Collagen, polycaprolactone and attapulgite composite scaffolds for in vivo bone repair in rabbit models

Although numerous materials have been explored as bone scaffolds, many of them are limited by their low osteoconductivity and high biodegradability. Therefore, new materials are desired to induce bone cell proliferation and facilitate bone formation. Attapulgite (ATP) is a hydrated silicate that exists in nature as a fibrillar clay mineral and is well known for its large specific surface area, high viscosity, and high absorption capacity, and therefore has the potential to be a new type of bone repair material due to its unique physicochemical properties. In this study, composite scaffolds composed of collagen/polycaprolactone/attapulgite (CPA) or collagen/polycaprolactone (CP) were fabricated through a salt-leaching method. The morphology, composition, microstructure, physical, and mechanical characteristics of the CPA and CP scaffolds were assessed. Cells from the mouse multipotent mesenchymal precursor cell line (D1 cells) were cocultured with the scaffolds, and cell adhesion, proliferation, and gene expression on the CPA and CP scaffolds were analyzed. Adult rabbits with radius defects were used to evaluate the performance of these scaffolds in repairing bone defects over 4–12 weeks. The experimental results showed that the cells demonstrated excellent attachment ability on the CPA scaffolds, as well as remarkable upregulation of the levels of osteoblastic markers such as Runx2, Osterix, collagen 1, osteopontin, and osteocalcin. Furthermore, results from radiography, micro-computed tomography, histological and immunohistochemical analysis demonstrated that abundant new bones were formed on the CPA scaffolds. Ultimately, these results demonstrated that CPA composite scaffolds show excellent potential in bone tissue engineering applications, with the capacity to be used as effective bone regeneration and repair scaffolds in clinical applications.


Introduction
Repairing bone defects caused by bone tumor resections, fracture defects or chronic infection is still challenging in orthopedic practice [1][2][3]. Autograft is currently the gold-standard approach to bone repair. However, this technique has several disadvantages, including limited supply of bone tissue, creating new trauma, and the potential for functional damage [4,5]. Allografts are also used clinically because of the availability of bone tissue supply, but these grafts carry the risk of infection, disease transmission, and inducing an immune response [6][7][8]. Advances in bone tissue engineering (BTE) have the potential to address these challenges and promote bone regeneration and repair [9][10][11][12]. However, fabricating 3D porous scaffolds for cell migration, proliferation, differentiation, and nutrient delivery is challenging.
An ideal scaffold for BTE should have (i) excellent mechanical properties to enable bone tissue remodeling; (ii) excellent biocompatibility to facilitate cell attachment and proliferation; (iii) appropriate biodegradability with controllable degradation and resorption rates to match the bone regeneration; (iv) appropriate porosity for cells to attach, migrate, proliferate and deposit the extracellular matrix (ECM) [1,13]. Interconnection of pores is also required to allow the transportation of nutrients, metabolites and the removal of waste products [1,14,15].
In general, three types of bone scaffold materials have been explored in the past: bioactive ceramics, biodegradable metals, and biodegradable polymers. Bioactive ceramics such as calcium phosphate (CaP) ceramics, including hydroxyapatite (HA) and tricalcium phosphate (TCP) have similar compositions to natural bone. These materials have good biocompatibility and osteoconductivity. The degradation of these materials releases ions that can enhance cell activity and facilitate bone repair. However, bioactive ceramics have low toughness and insufficient strength, and are therefore not ideal for use without the addition of other materials [1]. For example, HA is one of the major constituents of natural bone, accounting for 70% (by weight) of human bone [16]. HA material is biocompatible, osteoconductive, noncytotoxic, and nonimmunogenic and has a slow degradation rate [17][18][19]. Therefore, it has been widely used in BTE [5,20]. However, HA is fragile, inflexible, and is difficult to model, limiting its application in bone regeneration and repair [17,21]. CaP bioceramics have also been successfully applied in hard tissue regeneration as a gold standard. These bioceramics exhibit excellent biocompatibility and chemical similarity with hard tissues. They have therefore been used extensively in the clinical repair of bone defects. Recently, additive manufacturing has been employed for the fabrication of bioceramic scaffolds in addition to traditional methods such as solvent casting, freeze drying, phase separation etc. However, due to their brittleness, poor mechanical properties, and lack of other valuable trace metal ions (e.g. Si 4+ , Sr 2+ , and Mg 2+ ), these materials are only limited to treating the defects of non-loadbearing bone. To improve the mechanical properties, osteoconductivity, osteogenic ability and printability of CaP bioceramics, inorganic and organic materials have to be incorporated with CaP bioceramics [22]. In contrast, biodegradable metals such as Mg, Zn, Fe, and their alloys have greater toughness and processability than bioactive ceramics. However, biodegradable metals have a fast degradation process. Their rapid degradation leads to an equally rapid loss of mechanical integrity and stability 1 . Polymer materials are the most extensively used materials for BTE. So far, a considerable number of polymer materials such as polycaprolactone (PCL), poly-L-lactic acid (PLLA), polyglycolic acid (PGA), polylactic acid (PLA) and poly (lactic-co-glycolic) acid (PLGA) have been developed for bone regeneration [23][24][25][26][27][28][29] due to their wide availability and good mechanical properties. However, synthetic polymers show a lack of bioactivity, cell affinity, and hydrophilicity [1]. Therefore, when considering the three types of bone materials, each has its advantages and disadvantages. Composite scaffolds made from the above materials combine their advantages and eliminate their shortcomings , providing improved properties for BTE. Bioactive ceramics or biodegradable metals provide osteoconductivity, mechanical strength and the minerals required for BTE. Polymers can be used as binders to reinforce the mechanical property of bioactive ceramics and biodegradable metals and help model the architecture [30]. For example, chitosan has been used to make composite scaffolds with nanohydroxyapatite. It was found that the compression modulus increased with the amount of nanohydroxyapatite. The composite scaffolds were found to be superior in terms of cell attachment, proliferation, and morphology in relation to a chitosan scaffold [31]. PLGA and PCL have also been widely used as a binder for HA in recent decades, with PCL based composites attracting more attention than other synthetic polymer composites [2,15,31,32]. For example, composite scaffolds made from PCL and oyster shell powder showed improved crystallinity and mineralization ability for BTE application [33]. PCL has also been combined with octacalcium phosphate to fabricate electrospun scaffolds for BTE [32]. Enhanced performance of osteoblast human G-292 cells on the scaffolds was observed.
While a number of excellent polymers are available, the number of excellent bioactive ceramics is far fewer. Attapulgite (ATP) is a natural clay mineral that has been used in the environmental industry as an adsorbent [34], as a decolorizing agent [35], and for catalyst support [36]. ATP is composed of SiO 2 (55.03%), Al 2 O 3 (10.24%), MgO (10.49%), Fe 2 O 3 (3.53%), H 2 O + (10.13%), and H 2 O − (9.73%) [37][38][39], and is well known for its porosity, large specific surface area, high viscosity, and absorption ability. It possesses the physical properties as well as the minerals desired for BTE. Recently, it has been reported that electrospun ATP/PLGA nanofibers promote the differentiation of human mesenchymal stem cells into osteoblasts [40], making ATP a promising scaffold material for BTE. Combining the advantages of ATP and polymers will provide a new strategy for improved bone regeneration [38]. However, the use of ATP as BTE for bone repair is only beginning to emerge. Further studies are required to optimize its applications in BTE.
In this study, collagen (type 1)/polycaprolactone/attapulgite (COL1/PCL/ATP) composite scaffolds were fabricated for BTE. Collagen is an ECM protein of the main tissues (skin, blood vessels, and bone) [41,42]. A scaffold manufactured with collagen has excellent biocompatibility and biodegradability [43]. Collagen is often cross-linked by physical and chemical methods using ultraviolet light, glutaraldehyde, or genipin [7,44] to increase its Scheme 1. Schematic illustration of scaffold design. mechanical properties. PCL has been widely used in regenerative medicine because of its good biocompatibility, elasticity and low immunogenic response [2,18]. The purpose of this study was to compare a COL1/PCL/ATP (CPA) scaffold with a COL1/PCL (CP) scaffold to find out whether the addition of ATP can improve the differentiation of bone marrowderived stroma cells into osteoblast cells, thus promoting bone regeneration. The scaffolds were made through the salt leaching method as illustrated in Scheme 1, and were characterized by a combination of scanning electron microscopy (SEM), Fourier transform infrared spectroscopy (FTIR), and mechanism tests. Elemental composition, porosity, water absorption, contact angle, and degradation were investigated. Cell proliferation and morphology on the scaffolds, as well as the expression of bone regeneration related proteins such as Runx2, Osterix, alkaline phosphatase (ALP), collagen 1 (COL1), osteocalcin (OC), and osteopontin (OPN) were studied. The scaffolds were implanted into bone defects using a rabbit model. Radiography, micro-CT Scan, histological, and immunohistochemical staining were used to evaluate the in vivo bone regeneration. The current study demonstrated that CPA composite scaffolds have great potential in BTE applications.

Preparation of CP and CPA scaffolds
The porous cylindrical CP and CPA scaffolds were fabricated using a salt-leaching method [45]. Briefly, collagen type 1 (COL1) was dissolved in hexafluoroisopropanol (HFLP) to obtain an 8% (w/v) uniform solution. PCL polymer powder (Sigma-Aldrich, USA) was then added to the solution with magnetic stirring until a homogeneous solution was formed. The ratio of COL1: PCL was 4:1 (w/w). Various volumes of nanosized ATP (150-500 nm, Lanzhou Institute of Chemical Physics, Chinese Academy of Sciences) powder and sodium chloride particles (250-500 µm, Shanghai Chemical Industry, China) were mixed with the solution to form pastes with ATP concentrations of 0% (CP scaffold; COL1:PCL:ATP mass ratio of 40:40:0), 10% (CPA-low dose (CPA-L) scaffold; COL1:PCL:ATP mass ratio of 40:40:10), and 20% (CPA-high dose (CPA-H) scaffold; COL1:PCL:ATP mass ratio of 40:40:20). The pastes were then placed into molds that were 100 mm in length and 10 mm in diameter. The molds were left in the drying oven for 6-8 h, and the HFLP was completely evaporated in a vacuum condition until circular cylinders had formed. The resulting highly porous CP and CPA scaffolds had lengths of 45-50 mm and diameter of 4 mm. The scaffolds were then placed into deionized (DI) water for 72 h to elute all the sodium chloride. Samples were then tested with 0.1% AgNO 3 to conform the complete removal of NaCl.
EDC/NHS (in ethanol: water at 80:20 v/v) were used to crosslink the CP and CPA scaffolds. The airdried scaffolds were placed in 250 mM EDC/NHS solution for 24 h at room temperature before rinsing three times with DI water to remove the residual EDC/NHS. After this, the scaffolds were dried at room temperature. Each scaffold was cut with a surgical knife blade to 4 mm in diameter and 15 mm in length, sealed, and sterilized with Co 60 .

ATP.
The morphology and mineral content of the ATP were investigated using SEM (Hitachi S-3400 N system, Japan). ATP was deposited onto a carbon-coated copper grid and coated with gold for 1 h before observation. Each of the six samples was tested three times.

CP and CPA scaffolds.
SEM (JHitachi S 3400 N, Japan) was used to observe the interconnected porous structure and composition of the scaffolds. Before observation, scaffolds stored in 100% ethanol were dried using supercritical carbon dioxide.

Water absorption, porosity, contact angles, and degradation of scaffolds
The water absorption capacities of the CP and CPA scaffolds were determined by swelling the scaffolds in 0.2 M PBS at 37 • C (pH = 7.4). The scaffolds were cut to 10 mm in length and 4 mm in diameter, and separately immersed into 0.2 M PBS solution for 5 min and then for 1, 2, 3, 4, 5, 6, 24, 48, and 72 h. The water in each scaffold was then removed and weighed. The water absorption ratio was obtained with the following equation: where W H is the weight of the soaked scaffold and W D is the weight of the dry scaffold.
For the assessment of scaffold porosity, cylindrical samples were cut into pieces measuring 2 mm in length and 4 mm in diameter. The porosity of the scaffolds was measured based on gravimetry according to the following equation: where M 1 is the initial mass of a density bottle filled with ethanol, M s is the mass of the dry scaffold, M 2 is the mass of the scaffold submerged in absolute ethanol with the density bottle, and M 3 is the mass of the density bottle after the scaffold was gently removed.
The contact angle measurement of the scaffolds was carried out with a contact angle analyzer (FTA125, First Ten Angstroms, Portsmouth, VA, USA). The scaffolds were cut into 1 cm 2 square pieces (N = 4) and then placed on a testing plate. Subsequently, 0.01 ml of DI water was carefully dropped onto the test sample surface. After 10 s, the contact angles were recorded by video monitor.
For the assessment of scaffold degradation, cylindrical samples were cut into 10 mm lengths. The samples were placed into vials containing 5 ml PBS (pH = 7.4, 0.2 M), and these vials were placed into a shaking bath (37 • C) for 1-8 weeks. The PBS solution was exchanged weekly. The loss in mass of the scaffolds was then calculated using the following equation: where M 0 is the initial dry weight of the scaffold and M D is the dry weight after degradation.

FTIR
FTIR (Perkin-Elmer 1600 series, USA) was used to evaluate the functional groups of the scaffolds. Before testing, a total of 2 mg dried CP or CPA scaffold powder was mixed with 200 mg KBr and compressed into a disk. All spectra were obtained between 4000 and 400 cm −1 at a 4 cm −1 resolution with 32 scans. Six samples were examined for three times each.

Mechanical testing
Before testing, the scaffolds were cut so that the gauge length and diameter of all specimens measured 8 mm and 4 mm, respectively. For the assessment of flexural strength, the scaffolds were tested with a 3-point bend test on an Instron 4505 universal testing machine (Instron Pty Ltd, Norwood, MA, USA) at a rate of 1 mm min −1 . Compressive tests were performed on dry and wet scaffold (After immersion in the 0.02 M PBS (pH = 7.4) for 3 d) samples using a universal testing machine with a 1 kN load cell. Specimens were compressed to 60% of their total height at a rate of 2 mm min −1 to obtain the load/displacement curve. Stress-strain graphs were then drawn according to the load/deformation data.

Cell culture and seeding on scaffolds
Cells from the D1 mouse multipotent mesenchymal precursor cell line (CRL-12424, ATCC, USA) were cultured in DMEM/F12 (Gibco, USA) supplemented with 10% fetal calf serum (FCS) and 1% penicillin/streptomycin at 37 • C and 5% CO 2 . The scaffolds were sterilized with 70% ethanol for 1 h and then rinsed three times with sterile PBS before the cells were loaded. The scaffolds were soaked in the DMEM/F12 medium overnight. Cells (15 µl, approximately 1.6 × 10 5 cells ml −1 ) were seeded on both sides of the scaffold with a 2 h incubation step between additional seedings. The cell-seeded scaffolds were then placed into new 24-well plates containing 1 ml of DMEM/F12 complete medium. The plates were incubated at 37 • C with 5% CO 2 and were cultured for 7-21 d. During this period, the medium was changed twice weekly. At the desired time points, the cell-seeded scaffolds were removed from the medium and analyzed for cell adhesion, proliferation, and expression of osteoblast-related genes.

Cell proliferation and morphology
Cell proliferation was measured using the Cell Counting Kit-8 (CCK-8; Promega, USA). Briefly, the cell-seeded scaffold was removed from the original medium and placed into new cell culture plates after 1-7 d of culture. A total of 1 ml of medium containing 100 µl CCK-8 was added into each well, and the plates were incubated at 37 • C for 3 h. A total of 200 µl supernatant was then removed from each well to a new 96-well plate and tested for absorbance at 490 nm using a plate reader (BIOTEK, USA). Six parallel samples from each group were analyzed, and each test was run in triplicate. DMEM/F12 containing 100 µl CCK-8 was used as a control.
The microstructures of the scaffolds and the cellseeded scaffolds were assessed using SEM (JSM-5800LV). At days 3 and 7, the cell-seeded scaffolds were fixed in 2.5% glutaraldehyde in PBS for 3 additional days. The samples were then dehydrated by means of a series of alcohols at different concentrations (15%, 30%, 50%, 70%, 80%, 90%, 100%) followed by drying with supercritical carbon dioxide. Samples were sputter coated with gold for 10 min before the SEM measurement. Cell morphology, attachment on scaffolds, and cell density were assessed, respectively.

Fluorescence staining
At day 7, the cell-seeded scaffolds were rinsed in 0. 2 M PBS and fixed in 10% formaldehyde for 3 additional days. After dehydration, the scaffolds were embedded in paraffin, and 5 µm sections were used for fluorescence staining. Samples were stained with 3 µM 4 ′ ,6diamidino-2-phenylindole dihydrochloride (DAPI, Sigma, USA) for the nucleus and 10 µM phalloidin (Invitrogen Life Technologies, USA) for the F-actin filaments. The samples were then observed under a fluorescence microscope (Olympus, Japan).

Reverse transcriptase and quantitative real-time PCR
The total RNA of the cells was extracted using the Trizol reagent (Invitrogen, USA) following the manufacturer's procedure. Samples with 2 µg total RNA were reverse transcribed using the M-MLV Reverse Transcription kit with an oligo (dT) 18 primer (TaKaRa Company, Dalian, China). PCR amplification was carried out in a volume of 25 µl, which contained 200 ng cDNA, 20 µM each of the forward and reverse primers (Glyceraldehyde phosphate dehydrogenase (GAPDH), Runx2, Osterix, ALP, COL1, OC, and OPN), which were designed using Primer 3 software, and 12.5 µl of the Power SYBR Green PCR Master Mix (TaKaRa). The real-time PCR cycling conditions comprised an initial denaturation step at 95 • C for 10 min, 30 cycles of 30 s at 95 • C for denaturation, 30 s at 60 • C for annealing, 60 s at 72 • C for extension, and 10 min at 72 • C for a final extension. Relative gene expression was detected using a comparative Ct method (2 −ddCt ) using the following equation [46]: Assays were carried out in triplicate for each gene.

Examination of bone regeneration in rabbits
To assess the ability of the CPA-H and CP scaffolds to promote bone regeneration in a critical-sized defect, a 15 mm rabbit radius defect model was used. Male Japanese white rabbits (60 in total , 2 months old , weighing 1.5-2.0 kg) were used in the study and were divided into three groups: defect control, CP, and CPA-H. All animals were anesthetized by intravenous injection of pentobarbital sodium (0.3 ml kg −1 ). The bone radius was exposed with a straight incision approximately 3 cm in length along the forearm, and 15 mm defects were made in the bilateral radius using an electric drill. Bone debris was removed using physiologic saline irrigation. The radius defects were then implanted with CP or CPA-H scaffolds (CP and CPA-H groups), with the scaffolds fixed to the ulna using absorbable sutures. In the control group, no scaffolds were implanted in the defects. The wounds were then sutured. To prevent postoperative infections, penicillin injections (400 k units) were administered once a day for 7 d. The general conditions of the rabbits (diet, activity, energy, and wound healing) were continuously monitored for 2 weeks. The following experiments were carried out at 4, 8, and 12 weeks after surgery, respectively.

Radiological monitoring
Radiography was performed to identify new bone regeneration during the healing process. Images were captured at 4, 8, and 12 weeks after surgery using an xray machine (Faxitron MX-20, USA). Images were obtained at 45 kV and at 3-5 mAs. Evidence of callus formation and bone healing in the area of defects was also assessed.

Micro-CT scan
After radiographic examination, the rabbits were euthanized with an overdose of pentobarbital. The soft tissues attached to the bone defects were removed and the tibias were harvested and fixed in 10% formaldehyde for 48 h. The tibias were scanned using Micro-computed tomography (micro-CT) (GE Healthcare, USA) with the following parameters: voxel resolution of 15 µm with an integration time of 300 ms, at 50 keV with 850 µA current. Micro-CT software was used to transformed into 3D images and measure the bone mineral density of new bones. 3 rabbits in each group were selected for scanning at 4, 8, and 12 weeks after surgery ; each sample was scanned three times, and the amount of bone regeneration was calculated based on the average value.

Histological and immunohistochemical staining
The tibias were fixed in 10% formaldehyde for 3 d and were then decalcified in 5% EDTA at room temperature (pH = 7.0) for 5 weeks. After complete decalcification, the bone specimens were dehydrated in ascending ethanol and then embedded in paraffin. The paraffin blocks were sectioned into 5 µm thicknesses, stained with H&E and Masson staining, and analyzed with a microscope (Olympus, Japan).
The quantity of new bone area ratio was assessed using Image-Pro Plus 6.0 software (Media Cybernetics, Rockville, MD, USA). For the immunohistochemical analysis, the sections were deparaffinized and then rehydrated by boiling them for 10 min in a sodium citrate solution. The following protocols were performed according to the manual of the immunohistochemical kit: Sections were treated with several primary antibodies: rabbit monoclonal ALP antibody (1:1000, overnight at 4 • C; Abcam, USA), rabbit monoclonal COL1 (1:1000, overnight at 4 • C; Abcam, USA), rabbit monoclonal type 2 collagen (COL2) (1:1000, overnight at 4 • C; Abcam, USA), (1:1000, overnight at 4 • C; Abcam, USA), and rabbit monoclonal OPN antibody (1:1000, overnight at 4 • C; Abcam, USA); and then with a biotinylated goat anti-mouse or anti-rabbit secondary antibody for 30 min. Sections were stained with 3,3 ′diaminobenzidine tetrahydrochloride (DAB; Sigma, USA), rinsed with PBS, dehydrated, and examined with a microscope (Olympus). All histological and immunohistochemical images were photographed digitally with a microscope and analyzed with a digital image analysis system (DXM 1200, Nikon, Japan).

Study approval
All animal experiments in this study were approved by China's Animal Research Authority and Ethics Committee of the General Hospital of Lanzhou Military Command of the PLA. All animal studies were performed in compliance with the regulations and guidelines of the Orthopedics department of the General Hospital of Lanzhou Military Command of the PLA, and animal care was conducted according to the Association for Assessment and Accreditation of Laboratory Animal Care international (AAALAC) and Institutional Animal Care and Use Committee (IACUC) guidelines.

Statistical analyses
All data are presented as means ± standard deviations. A two-tailed unpaired Student's t test was used to assess for significant differences between groups; P value < 0.05 was considered statistically significant. SPSS 13.0 statistical software (SPSS, USA) was used to analyze all data.

Morphology and chemical composition of the ATP and CPA scaffolds
SEM measurements were carried out to reveal the morphology of the ATP, CP and CPA scaffolds with the data shown in figure 1. The ATP ( figure 1(A)) showed clusters of single rod structures with relatively smooth surfaces. The approximate length of the ATP rods was 0.2-1.8 µm with a diameter range from 15 to 50 nm. The ATP was found to be composed of SiO 2 (33.7%), MgO (9.0%), Al 2 O 3 (5.6%), Fe 2 O 3 (4.1%), CaO (1.5%), KBr (1.3%), and Ti (0.5%) (figure 1(Ac)). Figure 1(B) shows the SEM micrographs of CP and CPA scaffolds. The scaffolds showed porous structures, with pore sizes ranging from 250 to 500 µm, which were created by the sodium chloride particles. This porosity was beneficial for nutrient exchange and cell migration. Figure  1(C) shows the chemical composition of the CPA-H scaffolds. Strong signals were detected for Si, Mg, Al, and Fe, indicating the presence of ATP in the scaffolds. With an increase in ATP, Si was also increased (stacks.iop.org/BMM/15/045022/ figure S1), indicating the existence of an Si-containing material in the scaffolds. A strong carbon signal was also detected in the CPA scaffolds.

Characterization of the scaffolds: porosity, water absorption, contact angle, degradability, mechanical properties, and FTIR study
The CPA-H scaffolds, with and without cross-linking, were observed under SEM. Higher porosity was observed in CPA scaffolds that were not cross-linked ( figure 2(A)). When the scaffold was cross-linked by EDC/NHS, the structure became more compacted (figure 2(B)) with decreased porosity. However, after cross-linking, the porosity values for CP (87.7%), CPA-L (82.9%) and CPA-H (82.9%) were very similar ( figure 2(C)). The addition of the ATP only resulted in a small decrease in porosity. The water absorption values for the cross-linked scaffolds are shown in figure 2(D). The water absorption values decreased from 281.7% for CP to 234.9% for CPA-L and 150.7% for CPA-H, indicating that the water absorption capacity of the scaffolds decreased with increasing ATP concentration. The was due to the fact that ATP has less water adsorption ability. The contact angles (shown in figure 2(E)) of the scaffolds were 76 • for CP, 57.3 • for CPA-L and 42.5 • for CPA-H. These data suggest that the contact angle of the scaffolds was decreased when the amount of ATP was increased.     stretching vibration representing the Si-O vibrational peak at 800 cm −1 was also observed. Peaks that occurred at 3400 cm −1 and 1600 cm −1 represented amide A and amide I from collagen in both the CP and CPA scaffolds. In the CPA scaffolds, amide I and amide II appeared as peaks at 1735 cm −1 and 1456 cm −1 representing C = O and N-H stretching of the collagen polyether peaks, respectively. Figure 3 shows the (CRL-12424) cell morphology, adhesion, and proliferation on the scaffolds. After 3 d of culturing, a small number of cells was observed on the CP scaffolds, whereas large numbers of cells were observed on the CPA-L and CPA-H scaffolds. At day 7, the cell population density increased for all the scaffolds, and the deposition of ECM was observed on both CPA-L and CPA-H scaffolds. During the same time period, a very high cell density was observed on the CPA-H scaffold surface. The cells were more elongated on the CPA-H scaffold surface. The pores on the scaffold were covered by the cells and ECM secreted from the cells. In addition, a crystal-like ECM structure was observed on the lower-right side of the CPA-H scaffolds ( figure 3(A)). Immunofluorescence staining ( figure 3(B)) indicated substantially higher cell numbers on the CPA scaffolds than on the CP scaffolds at 7 d after inoculation.

Cell morphology, adhesion, and proliferation on the scaffolds
CCK-8 assay was performed to evaluate the cell proliferation of D1 cells on scaffolds with the data shown in figure 3(C). No significant difference in cell proliferation was seen for the CPA and CP scaffolds at

Expression of osteoblast molecular markers
Osteoblast molecular markers are important indicators for bone regeneration.

Implantation procedure, radiographic examination and micro-CT scan
The CP and CPA-H scaffolds were implanted into a 15 mm rabbit radius defect model with the implantation procedure shown in figure 4(A). X-ray experiments were performed at 4, 8, and 12 weeks after implantation ( figure 4(B)). Post-operative radiographs demonstrate that clear defects can be observed at the implant sites at week 4. However, some low-density developments were also observed in the bone defect areas. Bone regeneration with CPA-H is better than that of CPA-L and CP. At week 8, defects can still be seen in CP scaffold areas, while in CPA scaffold areas, the defects were filled in with low density new bone. At week 12, all the defects filled in with CPA scaffolds show higher density bone. In the CPA-H group, a bone marrow cavity had formed in the newly formed bone by week 12 ( figure 4(B)). Figure 4(C) shows the Micro-CT scan results of the scaffold areas. At 4 weeks, a small amount of bonelike tissue was observed within the bone defect area in Data are shown as mean ± standard deviation (n = 6). Scale bar = 100 µm. * P < 0.05, * * P < 0.01, * * * P < 0.01, # P < 0.05, ## P < 0.01. the CP scaffold group (figure 4(C)). However, the tissue mineral content was markedly lower in this group than that in the CPA-H scaffold group (figure 4(D)). After 8 weeks, bone-like tissue was observed in the CPA-H and CP groups, but the bone observed in the CP group was discontinuous. In addition, the tissue mineral content of the bone defect area was markedly lower in the CP group than that in the CPA-H group. A large amount of new bone-like tissue was observed in the bone defect area of the CPA-H scaffold group at 12 weeks after transplantation. The tissue mineral content was markedly increased in the CPA-H scaffold versus the control group. The tissue mineral content was shown to be dose and time dependent ( figure  4(D)). The more ATP presented in the scaffolds the higher tissue mineral content was detected. The tissue mineral content also increases with post-operative time.

New bone formation assessed by histologic analysis
Histologic Analysis was performed at week 4, 8, and 12 to evaluate the new bone formation with data shown in figure 5. 4 weeks after transplantation, monocytes, plasma cells, and lymphocytes were observed within all implant scaffolds. At week 8, the presence of fibrous connective tissue was seen in the control group. In the CP group, the scaffold materials partially disappeared, and more fibrous connective tissue grew into the scaffolds, indicating that the degradation of the scaffolds was faster than the desired degradation rate. In contrast, new bone was observed in the CPA-H scaffold group. At week 12, most of the area of bone defect was covered by fibrous connective tissue in the control group, indicating that new bone had not been formed. In the CP group, the scaffold was completely replaced by fibrous tissue. In contrast, in the CPA-H group, new bone formation was observed around the scaffold area. The original shape of the implant scaffolds had gradually disappeared and was visible only in the center of the scaffolds (figures 5(A) and (C)).
Masson staining ( figure 5(B)) demonstrated that the control group had fibrous tissue at the bone defect area 4 weeks after transplantation. The fibrous tissue filled in the central portion of the defect area. On the other hand, the reconnected fibrous tissue and small areas of bone-like tissue were found in the defect bone in the CP and CPA-H scaffold groups. In the CPA-H group, single bone fragments were observed 8 weeks after scaffold implantation. At 12 weeks, abundant new bone formation was seen in the center of the CPA-H scaffolds, but CPA-H materials had not been completely absorbed (figures 5(B) and (D)).

Immunohistochemical assessment
Data regarding immunohistochemical assessment are shown in figure 6. 4 weeks after implantation, positive expression of ALP, COL1, COL2, and OPN was observed in all three groups. For ALP, COL2 and OPN, the expression levels showed time and ATP dose dependent effects for all three groups, while the expression levels in the CPA-H scaffolds were higher than that in the CP scaffolds and the control group. For COL1, high expression was observed in the control group. The level of expression was found to be similar over a period of 4-12 weeks. In the CP group, COL1 expression increased with time, while in the CPA-H group, COL1 expression increased with time and leveled off at week 12. The high expression of COL1 in the control group demonstrated the presence of cartilage tissue within the control group, indicating that the bone regeneration was not good. The intensity of positive staining for ALP, COL1 COL2 and OPN in the CPA-H scaffolds increased over time. This expression was mainly seen in the border of the new bone. The results demonstrate that in the control group the new tissue formed was mainly fibrous tissue containing a lot of collagen type 1, whereas in the CP and CPA-H groups, new bone tissues were formed and the amount of new bone in the CPA-H group was much higher than that in the CP group.

Discussion
In this work, we investigated the bone regeneration efficacy of CPA composite scaffolds for segmental bone defect healing. Our results show that the water absorption, porosity, contact angle, and degradation of the scaffolds were improved with increasing ATP content. Pore sizes are important for BTE scaffolds and have been shown to affect the progression of osteogenesis [47]. Small pore sizes (<100 µm) not only reduce cell migration but also restrict nutrient delivery and waste removal. Additionally, they also limit the amount of new tissue regeneration [47]. It has been reported that pore sizes >100 µm with gradients are recommended [2,48]. In our study, the CP and CPA composite scaffolds were fabricated by the salt-leaching method. The scaffolds showed porous structures, with pore sizes ranging from 250 to 500 µm, which were created by sodium chloride particles. This porosity proved beneficial for nutrient exchange, cell migration, and bone regeneration.
Ideally, biomaterial implants for clinical applications should match the mechanical properties of the host tissue and exhibit strong interfacial bonds with hard and/or soft tissues. However, the major disadvantages of the inorganic and organic component are their low mechanical strength and fracture toughness, which typically restrict their use in BTE [3]. Our results showed that a higher concentration of ATP led to increased compressive strength, suggesting that increased ATP leads to improved mechanical strength. ATP can provide both strength and bioactivity, while collagen can add structural reinforcement and toughness. The mechanical strength of the CPA composite scaffolds is lower than that of natural bone. In the future, our strategy will be to produce enhanced composites and use novel fabrication methods (e.g. 3D printing) to overcome this drawback.
Cytocompatibility of the biomaterials used in scaffolds plays a vital role in sustaining cell adhesion, proliferation, and differentiation as well as enabling tissue regeneration after implantation [49]. In this study, the CPA scaffolds promoted colony morphology and induced cell proliferation when they were cocultured with cells for 1-7 d. The proliferation rate of cells on the CPA-H scaffold was significantly higher than the rate on the CPA-L and CP scaffolds after 7 d, suggesting that CPA scaffolds induce the proliferation of cells in a dose-dependent manner and the addition of ATP has a significant effect on cell growth. These results are consistent with the results reported by Wang et al [40], who found that the doping of ATP within PLGA nanofibers allows the osteoblastic differentiation of hMSCs to be induced. Our results also demonstrate that a CPA scaffold can increase cell proliferation and promote osteoblast related gene expression, and hence increase bone regeneration.
An immunofluorescence assay demonstrated that D1 cells were able to proliferate and migrate in the interior of the CPA scaffolds. Additionally, in vivo studies in a rat model demonstrated no evidence of severe immune response or fibrous capsule formation with CPA scaffolds, suggesting good biocompatibility. The D1 cells grown on CPA scaffolds had upregulated levels of osteoblastic markers, including upregulated levels of Runx2 at day 7 and Osterix at day 21. These results suggest that D1 cells can differentiate into osteoblasts on CPA scaffolds. We also found that the expressions of COL1, ALP, OC, and OPN were significantly higher on CPA scaffolds than that on CP scaffolds, and this increased expression occurred in a dose-dependent and time-dependent manner. These in vitro findings suggest that CPA has strong osteoinductive properties. CPA may promote D1 cell osteoblast differentiation through several mechanisms: (i), Elements contained in ATP (SiO 2 , Al 2 O 3 , MgO, Fe 2 O 3 , H 2 O + , and H 2 O − ), particularly SiO 2 , can induce D1 cells to differentiate into osteoblasts [1]. (ii), ATP has the desired physical properties, such as larger specific surface area, high viscosity, and high absorption ability, to make this mineral osteoinductive. (iii), The synergistic effect between the chemical and physical properties of ATP may induce D1 cells to differentiate into osteoblasts.
During in vivo bone tissue formation, the biomaterial should degrade slowly to match the perfect restoration of the bone defect [39,50]. In the rabbit model we used in this study, a large amount of bone mass was observed in the CPA scaffold groups at weeks 8 and 12. This was particularly obvious in the scaffold that contained CPA-H, in which the bone mass completely filled in the bone defect at week 12.
Immunohistochemical analysis also demonstrated a marked increase in the expression of ALP, COL1, COL2, and OPN in CPA scaffolds versus CP scaffolds at weeks 8 and 12. These findings again suggest that ATP has the capacity to induce the formation of vital bony structures. While ATP has not yet been widely used in tissue engineering, this study opens a door for ATP to be considered as a promising material for tissue engineering applications.

Conclusions
Salt leaching is an effective method of scaffold fabrication to create the macroporous scaffolds needed for BTE [51]. In this study, we used the salt-leaching method to fabricate macro-mesoporous scaffolds containing either CP or CPA as bone regeneration materials. We found that macropores (approximately 200-500 µm) had been created by sodium chloride in both the CP and CPA scaffolds, which is important for cell ingrowth and nutrition delivery. Cross-linking of collagen with EDC/NHS changed the structure of the collagen, which led to changes in porosity and water absorption. Cross-linking improved the compressive strength and decreased the porosity of the CPA scaffolds. These results suggest that cross-linked CPA scaffolds are more suitable for BTE. In conclusion, we found that CPA scaffolds prepared through a salt-leaching technique were more effective than CP scaffolds in inducing bone regeneration. These results suggest that CPA scaffolds, because of their specific physical and chemical properties, may have great potential for inducing bone regeneration and promoting bone repair. Further studies are needed to assess the use of these scaffolds in a clinical setting.