2 mm. The 1 mm defect was imaged by tiling three Z­stacks, spanning 1 500 mm along the long axis of the tibia from one side of the intact cortical bone to the other. DAPI images were acquired (425–475 nm filter) with 405 nm excitation, F4/80, CD31, sca-1 and Ki67 images were acquired (500–530 nm filter) with 488 nm excitation, Emcn, CD206, periostin and tdTomato images were acquired (552–617 nm filter) with 594 nm excitation, OSX, iNOS, periostin images were acquired (662–737 nm filter) with 647 nm excitation, and the second harmonic generation (SHG) of collagen fibers images were acquired (420–465 nm filter) with 860 nm excitation. The imaging was analyzed by Imaris software (Bitplane, USA).

Spatial correlation analysis

Distances between OSX+ cells and type H vessels were calculated as described previously.71 For each channel, a 3D-rendered image and its corresponding distance mask were created in the Imaris software. The distance mask encoded the distance of each pixel to the nearest surface of the channel as the brightness value of the pixel. The diameters of osteoprogenitors were approximately 2 0­50 μm. For the accurate estimation of the spatial correlation between cells and periostin, 10 μm was selected as the boundary condition. When the distance between OSX+ cell and type H vessel was less than 10 μm, OSX+ cells were determined to be coupled with vessels. The distance between OSX+ cells and vessels was then calculated by mapping the encoded distance values from one channel to the spatial coordinates of the other channel. Similarly, the distance between F4/80+ cells and periostin was also calculated as above. In brief, the 3D-rendered images and distance masks were created for F4/80 and periostin, respectively. The distance between F4/80+ cells to periostin was calculated by mapping the encoded distance values from one channel to the spatial coordinates of the other channel. When the distance between F4/80+ cells and periostin was less than 10 μm, periostin was determined to be secreted by this cell.

Microcomputed tomography

Freshly tibiae and femurs were dissected and stored in PBS at 4°C. These specimens were scanned at 6 μm resolution with a microcomputed tomography (Micro-CT) system (Skyscan 1172, Bruker, USA). The intact bone and defect regions were reconstructed using CTvox software (Materialize, USA). Newly formed bone within the defect was quantified for various parameters, including percent bone volume (BV/TV), trabecular number (Tb. N), trabecular thickness (Tb.Th), crossectional bone area per tissue area (B.Ar/T.Ar), maximum moment of inertia (MMI [max]), and connectivity density (Conn. Dn).

Strain field measurement

The mechanical properties of mice tibiae were measured using DIC according to established methods.72 The revealed periosteal region of the tibia was coated by a layer of matte, water-soluble white acrylic paint (XF-2, Tamiya Paint, USA). Minute black speckles were then applied using a high-precision airbrush with matte black water-based paint (0741, Haoshun, China). The painted tibiae were then placed on the loading receptacles of an electromagnetic mechanical test system. The test was conducted at a rate of 0.5 N/s up to a maximum axial load of 12 N. The tibiae were captured in the DIC system (XTDIC-Micro, XTOP, China). The strain distribution was calculated by XTDIC software.

Finite element analysis

The mouse tibia geometry data was imported into MIMICS 21.0 software to reconstruct a 3D solid (Materialize, Leuven, Belgium). The optimized 3D model of cortical bone was imported to Solidworks software 2018 (Dassault Systemes S.A, USA). A 1-mm circular defect was created on the anterior surface of the mouse tibia to simulate the regenerating tissue within the defect. The 3D model of the tibia assembly was imported into Abaqus software 6.1 (Dassault Systèmes Simulia Corp, Providence, RI). In this model, the cortical bone was assigned material properties of Young’s Modulus of 17 GPa and Poisson’s ratio of 0.3. The regenerating tissue was assigned Young’s Modulus of 0.3 GPa and Poisson’s ratio of 0.3. Tied contact was used to connect regenerating tissue within the circular hole on the cortical bone. Boundary conditions and loads were chosen to match the in vivo experimental conditions. The distal end of the tibia was completely fixed. Axial loading of 4 N or 12 N was applied on the metaphyseal surface of the tibia. After finishing the finite element models, the main outcomes of von Mises stress and the strains along the loading axis on the tibial cortical bone surface were calculated using Abaqus.

Four-point bending test

Femurs from WT and cKO mice were harvested and stored in PBS at −20 °C. Samples were returned to room temperature and fixed on an electromagnetic dynamic mechanical testing system (M-1000, CARE, China). This system utilized a 100-N load cell and a linear variable displacement transducer with a ± 40 mm maximum range. The four-point bending mold had a 4 mm upper- and 8 mm lower-span width. The parameters of bending modulus (GPa), ultimate strength (MPa), maximum load (N), and yield load (N) were calculated via a mapped displacement-loading curve.

Dynamic histomorphometry

Mice were intraperitoneally injected with calcein (5 mg/kg) on PSD 2 and 9 after MTD surgery. Tibiae were collected and embedded into methyl methacrylate as previous study.36 The images were captured using confocal microscopy with 488 nm excitation. Dynamic histomorphometry data were measured for periosteal surfaces. The calculation formulas included mineral apposition rate (MAR) and bone formation rate (BFR).

Flow Cytometry Analysis

Fresh tibiae of C57BL/6 J mice were collected and cut 1 mm above and below the defect on PSD 10. Bone marrow cells were flushed from the cut ends of tibiae with complete media (DMEM containing 10% FBS and 1% penicillin­ streptomycin). The cell suspension was then filtered through a 70-mm filter mesh. After centrifugation the cell supernatant, the red blood cells were lysed according to the instructions (C3702, Beyotime, CN). Cells were resuspended in 100 μL FACS buffer (PBS containing 1 nmol/L EDTA and 0.1% BSA). For analyzing M1 and M2 macrophages, cells were incubated with BV421 Rat anti-mouse F4/80 antibody (1:200, 565411, BD Biosciences, San Jose, CA) and PE-Cy7 Rat anti-mouse CD86 (M1 macrophage marker, 560582, BD Biosciences, San Jose, CA) for 30 min at 4 °C, followed by fixation and permeabilization. Alexa Fluor 647 Rat anti-mouse CD206 (M2 macrophage marker, 565250, BD Biosciences, San Jose, CA) was then used for intracellular staining. For analyzing periostin+ macrophage, cells were incubated with anti-periostin antibody (1:200; ab14041, Abcam, Cambridge, UK), followed by Alexa Fluor 488 secondary antibodies from donkey (1:400; Thermo Fisher Scientific, Waltham, MA, USA). Cells were then washed with FACS buffer and analyzed using a four-laser cell sorter (BD FACSCanto SORP, BD Biosciences, San Jose, CA). Flow cytometry data was analyzed with FlowJo software (version 10, BD Bioscience).

RNA-seq and analysis

Total RNA was extracted from mouse defect site using Trizol reagent following the manufacturer’s instructions. MGISEQ-2000RS Kit for DNBseq 2000 was used to purify poly-A+ transcripts and generate libraries with multiplexed barcode adapters following the manufacturer’s instructions. High-throughput sequencing (150 bp, two paired-end) was performed using the DNBseq 2000 in the BGI Genomics Co.,Ltd. All samples passed quality control analysis using Fast QC (Agilent). RNA-seq reads were aligned to the mouse genome (GRCm39) using HISAT2 with default parameters. DESeq2 utilized reads for counts per million (CPM), adjusted P-values (adj. P), and log2 fold changes (Log2FC) computation. Genes with P value < 0.05 and fold-change of at least 1.5 were identified as differentially expressed genes (DEGs) between conditions using the DESeq2 analysis of three RNA-seq biological replicates from different donors. Integrated pathway analysis was performed using Kyoto Encyclopedia of Genes and Genomes (KEGG) and gene ontology (GO) pathways in clusterProfiler. P and False discovery rate (FDR) values were calculated following the R program’s instructions.

Cell cultures

Mouse fibroblasts (L929) were purchased from Procell Life Science & Technology Co., Ltd. L929 cells were cultured in DMEM medium containing 10% FBS and 1% penicillin/streptomycin. The L929 cell culture supernatant was collected and filtered through a 0.22 μm membrane filter for culturing BMDMs. BMDMs were extracted from mice tibiae and femurs. In brief, mice were sacrificed and soaked in 75% ethanol. The skin and muscle surrounding bones were removed and kept on ice in PBS containing 1% penicillin/streptomycin until further dissection. A 1-mL syringe filled with growth medium (DMEM+ 20% FBS+ 1% penicillin/streptomycin+ 20% L929 cell culture supernatant) was inserted into the marrow cavity. The cells were then cultured in a growth medium for 7 days to harvest BMDMs.

Macrophage polarization

BMDMs were plated in 12­-well plates at the density of 4 × 104 cells/well in growth media. After overnight incubation, the growth medium was replaced with fresh growth media and maintained until the cell reached over 80% confluency. The medium was changed every 2 days. LPS was used to induce macrophage polarization to the M1 subtype, and IL-4 protein was used to induce macrophage polarization to the M2 subtype. As previous instruction, the culture medium was changed into M1 macrophage polarization media containing 100 ng/mL LPS (L2880, Sigma-Aldrich, German), the matured M1 macrophages were then harvested at 24 h. For M2 polarization, the culture medium was replaced with M2 macrophage polarization media containing 20 ng/mL IL-4 (PeproTech, Rocky Hill, NJ). Matured M2 macrophages were harvested at 24 h.

Fluid shear stress

BMDMs were plated in a 6­-well plate at a density of 3 × 105 per well in a growth medium. After overnight incubation, cells were exposed to orbital shear stress on the shaker at 1 Pa for 2 h. BMDMs were seeded on round coverslips for immunofluorescence staining. BMDMs in statics status were used as the control.

Real-time PCR

Mice tibiae were collected immediately after euthanasia, frozen in liquid nitrogen, and ground into powder. Total RNA from BMDMs and mouse tibiae were extracted by Trizol reagent (15596, Thermo Fisher Scientific, Waltham, MA, USA). The RNA was transcribed into cDNA using the RevertAid First Strand cDNA synthesis kit (K1622, Thermo Fisher Scientific, Waltham, MA, USA). Quantitative RT-PCR was performed using ABI QuantStudio 1 (A40426, Thermo Fisher Scientific, Waltham, MA, USA) using GoTaq® qPCR Master Mix (A6002, Promega, Madison, WI, USA). The primer sequences were displayed in Supporting Information (Table 1). The results were calculated as the 2−ΔΔCt method and normalized to the expression levels of the housekeeping gene 18S.

Western blotting

Total protein was extracted following the instructions for RIPA lysis buffer (P0013B, Beyotime, CN) in BMDMs. The concentration of total protein was measured using the BCA Protein Assay Kit (P0012, Beyotime, CN). Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to the nitrocellulose blotting membranes (MilliporeSigma, Burlington, MA). The primary antibody for Periostin (1:1000, ab14041, Abcam, Cambridge, UK), Smad2 (1:1 000, 5339S, Cell Signaling Technology, Danvers, MA), Smad3 (1:1 000, 9523S, Cell Signaling Technology, Danvers, MA), p-Smad2 (1:1 000, 18338S, Cell Signaling Technology, Danvers, MA), p-Smad3 (1:1 000, 9520S, Cell Signaling Technology, Danvers, MA), and β-Actin (1:1 000, 4970S, Cell Signaling Technology, Danvers, MA) was applied for incubation, followed by anti-Rabbit IgG (H+L) Secondary Antibody, HRP conjugate (1:10 000, 31460, Thermo Fisher Scientific, Waltham, MA, USA). The imaging was detected using Fusion SoloS.EDGE Chemiluminescence Imaging (Vilber Bio Imaging, France).

Scanning electron microscopy

BMDM morphology on the PCL/HA membrane (provided by Yan Wang, Southern University of Science and Technology) was analyzed using scanning electron microscopy (SEM). BMDMs were seeded on the PCL/HA membrane in a 6­well plate at a density of 3 × 105 per well. After 48 h incubation, samples with BMDMs were washed with double distilled water, fixed with 4% paraformaldehyde at room temperature, dehydrated in a graded ethanol solution, and sputter-coated with gold. The morphology of BMDMs was then observed by a scanning electron microscope (Regulus 8100, Hitachi, Japan).

Cell viability

The biocompatibility of the PCL/HA membrane was tested by seeding BMDMs on the membrane in a 96-well plate at a density of 1 × 104 cells per well. After 48 h incubation, tetrazolium salt-based colorimetric assay (CCK8 test) was performed according to the instruction (CCK8 kit, Dojindo, Japan). The absorbance of each well was measured by a microplate spectrophotometer (Bio-Tek, Winooski, VT, USA) at 450 nm.

TGF-β concentration evaluation

BMDMs were seeded in a 96-well plate at a density of 1 × 104 cells per well. After incubation overnight, cells were cultured in a growth medium containing 0, 2.5, 5.0, 10.0, 20.0, 40.0, 80.0, or 160.0 ng/mL of TGF-β for 24 h. The non-cytotoxic concentration was determined by CCK8 assay.

Artificial periosteum transplantation

PCL/HA membranes served as artificial periosteum. BMDMs extracted from C57BL/6J mice were seeded on PCL/HA membranes. C57BL/6J mice with MTD surgery as recipient mice were randomly divided into four groups: (1) defects for PCL/HA as blank controls; (2) defects for implantation of PCL/HA+BMDM; (3) defects for implantation of PCL/HA+BMDM-FSS; (4) defects for implantation of PCL/HA+BMDM-TGF-β. After MTD surgery, the periosteum around the mice defect was excised, and the PCL/HA membranes in different groups were curved to cover the defect site. Two sides of the membrane were sutured together with the surrounding muscles. Mice were sacrificed on PSD 10, and tibiae were collected for further investigation.

Subcutaneous implantation

PCL/HA membranes seeded with BMDMs, BMDMs-FSS, and BMDMs-TGF-β implanted subcutaneously, with PCL/HA membranes used as control. Mice were anesthetized by inhalation of 2.5% isoflurane and subcutaneously injected with meloxicam. After disinfection with povidone-iodine, a 1 cm dorsal midline incision was made over the thoracolumbar area. Each membrane was inserted on one side of the incision. All the incisions were sutured, reinforced with wound clip, and received erythromycin ointment. The membranes in different groups were removed for the following study.

Statistical analysis

Experimental data involving transplantations were validated using a minimum of six independent mice, and three separate cell groups. The presented data were aggregated from independent experiments. Statistical comparisons between the two groups were performed using a two-tailed Student’s t-test. Multiple comparisons with one variable were performed using two-tailed One-way ANOVA analysis with a Tukey’s Correction. Multiple comparisons with two or more variables were performed using two-tailed Two-way ANOVA analysis with a Tukey’s Correction. Data were presented as mean ± standard deviation (SD). Statistical significance was denoted as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.000 1. Statistical analysis was performed using Prism 8 (GraphPad Software).