Three dimensional printed polylactic acid-hydroxyapatite composite scaffolds for prefabricating vascularized tissue engineered bone: An in vivo bioreactor model

The repair of large bone defects with complex geometries remains a major clinical challenge. Here, we explored the feasibility of fabricating polylactic acid-hydroxyapatite (PLA-HA) composite scaffolds. These scaffolds were constructed from vascularized tissue engineered bone using an in vivo bioreactor (IVB) strategy with three-dimensional printing technology. Specifically, a rabbit model was established to prefabricate vascularized tissue engineered bone in two groups. An experimental group (EG) was designed using a tibial periosteum capsule filled with 3D printed (3DP) PLA-HA composite scaffolds seeded with bone marrow stromal cells (BMSCs) and crossed with a vascular bundle. 3DP PLA-HA scaffolds were also combined with autologous BMSCs and transplanted to tibial periosteum without blood vessel as a control group (CG). After four and eight weeks, neovascularisation and bone tissues were analysed by studying related genes, micro-computed tomography (Micro-CT) and histological examinations between groups. The results showed that our method capably generated vascularized tissue engineered bone in vivo. Furthermore, we observed significant differences in neovascular and new viable bone formation in the two groups. In this study, we demonstrated the feasibility of generating large vascularized bone tissues in vivo with 3DP PLA-HA composite scaffolds.

Quantitative detection genes expression. To better characterize the expression of angiogenesis and osteogenic biomarkers generated within the in vivo bioreactors, qRT-PCR was performed for VEGF and BMP-2 expressed during vascularisation and for OPN and COL-1, indicative of viable bone formation. As shown in Fig. 1, four genes related to angiogenesis and osteogenic were detected. The experimental group (EG) revealed significantly greater expression of all four genes compared with the control group (CG). Significant differences were observed between the two groups at 4 and 8 weeks after implantation (P < 0.05).
Micro-CT Measurement. The structures of microvasculature and newly formed bone tissues in bioreactor specimens were examined using microangiography and micro-CT. As shown in Fig. 2, microvascular architecture and angiogenesis could be observed in the two groups. Newly formed vessels were shown growing throughout the in vivo bioreactor and were found inside and outside the scaffold samples. 3D reconstructions of the micro-CT scans demonstrated that the average number of vessels and the average total vessel volume penetrating the in vivo bioreactor were higher in the EG than in the CG (Fig. 3). The number and volume of blood vessels increased at 8 weeks compared with at 4 weeks after implantation. The differences between the experimental and control samples were significantly different (p < 0.05, respectively).
From Fig. 4, newly formed bone tissue and a strengthened ossification trend could be observed at different time points. The architecture of the generated bone tissue was assessed, and the values of BV/TV, Tb.N and Tb.Th were significantly greater for the EG compared with the CG at 4 and 8 weeks after surgery. Tb.Sp values decreased in the EG compared with the CG. The differences between the experimental and control samples in vivo bioreactor were significantly different (p < 0.05, respectively) (Fig. 5).
Histological analysish. A histological examination of the specimens in the in vivo bioreactor revealed neovascular formation and ossified tissue generation in the two groups after 4 and 8 weeks of implantation in rabbits. As shown in Fig. 6, newly formed blood vessels were clearly observed in the EG; these few blood vessels were seen inside the bone tissue with infiltrating fibrous connective tissue. As shown in Fig. 7, new viable bone formation was demonstrated in the two groups. The lacunae of the neo-tissue, which were easily identified as osseous tissue, can be further visualized in the EG, whereas some areas of less mature woven bone within the neo-tissue were seen in the CG. Compared with at 4 weeks after implantation, at 8 weeks, stains of new blood vessels and bone tissue were more intense and uniform indicating the development of blood vessel distribution and mature osseous tissue.
Immunohistochemical analyses were performed to provide a more detailed analysis of the tissue formation process. The mean amounts and percentages of blood vessels and bone tissue positively stained for CD31 and OCN were higher in the EG compared with the CG. At 8 weeks after implantation compared with at 4 weeks, there were higher positive expression levels of CD31 in enhanced capillary formation and of OCN in new bone formation (Fig. 8). The differences between the experimental and control samples were significant (P < 0.05, respectively). These results correlated with the results determined by the micro-CT analysis.

Discussion
The reconstruction of large volume bony defects with customizable dimensions remains a significant clinical challenge. To generate vascularized large bone tissue as ideal functional bone grafts and to shape them as best as possible for aesthetic purposes, many studies have explored various reconstruction methods and biomaterial manufacturing techniques [21][22][23][24] . Rapid advances in bioreactor in vivo technology, tissue engineering practices, and 3D printing technologies represent promising strategies for bone defect reconstruction for reforming bones with shapes conforming to the geometry of the defect site.
A reason for poor bone formation may be insufficient vascularization of the scaffolds 25 . Vascularization is an essential prerequisite for bone development and for the regeneration of osteogenic cells. Furthermore, vascularization provides enhanced transport for oxygen, nutrients, and important bioactive factors to bone formation areas 26,27 . When compared with in vitro bone tissue engineering, these in vivo bioreactors more readily promote angiogenesis. As previously described 11 , to generate extensive vascularization and osteogenesis in the scaffolds, a saphenous vascular bundle was combined with periosteum. To prefabricate a vascularized bone graft, a saphenous vascular bundle was centrally inserted into a scaffold to provide transport for progenitor cells, cytokines, oxygen and nutrients and to remove waste products from cells 28 . The periosteum is a thin but highly vascularized tissue that provides a number of osteoblasts, pluripotent stem cells, and progenitor cells for vascularized bone growth 29 . A periosteal flap can be used to wrap the tissue-engineered construct to minimize tissue disturbance 30 . For these reasons, the combined use of a blood vessel and periosteum can take advantage of two well-established bone graft prefabrication strategies.
Seeding cells, growth factors and bioactive scaffolds are three essential elements for bone tissue engineering. Many approaches have involved various combinations of these elements to regenerate vascularized bone tissue 31 . However, conflicting studies have indicated contradictory uses of exogenous growth factors as core components for in vitro tissue engineering. In our study, BMSCs were selected as seed cells, whereas angioinductive or osteoblast growth factors were not used. Many studies have demonstrated that BMSCs can be used for in vivo graft colonization to enhance the effectiveness of bone formation 32,33 . Regarding the introduction of exogenous growth   factors to in vivo bioreactors, a few studies have suggested that these growth factors carry risks of tissue overgrowth and nerve impingement and may increase the complexity of the technology and associated regulatory burdens for clinical translation 34,35 . Thus, strategies utilizing in vivo bioreactors without exogenous growth factors are being explored. This represents a major advancement in the field and facilitates clinical translation.
To support viable bone tissue growth, in vivo bioreactors must be filled with appropriate scaffold materials 36 . Over the past decade, scaffold design remains the main challenge in tissue engineering due to the large number of requirements that need to be met, such as having the ability to deliver cells, supporting differentiation of regenerative cells, irregular geometries, biocompatibility, osteoconductivity and osteoinductivity 37,38 . To achieve geometric precision of the bone construct to conform to the contours of the recipient site, the fabrication of custom-contoured scaffolds is critical. With the aid of three-dimensional printing technology, the geometry and material properties of scaffolds can be controlled and tailored to the shape of defects 39 .
Among the various materials used in recent years, PLA-HA scaffolds have demonstrated their ability to promote new bone regeneration 40,41 . PLA is a polymer of lactic acid with high biocompatibility and degradability and can be easily fabricated into porous scaffolds 42 . HA can act as a reinforcing material to improve the osteoconductivity of the scaffold 43 . PLA and HA have been mixed to form a porous scaffold that exhibits the superimposed behaviour of ceramics in a polymeric matrix 19 . PLA-HA composite materials have been formed as scaffolds using three-dimensional printing technology and have been shown to have good biocompatibility and bioactivity in vitro 18 . In the present study, the good osteogenic capability and osteoinductive activity of a 3DP PLA-HA in vivo bioreactor were estimated and were shown to enhance bone formation.
In vivo bioreactors have been successfully used to create a series of complex tissues. Various types of scaffolds, orthotopic sites and vascular bundles have been considered for the vascularization and remodelling of regenerated bone tissue 28 . Studies have described in vivo bioreactors using vascular bundles filled with different  46 . Our previous research reported an in vivo bioreactor that used a saphenous vascular bundle threaded through beta tricalcium phosphate (β-TCP) granules wrapped by muscularis membrane 23 . However, in the present study, for the first time, we prefabricated vascularized bone grafts by inserting a saphenous vascular bundle into 3DP PLA-HA composite materials wrapped by a periosteal flap. The results showed a developed capillary network and new bone tissue at regenerated sites.
The mechanical stress is a critical property for estimating the strength of vascularized bone tissue. While there is a lack of detailed evidence on new bone formation due to mechanical stress, the micro-CT results in our study determined the mechanical strength of new bone tissue. We found that in the EG, the average values of BV/TV and Tb.N and Tb.Th were greater and that of Tb.Sp was smaller than those in the CG. According to Mittra, the measured micro-CT parameters and mechanical strengths of BV/TV, Tb.N, and Tb.Th were found to be positively correlated with increasing mechanical properties, whereas Tb.Sp was negatively correlated with increasing mechanical properties 47 . Therefore, the outcomes suggested that bone tissues grown within the in vivo bioreactor in the EG had greater mechanical strength than that in the CG. In future studies, additional mechanical analyses will be performed to further explore the nature of generated bone tissue.
Our study indicated that in vivo bioreactors and 3DP PLA-HA composite scaffolds are promising tools for the prefabrication of vascularized tissue engineered bone. We showed that such tools were sufficient to generate bone tissue capable of integrating with native tissues when transferred to a large bone defect. The use of these tissues in the reconstruction of large bone defects requires further study. Additional studies will need to be performed to further elucidate the mechanisms governing neovascularization and osteogenesis within in vivo bioreactors.

Conclusions
We demonstrated that large volume, customized, vascularized bone tissues can be prefabricated using 3DP PLA-HA composite scaffolds in an in vivo bioreactor without requiring the use of exogenous growth factors. Compared with in vivo bioreactors only implanted in the periosteum, the addition of a vascular bundle has a greater advantage in constructing large vascularized bone grafts. With reliable microsurgical transplantation, this novel approach may be a bridge for repairing large bone defects and facilitating clinical translation in bone tissue engineering.

Materials and Methods
Fabrication of the biomaterials. Poly(L-lactide) (PLA) was purchased from Sigma-Aldrich (Shanghai, China), and hydroxyapatite (Ca 10 (PO 4 ) 6 (OH) 2 ) (HA) was purchased from Sonac Company (The Netherlands) with a mean size of 2.1 ± 0.4 μm. As previously described, a new mini-deposition system (MDS) located in the Shanghai 3D printing centre was developed to fabricate scaffolds. PLA (85 wt.%) and HA (15 wt.%) were mixed and processed to produce PLA-HA composite scaffolds with the MDS method 48,49 . Based on three-dimensional computed tomography reconstruction, computer-aided design and computer-aided manufacturing systems, the 3D printer produced porous cylindrical scaffolds with a predetermined structure (5-mm diameter, 6-mm height; Fig. 9). These scaffolds with predefined configurations had a mean pore size of 500 μm and an average porosity was 60%. The 3D printed scaffolds were sterilized by ethylene oxide and prepared for study.
Animals. Twenty-four New Zealand white adult rabbits (6 months of age, ranging from 2.5 ± 0.2 kg in weight) were used in accordance with the guidelines of the Institutional Animal Experiment Department of Shanghai Jiao Tong University (Shanghai, China) and the principles of laboratory animal care (NIH publication number 85-23, revised 1985). The ethics committee of Shanghai Jiao Tong University specifically approved this study. All rabbit groups included one experimental group (EG) and one control group (CG) (Fig. 10). The general operating procedures are shown in Fig. 11. For each group, after implantation, six rabbits were euthanised at 4 and 8 weeks.

In vitro 3DP scaffolds combined with bone marrow stromal cells (BMSCs).
Bone marrow stromal cells (BMSCs) were harvested from New Zealand white adult rabbits. Bone marrow aspirates (2 mL) were taken from the lateral tibial tubercle using 5-mL injectors with 0.5 mL heparin sodium (2500 U/mL, Wanbang, China Chemical & Pharmaceutical Company, Jiangshu, China). The number of bone marrow aspirates was the same as the number of rabbits. Bone marrow was flushed with 20 mL Dulbecco's minimum Essential medium (DMEM; Gibco, Australia) supplemented with 10% foetal bovine serum (FBS; HyClone; America), penicillin (50 U/ml) and streptomycin (50 mg/ml).The mixture was centrifuged at 1000 rpm for 5 min, and the supernatant was removed. The cell pellet was re-suspended in DMEM and directly transferred to a 10-cm culture dish at 37 °C in an atmosphere of 5% CO 2 . The medium was exchanged twice a week. Before implanting BMSCs, 3DP scaffolds were transferred into 24-well cell culture plates and incubated with culture medium for 24 h. BMSCs were collected and suspended at a density of 5 × 10 6 /mL. BMSCs were seeded on each scaffold in a 24-well plate and allowed to adhere to the scaffold for 4 h to permeate into the material. Then, 10 mL DMEM was added. Implantation of the in vivo bioreactor. All New Zealand white rabbits were randomly divided into two groups, and an animal model of an in vivo bioreactor on adult New Zealand rabbits inside the tibia was established. Bioreactor implantation was performed with established methods 11,50,51 . Briefly, animals were anaesthetized and intubated. Under sterile operating conditions, a 4-cm incision was made on the medial aspect of the tibia of the hind legs of the rabbits. Skin and underlying subcutaneous tissue were elevated and extended. Then, a vessel bundle, including a saphenous artery and a saphenous vein, was completely exposed (Fig. 12A). The saphenous arteriovenous blood bundles were fully separated. Distal ends of the pedicles were tied and ligated, and the proximal portion of the pedicle remained attached to the circulatory. The periosteum on the surface of     the tibia was elevated and isolated (Fig. 12B). This operation should ensure the preservation of blood flow to the periosteum. Cuboid-shaped periosteal pockets 10 mm in length and 7.5 mm in diameter were created. Then, the pedicles were threaded through the central core of the cylindrical scaffolds and combined with autologous BMSCs (Fig. 12C) and secured by tying sutures under the periosteum capsule (Fig. 12D). After finishing all operations, vascular blood flow of the pedicles was confirmed (Video 1). An in vivo bioreactor in the experimental group (EG) was surgically inserted by the above operation, and an avascular bioreactor was designated the control group (CG). The soft tissue was sutured with 6-0 PDS sutures, and the skin was closed with a 5-0 PDS suture. The animals were returned to their cages and monitored post-operatively for signs of discomfort or adverse events. The rabbits were euthanised at 4 and 8 weeks, and the specimens were harvested.
Microfil perfusion. To demonstrate and visualize new blood vessel formation, as described by Bolland 52 , the lower abdominal aorta was perfused with Microfil (Flowtech, Carver, MA, USA) following euthanasia at 4 and 8 weeks post-operative. Briefly, a midline incision extending across the abdomen was made, and the lower part of the abdominal aorta was exposed. Then, 30 mL Microfil were perfused at a steady constant flow after clamping the upper part of the abdominal aorta. In vivo bioreactor-generated tissues were harvested from animals at 4 °C overnight. The tissues were stored in 10% neutral buffered formalin (NBF) for subsequent experiments.
RNA isolation and qRT-PCR. Quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR) was used to measure vascularisation and osteogenic differentiation. To determine angiogenesis and osteogenic gene expression levels, mRNA was isolated from biopsies surrounding the tissues generated in the in vivo bioreactors.
Vascular endothelial growth factor (VEGF) and bone morphogenetic protein-2 (BMP-2) were genes of interest because they are important regulators of angiogenesis and stimulate new vessel formation 53,54 . Osteogenic-related genes, osteopontin (OPN) and collagen type I (COL-1), were selected to estimate the levels of osteogenesis, respectively. QRT-PCR was performed on 3-mm biopsy samples. To recover and isolate mRNA, biopsy specimens were flash frozen in liquid nitrogen, crushed, and placed in TRIzol reagent (Invitrogen Pty, Ltd., Australia). mRNA was subsequently reverse-transcribed into cDNA using a PrimeScript TM RT reagent kit (Takara, Japan). Then, quantitative PCR was then performed with forward and reverse primers using SYBR Green detection reagent. The final analysis was performed on a 7500 Real Time PCR System (Applied Biosystems, USA). The relative expressions of the genes of interest were normalized against housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) 55 . As designed from PubMed, the forward and reverse primer sequences were as follows. For VEGF: forward 5′-CGA ACG TAC TTG CAG ATG TGA C-3′ and reverse 5′-CAA AGT GCT CAC GCA GTC TC-3′; for BMP-2: forward 5′-TTT GGT CAC GAT GGG AAG GG-3′ and reverse 5′-TGC ACG ATG  GCA TGG TTA GT-3′; for OPN: forward 5′-GTG TAC CCC ACT GAG GAT GC-3′ and reverse 5′-CAC GTG  TGA GCT GAG GTC TT-3′; for COL-1: forward 5′-TCG ATC CCA ACC AAG GAT GC-3′ and reverse 5′-CAA  ACT GGG TGC CAC CAT TG-3′; and for GAPDH: forward 5′-TGG AAT CCA CTG GCG TCT TC-3′ and  reverse 5′-GTC ATG AGC CCC TCC ACA AT-3′. The mean cycle threshold (Ct) value of each target gene was normalized against the Ct value of GAPDH. The relative expression level of each target gene was evaluated using the 2 −(normalized average Ct) ×10 method 56 .
Micro-computed tomography (micro-CT) analysis. All specimens were scanned using micro-CT (SCANCO MEDICAL AG, Switzerland) with a resolution of 10 μm to evaluate neurovascular and new bone formation. Then, 3D images were reconstructed using 3D Creator software (GE Healthcare BioSciences, Chalfont-St. Giles, UK). Blood vessels and bone tissues were separately reconstructed according to different grey value ranges. A region corresponding to vessels perfused with microfil on the micro-CT grey scale range was selected, and the parameters of vessel number (VN) and vessel volume (VV) were analysed. Three-dimensional measurements of the amounts of bone volume per total volume (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th) and trabecular spacing (Tb.Sp) in the bone tissue were also calculated by microCT.
Histological observation. All harvested tissues were stored in 10% neutral buffered formalin (NBF) and decalcified in 10% ethylenediaminetetraacetic acid (EDTA) for 4 weeks at 37 °C prior to paraffin embedding. The tissue specimens were sliced into 10-μm sections along the long axis of the tissue by a microtome with a diamond blade. Immunohistochemical staining was carried out to assess and analyse the formation of new blood vessel and new bone tissue, respectively. The sections were incubated with primary antibodies against CD31 and osteocalcin (OCN; Abcam, Cambridge, MA, USA) at 4 °C overnight. After rinsing with PBS buffer, HRP-conjugated secondary antibody and DAB solution were used to stain the sections. Then, cell nuclei were identified with haematoxylin. Images were observed by light microscopy (TE2000U, Nikon, Japan) and were acquired with BioQuant OSTEO II software (BioQuant Image Analysis Corporation, Nashville, TN). The CD31 and OCN-positive area was assessed using Image ProPlus 6.0 software to evaluate angiogenesis and osteogenesis.
Statistical analysis. The data are presented as the mean ± standard deviation. The normal distribution of all the data was proved by shapiro wilk test. Significant differences between the experimental and control bioreactors in each animal were examined with Student's t-test analysis. SPSS 17.0 software was employed for all statistical analyses, and P < 0.05 was considered significant.